research paper for health care system

Advertisement

Keeping pace with the healthcare transformation: a literature review and research agenda for a new decade of health information systems research

• Research Paper
• Open access
• Published: 17 July 2021
• Volume 31 , pages 901–921, ( 2021 )

Cite this article

You have full access to this open access article

• Nadine Ostern   ORCID: orcid.org/0000-0003-3867-3385 1 ,
• Guido Perscheid 2 ,
• Caroline Reelitz 2 &
• Jürgen Moormann 2  

7619 Accesses

11 Citations

1 Altmetric

Explore all metrics

A Correction to this article was published on 20 December 2021

This article has been updated

Accelerated by the coronavirus disease 2019 (Covid-19) pandemic, major and lasting changes are occuring in healthcare structures, impacting people's experiences and value creation in all aspects of their lives. Information systems (IS) research can support analysing and anticipating resulting effects.

The purpose of this study is to examine in what areas health information systems (HIS) researchers can assess changes in healthcare structures and, thus, be prepared to shape future developments.

A hermeneutic framework is applied to conduct a literature review and to identify the contributions that IS research makes in analysing and advancing the healthcare industry.

We draw an complexity theory by borrowing the concept of 'zooming-in and out', which provides us with a overview of the current, broad body of research in the HIS field. As a result of analysing almost 500 papers, we discovered various shortcomings of current HIS research.

Contribution

We derive future pathways and develop a research agenda that realigns IS research with the transformation of the healthcare industry already under way.

Similar content being viewed by others

Afterword: Moving Forward in Healthcare Systems in the Twenty-First Century

From Substantialist to Process Metaphysics – Exploring Shifts in IS Research

Background Literature on the Adoption of Health Information Technologies

Explore related subjects.

• Artificial Intelligence

Avoid common mistakes on your manuscript.

Introduction

Particularly since the last decade, IT has opened up new opportunities for ‘ehealth’ through telemedicine and remote patient monitoring, alongside potential improvements in the cost-effectiveness and accessibility of health care (Chiasson & Davidson, 2004 ). Accordingly, health information systems (HIS) research has come to focus on how healthcare organizations invest in and then assimilate HIS, looking in particular at the impact of digitalization on healthcare costs, healthcare quality, and patient privacy (Chen et al., 2019 ; Park, 2016 ).

Less attention has been paid to issues such as mobile health, health information interchange, digital health communities, and services that change customer expectations and may lead to major disruptions (Chen et al., 2019 ; Park, 2016 ). These topics, however, are becoming increasingly important due to the penetration of the user and health market by external players, especially tech companies, providing services such as fitness trackers, and surveillance software for patient monitoring in hospitals (Gantori et al., 2020 ). Modern IT, thus, becomes a catalyst to provide greater operational efficiency, offering new possibilities for tech companies to build new health-centred business models and services (Park, 2016 ).

The ways in which tech companies are entering the healthcare industry can be seen amid the spread of coronavirus disease 2019 (Covid-19), which is pushing healthcare systems to the edge of their capacities (Worldbank, 2020 ). In this extraordinary condition, the pandemic has provided an additional opportunity for tech companies that were hitherto not active or not allowed to enter the healthcare industry (Gantori et al., 2020 ).

We are currently seeing how entering the healthcare market is actually taking place, particularly in the USA, where tech companies are increasingly offering services to help address some of the problems associated with Covid-19. Google’s subsidiary Verily, for instance, facilitates the automation of coronavirus symptom screening and provides actionable, up-to-date information that supports community-based decision-making (Landi, 2020 ). Although the collaboration with Verily assists the US government in tracking cases to identify the spread of the virus, it is reasonable to suggest that Verily probably did not launch the screening tool out of altruism. In fact, to receive preliminary screening results via the Verily app, citizens have to log into their personal Google account (Greenwood, 2020 ). This allows Verily to gain immense value by aggregating huge, structured data sets and analyse them to come up with new health services, such as better tools for disease detection, new data infrastructures, and insurance offerings that – for better or for worse – may outplay current healthcare providers and even disrupt whole healthcare ecosystems (CB Insights, 2018 ). Similarly, Amazon has started to provide cloud space through Amazon Web Services to store health surveillance data for the Australian government’s tracing app (Tillett, 2020 ), and Amazon Care, a division initially responsible for handling internal staff care needs, now cooperates with the Bill and Melinda Gates Foundation to distribute Covid-19 testing kits to US residents (Lee & Nilsson, 2020 ).

Looking at information systems (IS) researchers’ previous assessments of state-of-the-art healthcare-related IS literature reveals that most scholars seem to have little or no concern for the beginning of those potentially long-lasting changes that are occurring in the healthcare industry (Chen et al., 2019 ). This is worrying, considering that it is already apparent that the years ahead will be marked by economic volatility and social upheaval as well as direct and indirect health consequences, including sweeping transformations in many of the world’s healthcare systems.

While it is clear that recent developments and the push of tech and platform companies into the healthcare sector can significantly improve the quality of life for billions of people around the world, it will be accompanied by serious challenges for healthcare industries, governments, and individuals (Park, 2016 ). Technological advances are, for instance, giving rise to a plethora of smart, connected products and services, combining sensors, software, data, analytics, and connectivity in all kinds of ways, which in turns leads to a restructuring of health industry boundaries and the empowerment of novel actors, especially tech and platform companies such as IBM, Google, and Amazon (Park, 2016 ).

Observing those changes, we need to develop a general understanding of long-term trends such as digitalization and blurring industry boundaries. As the pandemic is only an amplifier of longer-lasting trends, it is likely that the consequences and exogenous effects on the healthcare industry will go far beyond the time of the current pandemic. Given these observations, we wonder whether the IS research domain is ready to capture, understand, and accompany these developments, which require a holistic view of the healthcare industry, its structures, and the interdependencies between incumbents and new entrants. Thus, we argue that it is now time to develop a more comprehensive understanding of these developments and to determine the role that IS research can play by asking: How can we prepare HIS research to capture and anticipate current developments in the healthcare industry?

To find answers to this question, our paper provides a literature overview of HIS research by ‘zooming in and zooming out’ (Gaskin et al., 2014 ) and by drawing on complexity theory (Benbya et al., 2020 ). Since a healthcare system, like the industry as a whole, can be understood as a complex, digital socio-technical system (Kernick & Mitchell, 2009 ; Therrien et al., 2017 ), zooming in and zooming out is a way to view, capture, and theorize the causes, dynamics, and consequences of a system’s complexity. Complex systems are characterized by adaptiveness, openness (Cilliers, 2001 ), and the diversity of actors and their mutual dependency in the system, meaning that outcomes and research span various levels within these systems, although the boundaries of socio-technical systems are elusive. Assuming that HIS research is just as complex as the socio-technical system investigated, we first zoom in, focusing on concrete research outcomes across levels (i.e., what we can actually observe). Zooming in is followed by zooming out, which means abstracting from the concrete level and embracing the strengths and disparities of overall HIS research on a higher level in which concrete research outcomes are embedded (Benbya et al., 2020 ). Using this approach, we can capture and understand the complexity of HIS research without losing sight of concrete research issues and topics that drive research in this field.

To do this, we chose a hermeneutic framework to guide us in a thorough review and interpretation of HIS literature and lead us to the following overarching observations: (i) The literature review determines the unique contribution that IS research plays in analysing and advancing the healthcare industry. However, it also shows that we are hardly prepared to take up current developments and anticipate their consequences. (ii) The reason for this unpreparedness is that we currently neglect the ecosystem perspective and thus ignore holistic approaches to resolve the striking number of interrelated issues in HIS research. (iii) Based on the unique insights of this literature review, our paper provides a research agenda in which we use complexity theory to discuss the consequences of current developments. This theory assists IS researchers not only to better understand developments and implications thereof for the healthcare industry (and thus HIS research) but also to create a meaningful impact on the future of this industry. Since we have limited our research explicitly to the IS domain, our results may not be generally applicable to other healthcare research domains and we do not claim to provide an overview of the literature in the field of HIS research. However, while IS researchers cannot solve the pandemic directly, preparing them by providing a new research agenda will support them in developing concepts and applications, thereby helping them to overcome the negative effects of the pandemic. In our opinion, it is particularly important that IS research, and especially HIS-related research, obtains a deeper understanding of the needed transformation that is caused by digitalization and the emergence of new players catalysed by the current pandemic.

The remainder of this paper is structured as follows. The next section is concerned with the hermeneutic framework used to conduct the systematic literature review. After explaining the hermeneutic approach and the research steps, we elaborate on the key findings by zooming in; that is, we focus on the key results that emerge from analysing and interpreting the literature for each of the phases defined in the course of the literature sorting process. We then concentrate on zooming out, emphasizing the patterns and interdependencies across phases, which helps us determine the state of HIS research. The results of both parts of the literature review – i.e., zooming in and zooming out (Benbya et al., 2020 ; Gaskin et al., 2014 ) – support us in identifying strengths, as well as drawbacks, in HIS research. On this basis, we develop a research agenda that provides future directions for how HIS research can evolve to anticipate the impending transformation of the healthcare industry.

Literature review: a hermeneutic approach

To answer our research question, we conducted a literature review based on hermeneutic understanding. In particular, we followed Boell and Cecez-Kecmanovic ( 2014 ). They proposed a hermeneutic philosophy as a theoretical foundation and methodological approach that focuses on the inherently interpretive processes in which a reader engages in an ever-expanding and deepening understanding of a relevant body of literature. Adopting a comprehensive literature review approach that addresses well-known issues resulting from applying structured literature review approaches (e.g., Webster & Watson, 2002 ), we strive toward the dual purpose of hermeneutic analysis – i.e., to synthesize and critically assess the body of knowledge (Boell & Cecez-Kecmanovic, 2014 ). We would like to emphasize that the hermeneutic approach to literature reviews is not in opposition to structured approaches. Rather, it addresses the weaknesses of structured approaches (i.e., that they view engagement with the literature as a routine task rather than as a process of intellectual development) and complements them with the hermeneutic perspective to create a holistic approach for conducting literature reviews.

Theoretical underpinning and research method

A methodological means for engaging in reciprocal interpretation of a whole and its constituent elements is the hermeneutic cycle (Bleicher, 2017 ), which consists of a mutually intertwined search and acquisition circle (Circle 1 in Fig.  1 ) and the wider analysis and interpretation circle (Circle 2 in Fig.  1 ) (Boell & Cecez-Kecmanovic, 2014 ). Figure  1 depicts the steps associated with the hermeneutic literature review. The search and acquisition circle is shown on the left of the figure, while the analysis and interpretation circle containing steps of meta and content analysis is depicted on the right. The two circles should be understood as an iterative procedure, the nature of which will be explained in the following. 

Hermeneutic procedure applied to the literature review

Circle 1: Search and acquisition

The hermeneutic literature review starts with the search and acquisition circle, which is aimed at finding, acquiring, and sorting relevant publications. In line with holistic thinking, we began with the identification of a rather small set of highly relevant literature (Boell & Cecez-Kecmanovic, 2014 ) and went on to identify further literature on the basis of progressively emerging keywords. This step is central to the hermeneutic approach and addresses a criticism on structured literature reviews, namely that they downplay the importance of reading and dialogical interaction between the literature and the reader in the literature search process, reducing it to a formalistic search, stifling academic curiosity, and threatening quality and critique in scholarship and research (Boell & Cecez-Kecmanovic, 2014 ; MacLure, 2005 ). Thus, while the search process remains formalized, as in pure structured approaches, the hermeneutic approach allows us to acquire more information about the problem at hand and to identify more relevant sources of information (Boell & Cecez-Kecmanovic, 2014 ).

Given our initial research question and the scope of the review, we began by searching for papers in the Association for Information System’s (AIS’s) eLibrary over a period of 30 years (1990 to 2019). We consider this database to be a source of the most significant publications in the field of HIS research with a focus on the IS research domain. Using the keywords ‘digital health’ and ‘digital healthcare service’, we identified an initial set of 54 papers based on the title, abstract, and keyword search. Engaging in a first round of the hermeneutic search and acquisition circle, we extended and refined these keywords by identifying emerging topics within the literature, as well as using backward and forward search (Webster & Watson, 2002 ). In particular, with each additional paper identified through backward and forward search, we compared keyword references in the papers to our list of keywords and added them if there was sufficient content delimitation. The decision to add a keyword was discussed with all authors until consensus was reached. This led us to a set of 12 keywords, including ‘electronic health’, ‘ehealth’, ‘mobile health’, ‘mhealth’, ‘health apps’, ‘tech health’, ‘healthcare services’, ‘healthcare informatics’, ‘medical informatics’, and ‘health data’.

The selection of publications being considered for our research comprised all journals belonging to the AIS eLibrary, the Senior Scholars’ Basket of Eight Journals (e.g., European Journal of Information Systems, Information Systems Research , and MIS Quarterly ), well-regarded journals following the analyses of Chiasson and Davidson ( 2004 ) and Chen et al. ( 2019 ) (e.g., Business & Information Systems Engineering , Communications of the ACM, and Decision Support Systems ), and the proceedings of the major AIS conferences (e.g., Americas Conference on Information Systems (AMCIS), International Conference on Information Systems (ICIS)). An overview of the selected journals and proceedings is provided in Appendix 1 .

Using our set of keywords, we searched for each keyword individually in the AIS eLibrary and the databases of the respective journals. Subsequently, we created a dataset and filtered out the duplicates, yielding a total number of 1,789 papers to be screened in the search and acquisition circle (Circle 1 in Fig.  1 ). Figure  2 provides an overview of this process by listing the total number of articles identified for each journal individually.

Steps of the search process to create the data set

The resulting 1,789 papers progressively passed through the intertwined hermeneutic circles. Because of the large number, we divided the papers at random into four equally sized groups and assigned them to each of the authors. Each author then screened the paper in his or her group. In the course of several rounds of discussion, decisions on the inclusion of keywords and articles in the literature review were made by all authors, based on the original recommendations of the author responsible for the respective group. To ensure rigor and transparency of the analysis and results, we kept a logbook in which all decisions of the authors and steps of the literature review were recorded (Humphrey, 2011 ).

Given the abundance of topics that were already apparent from titles and abstracts, we began to sort the publications (Boell & Cecez-Kecmanovic, 2014 ). The process of sorting proved to be challenging, as HIS research is diverse and tends to be eclectic (Agarwal et al., 2010 ). This is why researchers have developed frameworks for clustering and analysing HIS research (LeRouge et al., 2007 ). So far, however, no consent on a unified framework has emerged, and sorting is often strongly influenced by the authors’ views on HIS research (Agarwal et al., 2010 ; Fichman et al., 2011 ). For instance, Agarwal et al. ( 2010 ) predetermined health IT adoption and health IT impact as major themes associated with health ITs, acknowledging that this pre-categorization of research topics made a systematic review of the growing and increasingly complex HIS literature unfeasible. Consequently, we decided to sort the articles we had identified into groups inspired by and loosely related to the phases of design science research (DSR) (Peffers et al., 2008 ), which is an essential step in hermeneutics – i.e., defining guidelines to facilitate interpretive explication (Cole & Avison, 2007 ). DSR can be understood as a cumulative endeavour and, therefore, we understood HIS research as accumulative knowledge that can be reconstructed and consolidated using DSR phases as guidance (vom Brocke et al., 2015 ; vom Brocke et al., 2009 ). In particular, this helped us to sort the articles without prejudice to expected HIS research topics and clusters (Grondin, 2016 ).

In the past, researchers have used the DSR process in the context of literature reviews to identify advances in design science-related research outcomes (Offermann et al., 2010 ). In this paper, we use the DSR phases – in the sense of a rough guideline – as a neutral lens to classify articles according to their research outcomes. We thereby assume that HIS literature can be seen as an overall process, where research results and progress are built upon each other and can be classified into phases of problem identification and research issues , definition of research objectives and possible solution space , design and development of solutions , demonstration of research effectiveness, innovativeness and acceptance , and evaluation . These phases served as a guide to achieve an outcome-oriented, first-hand sorting of articles, while this approach also gave us the opportunity to take a bird's-eye view on HIS research. Note that we intentionally omitted the last step of DSR – i.e., communication – as we regard communication as present in all published articles. Based on our initial reading, we assigned all 1,789 papers to the phases and discussed this sorting in multiple rounds until all authors agreed on the assignments.

Simultaneously, we applied criteria for the inclusion and exclusion of articles. We included full papers published in the journals and conference proceedings belonging to our selection. We excluded articles that were abstract-only papers, research-in-progress papers, panel formats, or workshop formats, as well as papers without direct thematic reference to our research objective. Additionally, during the acquisition stage we stored selected papers in a separate database whenever they fulfilled certain quality criteria (e.g., for separate studies using the same dataset, such as a conference publication and a subsequent journal publication, we only used the articles with the most comprehensive reporting of data to avoid over-representation).

The authors read the resulting 489 papers to identify new core terms and keywords that were used in subsequent searches, which not only provided the link to the analysis and interpretation circle but also informed the literature search. For this purpose, each author read the papers and kept notes in the logbook that supported us in systematically recording the review process and allowed us to shift from concentrating on particular papers to focusing on scientific concepts (Boell & Cecez-Kecmanovic, 2014 ; Webster & Watson, 2002 ).

Circle 2: Analysis and interpretation

The search and acquisition circle formed part of the iterative procedure of analysis and interpretation, whereby the reading of individual papers was the key activity linking Circle 1 to the steps of Circle 2 (Boell & Cecez-Kecmanovic, 2014 ). Through orientational reading we gained a general understanding of the literature, thus laying the foundation for the subsequent steps of analysis and interpretation (Boell & Cecez-Kecmanovic, 2014 ).

Within the analysis and interpretation circle, two types of reviews were conducted for all identified and sorted articles: in a first round a meta-review, and in a second round a content analysis of the papers was performed. Meta-reviews are a useful tool for capturing and analysing massive quantities of knowledge using systematic measures and metrics. We followed Palvia et al. ( 2015 ), who proposed a structured method that is integrated into the hermeneutic approach. In particular, having identified and sorted the relevant research articles, we applied proposed review features, including methodological approach, level of observation, sample size, and research focus (Humphrey, 2011 ; Palvia et al., 2015 ) to map, classify, and analyse the publications (Boell & Cecez-Kecmanovic, 2014 ). In doing so, we slightly adapted the classic meta-analysis by focusing on meta-synthesis, which is similar to meta-analysis but follows an interpretive rather than a deductive approach. Whereas a classic meta-analysis tries to increase certainty in cause-and-effect conclusions, meta-synthesis seeks to understand and explain the phenomena of mainly qualitative work (Walsh & Downe, 2005 ). The results of the meta-synthesis provided the basis for our subsequent critical assessment of content. Furthermore, we created a classification matrix based on particularly salient features of the meta-review (i.e., levels of observation and research foci), which facilitated and standardized the content analysis.

Within the matrix, the levels of observation comprised infrastructure (e.g., information exchange systems, electronic health records), individuals (patients and users of digital health services), professionals (e.g., nurses and general practitioners), organizations (hospitals and other medical institutions), and an ecosystem level. The latter is defined as individuals, professionals, organizations, and other stakeholders integrated via a digital infrastructure and aiming to create a digital environment for networked services and organizations with common resources and expectations (Leon et al., 2016 ). To identify the most important concepts used by researchers, we discussed a variety of approaches to the derivation of research foci – i.e., areas containing related or similar concepts that are frequently used in research on HIS. Finally, six research focus areas emerged, covering all relevant research areas. To describe the core HIS research issues addressed by these foci, we used the following questions:

HIS strategy: What are the prerequisites for configuring, implementing, using, maintaining, and finding value in HISs?

HIS creation: How are HISs composed or developed?

HIS implementation: How are HISs implemented and integrated?

HIS use and maintenance: How can HISs be used and maintained once in place?

Consequences and value of HIS: What are the consequences and the added value of HISs?

HIS theorization: What is the intellectual contribution of HIS research?

We used the classification matrix as a tool for assigning publications and finding patterns across research articles and phases. In particular, we used open, axial, and selective coding (Corbin & Strauss, 1990 ) to analyse the content of articles in a second round of the analysis and interpretation circle. Each author individually assigned open codes to text passages while reading the identified research articles, noting their thoughts in the shared digital logbook that was used for constant comparative analysis. Once all authors had agreed on the open codes, axial coding – which is the process of relating the categories and subcategories (including their properties) to each other (Wolfswinkel et al., 2013 ) – was conducted by each author and then discussed until consent on codes was reached. Next, we conducted selective coding and discussed the codes until theoretical saturation was achieved (Corbin & Strauss, 1990 ; Matavire & Brown, 2008 ). For the sake of consistent terminology, we borrowed terms from Chen et al. ( 2019 ), who used multimethod data analysis to investigate the intellectual structure of HIS research. In particular, they proposed 22 major research themes, which we assigned to the initial codes whenever possible. In two rounds of discussion in which we compared the assignment of codes, two additional codes emerged, which left us with a total of 24 theme labels (Appendix 2 ). By discussing the codes at all stages of coding, theoretical saturation emerged, which is the stage at which no additional data are being found or properties of selective codes can be developed (Glaser & Straus, 1968 ; Saunders et al., 2018 ). In fact, independent from each other, all authors saw similar instances occurring over and over again, resulting in the same codes, making us confident that we had reached theoretical saturation (Saunders et al., 2018 ).

Finally, we entered the codes into the classification matrix, which allowed us to identify patterns based on the meta and content analysis. This enabled us to provide insights into the strengths and weaknesses of current HIS research; these are presented in the following section.

Zooming-in: key findings of the phase-based literature analysis

In the following, we ‘zoom in’ (Gaskin et al., 2014 ) by presenting key findings of the literature review for each phase, illustrated by means of the classification matrices. We assigned selective codes that emerged from the content analysis to the fields of the matrices, with the numbers in brackets indicating the frequency with which codes emerged. Note that, for the sake of clarity, we displayed only the most relevant research themes in the matrices and indicated the number of further papers using the reference ‘other themes.’ A complete list of research themes for each phase can be found in the appendix (Appendix 2 ). In the following, each table shows the classification matrix and selective codes that resulted from the meta and content analysis of papers in the respective phase. The shaded areas in the matrix show focused research themes (i.e., selective codes) and characteristics of research articles that gave way to clusters (i.e., collections of themes that appear frequently and/or characteristically for the respective focus).

Phase 1: Problem identification and research issues

Within the first phase, a large body of literature was found (218 articles). This phase encompasses articles that identify problems and novel research issues as a main outcome, with the aim of pointing out shortcomings and provoking further research. For instance, besides behavioural issues such as missing user acceptances or trust in certain HISs, the design and effectiveness of national health programs and/or HIS is a frequently mentioned topic. It should be noted, however, that literature assigned to this phase is extremely diverse in terms of research foci, levels of observation, and research themes, and hardly any gaps can be identified (Table 1 ).

The first cluster (1a) encompasses the research focus of HIS strategy, spanning all levels of observation and totalling 24 publications. HIS strategy appears to be of particular relevance to the levels of organization and infrastructure. Content-wise, the theme of health information interchange is of particular interest, referring, for example, to the development of a common data infrastructure (Ure et al., 2009 ), consumer-oriented health websites (Fisher et al., 2007 ), and security risks of inter-organizational data sharing (Zhang & Pang, 2019 ). HIS productivity and HIS security are the second most salient themes, focusing, for example, on measuring the effectiveness of fitness apps (Babar et al., 2018 ) and presenting challenges with regard to the interoperability of medical devices (Sametinger et al., 2015 ).

The second cluster (1b), comprising 25 publications, represents the ecosystem level and focuses mainly on national and cross-national HIS-related issues such as the relation between ICT penetration and access to ehealth technologies across the European Union (Currie & Seddon, 2014 ), as well as on the collaboration and involvement of different stakeholders (Chang et al., 2009 ; King, 2009 ). Most important here is health information interchange – e.g., the provision, sharing, and transfer of information (Bhandari & Maheshwari, 2009 ; Blinn & Kühne, 2013 ).

Cluster 1c covers the research focus of HIS use and maintenance, as well as the consequences of HIS. Whereas most papers addressing the HIS acceptance theme focus on professionals’ or patients’ acceptance of specific technological solutions, such as telemedicine (Djamsbi et al., 2009 ) or electronic health records (Gabel et al., 2019 ), papers assigned to health information interchange focus on topics related to information disclosure, such as self-tracking applications (Gimpel et al., 2013 ). Finally, the HIS outsourcing and performance theme concentrates on financial aspects in organizations, including potential for quality improvements and cost reductions (Setia et al., 2011 ; Singh et al., 2011 ).

Finally, the fourth cluster (1d) focuses on HIS theorizing with respect to the individual and infrastructure levels of observation. Although this cluster represents a range of theme labels (15), those addressing HIS acceptance, HIS patient-centred care, as well as health analytics and data mining predominate. Papers within the theme label HIS acceptance cover a wide range of topics, such as the acceptance of telehealth (Tsai et al., 2019 ) up to the usage intentions of gamified systems (Hamari & Koivisto, 2015 ). The same applies to the papers assigned to the theme labels of health analytics and data mining. Focusing on the infrastructure level of observation, the identified papers mostly review academic research on data mining in healthcare in general (Werts & Adya, 2000 ), through to the review of articles on the usage of data mining with regard to diabetes self-management (Idrissi et al., 2019 ). Papers on HIS patient-centred care mostly address the challenges and opportunities of patient-centred ehealth applications (Sherer, 2014 ).

Apart from these clusters, quite a few research articles refer to the infrastructure level of observation, addressing information sharing in general (Li et al., 2008 ), electronic medical records (George & Kohnke, 2018 ; Wessel et al., 2017 ), and security and privacy issues (Zafar & Sneha, 2012 ).

Most common in terms of research methods within this phase are case studies (57), followed by quantitative data analyses (50), theoretical discussions (29), and literature studies (14). In particular, case studies dominate when referring to the ecosystem or infrastructure level of observation, whereas quantitative analyses are conducted when individuals or professionals are at the centre of the discussion. However, and unsurprisingly given the considerable diversity of research themes within this phase, the variety of research methods is also quite large, ranging from field studies (Paul & McDaniel, 2004 ), to interviews (Knight et al., 2008 ), to multimethod research designs (Motamarri et al., 2014 ).

Phase 2: Definition of research objectives and solution space

The second phase of HIS research yielded a lower number of articles (45) compared to the phase of problem identification and research issues. The second phase comprises articles that focus on proposing possible solutions to existing problems – i.e., introducing theory-driven, conceptual designs of health ecosystems including health information interchange, as well as scenario analyses anticipating the consequences of HIS implementation on an organizational level. Based on the research foci and levels of observation, we identified three specific thematic clusters, as shown in Table 2 .

The first cluster (2a) comprises the ecosystem level of observation and encompasses eight publications. Besides a strong tendency toward theory-driven research, health information interchange is the most common theme. We found that the need to enable cooperation within networks and to ensure accurate data input was addressed in most of the literature. While a majority of studies focus on the application of HIS in networks within specific boundaries, such as medical emergency coordination (Sujanto et al., 2008 ) or Singapore’s crisis management in the fight against the SARS outbreak in 2003 (Devadoss & Pan, 2004 ), other studies, such as that by Aanestad et al. ( 2019 ), take an overarching perspective, addressing the need to break down silo thinking and to start working in networks. Following the question of why action research fails to persist over time, Braa et al. ( 2004 ) highlighted the role of network alignment, criticizing action research projects for failing to move beyond the prototyping phase and, therefore, failing to have any real impact.

Cluster 2b, encompassing nine publications, was derived from the observation that studies within the organizational level concentrated strongly on HIS use and maintenance and the consequences of HIS research. Herein, a vast array of topics was observed, such as the potential for cost reduction through HIS (Byrd & Byrd, 2009 ), the impact of HIS on product and process innovation in European hospitals (Arvanitis & Loukis, 2014 ), and the perceived effectiveness of security risk management in healthcare (Zafar et al., 2012 ). Moreover, we found that practice-oriented methods, such as mixed-method approaches, surveys, data analyses, and case studies, are used predominantly within this cluster. Focusing on the latter, most studies analyse particular scenarios by using a rather small sample of cases, for instance, Al-Qirim ( 2003 ) analysed factors influencing telemedicine success in psychiatry and dermatology in Norway.

The third cluster (2c) was derived from analysis of the HIS creation research focus (nine publications). Although health information interchange is the most represented in this cluster, a large number of further themes can be observed. Studies within this cluster predominantly address design aspects of system interoperability, focusing on data processing and data interchange between the actors. HISs mostly serve as a tool for the development or enhancement of decision support systems, such as for real-time diagnostics combining knowledge management with specific patient information (Mitsa et al., 2007 ) or clinical learning models incorporating decision support systems in the dosing process of initial drug selection (Akcura & Ozdemir, 2008 ).

Phase 3: Design and development

The design and development phase comprises 84 research articles concerned with the creation of novel IS artefacts (e.g., theories, models, instantiations). We thereby refer to Lee et al.’s ( 2015 ) definition of the IS artefact – i.e., the information, technology, and social artefact that forms an IS artefact by interacting. We assigned to this phase papers that are explicitly concerned with developing solutions for information exchange (e.g., design of messaging systems or knowledge systems in hospitals), technological artefacts (e.g., hardware or software used for generating electronic health records), and social artefacts that relate to social objects (e.g., design of national or international institutions and policies to control specific health settings and patient-centred solutions). Within the design and development phase, the analysis revealed two clusters (Table 3 ).

The first cluster (3a) was identified in the research focus of HIS creation (31 articles). Here, the most frequent research theme is HIS innovation followed by HIS and patient-centred care, HIS productivity, and health analytics and data mining. The focus is on specific contexts, mostly medical conditions and artefacts developed for their treatment, such as in the context of mental health/psychotherapy (Neben et al., 2016 ; Patel et al., 2018 ), diabetes (Lichtenberg et al., 2019 ), or obesity (Pletikosa et al., 2014 ). Furthermore, information infrastructures or architectures – for instance, for the process of drug prescription (Rodon & Silva, 2015 ), or for communication between healthcare providers and patients (Volland et al., 2014 ) – are represented.

The second aggregation of research articles is found in cluster 3b, focusing on theoretical aspects of HIS (32 articles). Again, these studies span all levels of observation (including infrastructure, individual, professional, organization, and ecosystem). Topics in this theme are diverse, ranging from HIS on a national level (Preko et al., 2019 ), to knowledge management in healthcare (Wu & Hu, 2012 ) to security of HIS (Kenny & Connolly, 2016 ).

Beyond both clusters, it is evident that during design and development, researchers do not deal with the consequences of HIS, nor does HIS strategy play an important role. Furthermore, only in the research focus of theorization is the ecosystem level of some relevance to other levels (e.g., the individual level). It should be noted that ecosystems are mostly referred to in terms of nations or communities, without any transnational or global perspective. Furthermore, the term ‘ecosystem’ has not been used in research, and within the other research focus areas, the ecosystem level is barely represented. Moreover, articles combining different perspectives of the single levels of observation on HIS – namely individuals (i.e., patients), professionals (i.e., medical staff), and organizations (e.g., hospitals) – are rare. During design and development, potential users are not typically integrated, whereas it is quite common to derive requirements and an application design from theory, only involving users afterwards – e.g., in the form of a field experiment (e.g., Neben et al., 2016 ).

Surprisingly, theoretical papers outweigh papers on practical project work, whereby the latter mostly focus on a description of the infrastructure or artefact (e.g., Dehling & Sunyaev, 2012 ; Theobalt et al., 2013 ; Varshney, 2004 ) or are based on (mostly single) case studies (e.g., Hafermalz & Riemer, 2016 ; Klecun et al., 2019 ; Ryan et al., 2019 ). Within the design and development phase, the generation of frameworks, research models, or taxonomies is prevalent (e.g., Preko et al., 2019 ; Tokar et al., 2015 ; Yang & Varshney, 2016 ).

Phase 4: Demonstration

This phase includes 35 articles related to presenting and elaborating on proposed solutions – e.g., how HIS can be implemented organization-wide (e.g., via integration into existing hospital-wide information systems), proposed strategies and health policies, as well as novel solutions that focus on health treatment improvements. Within the demonstration phase, we identified two clusters that emerged from the meta and content analyses (Table 4 ).

Cluster 4a (10 articles) is characterized by articles that focus on HIS issues related to the infrastructure level, spanning the research foci of HIS strategy, creation, and deployment. Content-wise, the cluster deals mainly with technical feasibility and desirability of HISs, including topics such as the configuration of modular infrastructures that support a seamless exchange of HISs within and between hospitals (Dünnebeil et al., 2013 ). Moreover, papers in this cluster address HIS practicability by determining general criteria that are important for the design of health information systems (Maheshwari et al., 2006 ) or conduct HIS application tests by carrying out prototypical implementations of communication infrastructures. In particular, the latter are tested and proven to meet specific technical standards to guarantee the frictionless transmission of health information data (Schweiger et al., 2007 ). In contrast, Heine et al. ( 2003 ) upscaled existing HIS solutions and tested the infrastructure in large, realistic scenarios.

Conversely, cluster 4b (11 articles) is mainly concerned with HIS use and maintenance, spanning several levels of observation – i.e., infrastructure, individuals, professionals, and organizations. Interestingly, papers in this cluster aim at efficiency and added value when looking at the infrastructure and organizational levels, whereas researchers are more interested in acceptance when focusing on the individual and professional use of HISs. Overall, cluster 4b is primarily concerned with organizational performance (e.g., increases in efficiency due to better communication and seamless transfer of patient health information) as well as user acceptance of new HISs.

Although the two clusters constitute a diverse set of literature and themes, it is apparent that research taking an ecosystem perspective is very rarely represented. Across the papers, only three are concerned with issues related to the ecosystem level. In particular, Lebcir et al. ( 2008 ) applied computer simulations in a theoretical demonstration as a decision support system for policy and decision-makers in the healthcare ecosystem. Abouzahra and Tan ( 2014 ) used a mixed-methods approach to demonstrate a model that supports clinical health management. Findikoglu and Watson-Manheim ( 2016 ) addressed the consequences of the implementation of electronic health records (EHR) systems in developing countries.

Phase 5: Evaluation

The fifth phase includes 92 publications with a focus on assessing existing or newly introduced HIS artefacts – i.e., concepts, policies, applications, and programs – thereby proving their innovativeness, effectiveness, or user acceptance. As Table 5 shows, three clusters were identified.

The main focus of publications in the evaluation phase is on the infrastructure level, where most papers are related to HIS creation and HIS use and maintenance. Therefore, together with the publications pigeonholed to HIS deployment and consequences of HIS, these articles were summarized as the first cluster (5a, comprising 53 articles). The assessment of national HIS programs, as well as mobile health solutions, are a frequent focus (10 papers). Articles on HIS use and maintenance are largely related to the professional, organizational, and ecosystem levels and were thus grouped as cluster 5b (10 articles). A third cluster (5c – 11 articles) emerged from research articles in HIS theorization. Here, papers at all levels of observation were found. Research focusing on areas such as HIS strategy and consequences of HIS are, with a few exceptions, not covered in the evaluation phase. Methods used include interviews, focus groups, and observations (e.g., Romanow et al., 2018 ). Experiments and simulation are rarely applied (e.g., Mun & Lee, 2017 ). The number of interviews shows a huge spread, starting with 12 and reaching a maximum of 150 persons interviewed.

Under the evaluation lens, the ecosystem perspective is covered by seven articles, but only three papers look at cases, while the others focus on theorization or consequences in terms of costs. Overall, popular topics in the evaluation phase include mobile health and the fields of electronic medical records (EMR) and EHR, e.g., Huerta et al. ( 2013 ); Kim and Kwon ( 2019 ). The authors cover these themes mostly from an HIS creation perspective; thus, they deal with concrete concepts, prototypes, or even implemented systems. In the evaluation phase, just nine papers deal with HIS innovation – a good example being Bullinger et al. ( 2012 ), who investigated the adoption of open health platforms. We may conclude that, in most cases, evaluation is related to more established technologies of HIS. As expected, most articles in this phase rely on practice-oriented/empirical work (as opposed to theory-driven/conceptual work). Just two papers (Ghanvatkar & Rajan, 2019 ; Lin et al., 2017 ) deal with health analytics and data mining, one of the emerging topics of HIS.

Zooming out: key findings of the literature analysis across phases

Having elaborated on the key findings within each phase of HIS research, we now ‘zoom out’ (Benbya et al., 2020 ; Gaskin et al., 2014 ) to recognize the bigger picture. Thereby, we ‘black-box’ the concrete research themes (e.g., HIS implementation, health analytics, HIS innovation) to focus on clusters across phases, highlighting the breadth that HIS research encompasses (Leroy et al., 2013 ). In particular, while we focused on analysing the main topics within the different phases of HIS research in the zoom-in section, we now abstract from those to perform a comparative analysis of emerging clusters across those phases by zooming out. We do so by comparing the different clusters, taking into account the aspects of the level of observation and the research foci, which gave us the opportunity to identify areas of strong emphasis and potential gaps.

In particular, each author first conducted this comparative analysis on their own and then discussed and identified the potential weaknesses together. This was done in two rounds of discussion. In particular, it became obvious which areas hold immense potential for further research in healthcare (especially the penetration of new, initially non-healthcare actors, such as tech companies or other providers pushing into the industry). We summarize these potentials for research by proposing four pathways that can help HIS research to broaden its focus so that we can better understand and contribute to current developments. Notably, we expect that these insights will help to assess the state-of-the-art of HIS research and its preparedness for dealing with the consequences of Covid-19 and further pandemics, as well as for coping with associated exogenous shocks.

In zooming out, we identified discrepancies between phase 1 (problem identification and research issues) and the subsequent phases. In particular, the diversity of topics was considerably lower when it came to how researchers determined strategies; created, demonstrated, used, and maintained HISs; and coped with the consequences thereof. We observed that researchers pointed to a diverse set of issues that span all levels of observation, especially in HIS theorization, focusing on topics such as trust in HIS, data analytics, and problems associated with the carrying out of national health programs. Surprisingly, although we can assume that researchers recognized the multidimensionality of issues as a motivation to conduct HIS research, they did not seem to approach HIS research issues in a comprehensive and consistent way.

To illustrate this assertion, we point to the ‘shift of clusters’ that can be observed when comparing the single phases, from problem identification to the evaluation of HIS. We note that clusters increasingly migrate ‘downwards’ (i.e., from the ecosystem level down to the infrastructure level) and become even fewer. In line with Braa et al. ( 2004 ), we suggest that extant HIS research has identified a multitude of interrelated issues but has faced problems in translating these approaches into concrete and holistic solutions. This is reflected in the lower number of, and reduced diversity in, clusters across research themes when we move through the HIS research phases. Thus, we conclude that future HIS research can be broadened by taking into account the following pathway:

HIS research is well-prepared and able to identify and theorize on systemic problems related to the healthcare industry. Nonetheless, it has the potential to address these problems more thoroughly – i.e., to find solutions that are as diverse as the problems and, thus, suitable for coping with issues in the healthcare industry characterized by the involvement of multiple actors, such as governments, healthcare providers, tech companies, and their interactions in diverse ecosystems (pathway 1).

As we have seen, HIS research has tended to focus on important but incremental improvements to existing infrastructures, particularly in the phases of demonstration and evaluation, with the aim of presenting new IS artefacts and conceptual or practical solutions. For instance, Choi and Tulu ( 2017 ) considered improvements in user interfaces to decrease the complexity of mobile health applications using incremental interface design changes and altering touch techniques. Similarly, Roehrig and Knorr ( 2000 ) designed patient-centred access controls that can be implemented in existing infrastructures to increase the privacy and security of EHRs and avoid malicious access and misuse of patient health information by third parties.

While we sincerely acknowledge these contributions and wish to emphasize the multitude of papers that are concerned with enhancements to existing infrastructures, we would like to shift the view to the major challenges in HIS research. These challenges include combating global and fast-spreading diseases (e.g., malaria, tuberculosis, Covid-19) and tracking health statuses accurately and efficiently, especially in developing countries. All of these challenges necessitate global and comprehensive solutions, spanning individuals, organizations, and nations, and have to be embedded in a global ecosystem (Winter & Butler, 2011 ). Such grand challenges are, by nature, not easy to cope with, and the intention to develop a comprehensive solution from the perspective of IS researchers seems almost misguided. However, HIS research is currently missing the opportunity to make an impact, despite the discipline’s natural intersection with essential aspects of the healthcare industry (i.e., its infrastructures, technologies, and stakeholders, and the interdependencies between these components). Thus, we assert that:

HIS research has often focused on necessary and incremental improvements to existing IS artefacts and infrastructures. We see potential in shifting this focus to developing solutions that combine existing IS artefacts to allow for exchange of information and the creation of open systems, which will enhance support for and understanding of the emergence of ecosystems (pathway 2).

By focusing on incremental improvements, HIS research has become extraordinarily successful in solving isolated issues, especially in relation to the problems of patients and health service providers (e.g., hospitals and general practitioners). However, we observed during our analysis that spillover effects were seldom investigated. When, for example, a new decision support system in a hospital was introduced, positive consequences for patients, such as more accurate diagnoses, were rarely of interest to the research. In fact, our meta-analysis revealed that the level of observation for the majority of papers matched the level of analysed effects. While it is valid to investigate productivity and efficiency gains by introducing a hospital-wide decision support system, we are convinced that spillover effects (for instance, on patients) should also be within the focus of HIS research. Therein, we suggest that HIS research has not focused primarily on patients and their well-being but on IS infrastructures and artefacts. However, patient well-being is the ultimate direct (or indirect) goal of any HIS research (by increasing the accuracy and shortening the time of diagnosis, improving treatment success rates, etc.). Thus, we propose that:

HIS research is experienced in solving isolated issues related to the daily processes of healthcare providers; however, we see much potential in considering the value that is delivered by focusing on patient-centricity (pathway 3).

Putting the patient at the centre of HIS research implies shifting the focus of researchers to the patient’s own processes. The question remains as to how HIS researchers can support patient-centricity. While this is only possible by understanding patients’ processes, we also see the need to understand the whole system – i.e., the ecosystem in which patients’ processes are embedded. The ecosystem perspective needs to consider networked services and organizations, including resources and how they interact with stakeholders of the healthcare industry (including patients). To date, we observe, across phases the ecosystem perspective has largely been neglected. To be precise, although HIS research seems to be aware of the multilevel aspects of healthcare issues in the problem identification phase, researchers appear to stop or are hindered from developing solutions that go beyond the development of prototypes (Braa et al., 2004 ). Thus, we find that:

HIS research is capable of theorizing on an ecosystem level (i.e., capturing the complexity of the socio-technical health system), but would benefit from increasing the transfer of these insights into research so as to develop holistic solutions (pathway 4).

Looking at the strengths of HIS research, the reviewed papers accentuate the unique contribution that IS researchers can make to better understand and design IS artefacts for the healthcare context. This has been achieved by analysing empirical data and exploring contextual influences through the application and elaboration of IS theories (LeRouge et al., 2007 ). At the same time, our literature review shows the incredible diversity and high level of complexity of issues related to HISs, indicating that we need solutions characterized by holism and the inclusion of multiple actors (i.e., an integrative ecosystem perspective). So far, by concentrating on incremental improvements to existing infrastructures HIS research has widely failed to reach the necessary holistic level.

We would like to emphasize that we recognize the value of all previous approaches. Yet, it is necessary to ask whether we as IS researchers are in a position to identify current developments in the healthcare industry and to anticipate the consequences triggered by pandemics or other waves of disease. We acknowledge that this will be difficult unless we take a more holistic view and try to understand connections in the health ecosystems. Regarding whether HIS research is in a position to capture and anticipate consequences of the current push of tech companies in the healthcare industry catalysed, for example, by Covid-19, we assert that this is hardly the case, even if IS research is well-placed to interpret the expected socio-technical changes and adaptations within healthcare. Given the enormous potential for disruption caused by, for instance, pandemics and its consequences, such as the intrusion of technology companies into the market, it is now time to question and redefine the role of HIS research so that it can generate decisive impacts on the developments in this industry.

• Research agenda

To support HIS research for the transformation of the healthcare industry, we develop a research agenda that is informed by complexity theory. This theory implies that complex, socio-technical systems such as the healthcare industry can fluctuate between different states, ranging from homogenous forms of coevolution (i.e., a state where emergent structures and processes become similar to each other) to chaotic systems that are characterized by increasing levels of tension, which might result in extreme outcomes such as catastrophes or crises (Benbya et al., 2020 ).

While coevolution and chaos represent possible extreme states, the current situation – i.e., the penetration of tech companies into the healthcare industry – is best described by the dynamic process of emergence. Emergence is characterized by a disequilibrium, which implies unpredictability of outcomes that may lead to new structures, patterns, and properties within a system characterized by self-organization and bursts of amplification (Benbya et al., 2020 ; Kozlowski et al., 2013 ). Given the dynamics resulting from this, it seems impossible to predict the future; however, it is not impossible to prepare for it.

In particular, the current dynamics within the healthcare industry necessitate an understanding of exponential progress, not as the ability to foresee well-defined events in space and time, but as an anticipation of the consequences of emerging states and dynamic adaptive behaviours within the industry (Benbya et al., 2020 ). The following research agenda for HIS research is thus structured along three key issues: anticipating the range of actors’ behaviours, determining boundaries and fostering collaboration in the healthcare industry, and creating sustainable knowledge ecosystems.

According to these key issues, Table 6 offers guiding questions for HIS researchers. Addressing all issues will contribute to an understanding of the entire healthcare industry and the development of holistic solutions for a multitude of health issues by involving different actors (e.g., patients, hospitals, professionals, governments, NGOs). However, we propose approaching the agenda stepwise, in the order of the key issues, first looking at the range of behaviours and consequences of current developments for actors, then focusing on the blurring lines of the healthcare industry, and finally investigating the dissemination and sharing of knowledge, which we see as the ultimate means to connect actors and infrastructures to create a joint ecosystem. Table 6 thereby provides key guiding statements and exemplary research questions for future HIS research that support researchers in taking one of the aforementioned pathways. We structured guiding statements along three major areas of improvement. In addition, we offer exemplary research questions to these statements, as well as inspiring studies from other industries that have faced similar challenges and have been studied and supported by researchers.

Area of improvement 1: Anticipating the range of actor behaviours

As healthcare systems are becoming more open – for example, through the penetration of new market actors and the use of increasingly comprehensive and advanced health technologies – accurately determining the boundaries of an industry and its key actors is becoming more difficult. To model these systems, we must carefully model every interaction in them (Benbya et al., 2020 ), which first requires HIS researchers to identify potential actors in the ecosystem rather than predetermining assumed industry boundaries. As actors are not always evident, we follow Benbya et al. ( 2020 ) in proposing Salthe’s ( 1985 ) three-level specification, assisting researchers in identifying actors at the focal level of what is actually observed (e.g., hospitals, patients, and general practitioners) and its relations with the parts described at the lower level (e.g., administrators and legal professionals), taking into account entities or processes at a higher level in which actors at the focal level are embedded (e.g., national health system structures and supporting industries, such as the pharmaceutical or tech industries). These examples are only illustrative, and criteria for levels have to be suggested and discussed for each research endeavour.

To anticipate future developments in the healthcare industry, we also need to analyse the strategies and interests of actors for joining or staying in the healthcare industry. This is especially important because, like other complex socio-technical systems, the healthcare industry is made up of large numbers of actors that influence each other in nonlinear ways, continually adapting to internal or external tensions (Holland et al., 1996 ). If tension rises above a certain threshold, we might expect chaos or extreme outcomes. As these are not beneficial for the actors in the system, the eventual goal is to align actors’ interests and strategies across a specific range of behaviour to foster coevolution. This allows for multi-layered ecosystems that encourage joint business strategies in competitive landscapes, as well as the alignment of business processes and IT across actors (Lee et al., 2013 ).

Area of improvement 2: Determining boundaries and fostering collaboration

Actors build the cornerstones of the healthcare industry. Thus, if we want to understand and capture its blurring boundaries, there is a need to understand the complex causality of interactions among heterogeneous actors. In particular, scholars have emphasized that, in complex systems, outcomes rarely have a single cause but rather result from the interdependence of multiple conditions, implying that there exist multiple pathways from an input to an output (Benbya et al., 2020 ). To capture interaction, we follow Kozlowski et al. ( 2013 ), who envisioned a positive feedback process including bottom-up dynamic interaction among lower-level actors (upward causation), which over time manifests at higher, collective levels, while higher-level actors influence interaction at lower levels (downward causation). As these kinds of causalities shape interaction within healthcare ecosystems as well as at their boundaries, HIS researchers need to account for multi-directional causality in the form of upward, downward, and circular causality (Benbya et al., 2020 ; Kim, 1992 ).

Understanding casualties among actors in the healthcare industry is important for harnessing the advantages of the blurring of boundaries – e.g., by making use of the emergent ecosystem for launching innovation cycles (Hacklin, 2008 ). However, first, HIS researchers increasingly need to consider the ecosystem perspective by investigating interactions among actors and the role of IS infrastructures in fostering collaborative health innovations. We propose a focus on radical innovation, which is necessary to address the diversity and interdependence of issues present in the healthcare industry by putting the patient at the core of all innovation efforts. HIS researchers, however, need to break down the boundaries between different innovation phases and innovation agencies, including a higher level of unpredictability and overlap in their time horizons (Nambisan et al., 2017 ). Notably, this requires actors in the healthcare industry to discover new meaning around advanced technologies and IS infrastructures whose design needs to facilitate shared meaning among a diverse set of actors, thereby fuelling radical digital innovations (Nambisan et al., 2017 ).

Area of improvement 3: Creating sustainable knowledge ecosystems

We define knowledge dissemination and sharing as the ultimate means of connecting actors and aligning actions within common frameworks to shape an inclusive healthcare ecosystem. Paving the way for inclusive healthcare ecosystems is thus necessary to address the current shortcomings of HIS research as elaborated in the previous section.

Addressing knowledge dissemination and sharing is thereby of the utmost importance as we look at the healthcare industry in the current phase of emergence. This means that the industry might go through several transition phases in which existing actors, structures, and causal relationships dissipate and new ones emerge, resulting in a different set of causal relationships and eventually altering knowledge claims (Benbya et al., 2020 ). Creating a permeable and sustainable knowledge management system is necessary to ensure the transfer of knowledge for the best outcomes for the patient while securing the intellectual property rights and competitive advantages of diverse actors such as hospitals and other healthcare providers.

To be precise, we argue that to design sustainable knowledge management systems, HIS researchers need to implement systems with structures that create mutual benefits – i.e., encourage knowledge dissemination and sharing (e.g., open innovation) by actors in the healthcare industry. In a comprehensive and sustainable knowledge management system, however, not only corporations but also patients should be encouraged to share knowledge. Using this information, researchers and health service providers will be enabled to create optimized infrastructures, processes, and products (e.g., for predictive algorithms that improve treatment accuracy, or for assessing the likelihood of the occurrence of certain diseases and even of pandemics). At the same time, the trustworthiness of predictions and the anonymity of health information (and thus privacy) must be ensured. Bridging this duality of data sharing and knowledge dissemination, on the one hand, and protection of health information, on the other, is therefore essential for future HIS research.

This paper analyses the HIS literature within the IS research domain, prompted by the question of whether IS researchers are prepared to capture and anticipate exogenous changes and the consequences of current developments in the healthcare industry. While this review is limited to insights into the IS research domain and does not claim to offer insights into the health literature in general or related publications (e.g., governmental publications), we disclose several shortcomings and three key issues. Based on these, we provide initial guidance on how IS research can develop so that it is prepared to capture the expected large and long-lasting changes from current and possible future pandemics as well as the necessary adaptation of global healthcare industries affecting human agencies and experiences in all dimensions. Thus, while adaptations in the healthcare industry are already emerging, IS researchers have yet to develop a more comprehensive view of the healthcare industry. For this purpose, we provide a research agenda that is structured in terms of three areas of improvement: anticipating the range of actors’ behaviours, determining boundaries and fostering collaborations among actors in the healthcare industry, and creating sustainable knowledge management systems. In particular, addressing these areas will assist IS researchers in balancing the shortcomings of current HIS research with the unique contribution that IS research plays in analysing, advancing, and managing the healthcare industry. We are confident that IS research is not only capable of anticipating changes and consequences but also of actively shaping the future of the healthcare industry by promoting sustainable healthcare ecosystems, cultivating structures of mutual benefit and cooperation between actors, and realigning IS research to face the imminent transformation of the healthcare industry. IS research cannot contribute directly to solving the current pandemic problems; however, it can contribute indirectly triggering timely adaptations of novel technologies in global health systems, and proposing new processes, business models, and systematic changes that will prepare health systems to cope with increasing digitalization and emerging players whose push into the market enabled by the exogenous effects triggered by the pandemic.

While we are confident that the proposed research agenda based on the analysis of HIS literature provides fruitful arrays for being prepared in anticipating the future role of IS research for the healthcare industry, our results need to be reflected in light of their shortcomings. First and foremost, we recognize that the selection of literature, which is limited to the IS research domain, excludes other contextual factors that are not primarily considered by IS researchers. Thus, we cannot assume completeness, providing instead a broad overview of current issues in HIS research. In addition, possible biases may have arisen due to the qualitative analysis approach used. By independently coding and discussing codes to the point of theoretical saturation, we are confident that we largely eliminated biases in the thematic analysis. However, data saturation could not be achieved. This means that further insights could have emerged through the addition of other database searches and journals with a broader scope. Additionally, the initial sorting of papers into single defined phases of DSR research restricted multiple assignments that could have led to different results. However, we consider sorting as a necessary step of abstraction, especially given the large number of papers analysed.

We deliberately considered IS research, for which we have developed an agenda for potential future research avenues. For each of those avenues, researchers should go deeper into the subject matter in order to examine the complexity of the paths shown and to include them in the analysis (e.g., through in-depth case studies). However, it is also clear from the issues identified that IS researchers cannot solve current challenges by working on the pathways alone. In fact, the issues identified in the research agenda are only the starting point for further research, which should address the proposed issues step by step and in cooperation with other research disciplines. The latter is likely to generate further and deeper-rooted problems, as well as, in turn, future paths for research. Nevertheless, we are confident that this paper provides an important first step in opening up HIS research to better understand current developments in the healthcare industry. Further, by following and enhancing the proposed research pathways, we believe that HIS research can contribute to and support changes already taking place in the healthcare industry.

Change history

20 december 2021.

A Correction to this paper has been published: https://doi.org/10.1007/s12525-021-00518-8

Aanestad, M., Vassilakopoulou, P., & Ovrelid, E. (2019). Collaborative innovation in healthcare: boundary resources for peripheral actors. International Conference on Information Systems (ICIS), 1 , 1–17.

Google Scholar  

Abouzahra, M., & Tan, J. (2014). The multi-level impact of clinical decision support system: a framework and a call for mixed methods. Pacific Asia Conference on Information Systems (PACIS), 1–17.

Agarwal, R., Gao, G., DesRoches, C., & Jha, A. K. (2010). Digital transformation of healthcare: current status and the road ahead. Information Systems Research, 21 (4), 796–809. https://doi.org/10.1287/isre.1100.0327

Article   Google Scholar  

Akcura, M. T., & Ozdemir, Z. D. (2008). Physician learning and clinical decision support systems. Americas Conference on Information Systems  (AMCIS) , 1–10.

Akkerman, S. F., & Bakker, A. (2011). Boundary crossing and boundary objects. Review of Educational Research, 81 (2), 132–169. https://doi.org/10.3102/0034654311404435

Al-Qirim, N. (2003). Championing telemedicine in New Zealand: the case of utilizing video conferencing in psychiatry and dermatology. Americas Conference on Information Systems (AMCIS),  1–10.

Arvanitis, S., & Loukis, E. (2014). An empirical investigation of the impact of ICT on innovation in European hospitals. European Conference on Information Systems (ECIS),  1–13.

Babar, Y., Chan, J., & Choi, B. (2018). "Run Forrest Run!": measuring the impact of app-enabled performance and social feedback on running performance. Pacific Asia Conference on Information Systems (PACIS 2018),  160.

Benbya, H., Nan, N., Tandriverdi, H., & Yoo, Y. (2020). Complexity and information systems research in the emerging digital world. MIS Quarterly , 44 (1), 1–17. https://doi.org/10.25300/MISQ/2020/13304

Bhandari, G., & Maheshwari, B. (2009). Toward an integrated health information system for collaborative decision making and resource sharing: findings from a Canadian study.  Americas Conference on Information Systems (AMCIS),  648.

Bleicher, J. (2017). Contemporary Hermeneutics: Hermeneutics as method, philosophy, and critique (2nd ed.). Routledge.

Book   Google Scholar  

Blinn, N., & Kühne, M. (2013). Health information on the internet - state of the art and analysis. Business & Information Systems Engineering, 5 (4), 259–274. https://doi.org/10.1007/s12599-013-0274-4

Boell, S., & Cecez-Kecmanovic, D. (2014). A hermeneutic approach for conducting literature reviews and literature searches. Communication of the Association for Information Systems, 34 (12), 257–286.  https://doi.org/10.17705/1CAIS.03412

Braa, J., Monteiro, E., & Sahay, S. (2004). Networks of action: sustainable health information systems across developing countries. MIS Quarterly, 28 (3), 337–362. https://doi.org/10.2307/25148643

Bullinger, A., Rass, M., & Moeslein, K. (2012). Towards open innovation in health care. European Conference on Information Systems (ECIS),  1–13.

Byrd, L., & Byrd, T. (2009). Examining the effects of healthcare technology on operational cost. Americas Conference on Information Systems (AMCIS),  1–10.

CB Insights. (2018). How Google Plans to Use AI to Reinvent the $3 Trillion US Healthcare Industry . https://www.distilnfo.com/lifesciences/files/2018/11/CB-Insights_Google-Strategy-Healthcare.pdf

Chang, I. C., Hwang, H. G., Hung, M. C., Kuo, K. M., & Yen, D. C. (2009). Factors affecting cross-hospital exchange of electronic medical records. Information & Management, 46 (2), 109–115. https://doi.org/10.1016/j.im.2008.12.004

Chen, L., Baird, A., & Straub, D. (2019). An analysis of the evolving intellectual structure of health information systems research in the information systems discipline. Journal of the Association for Information Systems , 20 (8), 1023–1074. https://doi.org/10.17705/1jais.00561

Chiasson, M. W., & Davidson, E. (2004). Pushing the contextual envelope: developing and diffusing IS theory for health information systems research. Information and Organization, 14 (3), 155–188. https://doi.org/10.1016/j.infoandorg.2004.02.001

Choi, W., & Tulu, B. (2017). Effective use of user interfaces and user experience in an mHealth application. Hawaii International Conference on System Sciences (HICSS),  3803–3812. https://doi.org/10.24251/HICSS.2017.460

Cole, M., & Avison, D. (2007). The potential of hermeneutics in information systems research. European Journal of Information Systems, 16 (6), 820–833.  https://doi.org/10.1057/palgrave.ejis.3000725

Cilliers, P. (2001). Boundaries, hierarchies and networks in complex systems. International Journal of Innovation Management, 5 (2), 135–147. https://doi.org/10.1515/9781501502590-009

Corbin, J. M., & Strauss, A. (1990). Grounded theory research: procedures, canons, and evaluative criteria. Qualitative Sociology, 13 (1), 3–21. https://doi.org/10.1007/BF00988593

Currie, W. L., & Seddon, J. J. (2014). A cross-national analysis of eHealth in the European Union: some policy and research directions. Information & Management, 51 (6), 783–797. https://doi.org/10.1016/j.im.2014.04.004

Dehling, T., & Sunyaev, A. (2012). Architecture and design of a patient-friendly eHealth web application: patient information leaflets and supplementary services.  Americas Conference on Information Systems (AMCIS),  1–8.

Devadoss, P., & Pan, S. L. (2004). Leveraging eGovernment infrastructure for crisis management: lessons from managing SARS outbreak in Singapore. Americas Conference on Information Systems (AMCIS),  1–10.

Djamsbi, S., Fruhling, A., & Loiacono, E. (2009). The influence of affect, attitude and usefulness in the acceptance of telemedicine systems. Journal of Information Technology Theory and Application, 10 (1), 1–38.

Dünnebeil, S., Krcmar, H., Sunyaev, A., Leimeister, J., & M. (2013). Modular architecture of value-added applications for German healthcare telematics. Business & Information Systems Engineering, 5, 3–16. https://doi.org/10.1007/s12599-012-0243-3

Enkel, E., & Gassmann, O. (2010). Creative imitation: exploring the case of cross-industry innovation. R&d Management, 40 (3), 256–270. https://doi.org/10.1111/j.1467-9310.2010.00591.x

Fichman, R. G., Kohli, R., & Krishnan, R. (2011). The role of information systems in healthcare: current research and future trends. Information Systems Research, 22 (3), 419–428. https://doi.org/10.2307/23015587

Findikoglu, M., & Watson-Manheim, M. B. (2016). Linking macro-level goals to micro-level routines: EHR-enabled transformation of primary care services. Journal of Information Technology, 31 (4), 382–400. https://doi.org/10.1057/s41265-016-0023-5

Fisher, J., Burstein, F., Lynch, K., Lazarenko, K., & McKemmish, S. (2007). Health information websites: is the health consumer being well-served? Americas Conference on Information Systems (AMCIS),  1–10.

Fisman, R., Branstetter, L. G., & Foley, C. F. (2004). Do stronger intellectual property rights increase international technology transfer? Empirical evidence from US firm-level panel data . The World Bank, Washington, DC. https://doi.org/10.1596/1813-9450-3305

Gabel, M., Foege, J. N., & Nuesch, S. (2019). The (In)Effectiveness of incentives - a field experiment on the adoption of personal electronic health records.  International Conference on Information Systems (ICIS),  1–17.

Gantori, S., Issel, H., Donovan, P., Rose, B., Kane, L., Dennean, K., Ganter R., Sariyska, A., Wayne, G., Hyde, C., & Lee, A. (2020). Future of the Tech Economy . https://www.ubs.com/global/en/wealth-management/chief-investment-office/investment-opportunities/investing-in-the-future/2020/future-of-tech-economy.html

Gaskin, J., Berente, N., Lyytinen, K., & Yoo, Y. (2014). Toward generalizable sociomaterial inquiry. MIS Quarterly , 38 (3), 849–872. https://doi.org/10.25300/MISQ/2014/38.3.10

George, J. F., & Kohnke, E. (2018). Personal health record systems as boundary objects. Communication of the Association for Information Systems , 42 (1), 21–50. https://doi.org/10.17705/1CAIS.04202

Ghanvatkar, S., & Rajan, V. (2019). Deep recurrent neural networks for mortality prediction in intensive care using clinical time series at multiple resolution.  International Conference on Information Systems (ICIS),  1–18.

Gimpel, H., Nißen, M., & Görlitz, R. A. (2013). Quantifying the quantified self: a study on the motivation of patients to track their own health. International Conference on Information Systems (ICIS),  1–17.

Glaser, B. G., Strauss, A. L., & Strutzel, E. (1968). The discovery of grounded theory; strategies for qualitative research. Nursing Research, 17 (4), 364.

Greenwood, F. (2020, March). Google Want Your Data in Exchange for a Coronavirus Test. Foreign Policy. https://foreignpolicy.com/2020/03/30/google-personal-health-data-coronavirus-test-privacy-surveillance-silicon-valley/

Grondin, J. (2016). What is the hermeneutical cycle? In N. Keane & C. Lawn (Eds.), The Blackwell Companion to Hermeneutics,  299–305. Oxford, UK: Blackwell Publishing.

Hacklin, F. (2008). Management of Convergence in Innovation: Strategies and Capabilities for Value Creation Beyond Blurring Industry Boundaries . Zurich, Switzerland: Physica.

Hafermalz, E., & Riemer, K. (2016). Negotiating distance: "presencing work” in a case of remote telenursing. International Conference on Information Systems(ICIS), 1–13.

Hamari, J., & Koivisto, J. (2015). Why do people use gamification services? International Journal of Information Management, 35 (4), 419–431. https://doi.org/10.1016/j.ijinfomgt.2015.04.006

Heine, C., Herrler, R., Petsch, M., & Anhalt, C. (2003). ADAPT: Adaptive Multi-Agent Process Planning and Coordination of Clinical Trials. Americas Conference on Information Systems (AMCIS),  1823–1834.

Holland, S., Phimphachanh, C., Conn, C., & Segall, M. (1996). Impact of economic and institutional reforms on the health sector in Laos: implications for health system management. IDS Publications, Institute of Development Studies, 28 , 133–154.

Huerta, T. R., Thompson, M. A., Ford, E. W., & Ford, W. F. (2013). Electronic health record implementation and hospitals’ total factor productivity. Decision Support Systems, 55 (2), 450–458. https://doi.org/10.1016/j.dss.2012.10.004

Humphrey, S. E. (2011). What does great meta-analysis look like? Organizational Psychology Review, 1 (2), 99–103. https://doi.org/10.1177/2041386611401273

Idrissi, T. E., Idri, A., & Bakkoury, Z. (2019). Systematic map and review of predictive techniques in diabetes self-management. International Journal of Information Management, 46 , 263–277. https://doi.org/10.1016/j.ijinfomgt.2018.09.011

Kenny, C., & Connolly, R. (2016). Drivers of health information privacy concern: a comparison study.  Americas Conference on Information Systems (AMCIS),  1–10.

Kernick, D., & Mitchell, A. (2009). Working with lay people in health service research: a model of co-evolution based on complexity theory. Journal of Interprofessional Care, 24 (1), 31–40. https://doi.org/10.3109/13561820903012073

Kim, S. H., & Kwon, J. (2019). How Do EHRs and a meaningful use initiative affect breaches of patient information? Information Systems Research, 30 (4), 1107–1452. https://doi.org/10.1287/isre.2019.0858

Kim, S. (1992). Downward causation in emergentism and non-reductive physicalism. In A. Beckermann, H. Flohr, & J. Kim (Eds.). Emergence or Reduction. Essays on the Prospects of Nonreductive Physicalism (118–138). Berlin, Germany: de Gryuter.

King, N. (2009). An initial exploration of stakeholder benefit dependencies in ambulatory ePrescribing. Americas Conference on Information Systems (AMCIS),  1–10.

Klecun, E., Zhou, Y., Kankanhalli, A., Wee, Y. H., & Hibberd, R. (2019). The dynamics of institutional pressures and stakeholder behavior in national electronic health record implementations: a tale of two countries. Journal of Information Technology, 34 (4), 292–332. https://doi.org/10.1177/0268396218822478

Knight, J., Patrickson, M., & Gurd, B. (2008). Towards understanding apparent South Australian GP resistance to adopting Health Informatics systems.  Australasian Conference on Information Systems (ACIS),  492–501.

Kozlowski, S. W. J., Chao, G. T., Grand, J. A., Braun, M. T., & Kulijanin, G. (2013). Advancing multilevel research design: capturing the dynamics of emergence. Organizational Research Methods, 16 (4), 581–615. https://doi.org/10.1177/1094428113493119

Landi, H. (2020, April). Alphabet’s Verily rolls out COVID screening tool for health systems . FierceHealthcare. https://www.fiercehealthcare.com/tech/alphabet-s-verily-rolls-out-covid-screening-tool-for-health-systems

Lebcir, R. M., Choudrie, J., Atum, R. A., & Corker, R. J. (2008). Examining HIV and tuberculosis using a decision support systems computer simulation model: the case of the Russian Federation. Americas Conference on Information Systems (AMCIS),  1–18.

Lee, D., & Nilsson, P. (2020, March). Amazon auditions to be “the new Red Cross” in Covid-19 crisis . Financial Times. https://www.ft.com/content/220bf850-726c-11ea-ad98-044200cb277f

Lee, A. S., Thomas, M., & Baskerville, R. L. (2015). Going back to basics in design science: from the information technology artifact to the information systems artifact. Information Systems Journal, 25 (1), 5–21. https://doi.org/10.1111/isj.12054

Lee, C. H., Venkatraman, N., Tanriverdi, H., & Iyer, B. (2013). Complementary-based hypercompetition in the software industry. Strategic Management Journal, 31 (13), 1431–1356. https://doi.org/10.1002/smj.895

Leon, M. C., Nieto-Hipolito, J. I., Garibaldi-Beltran, J., Amaya-Parra, G., Luque-Morales, P., Magana-Espinoza, P., & Aquilar-Velazco, J. (2016). Designing a model of a digital ecosystem for healthcare and wellness using the BM canvas. Journal of Medical Systems, 40 (6), 144–154. https://doi.org/10.1007/s10916-016-0488-3

LeRouge, C., Mantzana, V., & Wilson, E. V. (2007). Healthcare information systems research, revelations and visions. European Journal of Information Systems, 16 (6), 669–671. https://doi.org/10.1057/palgrave.ejis.3000712

Leroy, J., Cova, B., & Salle, R. (2013). Zooming in VS zooming out on value co-creation: consequences for BtoB research. Industrial Marketing Management, 42 (7), 1102–1111. https://doi.org/10.1016/j.indmarman.2013.07.006

Li, L., Jeng, L., Naik, H. A., Allen, T., & Frontini, M. (2008). Creation of environmental health information system for public health service: a pilot study. Information Systems Frontiers, 10 (5), 531–542. https://doi.org/10.1007/s10796-008-9108-1

Lichtenberg, S., Greve, M., Brendel, A. B., & Kolbe, L. M. (2019). Towards the design of a mobile application to support decentralized healthcare in developing countries – The case of diabetes care in eSwatini. Americas Conference on Information Systems (AMCIS),  1–10.

Lin, Y.-K., Chen, H., Brown, R. A., Li, S.-H., & Yang, H.-J. (2017). Healthcare predictive analytics for risk profiling in chronic care: a Bayesian multitask learning approach. MIS Quarterly , 41 (2), 473–495. https://doi.org/10.25300/MISQ/2017/41.2.07

MacLure, M. (2005). Clarity bordering on stupidity: where’s the quality in systematic review? Journal of Education Policy, 20 (4), 393–416. https://doi.org/10.4324/9780203609156

Maheshwari, M., Hassan, T., & Chatterjee, S. (2006). A framework for designing healthy lifestyle management information system.  Americas Conference on Information Systems (AMCIS),  2811–2817.

Matavire, R., & Brown, I. (2008). Investigating the use of “Grounded Theory” in information systems research. Proceedings of the 2008 annual research conference of the South African Institute of Computer Scientists and Information Technologists on IT research in developing countries,  139–147.  https://doi.org/10.1145/1456659.1456676

Mitsa, T., Fortier, P. J., Shrestha, A., Yang, G., & Dluhy, N. M. (2007). Information systems and healthcare XXI: a dynamic, client-centric, point-of-care system for the novice nurse. Communication of the Association for Information Systems , 19 , 740–761. https://doi.org/10.17705/1CAIS.01936

Motamarri, S., Akter, S., Ray, P., & Tseng, C.-L. (2014). Distinguishing “mHealth” from other healthcare services in a developing country: a study from the service quality perspective. Communication of the Association for Information Systems , 34 (1), 669–692. https://doi.org/10.17705/1CAIS.03434

Mun, C., & Lee, O. (2017). Integrated supporting platform for the visually impaired: using smart devices. International Conference on Information Systems (ICIS),  1–18.

Nambisan, S., Lyytinen, K., Majchrzak, A., & Song, M. (2017). Digital innovation management: reinventing innovation management research in a digital world. MIS Quarterly , 41 (1), 223–238. https://doi.org/10.25300/MISQ/2017/41:1.03

Neben, T., Seeger, A. M., Kramer, T., Knigge, S., White, A. J., & Alpers, G. W. (2016). Make the most of waiting: theory-driven design of a pre-psychotherapy mobile health application.  Americas Conference on Information Systems (AMCIS),  1–10.

Nicholls-Nixon, C. L., & Jasinski, D. (1995). The blurring of industry boundaries: an explanatory model applied to telecommunications. Industrial and Corporate Change, 4 (4), 755–768. https://doi.org/10.1093/icc/4.4.755

Offermann, P., Blom, S., Schönherr, M., & Bub, U. (2010). Artifact types in information systems design science – a literature review. International Conference on Design Science Research in Information Systems,  77–92. https://doi.org/10.1007/978-3-642-13335-0_6

Palvia, P., Kakhki, M. D., Ghoshal, T., Uppala, V., & Wang, W. (2015). Methodological and topic trends in information systems research: a meta-analysis of IS journals. Communication of the Association for Information Systems , 37 (30), 630–650. https://doi.org/10.17705/1CAIS.03730

Park, H. A. (2016). Are we ready for the fourth industrial revolution? Yearbook of Medical Informatics , (1), 1–3. https://doi.org/10.15265/IY-2016-052

Patel, M., Shah, A., Shah, K., & Plachkinova, M. (2018). Designing a mobile app to help young adults develop and maintain mental well-being.  Americas Conference on Information Systems (AMCIS),  1–10.

Paul, L. D., & McDaniel, R. R. J. (2004). A field study of the effect of interpersonal trust on virtual collaborative relationship performance. MIS Quarterly, 28 (2), 183–227.  https://doi.org/10.2307/25148633

Peffers, K., Tuunanen, T., Rothenberger, M. A., & Chatterjee, S. (2008). A design science research methodology for information systems research. Journal of Management Information Systems, 24 (3), 45–78. https://doi.org/10.2753/MIS0742-1222240302

Pletikosa, I., Kowatsch, T., Büchter, D., Brogle, B., Dintheer, A., Wiegand, D., Durrer, D., l'Allemand-Jander, D., Schutz, Y., Maass, W. (2014). Health information system for obesity prevention and treatment of children and adolescents. European Conference on Information Systems (ECIS),  1–13.

Preko, M., Boateng, R., & Effah, J. (2019). Health informatics and brain drain mitigation in Ghana.  Americas Conference on Information Systems (AMCIS),  1–10.

Rivard, P. E., Rosen, A. K., & Carroll, J. S. (2006). Enhancing patient safety through organizational learning: are patient safety indicators a step in the right direction?  Health Services Research , 41(4p2), 1633–1653. https://doi.org/10.1111/j.1475-6773.2006.00569.x

Rodon, J., & Silva, L. (2015). Exploring the formation of a healthcare information infrastructure: hierarchy or meshwork? Journal of the Association for Information System, 16 (5), 394–417. https://doi.org/10.17705/1JAIS.00395

Roehrig, S., & Knorr, K. (2000). Toward a secure web based health care application. European Conference on Information Systems (ECIS),  1–8. https://doi.org/10.4018/978-1-930708-13-6.ch007

Romanow, D., Rai, A., & Keil, M. (2018). CPOE-Enabled Coordination: appropriation for deep structure use and impacts on patient outcomes. MIS Quarterly , 42 (1), 189–212. https://doi.org/10.25300/MISQ/2018/13275

Ryan, J., Doster, B., Daily, S., & Lewis, C. (2019). Seeking operational excellence via the digital transformation of perioperative scheduling. Americas Conference on Information Systems (AMCIS),  1–10.

Rycroft, R. W., & Kash, D. E. (2004). Self-organizing innovation networks: implications for globalization. Technovation, 24 (3), 187–197. https://doi.org/10.1016/S0166-4972(03)00092-0

Salthe, S. N. (1985). Evolving Hierarchical Systems . Columbia University Press.  https://doi.org/10.7312/salt91068

Samentinger, J., Rozenblit, J., Lysecky, R., & Ott, P. (2015). Security challenges for medical devices. Communications of the Association for Information Systems, 58 (4), 74–82.  https://doi.org/10.1145/2667218

Saunders, B., Sim, J., Kingstone, T., Baker, S., Waterfield, J., Bartlam, B., Burroughs, H., & Jinks, C. (2018). Saturation in qualitative research: exploring its conceptualization and operationalization. Quality & Quantity, 52 (4), 1893–1907. https://doi.org/10.1007/s11135-017-0574-8

Schweiger, A., Sunyaev, A., Leimeister, J. M., & Krcmar, H. (2007). Information systems and healthcare XX: toward seamless healthcare with software agents. Communications of the Association for Information Systems , 19 (33), 392–710. https://doi.org/10.17705/1CAIS.01933

Schwetschke, S., & Durugbo, C. (2018). How firms synergise: understanding motives and management of co-creation for business-to-business services. International Journal of Technology Management, 76 (3–4), 258–284. https://doi.org/10.1504/IJTM.2018.091289

Setia, P., Setia, M., Krishnan, R., & Sambamurthy, V. (2011). The effects of the assimilation and use of IT applications on financial performance in healthcare organizations. Journal of the Association for Information System , 12 (Special Issue), 274–298. https://doi.org/10.17705/1jais.0060

Shahmoradi, L., Safadari, R., & Jimma, W. (2017). Knowledge management implementation and the tools utilized in healthcare for evidence-based decision making: A systematic review. Ethiopian Journal of Health Sciences, 27 (5), 541–558. https://doi.org/10.4314/ejhs.v27i5.13

Sherer, S. A. (2014). Patients are not simply health it users or consumers: the case for “e Healthicant” applications. Communications of the Association for Information systems , 34(1), 17. https://doi.org/10.17705/1CAIS.03417 .

Singh, R., Mathiassen, L., Stachura, M. E., & Astapova, E. V. (2011). Dynamic capabilities in home health: IT-enabled transformation of post-acute care. Journal of the Association for Information System , 12 (2), 163–188. https://doi.org/10.17705/1jais.00257

Sujanto, F., F., B., Ceglowski, A., & Churilov, L. (2008). Application of domain ontology for decision support in medical emergency coordination. Americas Conference on Information Systems (AMCIS),  1–10.

Theobalt, A., Emrich, A., Werth, D., & Loos, P. (2013). A conceptual architecture for an ICT-based personal health system for cardiac rehabilitation. Americas Conference on Information Systems (AMCIS),  1–10.

Therrien, M.-C., Normandin, J.-M., & Denis, J.-L. (2017). Bridging complexity theory and resilience to develop surge capacity in health systems. Journal of Health Organisation and Management, 31 (1), 96–109. https://doi.org/10.1108/JHOM-04-2016-0067

Tillett, A. (2020, April). Amazon to store data from virus tracing app. Financial Review. https://www.afr.com/politics/federal/amazon-to-store-data-from-virus-tracing-app-20200424-p54mwq

Tokar, O., Batoroev, K., & Böhmann, T. (2015). A framework for analyzing patient-centered mobile applications for mental health. Americas Conference on Information Systems (AMCIS),  1–15.

Tsai, J. M., Cheng, M. J., Tsai, H. H., Hung, S. W., & Chen, Y. L. (2019). Acceptance and resistance of telehealth: the perspective of dual-factor concepts in technology adoption. International Journal of Information Management, 49 , 34–44. https://doi.org/10.1016/j.ijinfomgt.2019.03.003

Ure, J., Procter, R., Lin, Y. W., Hartswood, M., Anderson, S., Lloyd, S., Wardlaw, J., Gonzalez-Velez, H., & Ho, K. (2009). The development of data infrastructures for eHealth: a socio-technical perspective. Journal of the Association for Information Systems , 10 (5), 415–429. https://doi.org/10.17705/1jais.00197

Varshney, U. (2004). Using wireless networks for enhanced monitoring of patients.  Americas Conference on Information Systems (AMCIS),  1–10. https://doi.org/10.1504/IJHTM.2005.007009

Volland, D., Korak, K., & Kowatsch, T. (2014). A health information system that extends healthcare professional-patient communication.  European Conference on Information Systems (ECIS),  1–10.

vom Brocke, J., Simons, A., Reimer, K., Niehaves, B., Plattfaut, R., & Cleven, A. (2015). Standing on the shoulders of giants: challenges and recommendations of literature search in information systems research. Communications of the Association for Information Systems, 37 (1), 205–224.  https://doi.org/10.17705/1CAIS.03709

vom Brocke, J., Simons, A., Niehaves, B., Reimer, K., Plattfaut, R., & Cleven, A. (2009): Reconstructing the giant: on the importance of rigour in documenting the literature search process. European Conference on Information Systems (ECIS) , 1–12.

Walsh, D., & Downe, S. (2005). Meta-synthesis method for qualitative research: a literature review, methodological issues in nursing. Journal of Advanced Nursing, 50 (2), 204–211. https://doi.org/10.1111/j.1365-2648.2005.03380.x

Webster, J., & Watson, R. T. (2002). Analyzing the past to prepare for the future: writing a literature review. MIS Quarterly, 26 (2), 13–23. https://doi.org/10.2307/4132319

Werts, N., & Adya, M. (2000). Data mining in healthcare: issues and a research agenda. Americas Conference on Information Systems (AMCIS),  98.

Wessel, L., Gersch, M., & Harloff, E. (2017). Talking past each other - a discursive approach to the formation of societal-level information pathologies in the context of the electronic health card in Germany. Business & Information Systems Engineering, 59 (1), 23–40. https://doi.org/10.1007/s12599-016-0462-0

Winter, S. J., & Butler, B. S. (2011). Creating bigger problems: grand challenges as boundary objects and the legitimacy of the information systems field. Journal of Information Technology, 26 (2), 99–108. https://doi.org/10.1057/jit.2011.6

Wolfswinkel, J. F., Furtmueller, E., & Wilderom, C. P. M. (2013). Using grounded theory as a method for rigorously reviewing literature. European Journal of Information Systems, 22 (1), 45–55. https://doi.org/10.1057/ejis.2011.51

Malpass, D., (2020, March). Coronavirus highlights the need to strengthen health systems. Worldbank Blogs. https://blogs.worldbank.org/voices/coronavirus-covid19-highlights-need-strengthen-health-systems

Wu, I.-L., & Hu, Y.-P. (2012). Examining knowledge management enabled performance for hospital professionals: a dynamic capability view and the mediating role of process capability. Journal of the Association for Information System , 13 (12), 976–999. https://doi.org/10.17705/1jais.00319

Yang, A., & Varshney, U. (2016). A taxonomy for mobile health implementation and evaluation. Americas Conference on Information Systems (AMCIS),  1–10.

Zafar, H., & Sneha, S. (2012). Ubiquitous healthcare information system: toward crossing the security chasm. Communication of the Association for Information Systems , 31 (9), 193–206. https://doi.org/10.17705/1CAIS.03109

Zhang, L., & Pang, M. S. (2019). Does sharing make my data more insecure? An empirical study on health information exchange and data breaches. International Conference on Information Systems (ICIS),  1–14.

Download references

Open Access funding enabled and organized by Projekt DEAL.

Author information

Authors and affiliations.

Chair for Digitalization and Process Management, Philipps-University Marburg, Universitätsstraße 24, 35037, Marburg, Germany

Nadine Ostern

Frankfurt School of Finance & Management, ProcessLab, Adickesallee 32-34, 60322, Frankfurt am Main, Germany

Guido Perscheid, Caroline Reelitz & Jürgen Moormann

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Nadine Ostern .

Additional information

Responsible Editor: Shengnan Han

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The original online version of this article was revised: In this article the affiliation of authors are incorrect. Given here are the correct affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 195 KB)

Rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Ostern, N., Perscheid, G., Reelitz, C. et al. Keeping pace with the healthcare transformation: a literature review and research agenda for a new decade of health information systems research. Electron Markets 31 , 901–921 (2021). https://doi.org/10.1007/s12525-021-00484-1

Download citation

Received : 11 December 2020

Accepted : 26 May 2021

Published : 17 July 2021

Issue Date : December 2021

DOI : https://doi.org/10.1007/s12525-021-00484-1

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Health information systems research

JEL Classifications

• Find a journal
• Publish with us
• Track your research

Volume 20 Supplement 2

The Physician and Professionalism Today: Challenges to and strategies for ethical professional medical practice

• Open access
• Published: 09 December 2020

Objectives, methods, and results in critical health systems and policy research: evaluating the healthcare market

• Jean-Pierre Unger 1 ,
• Ingrid Morales 2 &
• Pierre De Paepe 1  

BMC Health Services Research volume  20 , Article number:  1072 ( 2020 ) Cite this article

7823 Accesses

3 Citations

Metrics details

Since the 1980s, markets have turned increasingly to intangible goods – healthcare, education, the arts, and justice. Over 40 years, the authors investigated healthcare commoditisation to produce policy knowledge relevant to patients, physicians, health professionals, and taxpayers. This paper revisits their objectives, methods, and results to enlighten healthcare policy design and research.

This paper meta-analyses the authors’ research that evaluated the markets impact on healthcare and professional culture and investigated how they influenced patients’ timely access to quality care and physicians’ working conditions. Based on these findings, they explored the political economic of healthcare.

In low-income countries the analysed research showed that, through loans and cooperation, multilateral agencies restricted the function of public services to disease control, with subsequent catastrophic reductions in access to care, health de-medicalisation, increased avoidable mortality, and failure to attain the narrow MDGs in Africa.

The pro-market reforms enacted in middle-income countries entailed the purchaser-provider split, privatisation of healthcare pre-financing, and government contracting of health finance management to private insurance companies. To establish the materiality of a cause-and-effect relationship, the authors compared the efficiency of Latin American national health systems according to whether or not they were pro-market and complied with international policy standards.

While pro-market health economists acknowledge that no market can offer equitable access to healthcare without effective regulation and control, the authors showed that both regulation and control were severely constrained in Asia by governance and medical secrecy issues.

In high-income countries they questioned the interest for population health of healthcare insurance companies, whilst comparing access to care and health expenditures in the European Union vs. the U.S., the Netherlands, and Switzerland. They demonstrated that commoditising healthcare increases mortality and suffering amenable to care considerably and carries professional, cultural, and ethical risks for doctors and health professionals. Pro-market policies systems cause health systems inefficiency, inequity in access to care and strain professionals’ ethics.

Policy research methodologies benefit from being inductive, as health services and systems evaluations, and population health studies are prerequisites to challenge official discourse and to explore the historical, economic, sociocultural, and political determinants of public policies.

Since the 1980s, markets have turned increasingly to intangible goods – health care, education, arts, and justice. Political changes have accompanied the transformation of health systems. After World War II, the WHO was founded to counter the health effects of devastating destruction, but over the last decades its funding by the World Health Assembly dropped to 25% of its budget. Foundations and industrial countries funded the rest, that is, their preferred programmes. The 1978 Alma Ata Declaration establishing the Primary Health Care policy had resulted from able WHO leadership and a growing social movement demanding health for all. One year later, the Selective Primary Health Care movement promoted by the Rockefeller Foundation undermined its foundations. It led the international policy exclusively to support disease control programmes in LMICs and to turn their first-line health services into epidemiological units allegedly because comprehensive primary health care was costly.

After the collapse of the “socialist” camp in 1989, the Washington Consensus, WB, and IMF conditioned low-interest loans on moves to market economy and government withdrawal from health care provision and financing. Since the 2000s, governments in industrialized countries and their private sector set up international disease control programmes called Global Health Initiatives. These were actually epidemiological public-private partnerships that replaced international cooperation in the health sector.

With the Millennium Development Goals (MDGs) and subsequent Sustainable Development Goals (SDGs), the United Nations set quite unambitious global health goals. They assigned donor-driven targets to LMIC governments, that is, controlling a limited number of pathologies, first transmissible and then increasingly chronic ones.

Over this period the authors evaluated pro-market reforms and policies and identified their determinants through the lenses of patients, physicians and health professionals, and taxpayers. Patients are concerned about accessibility to healthcare services and the price and quality of care. Physicians’ interests are, or should also be, their problem-solving capacity, professional freedom, intellectual progress, medical ethics, work environment, and income. These were the authors’ yardsticks to assess health systems and conduct policy research. These studies thus covered curative medicine, preventive medical care, and medical education but not the important field of inter-sectoral public health policies.

It all started in 1982, when J.-P. Unger discovered in Boston the Rockefeller Foundation’s long-term strategy to commoditise healthcare financing worldwide. In investigating the health marketisation motives of the “Selective Primary Healthcare” strategy in low- and middle-income countries (LMICs), he interviewed J. Walsh and K. Warren, the authors of a publication released just 1 year after the Alma Ata conference that advocated an alternative to the Primary Healthcare strategy called “selective primary healthcare” [ 1 ]. Their message was that the Primary Healthcare Strategy endorsed by the World Health Assembly in Alma Ata in 1978 was unaffordable. Instead, the Rockefeller Foundation promoted a policy that would turn low-income countries’ (LICs) public health systems into structures fit to host disease control programmes – “like Christmas ornaments festooning a Christmas tree” – rather than delivering individual health care. A field experiment in Deschapelles, Haiti [ 2 ], was a central piece of evidence supporting this strategy to make LIC health centres in public services mere disease control structures. The scenario pushed by the Rockefeller Foundation eventually came to be in LICs in the 1980s, albeit with major variants.

In 1986, we invalidated the efficiency alibi of this strategy. As an answer to the Rockefeller strategy, an action research project covering 180,000 persons in Kasongo, Congo [ 3 ] (then Zaire), enabled us to show that the cost of delivering individual health care and a few disease control and other public health interventions under a single administration could be similar to those of first-line services providing just five disease control programmes, because the former solution made it possible to keep its administration simple [ 4 ]. That prompted us to study the economic motives and public health consequences of healthcare insurance commercialisation, healthcare commoditisation, and health service privatisation and to build a case with coordinated studies. This paper meta-analyses the objectives, methods, and results of evaluations and research into market based health systems and policies spanning over 35 years ( https://pure.itg.be/en/persons/jeanpierre-unger(92d91a56-f267-4b85-82e7-9e4f8a8cffed).html ). Specifically, it aims to make sense of an array of policy studies that all relied on the same medical and public health ethical criteria already formulated in 1972 [ 5 ]; and to delineate health policy research standards relevant to physicians, health professionals, and patients’ representatives committed to the human right to health, i.e., the right to access professional care in universal health systems [ 6 ].

Research strategy

On the grounds of the Kasongo experience and aforementioned Walsh and Warren interview, we formulated in 1983 the overarching hypothesis of our decades-long policy research: Pro-market reforms of healthcare financing and management expand the healthcare delivery and disease control market to the detriment of patients, populations, doctors, health professionals, and taxpayers.

To confirm or overturn this hypothesis, we tested four secondary hypotheses and tried to show a causal relationship between pro-market policies’ characteristics and the following phenomena:

Regarding the access of patients and persons with health risks to professionally delivered healthcare, we tried to verify whether the market tended to allocate individual, “discretional” health care to the rich and public health interventions to the poor, thereby reducing the general population’s access to care significantly.

Regarding disease control, we checked whether public health programmes often failed because the market assigned them a vertical structure to be better suited to absorbing medical equipment and pharmaceuticals with public financing.

Regarding fiscal justice, we strove to determine whether health markets ran counter to social justice in health as they precluded the efficient and equitable use of taxes in the care sector.

About professionals’ ethics and personal development, we aimed to verify if care commoditisation was compatible with the physician’s reliance on professional ethics and investments in medical equipment and pharmaceuticals might antagonise the conditions of doctors’ and teams professional development.

This paper meta-analyses the authors’ research evaluating the impact of markets on health care and professional culture and investigating how they influenced patients’ timely access to quality care and physicians’ working conditions. Based on these findings, they explored the political economy of health care. However, there was no early design of a long-term research strategy. They conducted the studies according to opportunities, although some principles were adopted from the start:

Interdisciplinarity

Testing the above hypotheses required ad hoc, interdisciplinary research methods in order to build a good case for a causal relationship.

Heterogeneous research setting

The hypotheses had to be tested in a large array of health systems, from low- to high-income countries. To allow generalisations about the healthcare environment, countries and regions would be key policy analysis units.

Inductive reasoning

Historical studies would be based on public health evaluation of healthcare systems. Interpreting policy decisions critically required previous ex-post demonstration of ill-functioning services.

The authors approached qualitative research in medical care and public health policies by making use of the concept of praxeology that Bourdieu developed and adapted to sociology in his “Outline of a Theory of Practice.” [ 7 ] They took this approach because both medicine and public health, like sociology research, are combinations of practice and theory [ 8 ]. They believed that the failure to connect them was a frequent weakness of contemporary medical and public health research. An important aspect of praxeology is inductive reasoning. It builds on and evaluates propositions that are abstractions of observations of individual instances of members of the same class. In this regard, the policy evaluations were problem-based and relied on paradoxical observations of care delivery and health service management. They were the raw material of the research and prerequisites for assessing health systems and policies and then exploring the social, political, and economic determinants of faulty ones. Figure 1 depicts the inductive chain generally used in these policy analyses.

Sequencing the authors’ research on (inter-)national health policies

Deconstruction of the policy discourse

Deconstruction is a form of critical analysis concerned with the relationship between text and meaning. Jacques Derrida’s 1967 work on grammatology introduced the majority of its influential concepts. The authors set out to deconstruct public policies with qualitative, interpretative research and nested probabilistic studies. Their goal in this respect was to verify the evidence sustaining pro-market reforms in LMIC and high-income country (HIC) settings; based on these findings, expose their practical, political economy rationale; and then tentatively deconstruct the pro-market discourse of multilateral agencies and commercial organisations. Case studies of national healthcare policies and disease control programmes would provide the material required to analyse international policies and national health sector reforms [ 9 ].

Explicit research values

The authors made explicit their ethical values of social justice and medical professionalism because research methodology, policy evaluation, and interpretation depend on social, economic, and professional standards. These values, published elsewhere, were conceived of for healthcare delivery, management, planning, financing, and disease control. In particular, the authors relied on three healthcare standards with policy implications formulated in 1971, namely, holistic (biopsychosocial and patient-centred), continuous, and integrated care [ 5 ]. In Belgium, they served as an ideology to cement alliances of professionals concerned about quality and equitable access to care for more than 40 years [ 10 , 11 ]. The authors also relied on another key standard of medical practice, the Hippocratic “self-effacement” tenet (“Into whatsoever houses I enter, I will enter to help the sick, and I will abstain from all intentional wrong-doing and harm”) that is expected to deter physicians from making self-interested clinical decisions and maximising their profits with ad hoc clinical decisions, i.e., practising commercial medicine.

Evaluating disease control programmes, the hub of international and national health policies in LICs

By 2015, Africa still had not attained the modest MDGs in health. In 2007, we reviewed the grey literature issued by the main multilateral agencies active in the LIC health sector. Under the aegis of the MDGs, disease control was the conceptual and operational hub of health system reform in LICs. Our review revealed that over the preceding 25 years, virtually all the multilateral agencies active in the health sector had adopted policies restricting the function of LICs’ public services to disease control, thereby allocating individual healthcare delivery to commercial services (and private, non-profit facilities where they existed) [ 12 ].

To convince physicians and policy makers in LICs to adhere to sectoral reforms and to replace individual care delivery by disease control in public services, the Bretton Wood agencies attached conditions to their loans and projects and financed a host of local experts to produce the “scientific” justifications of this policy.

For LIC populations, the avoidable mortality, suffering, and anxiety that followed the loss of access to individual care proved immense. In Africa, virtually none of the MDGs were attained, regardless of their limited scope, precisely because in failing to deliver individual healthcare, African public services could no longer implement disease control initiatives satisfactorily.

To explain why a huge financial effort (AIDS control funds, for instance, were multiplied twentyfold between 1997 and 2007) could not achieve the MDGs in Africa, the authors

showed mathematically that successful disease control programmes required health facilities to be used by patients with various symptoms, as they represented the pool of users that these programmes needed for early case detection and follow-up [ 15 ].

studied the mechanisms whereby integrated disease control interventions hampered access to care in the services in which they were integrated and so undermined public services. Although a few AIDS and under-five programmes had been known to deliver bio-psychosocial care, disease control programmes in Africa have reduced the problem-solving capacities of health services; shrunk the professional identity and skills of physicians and nurses; reduced access to drugs to those managed by Global Health Initiatives; and limited in-service training to collective care delivery [ 16 ].

showed this to be a “catch 22” situation, with disease control programmes drastically reducing the number of users in the (public service) facilities where such programmes were implemented [ 17 ].

analysed the evidence of pro-market policies for other characteristics, such as equitable access to quality health care; mismatch of commercial healthcare delivery with medical ethics [ 18 ]; the inability of public services focusing on disease control to respond to people’s demands for individual care, thus preventing community participation; and undue restrictions on professional autonomy in health services designed as “machine bureaucracies.” [ 19 ]

The authors concluded that Hypothesis 2 was plausible because of the following:

Disease control-based reforms strained access to care in LICs without achieving their alleged epidemiological goals.

Replacing individual health care by disease control interventions in LIC public services could be the real motive of the related (inter-)national policies. This was because these reforms practically, albeit tacitly, ushered in a situation in which competition between public and private providers in delivering individual care was made impossible. Multinationals linked to charities that were focusing LMIC public services on disease control took advantage of the disappearance of publicly delivered health care to sell medical care to LMIC middle and upper classes without having to face public sector competition. International disease control programmes not only permitted the use of cooperation funds to purchase drugs and medical equipment manufactured by HIC industries, so fomenting aid-dependent pharmaceutical markets in LICs, but were also structured to foster the healthcare market in urban settings.

How do health-financing markets perform in middle-income countries? Comparing Latin American national healthcare policies and evaluating healthcare regulation in Asia

In MICs, pro-market health system reforms focused on national health care financing. Starting in Chile in the 1980s (under a military government) and in Colombia in 1993 (under an authoritarian government), the privatisation of health financing in Latin America occurred in virtually every country, even those with “socialist” governments. The two exceptions that did not undergo market reforms, Costa Rica [ 20 ] and Cuba [ 21 ], were performance outliers. However, the reform scenarios and organisation of health systems were not identical across the continent. Schematically, Insurance companies made profits whilst managing government funds, capturing the health expenditures of the healthy and wealthy middle class, and employing or contracting physicians. The political economics of health sector reforms in MICs consisted of variable combinations of

under-financing public services;

unduly favouring investments in public services over their recurrent operating costs;

putting the physicians working for publicly-oriented institutions under economic and workload stress;

separating purchasers and providers by law so as to create a niche for commercial insurance banks;

allowing commercial entities to manage public funds and possibly making this scheme mandatory;

privatising public hospitals or imposing commercial competition rules on them (the so-called “management property split”) and on contracted, self-employed physicians, too;

stimulating private financing of public hospitals (“private finance initiatives”);

limiting public services’ activities to unprofitable care, e.g., for the poor (Medicaid) and the elderly (Medicare) in HICs, and to disease control programmes in LMICs; and

liberalising investments in health care under the aegis of international trade treaties.

Given the many cultural and political similarities across Latin American countries, their health systems offered a good setting to explore strategic variants of care commoditisation. The authors assessed primarily the effects of pro-market reforms in Latin America by comparing the performance of systems abiding by international (World Bank, International Monetary Fund, Inter-American Development Bank, etc.) health policies with those that did not [ 9 ]. They thus studied the histories and functioning of some national health systems and the impact of financing options on their management, care quality, and access to care. To study health systems’ productivity, they relied on aggregated production data, population-based care accessibility and continuity rates and ratios; direct observations in healthcare services and administration; and interviews of patients, physicians, health professionals, policy makers, and public health experts.

They studied the health care and outcomes of large-scale, nationwide, in vivo experiments of care commoditisation. The ones they studied did not show any benefits for patients, physicians, health professionals, and/or public finances:

Colombia, which had been a good student, by international standards, since 1993, had a deplorable health record [ 22 ]. In our interviews we studied and compared the barriers to access to care erected in Colombia by a managed competition model with the barriers in north-eastern Brazil, where public services were severely under-financed [ 23 , 24 ]. As expected, both had very poor results.

In 2006, Chile’s public services [ 25 ], which had survived the dictatorship, covered 84% of the population with half of the country’s health expenditure. However, with just 50% of the country’s health expenditures, the public services managed to make the country a positive outlier in Latin America on many health indicators. The technical challenge of this study was to relate health system features to indicators of output (utilisation and coverage rates, for instance) and outcome (maternal mortality, for instance).

Finally, in 2001, Costa Rica, with its publicly-oriented healthcare services and financing, had about the same demographic and epidemiological features as the United States, although it spent nine times less per capita on health than the U.S. [ 20 ]

To fuel the legal and institutional dynamics of health insurance privatisation, the WHO and other UN agencies promoted a strategy called “Universal Health Coverage” (UHC) [ 26 ], that is, universal access to health insurance. Its pro-market discourse endorsed the idea that only insured populations could access health care [ 27 ], despite evidence that expanding insurance coverage might reduce service utilisation, e.g., when public-private insurance mixes were supposed to achieve universal coverage of health risks [ 28 , 29 , 30 ] and evidence of the superior effectiveness, fairness, and efficiency of Latin American off-market health systems [ 20 , 21 ].

The findings of these international comparisons led the authors to question the UHC strategy as a way to secure universal access to care. This was not only because public-private mixes in healthcare financing give rise to severe inefficiency in health systems, but because access to care was shown to be highly dependent on non-financial factors (geographical and psychological accessibility of health services, for instance) [ 31 ] otherwise neglected by the UHC strategy and possibly even undermined by it. In the absence of performance-based evidence supporting health-financing marketisation, the hypothesised centrality of an economic agenda in Latin American health reforms became plausible.

In sum, these comparisons of the Chilean, Colombian, Costa Rican, and Brazilian health systems and historical studies of Bolivia and Ecuador support Hypothesis 1 regarding the negative impact of pro-market policies on access to care and quality of care and Hypothesis 3 regarding fiscal injustice and inefficient use of public funds by commercial health services and insurance companies.

In addition, the authors’ studies of Asian health systems showed that the health care market was structurally flawed by the impossibility of regulating and controlling the activities of the private but also public health care sector in MICs properly. Whilst the Rockefeller Foundation had tacitly admitted that without regulation and control, privatising health services could not produce equitable access to care [ 32 ], the authors showed through their observations in nine (maternal health) case studies of regulations in China, India, and Vietnam [ 33 ] and theoretical discussion [ 34 ] that regulation and control of for-profit care delivery were most likely to be ineffectual in the MIC care sector.

In Vietnam, for instance, sex-selective abortion was responsible for a serious gender imbalance in spite of a decade of State regulation and control. Although a regulation against the practice had been passed in 2003 and implemented since 2006, regional disparities in gender-specific birth rates increased between 2006 and 2011. As a “critical incident”, the number of ultrasound violations detected in 2011 had been 1 positive out of 83,192 controls done in the health districts under study. And in 2016, the gender ratio still was 112 females/100 males in Vietnam. Against a background of strong social demand for sex-selective abortion in the middle class, selective abortions remained undetected in spite of the regulation and inspections because of the policy-makers’ failure to allocate sufficient resources to this exercise, weak governance, medical secrecy, conflicts of interests, dual physicians’ employment (in public and private healthcare services), the opacity of the medical market, and difficulties specifying contingency in clinical situations [ 33 ].

This set of nine studies in China, India, and Vietnam thus supports the plausibility of Hypothesis 4, as they confirm the vulnerability of medical ethics to care commoditisation policies when regulation and control of medical practice are ineffectual, which actually they are because of the socio-political and technical characteristics of middle income countries.

Assessing the impact of health markets on access to care in Europe

At the end of World War II, unionised blue-collar workers imposed social protection schemes in health. In a bipolar world, the workers’ organisations took advantage of progressive ideologies that were gathering strong followings in Europe. Whilst the weakened employers’ organisations prevailed upon the Social Democrats and Social Christians to join them in the anti-Communist struggle, they conceded the pillarization of European States. Workers’ trade unions, mutual societies, and political parties entered the parliaments (as was the case before Word War II), but also the State’s executive branches, judiciary, and social and health services, education, the police, and the army. That is how workers’ representatives limited the impact of corruption in State constituencies, i.e., preventing those who had the will and resources from buying the State’s policy and administrative decisions. They locked the sustainability of social security into government structures and secured access to professional health care in universal health systems as a human right. Admittedly, the users of healthcare services paid for this State pillarization with a dose of nepotism and its consequences. Still, European States had acquired key democratic features. Heated negotiations between representatives of social classes with opposing interests produced sectoral priorities within the overarching framework of national health budgets. In Belgium, for instance, this debate was institutionalised in the national social security organisation. Footnote 1

The macroeconomic result of this pillarization of the State can be seen in two inversely proportional numbers that show the importance of risk-pooling and solidarity in European health care financing, namely:

a government share in total health spending that long exceeded 80% and

total health expenditures that were high enough (about US$4000 per capita in 2014, of which approximately 10% was for the commercial sector) to make the publicly-oriented healthcare services Footnote 2 effective but sufficiently modest (10% of European GDP versus 17.1% of U.S. GDP in 2014) to favour economic growth outside the health sector.

That is how employees’ and employers’ taxes and social contributions made it possible to limit household expenditures on health care whilst securing one of the best geographically, financially, psychologically, and technically accessible forms of professional health care. Importantly, these schemes gave physicians sufficient professional autonomy. Access to professional care was equitable thanks to cost-redistributing, non-profit, non-actuarial health care financing and a sufficiently large proportion of non-material investments in the health sector.

With government social security schemes that included fairly comprehensive universal health insurance, Europeans enjoyed a high degree of social protection from 1945 to 1989 in Eastern Europe, until the 2008 financial crisis in Southern Europe, and even later in other countries.

Unfortunately, the institutional pillarization did not prevail at the European Union and Commission level. Rather, European politicians, civil servants, and political parties were the targets of more than 30,000 commercial lobbyists (1.4 per European Commission (EC) civil servant) [ 35 ] working to foster the interests of the international insurance banks that were investing in health, amongst other things. In contradiction to the provisions of the Treaty of Rome [ 36 ], the EC intervened in the Member States’ health care systems by negotiating international trade treaties involving investments in health care that could make healthcare management and medical practice subject to a commercial rationale. In addition, the 3% budget deficit rule that the Maastricht Treaty imposed on Member States gave political parties an opportunity and a plausible reason to cut public expenditures on health until healthcare financing would be sufficiently privatised, as the WB and IMF had done earlier in Latin America.

Public expenditure on health care was severely constrained but once health laws and regulations had been modified, as shown by the history of Dutch, Swiss, and Colombian health systems, insurance banks strove to maximise public and private expenditures on health care and governments found the needed resources through inter-branch arbitration.

In the U.S., where the health market was mature, the wealthy faced more problems accessing health care than the poor in most OECD countries, whilst the U.S. government alone spent more on health per capita than the total (public and private) per capita spending of most European countries [ 37 ]. Nevertheless, over the last 10 years, the number of uninsured Americans varied between 35 and 50 million. Many more were poorly insured. If the U.S. insurance coverage rate were applied to Europe, the number of uninsured Europeans would reach about 75 million. If the European ratio of mortality amenable to care became that of the United States, avoidable mortality would increase by up to 100,000 deaths per year.

In Latin American countries, the same financial structure yielded the same health effects as in the U.S. but, admittedly, not in the Netherlands and Switzerland. The sustainable performance of these two health systems is central to policy debates in Europe and, expectedly, insurance banks praise their functioning, except for one small detail: Since health care financing has been marketed (respectively in 1996 and 2006), the Swiss and Dutch health expenditures have skyrocketed [ 38 ].

What are the reasons to believe that health insurance markets are environments hostile to the universal right to care? The authors evaluated [ 39 ] the performances of the U.S., the Netherlands, and Switzerland, three industrial nations that pursued market-based financing models, with a focus on equity in access to care, care quality, health status, and efficiency. They then assessed the consistency of their findings with those of various research teams. Using secondary data obtained from a semi-structured review of articles from 2000 to 2017, inter alia, they discussed the hypothesis that commercial health care insurance was detrimental to access to professional health care and population health status.

The findings can be summarised as follows:

In 2010, poor Americans had twice the unmet care needs of Americans with above-average incomes and ten times more than the UK poor. The unmet care needs of the rich in the U.S. exceeded those of the poor in several industrial countries [ 40 ]. The number of Dutchmen and -women experiencing financial obstacles to health care quadrupled between 2007 and 2013 [ 41 ]. Switzerland ranked second worst in a 2016 survey of 11 countries, just ahead of the USA, with 22% of Swiss adults likely to skip needed care [ 42 ].

The most negative impacts of “managed care” on care quality were its tight constraints on physicians’ professional autonomy, large reliance on the physicians’ material motivation, the fragmentation of health services, and a tendency to apply evidence-based medicine too rigidly. In requiring strict application of clinical protocols, commercially managed care was less likely to be favourable to care quality than systems giving physicians sufficient freedom to rely on professional decision-making and medical ethics.

The prevalence of burnout amongst MDs made medical practice the riskiest occupation in the United States and one of the riskiest occupations in Europe [ 43 ]. This burnout was not related to insufficient income but to excessive workloads and to perceiving existential threats to their professional identity, ethics, and autonomy in the way health care was organised. This observation supports Hypothesis 4 because these psychological and professional status threats actually result from the commoditisation of care [ 44 ].

Countries with a commercial insurance monopoly generally remained above the maternal, infant, and neonatal mortality rates v. the health-spending regression line [ 45 ]. And the growth rates of health expenditure were the highest in the U.S. and Switzerland, with the Netherlands not far behind [ 46 ].

Controlling for the impact of the obesity confounding factor, these studies reveal that the industrialisation of care contributes to the comparatively poor performance of the U.S., Dutch, and Swiss health systems, with the Dutch first-line services being an exception made possible by the GPs’ medical culture and the low cost to patient.

International trade treaties may further worsen the mortality rates of cardiovascular and cerebrovascular conditions, diabetes, and cancers in Europe, since they favour the food industry’s market penetration [ 47 ].

These findings admittedly conflict with recent influential health system rankings, perhaps because of the ways their health indicators are constructed and a bias towards assessing first-line healthcare services.

In conclusion, the comparison of US, Dutch, and Swiss health systems with the others in Europe supports the validity of Hypotheses 1 and 3. The most inefficient system is where the insurance market has achieved its maximal development, that is, in the U.S. In general, healthcare expenditures rose faster where health insurance was commoditised. The Netherlands and Switzerland reveal that increasing expenditure on health care enables health systems based on commercial insurance to maintain decent access to professionally-delivered health care for a few years.

The sizeably better, much more equitable access to health care in Western Europe (and its demographic and epidemiological superiority over the U.S.) and its much lower cost is generally explained by redistributive laws and regulations (tax-based or mandatory social security) channelled through health care public services or mutual societies that permit solidarity in health care financing.

The analysis of the U.S. health system’s disappointing performance reveals that actuarial management of health finances and the commercial management of health services are responsible for deficient accessibility to care and services. In particular, actuarial management of health care reduces risk pooling and solidarity in health financing between men and women, the young and the elderly, the sick and the well, high and low risks, and rich and poor.

Methodological lessons for descriptive, policy studies

Identifying health services productivity shortfalls and dysfunctional structures

The authors tried to provide patients’ and physicians’ organisations with the evidence and clues about policies from the angle of the human right to care and professional endeavour. Their research assessed the influence of policies on health services’ productivity in defined historical contexts from various standpoints: those of patients (e.g., care quality and accessibility); physicians (e.g., continuing medical education and teamwork); taxpayers (efficiency and equity in use of public monies); and public health specialists (health care and disease control management).

From an inductive study perspective, documenting health services’ structural and functional deficiencies provided the raw material for assessing health systems and possibly challenging policy decisions and official discourse.

To gauge the quality of health care, the authors used medical knowledge to observe clinical practice (sometimes as mock patients) [ 48 ]. For instance, to assess the impact of managed care techniques on care quality in Costa Rica, they sat in on consultations. The research hypotheses had been formulated by the Limon region’s GPs, who suggested that there was a relationship between managed care ( compromisos de gestión ) and the lack of time available for interpersonal communication and deficient care accessibility [ 49 ]. In addition, they collected data on disease-specific indicators to explore the extent to which managed care techniques were responsible for decreasing care quality and data reliability.

To assess care accessibility, they often used the services’ routine production data, with indicators such as population-based utilisation rates of curative care in first-line services and hospital admission rates [ 31 ], referral completion rates, and preventive (vaccination, antenatal clinics, etc.) coverage rates, and then they validated them by triangulation when possible. As a proxy for the financial accessibility of health care, they used “catastrophic health expenditures.” [ 50 ] Routine data proved cheaper, readily available, and a good reflection of the services’ operations in large geographical areas, but the method had limits even when it was combined with data triangulation and controls:

In Colombia, semi-structured interviews of patients and professionals proved indispensable to gauge care accessibility [ 51 , 52 , 53 ] because networks of “sentinel physicians” were not organised to collect service utilisation and epidemiological data; population-based statistics were not available and the denominators would have consisted of populations affiliated with a myriad of health insurers and care providers; and private insurance companies were reluctant to provide data that could undermine their reputations.

The routine data were sometimes biased, such as in the case of a state administration in charge of determining regional maternal mortality rates in Asia. Aside from the technical difficulties of establishing the maternal mortality rate (MMR), middle line managers were likely to be penalised when this indicator was too high but also too low, because in the latter case the administration did not trust the data’s validity [ 33 ]. Hence a regression to the mean …

In general, the authors relied on output indicators rather than on population outcome. However, two demographic indicators proved particularly interesting for critical assessment of healthcare systems:

The Maternal Mortality Rate (MMR) reflects access to the entire healthcare system pyramid [ 54 ], particularly in LMICs [ 55 ] and probably in any situation where it exceeds 40 per thousand. This is in contrast to the Infant Mortality Rate (IMR), which in LICs often mirrors low-cost interventions that may reduce access to care (such as immunisation campaigns) [ 56 ] and biomedical/sociocultural health determinants (such as the availability of food and clean water and women’s education, respectively). Since the lower the per capita GDP, the cheaper and less reliable the demographic indicators used [ 57 ], the authors retained in practice only the gross differences when comparing the health systems’ performances in terms of MMR. In 2010, for instance, Moldova, the poorest country in Europe, had the same MMR as the U.S., despite spending 1/20 as much on health per capita.

In HICs, life expectancy and population mortality rates mirror obesity-associated pathologies but, just as importantly, access to quality health care. Up to 80% of premature deaths in Poland were explained by unsatisfactory access to health care [ 58 ]. According to Kruk and co-workers, 15.6 million excess deaths from 61 conditions occurred in LMIC in 2016. This research compared case fatality between each LMIC with corresponding numbers from 23 high-income reference countries with strong health systems. After excluding deaths that were preventable by public health measures, the authors found that 55% of excess deaths were amenable to health care and could be put down to either the receipt of poor-quality care or the non-utilisation of health care [ 59 ].

To evaluate health systems by the design and performance of their disease control programmes, the authors relied on two models:

An all-purpose disease control model (“ vertical analysis” ), designed by P. Mercenier [ 60 ] to provide standards for the design of disease-specific control programmes. It was based on the systemic representation of the disease-specific syndromes and vector development stages and biomedical and socio-cultural interventions to interrupt the disease chain in the field, from aetiology to patient death.

M. Piot’s model [ 61 ] to assess care continuity for any defined disease. It establishes the disease-specific cure rate as the product of coefficients measuring detection, diagnosis, and treatment activities. As the model reveals the health system characteristics needed to secure, say, early detection and care continuity, they used it to contrast the performances of public and private sectors in tuberculosis control in India [ 62 ] and to evaluate malaria control programmes in Mali and Sub-Saharan Africa in general [ 15 ].

Once health system productivity had been studied, the authors analysed the organisation of health services and systems. For this they relied on managerial models and standards specific to

publicly-minded care management (e.g., concerned with access to professional health care, professional autonomy and well-being, professional ethics, and public health) [ 63 ];

the systemic management of hospital(s) and first-line facilities networks [ 18 ]; and

“divisionalised adhocracy”, an organisation pattern that favours knowledge management and teamwork [ 19 ] and is suited to systems whose end-line producers are highly skilled and sufficiently autonomous professionals (as are physicians) rather than workers and technicians, as assumed by the classic generic management theories.

Studies of health financing and systems characteristics that cause low services productivity

Health system case studies and the existence of large databases in the health sector provided the opportunity to single out natural, quasi-experimental study designs – time series and non-equivalent comparisons – to contrast health systems with and without or before and after pro-market health reforms:

For non-equivalent groups (countries, regions, etc.), the authors compared the performances of national/regional health systems in Latin America compliant with the international policy standards with those of “disobedient” ones [ 9 ] and established a typology of reforms.

With time series, we showed long-standing, substandard performances in the quality, accessibility, and financing of health care (for instance, after the privatisation of health insurance in Colombia).

Time series of health services’ routine data also proved useful to reveal contradictory interactions of health activities in populations. For instance, in the late 1980s, the utilisation of medical consultations decreased steadily in Senegal whilst immunisation campaigns were implemented in health care services [ 64 ]. The challenge of the study consisted in demonstrating a causal relationship between these campaigns and the subsequent sustained deterioration of care accessibility in public services.

Beyond substandard care performances: political economics

Inductive research made it possible to deconstruct official self-apologetic discourses. The authors were then able to seek the real motives for ill-conceived policies whose results belied the stated objectives. Their entry point in the complex socio-cultural and political determinants of health policies was political economics because of the huge weight of health expenditures in the global economy (up to 17% of U.S. GDP and 11.3% of Germany’s GDP) and the political leverage acquired by the economic players. The economic determinism of health care policies was so powerful that these players did not even need to be coordinated to gear health systems towards care markets [ 65 ].

From corruption [ 66 ], political leverage, and lobbying to trade, it takes time for relationships between commercial organisations and public institutions to result in health systems’ structures and new professional practice. Some studies thus adopted an historical viewpoint [ 12 , 65 , 67 , 68 ] to probe the care commoditisation mechanisms. Even in non-profit organisations, the main determinant of poor healthcare accessibility proved to be the business mission of health financing, management, and medical practice.

However, correlations between events, sequences, sociological observations, and relationships between historical times enabled us to identify professional, cultural, and geostrategic determinants of health policies alongside economic ones. The prevailing order was reflected in professional culture thanks to education, information, scientific ideology, and advertising. The resulting personal characteristics, identity, and knowledge of physicians and professionals were the conditions of health systems’ reproducibility. Bourdieu calls these internal features “habitus,” i.e., ways of doing and being, and “representations”.

Since 1985, the trend has been towards the privatisation of health financing, public subsidies for private health care providers, commercial management of health services, and for-profit medical practice, in spite of the wealth of evidence pointing to the risks of large-scale mortality and morbidity and threats to professional ethics associated with the commoditisation of care.

Governments and multilateral agencies ought to be held accountable when health policies cause avoidable mortality and suffering and thus human rights violations, or at least “be shamed”, as Sir Michael Marmot once said. Therefore, with States being fields “structured according to oppositions linked to specific forms of capital” [ 69 ], health system and policy research should not so much address the knowledge needs of policy makers directly as those of physicians, socially-minded professionals, and patients’ organisations that could leverage them. Political indictments on the impact of health policies require these organisations to access the relevant scientific and professional information in order to question and challenge public policies in the health sector.

The studies analysed here stemmed from the human right to access professional care in universal health systems and the knowledge they produced was directed at physicians, health professionals, and patients’ organisations sharing moral values and interested in lobbying health policies. The present meta-analysis sheds light on the requirements of this type of research:

Inductive, multidisciplinary policy research is time-consuming but often a condition to study health policies independently:

International health policies assessments benefit from analysing national healthcare policies and disease control programmes.

National health policies should be studied with political economy and medical concepts, and through the lenses of political science and history, but importantly on the grounds of health systems and services productivity assessment.

Medical concepts, public health models, and indicators of professional care delivery and non-profit health management make it possible to evaluate health systems from a professionally- and socially-driven, problem-based perspective.

Health systems and policy researchers need scientific and professional knowledge. Academics should engage in medical, managerial, and policy-making work alongside their research and teaching activities. Therefore, medical and public health schools should learn to assess the academic’s professional proficiency and ability to derive validated theory from their practice.

Professional ethics should be a criterion for evaluating care quality:

Although values are an obstacle to Weberian axiological neutrality in medical, public health, and education policy studies, they are indispensable to assess care quality, health services, and healthcare systems. From a phenomenological perspective, they ought to be made available to the reader.

Health systems have evolved rapidly over the last three decades. Long-term reliance on the same set of explicit ethical and technical criteria applicable to medical practice and health services organisation is what allows valid conclusions to be drawn from time series and comparative or historical studies of health systems that belong to different eras.

Such studies ought not to be only descriptive and critical but also designed as proposals to improve health systems and policies. Those analysed here reveal many nationwide experiences to improve access to professional care. Some countries (Costa Rica, Cuba, Spain, Sri Lanka, Thailand, and Italy), states (e.g. , Kerala), regions or cities (e.g. Rosario, Argentina), and health systems (Chilean public services) acquired collective knowledge to develop non-commercial care delivery and promote ethical, medical practice. There is no doubt that decades of neoliberal policy have compromised their professional achievements, to the point that they are often no longer perceptible.

Medical journals ought to be devoted to professional practice and not only to science, and be independent and publicly financed. Given the undeclared conflict of interest created by the presence of insurance banks in the shareholding of top impact-factor medical journals, physicians’ and patients’ organisations should lobby public universities to stop relying on the researcher’s bibliometrics and the impact factor to decide on scientists’ careers.

The hypothesis that the authors formulated in 1983 can reasonably be accepted. Health markets most likely undermine patients’ health, physicians and professionals’ status and morale, and taxpayers’ interests. The key function of health sector reforms is not public health but economic: they aim to privatise the profitable part of health care financing; maximise the return on health care with commercial healthcare management of services and for-profit care delivery; prevent public services from being involved in a competition with the private sector for health care delivery, management, and financing; and open markets in LMICs with public aid funds to medical and pharmaceutical goods preferably manufactured in industrialised countries.

The studies analysed here show physicians and their organisations that commercial healthcare financing is incompatible with ethical, medical practice because, with or without vertical integration (in HMOs or PPOs), whether through contracts or wages, it imposes the goal of maximising shareholders’ profits on physicians and health professionals, whereas this commercial mission goes against the grain of Hippocratic ethics.

To patients’ organisations, the studies analysed here prove worldwide that care commercialisation prevents solidarity in healthcare financing and obstructs equal access to care. Markets segment health systems, they foment competition between physicians, whilst cooperation among them is essential to peoples’ health [ 13 ]. Moreover, they use public expenditure on healthcare inefficiently.

This research thus opens avenues for joint political action by patients’ and physicians’ organisations to defend and promote social protection in health because it shows that both doctors and patients benefit from professional care delivery and publicly-oriented care financing and management; the major contemporary threat to care accessibility and quality, namely, the privatisation of health care financing, also jeopardises the physicians’ autonomy, ethics, and incomes.

Finally, this research shows that competition prevails between not only commercial entities but also sectors. The interests of insurance banks investing in health and those of all the other economic actors are contradictory: Inter-country comparisons of total health expenditures reveal that the commodisation of care is accompanied by broad inter-sectoral, macro-economic redistribution. Economic agents that do not invest in health insurances would do better to learn from this.

Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Whilst these schemes had been forced upon employers, they unexpectedly proved to be highly favourable to economic growth.

Referring to health services and systems, we use the terms “publicly oriented”, “publicly minded”, “socially driven”, “non commercial”, and “not for profit” interchangeably

Abbreviations

Disability-adjusted life years

European Commission

Global health initiatives

General practitioner

High income countries

Infant mortality rate

Low and middle income countries

Medical doctor

Millennium development goals

Maternal mortality rate

Quality-adjusted life years

Sustainable development goals

Walsh JA, Warren KS. Selective primary healthcare: an interim strategy for disease control in developing countries. N Engl J Med. 1979;301(18):967–74.

Article   CAS   PubMed   Google Scholar  

Berggren WL, Ewbank D, Berggren G. Reduction of mortality in rural Haiti through a primary-health-care program. N Engl J Med. 1981;304:1324–30.

The Kasongo project. Annales belges de médecine tropicale. 1981. 60, suppl.

Google Scholar  

Unger JP, Killingsworth JR. Selective primary healthcare: methods and results. Soc Sci Med. 1986;22:1001–13.

G.E.R.M. Pour une politique de la santé. Bruxelles: La Revue Nouvelle; 1971.

Unger JP, Morales I, De Paepe P, Roland M. Medical heuristics and action-research. Professionalism versus science. Forthcoming as part of BMC Health Services Research Volume 20 Supplement 2, 2020: “ The Physician and Professionalism Today: Challenges to and strategies for ethical professional medical practice ." The full contents of the supplement are available online at https://bmchealthservres.biomedcentral.com/articles/supplements/volume-20-supplement-2 .

Bourdieu P. Esquisse d'une théorie de la pratique. Paris: Les Editions du Seuil; 2000.

Unger JP, De Paepe P, Van Dessel P, Stolkiner A. The production of critical theories in Health Systems Research and Education. An epistemological approach to emancipating public research and education from private interests. Health Cult Soc. 2011;1,1:ISSN 2161–6590. https://doi.org/10.5195/hcs.2011.50 http://hcs.pitt.edu . (online).

Article   Google Scholar  

International Health and Aid Policies. Unger J.P., De Paepe P., Sen S. & Soors W., editors. Cambridge University Press; 2010. Available: http://www.cambridge.org/us/catalogue/catalogue.asp?isbn=9780521174268 . Accessed 13 Sept 2020.

http://www.maisonmedicale.org . Accessed 13 Sept 2020.

http://www.sante-solidarite.be . Accessed 13 Sept 2020.

De Paepe P, Soors W, Unger JP. International aid policy: public disease control and private curative care? Cad Saude Publica. 2007;23(Suppl.2):S273–81.

Article   PubMed   Google Scholar  

Segall M. District health systems in a neoliberal world: a review of five keypolicy areas. Int J Health Plann Manag. 2003;18:S5–S26.

Van der Stuyft P, Unger JP. Improving the performance of health systems: the world health report as go-between for scientific evidence and ideological discourse. Tropical Med Int Health. 2000;5(10):675–7.

Unger JP, d’Alessandro U, De Paepe P, Green A. Can malaria be controlled where basic health services are not used? Tropical Med Int Health. 2006;11(3):314–22.

Unger JP, De Paepe P, Green A. A code of best practice for disease control programs to avoid damaging healthcare services in developing countries. Int J Health Plann Manag. 2003;18:S27–39.

Unger JP, De Paepe P, Ghilbert P, Soors W, Green A. Disintegrated care: the Achilles heel of international health policies. In low and middle income countries. Int J Integr Care. 2006;6:e14.

PubMed   PubMed Central   Google Scholar  

Unger JP, De Paepe P, Ghilbert P, Soors W, Green A. Integrated care: a fresh perspective for international health policies in low and middle-income countries. Int J Integr Care. 2006;6:e15.

Unger JP, Macq J, Bredo F, Boelaert M. Through Mintzberg's glasses: a fresh look at the organisation of ministries of health. Bull World Health Organ. 2000;78(8):1005–14.

CAS   PubMed   PubMed Central   Google Scholar  

Unger JP, De Paepe P, Buitrón R, Soors W. Achievements of a heterodox health policy. Am J Public Health. 2008;98(4):636–43.

Article   PubMed   PubMed Central   Google Scholar  

De Vos P, De Ceukelaire W, Van der Stuyft P. Colombia and Cuba, contrasting models in Latin America’s health sector reform. Tropical Med Int Health. 2006;11(10):1604–12.

De Groote T, De Paepe P, Unger JP. Colombia: in vivo test of health sector privatisation in the developing world. Int J Health Serv. 2005;35(1):125–41.

Vargas I, Vázquez ML, Mogollón-Perez AS, Unger JP. Barriers of access to care in a managed competition model: lessons from Colombia. BMC Health Serv Res. 2010;10(297):1–12.

Garcia-Subirats I, Vargas I, Mogollón-Perez AS, De Paepe P, Ferreira da Silva MR, Unger JP, et al. Inequities in access to healthcare in different health systems. A study in municipalities of central Colombia and north-eastern Brazil. Int J Equity Health. 2014;13(10):1–15.

Unger JP, De Paepe P, Arteaga HO, Solimano CG. Chile’s neoliberal health reform: an assessment and a critique. PLoS Med. 2008;5:4e79:0001–6.

http://www.bmg.bund.de/fileadmin/redaktion/pdf_who/Information_Ministerial_Conference_.pdf . Accessed 22 Apr 2016.

https://oi-files-d8-prod.s3.eu-west-2.amazonaws.com/s3fs-public/file_attachments/bp176-universal-health-coverage-091013-summ-en__1.pdf . Accessed 13 Sept 2020.

Andoh-Adjei FX. Assessing the performance of district mutual health insurance schemes in Ghana. International course in health development 46;2009/2010. Amsterdam: Royal Tropical Institute; 2010.

Guarnizo-Herreño CC, Agudelo C. Equidad de Género en el Acceso a los Servicios de Salud en Colombia. Rev Salud Pública Colomb. 2008;10(Sup1):44–57.

Consorcio de Investigación Economica y Social. Investigaciones sobre salud. 2011. http://www.cies.org.pe/investigaciones/salud .

De Paepe P, Rojas E, Abad L, Van Dessel P, Unger JP. Improving access. In: Unger JP, De Paepe P, Sen K, Soors W, editors. International health and aid policies; the need for alternatives. Cambridge: Cambridge University Press; 2010. p. p210–24.

Chapter   Google Scholar  

Lagomarsino G, Nachuk S, Singh KS. Public stewardship of private providers in mixed health systems. Washington DC: Synthesis from the Rockefeller Foundation. Results for Development Institute; 2009.

Unger J.P., Van Dessel P., Van der Veer C. & Shelmerdine S. Maternal health regulations in Vietnam, India and China. A comparison across case studies and countries. Deliverable 5.1. HESVIC project ‘Health system stewardship and regulation in Vietnam, India and China’. Institute of Tropical Medicine, Antwerp. A project financed by the European Commission. 154 pages. 2012. Available: https://medicinehealth.leeds.ac.uk/downloads/download/122/hesvic_-_comparative_report_d5_1_120713_final . Accessed 13 Sept 2020.

Shuftan C, Unger JP. The Rockefeller Foundation’s public stewardship of private providers in mixed health systems: a point-by-point critique. Soc Med. 2011;6(2):128–36.

The Guardian. 30,000 lobbyists and counting: is Brussels under corporate sway? 2014.

In Article 152 (consolidated, Amsterdam version of the Rome Treaty): “Community action in the field of public health shall fully respect the responsibilities of the Member States for the organisation and delivery of health services and medical care. In particular, measures referred to in paragraph 4(a) shall not affect national provisions on the donation or medical use of organs and blood”.

OECD. Health data set, the Commonwealth Fund, and the 2012 WHO Global Health expenditure database as in Gapminder. 2012.

http://www.oecd.org/els/health-systems/health-data.htm . Accessed 25 Mar 2018.

Unger J-P, De Paepe P. Commercial health care financing: the cause of U.S., Dutch, and Swiss health systems inefficiency? Int J Health Serv. 2019;9(3):431–56 https://doi.org/10.1177/0020731419847113 .

Unmet care needs due to costs in eleven OECD countries by income group. Commonwealth Fund Health Survey. 2010.

Experienced cost-related access problem, 2007 and 2013. The Commonwealth Fund. Interactives and data. 2014. Available: http://www.commonwealthfund.org/interactivesand/international-survey-data .

Osborn R., Squires D., Doty M.M., Sarnak D.O. & Schneider E.C. In new survey of eleven countries, US adults still struggle with access to and affordability of healthcare Health Aff. Published online November 16, 2016.

Shanafelt TD, Boone S, Tan L, et al. Burnout and satisfaction with work-life balance among US physicians relative to the general US population. Arch Intern Med. 2012;172(18):1377–85. https://doi.org/10.1001/archinternmed.2012.3199 .

Unger J-P. Physicians’ burnout (and that of psychologists, nurses, magistrates, researchers, and professors). For a Control Program. Int J Health Serv. 2019; https://doi.org/10.1177/0020731419883525 .

Gapminder data drawn from the OECD’s 2012 and 2018 health data sets. OECD QWIDS through www.gapminder.org .

OECD Health Statistics. 2018 https://stats.oecd.org/Index.aspx?DataSetCode=SHA . Accessed 5 Nov 2020.

Thow AM, Snowdon W, Labonté R, Gleeson D, Stuckler D, Hattersley L, et al. Will the next generation of preferential trade and investment agreements undermine prevention of non communicable diseases? A prospective policy analysis of the trans Pacific partnership agreement. Health Pol. 2015;119:88–96.

Unger J-P, Marchal B, Dugas S, Wuidar MJ, Burdet D, Leemans P, Unger J. Interface flow process audit: using the patient's career as a tracer of quality of care and of system organisation. Int J Integr Care. 2004;4:ISSN 1568–4156 http://www.ijic.org/ .

Soors W, De Paepe P, Unger JP. Management commitments and primary care: another lesson from Costa Rica for the world? Int J Health Serv. 2014;44(2):337–53.

Xu K, Evans DB, Kawabata K, Zeramdini, Klavus J, Murray CJL. Household catastrophic health expenditure: a multicountry analysis. Lancet. 2003;362:111–7.

Vargas I, Unger JP, Mogollon A, Vazquez ML. Effects of managed care mechanisms on access to healthcare: results from a qualitative study in Colombia. Int J Health Plann Manag. 2013;28(1):e13–33.

Garcia-Subirats I, Vargas I, Mogollón AS, De Paepe P, Ferreira da Silva MR, Unger JP, Vázquez ML. Barriers in access to healthcare in countries with different health systems. A cross-sectional study in municipalities of central Colombia and north-eastern Brazil. Soc Sci Med. 2014;106C:204–13.

Vargas I, Mogollón AS, De Paepe P, Ferreira da Silva MR, Unger JP, Vázquez ML. Do existing mechanisms contribute to improvements in care coordination across levels of care in health services networks? Opinions of the health personnel in Colombia and Brazil. BMC Health Serv Res. 2015;15:213. https://doi.org/10.1186/s12913-015-0882-4 .

Filippi V, Ronsmans C, Campbell OMR, Graham WJ, Mills A, Borghi J, et al. Maternal health in poor countries: the broader context and a call for action. Lancet. 2006;368,9546:1535–41.

Unger JP, Van Dessel P, Sen K, De Paepe P. International health policy and stagnating maternal mortality. Is there a causal link? Reprod Health Matters. 2009;17,33:91–104.

Unger JP. Can intensive campaigns dynamise front line health services ? The evaluation of a vaccination campaign in Thiès Medical District, Senegal. Soc Sci Med. 1991;32(3):249–59.

Hogan MC, Foreman KJ, Naghavi M, et al. Maternal mortality for 181 countries, 1980–2008: a systematic analysis of progress towards millennium development goal 5. Lancet. 2010;375(9726):1609–23.

Nolte E, Scholz R, Shkolnikov V, McKee M. The contribution of medical care to changing life expectancy in Germany and Poland. Soc Sci Med. 2002;55(11):1905–21 Available: http://www.demogr.mpg.de/publications/files/1257_1042711497_1_Avoid-Germ-Poland.pdf .

Kruk ME, Gage AD, Joseph NT, Danaei G, García-Saisó S, Salomon JA. Mortality due to low-quality health systems in the universal health coverage era: a systematic analysis of amenable deaths in 137 countries. Lancet. 2018;392:2203–12.

Unger JP, Criel B, Mercenier P. L'approche verticale: une méthodologie d'identification des priorités stratégiques du contrôle des maladies tropicales. In: Van Lerberghe W, de Béthune X, editors. Intégration et recherche Antwerp Institute of Tropical Medicine; 1998. p. 17–43.

Piot MA. A simulation model of case finding and treatment in tuberculosis control programmes. Geneva: WHO; 1967.

Unger JP, De Paepe P, Ghilbert P, Zocchi W, Van Dessel P, Qadeer I, et al. Privatisation (PPM-DOTS) strategy for tuberculosis control: how evidence-based is it? In: Unger JP, De Paepe P, Sen K, Soors W, editors. International health and aid policies; the need for alternatives. Cambridge: Cambridge University Press; 2010. p. 57–66.

Unger J-P, Marchal B, Green A. Quality standards for health care delivery and management in publicly-oriented health services. Int J Health Plann Manag. 2003;18:S79–88.

Unger JP, Mbaye A, Diao M. Can intensive campaigns dynamise front line health services ? The evaluation of a vaccination campaign in Thiès Medical District, Senegal. Soc Sci Med. 1991;32(3):249–59.

De Paepe P, Echeverria RE, Aguilar SE, Unger JP. Ecuador’s silent health reform. Int J Health Serv. 2012;42(2):219–33.

Lewis M. Governance and corruption in public healthcare systems: Center for Global Development. The World Bank; 2006.

Tejerina H, De Paepe P, Closon MC, Van Dessel P, Darras C, Unger JP. Forty years of USAID health cooperation in Bolivia. A lose–lose game? Int J Health Plann Manag. 2014;29(1):90–107.

Tejerina H, De Paepe P, Soors W, Lanza OV, Closon MC, Van Dessel P, Unger JP. Revisiting health policy and the World Bank in Bolivia. Global Soc Policy. 2011;11:22–44 Available: http://gsp.sagepub.com/cgi/content/abstract/11/1/22 .

Bourdieu P. Sur l'État. Cours au Collège de France (1989–1992). Paris: Seuil; 2012.

Download references

Acknowledgements

We are indebted to Professors Charlene Harrington (Department of Social & Behavioral Sciences, University of California San Francisco), Antonio Ugalde (University of Texas at Austin, College of Liberal Arts), and Matt Anderson (Albert Einstein College of Medicine, New York) for their indispensable comments. Gaby Leyden edited the manuscript thoroughly. No error can be attributed to them.

About this supplement

This article has been published as part of BMC Health Services Research Volume 20 Supplement 2, 2020: “The Physician and Professionalism Today: Challenges to and strategies for ethical professional medical practice”. The full contents of the supplement are available online at https://bmchealthservres.biomedcentral.com/articles/supplements/volume-20-supplement-2 .

Publication of the supplement has been funded by the Institute of Tropical Medicine, Antwerp, Belgium.

Author information

Authors and affiliations.

Department of Public Health, Institute of Tropical Medicine, Nationalestraat 155, B-2000, Antwerp, Belgium

Jean-Pierre Unger & Pierre De Paepe

Medical Director, Office de la Naissance et de l’Enfance, French Community of Belgium, Chaussée de Charleroi 95, B-1060, Brussels, Belgium

Ingrid Morales

You can also search for this author in PubMed   Google Scholar

Contributions

The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Jean-Pierre Unger .

Ethics declarations

Ethics approval and consent to participate.

Not Applicable (this manuscript does not involve human participants, human data or human tissue)

Consent for publication

Not Applicable (this manuscript does not contain any individual person’s data)

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Unger, JP., Morales, I. & De Paepe, P. Objectives, methods, and results in critical health systems and policy research: evaluating the healthcare market. BMC Health Serv Res 20 (Suppl 2), 1072 (2020). https://doi.org/10.1186/s12913-020-05889-w

Download citation

Published : 09 December 2020

DOI : https://doi.org/10.1186/s12913-020-05889-w

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Health policy
• Health systems research
• Research methodology

BMC Health Services Research

ISSN: 1472-6963

• General enquiries: [email protected]

Loading metrics

Open Access

Peer-reviewed

Research Article

Assessing the impact of healthcare research: A systematic review of methodological frameworks

Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Validation, Writing – original draft, Writing – review & editing

Affiliation Centre for Patient Reported Outcomes Research, Institute of Applied Health Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham, United Kingdom

Roles Conceptualization, Formal analysis, Funding acquisition, Methodology, Project administration, Supervision, Validation, Writing – review & editing

* E-mail: [email protected]

Roles Data curation, Formal analysis, Methodology, Validation, Writing – review & editing

Roles Formal analysis, Methodology, Supervision, Validation, Writing – review & editing

• Samantha Cruz Rivera, 
• Derek G. Kyte, 
• Olalekan Lee Aiyegbusi, 
• Thomas J. Keeley, 
• Melanie J. Calvert

• Published: August 9, 2017
• https://doi.org/10.1371/journal.pmed.1002370
• Reader Comments

Increasingly, researchers need to demonstrate the impact of their research to their sponsors, funders, and fellow academics. However, the most appropriate way of measuring the impact of healthcare research is subject to debate. We aimed to identify the existing methodological frameworks used to measure healthcare research impact and to summarise the common themes and metrics in an impact matrix.

Methods and findings

Two independent investigators systematically searched the Medical Literature Analysis and Retrieval System Online (MEDLINE), the Excerpta Medica Database (EMBASE), the Cumulative Index to Nursing and Allied Health Literature (CINAHL+), the Health Management Information Consortium, and the Journal of Research Evaluation from inception until May 2017 for publications that presented a methodological framework for research impact. We then summarised the common concepts and themes across methodological frameworks and identified the metrics used to evaluate differing forms of impact. Twenty-four unique methodological frameworks were identified, addressing 5 broad categories of impact: (1) ‘primary research-related impact’, (2) ‘influence on policy making’, (3) ‘health and health systems impact’, (4) ‘health-related and societal impact’, and (5) ‘broader economic impact’. These categories were subdivided into 16 common impact subgroups. Authors of the included publications proposed 80 different metrics aimed at measuring impact in these areas. The main limitation of the study was the potential exclusion of relevant articles, as a consequence of the poor indexing of the databases searched.

Conclusions

The measurement of research impact is an essential exercise to help direct the allocation of limited research resources, to maximise research benefit, and to help minimise research waste. This review provides a collective summary of existing methodological frameworks for research impact, which funders may use to inform the measurement of research impact and researchers may use to inform study design decisions aimed at maximising the short-, medium-, and long-term impact of their research.

Author summary

Why was this study done.

• There is a growing interest in demonstrating the impact of research in order to minimise research waste, allocate resources efficiently, and maximise the benefit of research. However, there is no consensus on which is the most appropriate tool to measure the impact of research.
• To our knowledge, this review is the first to synthesise existing methodological frameworks for healthcare research impact, and the associated impact metrics by which various authors have proposed impact should be measured, into a unified matrix.

What did the researchers do and find?

• We conducted a systematic review identifying 24 existing methodological research impact frameworks.
• We scrutinised the sample, identifying and summarising 5 proposed impact categories, 16 impact subcategories, and over 80 metrics into an impact matrix and methodological framework.

What do these findings mean?

• This simplified consolidated methodological framework will help researchers to understand how a research study may give rise to differing forms of impact, as well as in what ways and at which time points these potential impacts might be measured.
• Incorporating these insights into the design of a study could enhance impact, optimizing the use of research resources.

Citation: Cruz Rivera S, Kyte DG, Aiyegbusi OL, Keeley TJ, Calvert MJ (2017) Assessing the impact of healthcare research: A systematic review of methodological frameworks. PLoS Med 14(8): e1002370. https://doi.org/10.1371/journal.pmed.1002370

Academic Editor: Mike Clarke, Queens University Belfast, UNITED KINGDOM

Received: February 28, 2017; Accepted: July 7, 2017; Published: August 9, 2017

Copyright: © 2017 Cruz Rivera et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All relevant data are within the paper and supporting files.

Funding: Funding was received from Consejo Nacional de Ciencia y Tecnología (CONACYT). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript ( http://www.conacyt.mx/ ).

Competing interests: I have read the journal's policy and the authors of this manuscript have the following competing interests: MJC has received consultancy fees from Astellas and Ferring pharma and travel fees from the European Society of Cardiology outside the submitted work. TJK is in full-time paid employment for PAREXEL International.

Abbreviations: AIHS, Alberta Innovates—Health Solutions; CAHS, Canadian Academy of Health Sciences; CIHR, Canadian Institutes of Health Research; CINAHL+, Cumulative Index to Nursing and Allied Health Literature; EMBASE, Excerpta Medica Database; ERA, Excellence in Research for Australia; HEFCE, Higher Education Funding Council for England; HMIC, Health Management Information Consortium; HTA, Health Technology Assessment; IOM, Impact Oriented Monitoring; MDG, Millennium Development Goal; NHS, National Health Service; MEDLINE, Medical Literature Analysis and Retrieval System Online; PHC RIS, Primary Health Care Research & Information Service; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; PROM, patient-reported outcome measures; QALY, quality-adjusted life year; R&D, research and development; RAE, Research Assessment Exercise; REF, Research Excellence Framework; RIF, Research Impact Framework; RQF, Research Quality Framework; SDG, Sustainable Development Goal; SIAMPI, Social Impact Assessment Methods for research and funding instruments through the study of Productive Interactions between science and society

Introduction

In 2010, approximately US$240 billion was invested in healthcare research worldwide [ 1 ]. Such research is utilised by policy makers, healthcare providers, and clinicians to make important evidence-based decisions aimed at maximising patient benefit, whilst ensuring that limited healthcare resources are used as efficiently as possible to facilitate effective and sustainable service delivery. It is therefore essential that this research is of high quality and that it is impactful—i.e., it delivers demonstrable benefits to society and the wider economy whilst minimising research waste [ 1 , 2 ]. Research impact can be defined as ‘any identifiable ‘benefit to, or positive influence on the economy, society, public policy or services, health, the environment, quality of life or academia’ (p. 26) [ 3 ].

There are many purported benefits associated with the measurement of research impact, including the ability to (1) assess the quality of the research and its subsequent benefits to society; (2) inform and influence optimal policy and funding allocation; (3) demonstrate accountability, the value of research in terms of efficiency and effectiveness to the government, stakeholders, and society; and (4) maximise impact through better understanding the concept and pathways to impact [ 4 – 7 ].

Measuring and monitoring the impact of healthcare research has become increasingly common in the United Kingdom [ 5 ], Australia [ 5 ], and Canada [ 8 ], as governments, organisations, and higher education institutions seek a framework to allocate funds to projects that are more likely to bring the most benefit to society and the economy [ 5 ]. For example, in the UK, the 2014 Research Excellence Framework (REF) has recently been used to assess the quality and impact of research in higher education institutions, through the assessment of impact cases studies and selected qualitative impact metrics [ 9 ]. This is the first initiative to allocate research funding based on the economic, societal, and cultural impact of research, although it should be noted that research impact only drives a proportion of this allocation (approximately 20%) [ 9 ].

In the UK REF, the measurement of research impact is seen as increasingly important. However, the impact element of the REF has been criticised in some quarters [ 10 , 11 ]. Critics deride the fact that REF impact is determined in a relatively simplistic way, utilising researcher-generated case studies, which commonly attempt to link a particular research outcome to an associated policy or health improvement despite the fact that the wider literature highlights great diversity in the way research impact may be demonstrated [ 12 , 13 ]. This led to the current debate about the optimal method of measuring impact in the future REF [ 10 , 14 ]. The Stern review suggested that research impact should not only focus on socioeconomic impact but should also include impact on government policy, public engagement, academic impacts outside the field, and teaching to showcase interdisciplinary collaborative impact [ 10 , 11 ]. The Higher Education Funding Council for England (HEFCE) has recently set out the proposals for the REF 2021 exercise, confirming that the measurement of such impact will continue to form an important part of the process [ 15 ].

With increasing pressure for healthcare research to lead to demonstrable health, economic, and societal impact, there is a need for researchers to understand existing methodological impact frameworks and the means by which impact may be quantified (i.e., impact metrics; see Box 1 , 'Definitions’) to better inform research activities and funding decisions. From a researcher’s perspective, understanding the optimal pathways to impact can help inform study design aimed at maximising the impact of the project. At the same time, funders need to understand which aspects of impact they should focus on when allocating awards so they can make the most of their investment and bring the greatest benefit to patients and society [ 2 , 4 , 5 , 16 , 17 ].

Box 1. Definitions

• Research impact: ‘any identifiable benefit to, or positive influence on, the economy, society, public policy or services, health, the environment, quality of life, or academia’ (p. 26) [ 3 ].
• Methodological framework: ‘a body of methods, rules and postulates employed by a particular procedure or set of procedures (i.e., framework characteristics and development)’ [ 18 ].
• Pathway: ‘a way of achieving a specified result; a course of action’ [ 19 ].
• Quantitative metrics: ‘a system or standard of [quantitative] measurement’ [ 20 ].
• Narrative metrics: ‘a spoken or written account of connected events; a story’ [ 21 ].

Whilst previous researchers have summarised existing methodological frameworks and impact case studies [ 4 , 22 – 27 ], they have not summarised the metrics for use by researchers, funders, and policy makers. The aim of this review was therefore to (1) identify the methodological frameworks used to measure healthcare research impact using systematic methods, (2) summarise common impact themes and metrics in an impact matrix, and (3) provide a simplified consolidated resource for use by funders, researchers, and policy makers.

Search strategy and selection criteria

Initially, a search strategy was developed to identify the available literature regarding the different methods to measure research impact. The following keywords: ‘Impact’, ‘Framework’, and ‘Research’, and their synonyms, were used during the search of the Medical Literature Analysis and Retrieval System Online (MEDLINE; Ovid) database, the Excerpta Medica Database (EMBASE), the Health Management Information Consortium (HMIC) database, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL+) database (inception to May 2017; see S1 Appendix for the full search strategy). Additionally, the nonindexed Journal of Research Evaluation was hand searched during the same timeframe using the keyword ‘Impact’. Other relevant articles were identified through 3 Internet search engines (Google, Google Scholar, and Google Images) using the keywords ‘Impact’, ‘Framework’, and ‘Research’, with the first 50 results screened. Google Images was searched because different methodological frameworks are summarised in a single image and can easily be identified through this search engine. Finally, additional publications were sought through communication with experts.

Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (see S1 PRISMA Checklist ), 2 independent investigators systematically screened for publications describing, evaluating, or utilising a methodological research impact framework within the context of healthcare research [ 28 ]. Papers were eligible if they included full or partial methodological frameworks or pathways to research impact; both primary research and systematic reviews fitting these criteria were included. We included any methodological framework identified (original or modified versions) at the point of first occurrence. In addition, methodological frameworks were included if they were applicable to the healthcare discipline with no need of modification within their structure. We defined ‘methodological framework’ as ‘a body of methods, rules and postulates employed by a particular procedure or set of procedures (i.e., framework characteristics and development)’ [ 18 ], whereas we defined ‘pathway’ as ‘a way of achieving a specified result; a course of action’ [ 19 ]. Studies were excluded if they presented an existing (unmodified) methodological framework previously available elsewhere, did not explicitly describe a methodological framework but rather focused on a single metric (e.g., bibliometric analysis), focused on the impact or effectiveness of interventions rather than that of the research, or presented case study data only. There were no language restrictions.

Data screening

Records were downloaded into Endnote (version X7.3.1), and duplicates were removed. Two independent investigators (SCR and OLA) conducted all screening following a pilot aimed at refining the process. The records were screened by title and abstract before full-text articles of potentially eligible publications were retrieved for evaluation. A full-text screening identified the publications included for data extraction. Discrepancies were resolved through discussion, with the involvement of a third reviewer (MJC, DGK, and TJK) when necessary.

Data extraction and analysis

Data extraction occurred after the final selection of included articles. SCR and OLA independently extracted details of impact methodological frameworks, the country of origin, and the year of publication, as well as the source, the framework description, and the methodology used to develop the framework. Information regarding the methodology used to develop each methodological framework was also extracted from framework webpages where available. Investigators also extracted details regarding each framework’s impact categories and subgroups, along with their proposed time to impact (‘short-term’, ‘mid-term’, or ‘long-term’) and the details of any metrics that had been proposed to measure impact, which are depicted in an impact matrix. The structure of the matrix was informed by the work of M. Buxton and S. Hanney [ 2 ], P. Buykx et al. [ 5 ], S. Kuruvila et al. [ 29 ], and A. Weiss [ 30 ], with the intention of mapping metrics presented in previous methodological frameworks in a concise way. A consensus meeting with MJC, DGK, and TJK was held to solve disagreements and finalise the data extraction process.

Included studies

Our original search strategy identified 359 citations from MEDLINE (Ovid), EMBASE, CINAHL+, HMIC, and the Journal of Research Evaluation, and 101 citations were returned using other sources (Google, Google Images, Google Scholar, and expert communication) (see Fig 1 ) [ 28 ]. In total, we retrieved 54 full-text articles for review. At this stage, 39 articles were excluded, as they did not propose new or modified methodological frameworks. An additional 15 articles were included following the backward and forward citation method. A total of 31 relevant articles were included in the final analysis, of which 24 were articles presenting unique frameworks and the remaining 7 were systematic reviews [ 4 , 22 – 27 ]. The search strategy was rerun on 15 May 2017. A further 19 publications were screened, and 2 were taken forward to full-text screening but were ineligible for inclusion.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

https://doi.org/10.1371/journal.pmed.1002370.g001

Methodological framework characteristics

The characteristics of the 24 included methodological frameworks are summarised in Table 1 , 'Methodological framework characteristics’. Fourteen publications proposed academic-orientated frameworks, which focused on measuring academic, societal, economic, and cultural impact using narrative and quantitative metrics [ 2 , 3 , 5 , 8 , 29 , 31 – 39 ]. Five publications focused on assessing the impact of research by focusing on the interaction process between stakeholders and researchers (‘productive interactions’), which is a requirement to achieve research impact. This approach tries to address the issue of attributing research impact to metrics [ 7 , 40 – 43 ]. Two frameworks focused on the importance of partnerships between researchers and policy makers, as a core element to accomplish research impact [ 44 , 45 ]. An additional 2 frameworks focused on evaluating the pathways to impact, i.e., linking processes between research and impact [ 30 , 46 ]. One framework assessed the ability of health technology to influence efficiency of healthcare systems [ 47 ]. Eight frameworks were developed in the UK [ 2 , 3 , 29 , 37 , 39 , 42 , 43 , 45 ], 6 in Canada [ 8 , 33 , 34 , 44 , 46 , 47 ], 4 in Australia [ 5 , 31 , 35 , 38 ], 3 in the Netherlands [ 7 , 40 , 41 ], and 2 in the United States [ 30 , 36 ], with 1 model developed with input from various countries [ 32 ].

https://doi.org/10.1371/journal.pmed.1002370.t001

Methodological framework development

The included methodological frameworks varied in their development process, but there were some common approaches employed. Most included a literature review [ 2 , 5 , 7 , 8 , 31 , 33 , 36 , 37 , 40 – 46 ], although none of them used a recognised systematic method. Most also consulted with various stakeholders [ 3 , 8 , 29 , 31 , 33 , 35 – 38 , 43 , 44 , 46 , 47 ] but used differing methods to incorporate their views, including quantitative surveys [ 32 , 35 , 43 , 46 ], face-to-face interviews [ 7 , 29 , 33 , 35 , 37 , 42 , 43 ], telephone interviews [ 31 , 46 ], consultation [ 3 , 7 , 36 ], and focus groups [ 39 , 43 ]. A range of stakeholder groups were approached across the sample, including principal investigators [ 7 , 29 , 43 ], research end users [ 7 , 42 , 43 ], academics [ 3 , 8 , 39 , 40 , 43 , 46 ], award holders [ 43 ], experts [ 33 , 38 , 39 ], sponsors [ 33 , 39 ], project coordinators [ 32 , 42 ], and chief investigators [ 31 , 35 ]. However, some authors failed to identify the stakeholders involved in the development of their frameworks [ 2 , 5 , 34 , 41 , 45 ], making it difficult to assess their appropriateness. In addition, only 4 of the included papers reported using formal analytic methods to interpret stakeholder responses. These included the Canadian Academy of Health Sciences framework, which used conceptual cluster analysis [ 33 ]. The Research Contribution [ 42 ], Research Impact [ 29 ], and Primary Health Care & Information Service [ 31 ] used a thematic analysis approach. Finally, some authors went on to pilot their framework, which shaped refinements on the methodological frameworks until approval. Methods used to pilot the frameworks included a case study approach [ 2 , 3 , 30 , 32 , 33 , 36 , 40 , 42 , 44 , 45 ], contrasting results against available literature [ 29 ], the use of stakeholders’ feedback [ 7 ], and assessment tools [ 35 , 46 ].

Major impact categories

1. primary research-related impact..

A number of methodological frameworks advocated the evaluation of ‘research-related impact’. This encompassed content related to the generation of new knowledge, knowledge dissemination, capacity building, training, leadership, and the development of research networks. These outcomes were considered the direct or primary impacts of a research project, as these are often the first evidenced returns [ 30 , 62 ].

A number of subgroups were identified within this category, with frameworks supporting the collection of impact data across the following constructs: ‘research and innovation outcomes’; ‘dissemination and knowledge transfer’; ‘capacity building, training, and leadership’; and ‘academic collaborations, research networks, and data sharing’.

1 . 1 . Research and innovation outcomes . Twenty of the 24 frameworks advocated the evaluation of ‘research and innovation outcomes’ [ 2 , 3 , 5 , 7 , 8 , 29 – 39 , 41 , 43 , 44 , 46 ]. This subgroup included the following metrics: number of publications; number of peer-reviewed articles (including journal impact factor); citation rates; requests for reprints, number of reviews, and meta-analysis; and new or changes in existing products (interventions or technology), patents, and research. Additionally, some frameworks also sought to gather information regarding ‘methods/methodological contributions’. These advocated the collection of systematic reviews and appraisals in order to identify gaps in knowledge and determine whether the knowledge generated had been assessed before being put into practice [ 29 ].

1 . 2 . Dissemination and knowledge transfer . Nineteen of the 24 frameworks advocated the assessment of ‘dissemination and knowledge transfer’ [ 2 , 3 , 5 , 7 , 29 – 32 , 34 – 43 , 46 ]. This comprised collection of the following information: number of conferences, seminars, workshops, and presentations; teaching output (i.e., number of lectures given to disseminate the research findings); number of reads for published articles; article download rate and number of journal webpage visits; and citations rates in nonjournal media such as newspapers and mass and social media (i.e., Twitter and blogs). Furthermore, this impact subgroup considered the measurement of research uptake and translatability and the adoption of research findings in technological and clinical applications and by different fields. These can be measured through patents, clinical trials, and partnerships between industry and business, government and nongovernmental organisations, and university research units and researchers [ 29 ].

1 . 3 . Capacity building , training , and leadership . Fourteen of 24 frameworks suggested the evaluation of ‘capacity building, training, and leadership’ [ 2 , 3 , 5 , 8 , 29 , 31 – 35 , 39 – 41 , 43 ]. This involved collecting information regarding the number of doctoral and postdoctoral studentships (including those generated as a result of the research findings and those appointed to conduct the research), as well as the number of researchers and research-related staff involved in the research projects. In addition, authors advocated the collection of ‘leadership’ metrics, including the number of research projects managed and coordinated and the membership of boards and funding bodies, journal editorial boards, and advisory committees [ 29 ]. Additional metrics in this category included public recognition (number of fellowships and awards for significant research achievements), academic career advancement, and subsequent grants received. Lastly, the impact metric ‘research system management’ comprised the collection of information that can lead to preserving the health of the population, such as modifying research priorities, resource allocation strategies, and linking health research to other disciplines to maximise benefits [ 29 ].

1 . 4 . Academic collaborations , research networks , and data sharing . Lastly, 10 of the 24 frameworks advocated the collection of impact data regarding ‘academic collaborations (internal and external collaborations to complete a research project), research networks, and data sharing’ [ 2 , 3 , 5 , 7 , 29 , 34 , 37 , 39 , 41 , 43 ].

2. Influence on policy making.

Methodological frameworks addressing this major impact category focused on measurable improvements within a given knowledge base and on interactions between academics and policy makers, which may influence policy-making development and implementation. The returns generated in this impact category are generally considered as intermediate or midterm (1 to 3 years). These represent an important interim stage in the process towards the final expected impacts, such as quantifiable health improvements and economic benefits, without which policy change may not occur [ 30 , 62 ]. The following impact subgroups were identified within this category: ‘type and nature of policy impact’, ‘level of policy making’, and ‘policy networks’.

2 . 1 . Type and nature of policy impact . The most common impact subgroup, mentioned in 18 of the 24 frameworks, was ‘type and nature of policy impact’ [ 2 , 7 , 29 – 38 , 41 – 43 , 45 – 47 ]. Methodological frameworks addressing this subgroup stressed the importance of collecting information regarding the influence of research on policy (i.e., changes in practice or terminology). For instance, a project looking at trafficked adolescents and women (2003) influenced the WHO guidelines (2003) on ethics regarding this particular group [ 17 , 21 , 63 ].

2 . 2 . Level of policy impact . Thirteen of 24 frameworks addressed aspects surrounding the need to record the ‘level of policy impact’ (international, national, or local) and the organisations within a level that were influenced (local policy makers, clinical commissioning groups, and health and wellbeing trusts) [ 2 , 5 , 8 , 29 , 31 , 34 , 38 , 41 , 43 – 47 ]. Authors considered it important to measure the ‘level of policy impact’ to provide evidence of collaboration, coordination, and efficiency within health organisations and between researchers and health organisations [ 29 , 31 ].

2 . 3 . Policy networks . Five methodological frameworks highlighted the need to collect information regarding collaborative research with industry and staff movement between academia and industry [ 5 , 7 , 29 , 41 , 43 ]. A policy network emphasises the relationship between policy communities, researchers, and policy makers. This relationship can influence and lead to incremental changes in policy processes [ 62 ].

3. Health and health systems impact.

A number of methodological frameworks advocated the measurement of impacts on health and healthcare systems across the following impact subgroups: ‘quality of care and service delivering’, ‘evidence-based practice’, ‘improved information and health information management’, ‘cost containment and effectiveness’, ‘resource allocation’, and ‘health workforce’.

3 . 1 . Quality of care and service delivery . Twelve of the 24 frameworks highlighted the importance of evaluating ‘quality of care and service delivery’ [ 2 , 5 , 8 , 29 – 31 , 33 – 36 , 41 , 47 ]. There were a number of suggested metrics that could be potentially used for this purpose, including health outcomes such as quality-adjusted life years (QALYs), patient-reported outcome measures (PROMs), patient satisfaction and experience surveys, and qualitative data on waiting times and service accessibility.

3 . 2 . Evidence-based practice . ‘Evidence-based practice’, mentioned in 5 of the 24 frameworks, refers to making changes in clinical diagnosis, clinical practice, treatment decisions, or decision making based on research evidence [ 5 , 8 , 29 , 31 , 33 ]. The suggested metrics to demonstrate evidence-based practice were adoption of health technologies and research outcomes to improve the healthcare systems and inform policies and guidelines [ 29 ].

3 . 3 . Improved information and health information management . This impact subcategory, mentioned in 5 of the 24 frameworks, refers to the influence of research on the provision of health services and management of the health system to prevent additional costs [ 5 , 29 , 33 , 34 , 38 ]. Methodological frameworks advocated the collection of health system financial, nonfinancial (i.e., transport and sociopolitical implications), and insurance information in order to determine constraints within a health system.

3 . 4 . Cost containment and cost-effectiveness . Six of the 24 frameworks advocated the subcategory ‘cost containment and cost-effectiveness’ [ 2 , 5 , 8 , 17 , 33 , 36 ]. ‘Cost containment’ comprised the collection of information regarding how research has influenced the provision and management of health services and its implication in healthcare resource allocation and use [ 29 ]. ‘Cost-effectiveness’ refers to information concerning economic evaluations to assess improvements in effectiveness and health outcomes—for instance, the cost-effectiveness (cost and health outcome benefits) assessment of introducing a new health technology to replace an older one [ 29 , 31 , 64 ].

3 . 5 . Resource allocation . ‘Resource allocation’, mentioned in 6frameworks, can be measured through 2 impact metrics: new funding attributed to the intervention in question and equity while allocating resources, such as improved allocation of resources at an area level; better targeting, accessibility, and utilisation; and coverage of health services [ 2 , 5 , 29 , 31 , 45 , 47 ]. The allocation of resources and targeting can be measured through health services research reports, with the utilisation of health services measured by the probability of providing an intervention when needed, the probability of requiring it again in the future, and the probability of receiving an intervention based on previous experience [ 29 , 31 ].

3 . 6 . Health workforce . Lastly, ‘health workforce’, present in 3 methodological frameworks, refers to the reduction in the days of work lost because of a particular illness [ 2 , 5 , 31 ].

4. Health-related and societal impact.

Three subgroups were included in this category: ‘health literacy’; ‘health knowledge, attitudes, and behaviours’; and ‘improved social equity, inclusion, or cohesion’.

4 . 1 . Health knowledge , attitudes , and behaviours . Eight of the 24 frameworks suggested the assessment of ‘health knowledge, attitudes, behaviours, and outcomes’, which could be measured through the evaluation of levels of public engagement with science and research (e.g., National Health Service (NHS) Choices end-user visit rate) or by using focus groups to analyse changes in knowledge, attitudes, and behaviour among society [ 2 , 5 , 29 , 33 – 35 , 38 , 43 ].

4 . 2 . Improved equity , inclusion , or cohesion and human rights . Other methodological frameworks, 4 of the 24, suggested capturing improvements in equity, inclusion, or cohesion and human rights. Authors suggested these could be using a resource like the United Nations Millennium Development Goals (MDGs) (superseded by Sustainable Development Goals [SDGs] in 2015) and human rights [ 29 , 33 , 34 , 38 ]. For instance, a cluster-randomised controlled trial in Nepal, which had female participants, has demonstrated the reduction of neonatal mortality through the introduction of maternity health care, distribution of delivery kits, and home visits. This illustrates how research can target vulnerable and disadvantaged groups. Additionally, this research has been introduced by the World Health Organisation to achieve the MDG ‘improve maternal health’ [ 16 , 29 , 65 ].

4 . 3 . Health literacy . Some methodological frameworks, 3 of the 24, focused on tracking changes in the ability of patients to make informed healthcare decisions, reduce health risks, and improve quality of life, which were demonstrably linked to a particular programme of research [ 5 , 29 , 43 ]. For example, a systematic review showed that when HIV health literacy/knowledge is spread among people living with the condition, antiretroviral adherence and quality of life improve [ 66 ].

5. Broader economic impacts.

Some methodological frameworks, 9 of 24, included aspects related to the broader economic impacts of health research—for example, the economic benefits emerging from the commercialisation of research outputs [ 2 , 5 , 29 , 31 , 33 , 35 , 36 , 38 , 67 ]. Suggested metrics included the amount of funding for research and development (R&D) that was competitively awarded by the NHS, medical charities, and overseas companies. Additional metrics were income from intellectual property, spillover effects (any secondary benefit gained as a repercussion of investing directly in a primary activity, i.e., the social and economic returns of investing on R&D) [ 33 ], patents granted, licences awarded and brought to the market, the development and sales of spinout companies, research contracts, and income from industry.

The benefits contained within the categories ‘health and health systems impact’, ‘health-related and societal impact’, and ‘broader economic impacts’ are considered the expected and final returns of the resources allocated in healthcare research [ 30 , 62 ]. These benefits commonly arise in the long term, beyond 5 years according to some authors, but there was a recognition that this could differ depending on the project and its associated research area [ 4 ].

Data synthesis

Five major impact categories were identified across the 24 included methodological frameworks: (1) ‘primary research-related impact’, (2) ‘influence on policy making’, (3) ‘health and health systems impact’, (4) ‘health-related and societal impact’, and (5) ‘broader economic impact’. These major impact categories were further subdivided into 16 impact subgroups. The included publications proposed 80 different metrics to measure research impact. This impact typology synthesis is depicted in ‘the impact matrix’ ( Fig 2 and Fig 3 ).

CIHR, Canadian Institutes of Health Research; HTA, Health Technology Assessment; PHC RIS, Primary Health Care Research & Information Service; RAE, Research Assessment Exercise; RQF, Research Quality Framework.

https://doi.org/10.1371/journal.pmed.1002370.g002

AIHS, Alberta Innovates—Health Solutions; CAHS, Canadian Institutes of Health Research; IOM, Impact Oriented Monitoring; REF, Research Excellence Framework; SIAMPI, Social Impact Assessment Methods for research and funding instruments through the study of Productive Interactions between science and society.

https://doi.org/10.1371/journal.pmed.1002370.g003

Commonality and differences across frameworks

The ‘Research Impact Framework’ and the ‘Health Services Research Impact Framework’ were the models that encompassed the largest number of the metrics extracted. The most dominant methodological framework was the Payback Framework; 7 other methodological framework models used the Payback Framework as a starting point for development [ 8 , 29 , 31 – 35 ]. Additional methodological frameworks that were commonly incorporated into other tools included the CIHR framework, the CAHS model, the AIHS framework, and the Exchange model [ 8 , 33 , 34 , 44 ]. The capture of ‘research-related impact’ was the most widely advocated concept across methodological frameworks, illustrating the importance with which primary short-term impact outcomes were viewed by the included papers. Thus, measurement of impact via number of publications, citations, and peer-reviewed articles was the most common. ‘Influence on policy making’ was the predominant midterm impact category, specifically the subgroup ‘type and nature of policy impact’, in which frameworks advocated the measurement of (i) changes to legislation, regulations, and government policy; (ii) influence and involvement in decision-making processes; and (iii) changes to clinical or healthcare training, practice, or guidelines. Within more long-term impact measurement, the evaluations of changes in the ‘quality of care and service delivery’ were commonly advocated.

In light of the commonalities and differences among the methodological frameworks, the ‘pathways to research impact’ diagram ( Fig 4 ) was developed to provide researchers, funders, and policy makers a more comprehensive and exhaustive way to measure healthcare research impact. The diagram has the advantage of assorting all the impact metrics proposed by previous frameworks and grouping them into different impact subgroups and categories. Prospectively, this global picture will help researchers, funders, and policy makers plan strategies to achieve multiple pathways to impact before carrying the research out. The analysis of the data extraction and construction of the impact matrix led to the development of the ‘pathways to research impact’ diagram ( Fig 4 ). The diagram aims to provide an exhaustive and comprehensive way of tracing research impact by combining all the impact metrics presented by the different 24 frameworks, grouping those metrics into different impact subgroups, and grouping these into broader impact categories.

NHS, National Health Service; PROM, patient-reported outcome measure; QALY, quality-adjusted life year; R&D, research and development.

https://doi.org/10.1371/journal.pmed.1002370.g004

This review has summarised existing methodological impact frameworks together for the first time using systematic methods ( Fig 4 ). It allows researchers and funders to consider pathways to impact at the design stage of a study and to understand the elements and metrics that need to be considered to facilitate prospective assessment of impact. Users do not necessarily need to cover all the aspects of the methodological framework, as every research project can impact on different categories and subgroups. This review provides information that can assist researchers to better demonstrate impact, potentially increasing the likelihood of conducting impactful research and reducing research waste. Existing reviews have not presented a methodological framework that includes different pathways to impact, health impact categories, subgroups, and metrics in a single methodological framework.

Academic-orientated frameworks included in this review advocated the measurement of impact predominantly using so-called ‘quantitative’ metrics—for example, the number of peer-reviewed articles, journal impact factor, and citation rates. This may be because they are well-established measures, relatively easy to capture and objective, and are supported by research funding systems. However, these metrics primarily measure the dissemination of research finding rather than its impact [ 30 , 68 ]. Whilst it is true that wider dissemination, especially when delivered via world-leading international journals, may well lead eventually to changes in healthcare, this is by no means certain. For instance, case studies evaluated by Flinders University of Australia demonstrated that some research projects with non-peer-reviewed publications led to significant changes in health policy, whilst the studies with peer-reviewed publications did not result in any type of impact [ 68 ]. As a result, contemporary literature has tended to advocate the collection of information regarding a variety of different potential forms of impact alongside publication/citations metrics [ 2 , 3 , 5 , 7 , 8 , 29 – 47 ], as outlined in this review.

The 2014 REF exercise adjusted UK university research funding allocation based on evidence of the wider impact of research (through case narrative studies and quantitative metrics), rather than simply according to the quality of research [ 12 ]. The intention was to ensure funds were directed to high-quality research that could demonstrate actual realised benefit. The inclusion of a mixed-method approach to the measurement of impact in the REF (narrative and quantitative metrics) reflects a widespread belief—expressed by the majority of authors of the included methodological frameworks in the review—that individual quantitative impact metrics (e.g., number of citations and publications) do not necessary capture the complexity of the relationships involved in a research project and may exclude measurement of specific aspects of the research pathway [ 10 , 12 ].

Many of the frameworks included in this review advocated the collection of a range of academic, societal, economic, and cultural impact metrics; this is consistent with recent recommendations from the Stern review [ 10 ]. However, a number of these metrics encounter research ‘lag’: i.e., the time between the point at which the research is conducted and when the actual benefits arise [ 69 ]. For instance, some cardiovascular research has taken up to 25 years to generate impact [ 70 ]. Likewise, the impact may not arise exclusively from a single piece of research. Different processes (such as networking interactions and knowledge and research translation) and multiple individuals and organisations are often involved [ 4 , 71 ]. Therefore, attributing the contribution made by each of the different actors involved in the process can be a challenge [ 4 ]. An additional problem associated to attribution is the lack of evidence to link research and impact. The outcomes of research may emerge slowly and be absorbed gradually. Consequently, it is difficult to determine the influence of research in the development of a new policy, practice, or guidelines [ 4 , 23 ].

A further problem is that impact evaluation is conducted ‘ex post’, after the research has concluded. Collecting information retrospectively can be an issue, as the data required might not be available. ‘ex ante’ assessment is vital for funding allocation, as it is necessary to determine the potential forthcoming impact before research is carried out [ 69 ]. Additionally, ex ante evaluation of potential benefit can overcome the issues regarding identifying and capturing evidence, which can be used in the future [ 4 ]. In order to conduct ex ante evaluation of potential benefit, some authors suggest the early involvement of policy makers in a research project coupled with a well-designed strategy of dissemination [ 40 , 69 ].

Providing an alternate view, the authors of methodological frameworks such as the SIAMPI, Contribution Mapping, Research Contribution, and the Exchange model suggest that the problems of attribution are a consequence of assigning the impact of research to a particular impact metric [ 7 , 40 , 42 , 44 ]. To address these issues, these authors propose focusing on the contribution of research through assessing the processes and interactions between stakeholders and researchers, which arguably take into consideration all the processes and actors involved in a research project [ 7 , 40 , 42 , 43 ]. Additionally, contributions highlight the importance of the interactions between stakeholders and researchers from an early stage in the research process, leading to a successful ex ante and ex post evaluation by setting expected impacts and determining how the research outcomes have been utilised, respectively [ 7 , 40 , 42 , 43 ]. However, contribution metrics are generally harder to measure in comparison to academic-orientated indicators [ 72 ].

Currently, there is a debate surrounding the optimal methodological impact framework, and no tool has proven superior to another. The most appropriate methodological framework for a given study will likely depend on stakeholder needs, as each employs different methodologies to assess research impact [ 4 , 37 , 41 ]. This review allows researchers to select individual existing methodological framework components to create a bespoke tool with which to facilitate optimal study design and maximise the potential for impact depending on the characteristic of their study ( Fig 2 and Fig 3 ). For instance, if researchers are interested in assessing how influential their research is on policy making, perhaps considering a suite of the appropriate metrics drawn from multiple methodological frameworks may provide a more comprehensive method than adopting a single methodological framework. In addition, research teams may wish to use a multidimensional approach to methodological framework development, adopting existing narratives and quantitative metrics, as well as elements from contribution frameworks. This approach would arguably present a more comprehensive method of impact assessment; however, further research is warranted to determine its effectiveness [ 4 , 69 , 72 , 73 ].

Finally, it became clear during this review that the included methodological frameworks had been constructed using varied methodological processes. At present, there are no guidelines or consensus around the optimal pathway that should be followed to develop a robust methodological framework. The authors believe this is an area that should be addressed by the research community, to ensure future frameworks are developed using best-practice methodology.

For instance, the Payback Framework drew upon a literature review and was refined through a case study approach. Arguably, this approach could be considered inferior to other methods that involved extensive stakeholder involvement, such as the CIHR framework [ 8 ]. Nonetheless, 7 methodological frameworks were developed based upon the Payback Framework [ 8 , 29 , 31 – 35 ].

Limitations

The present review is the first to summarise systematically existing impact methodological frameworks and metrics. The main limitation is that 50% of the included publications were found through methods other than bibliographic databases searching, indicating poor indexing. Therefore, some relevant articles may not have been included in this review if they failed to indicate the inclusion of a methodological impact framework in their title/abstract. We did, however, make every effort to try to find these potentially hard-to-reach publications, e.g., through forwards/backwards citation searching, hand searching reference lists, and expert communication. Additionally, this review only extracted information regarding the methodology followed to develop each framework from the main publication source or framework webpage. Therefore, further evaluations may not have been included, as they are beyond the scope of the current paper. A further limitation was that although our search strategy did not include language restrictions, we did not specifically search non-English language databases. Thus, we may have failed to identify potentially relevant methodological frameworks that were developed in a non-English language setting.

In conclusion, the measurement of research impact is an essential exercise to help direct the allocation of limited research resources, to maximise benefit, and to help minimise research waste. This review provides a collective summary of existing methodological impact frameworks and metrics, which funders may use to inform the measurement of research impact and researchers may use to inform study design decisions aimed at maximising the short-, medium-, and long-term impact of their research.

Supporting information

S1 appendix. search strategy..

https://doi.org/10.1371/journal.pmed.1002370.s001

S1 PRISMA Checklist. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist.

https://doi.org/10.1371/journal.pmed.1002370.s002

Acknowledgments

We would also like to thank Mrs Susan Bayliss, Information Specialist, University of Birmingham, and Mrs Karen Biddle, Research Secretary, University of Birmingham.

• View Article
• PubMed/NCBI
• Google Scholar
• 3. HEFCE. REF 2014: Assessment framework and guidance on submissions 2011 [cited 2016 15 Feb]. Available from: http://www.ref.ac.uk/media/ref/content/pub/assessmentframeworkandguidanceonsubmissions/GOS%20including%20addendum.pdf .
• 8. Canadian Institutes of Health Research. Developing a CIHR framework to measure the impact of health research 2005 [cited 2016 26 Feb]. Available from: http://publications.gc.ca/collections/Collection/MR21-65-2005E.pdf .
• 9. HEFCE. HEFCE allocates £3.97 billion to universities and colleges in England for 2015–1 2015. Available from: http://www.hefce.ac.uk/news/newsarchive/2015/Name,103785,en.html .
• 10. Stern N. Building on Success and Learning from Experience—An Independent Review of the Research Excellence Framework 2016 [cited 2016 05 Aug]. Available from: https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/541338/ind-16-9-ref-stern-review.pdf .
• 11. Matthews D. REF sceptic to lead review into research assessment: Times Higher Education; 2015 [cited 2016 21 Apr]. Available from: https://www.timeshighereducation.com/news/ref-sceptic-lead-review-research-assessment .
• 12. HEFCE. The Metric Tide—Report of the Independent Review of the Role of Metrics in Research Assessment and Management 2015 [cited 2016 11 Aug]. Available from: http://www.hefce.ac.uk/media/HEFCE,2014/Content/Pubs/Independentresearch/2015/The,Metric,Tide/2015_metric_tide.pdf .
• 14. LSE Public Policy Group. Maximizing the impacts of your research: A handbook for social scientists. http://www.lse.ac.uk/government/research/resgroups/LSEPublicPolicy/Docs/LSE_Impact_Handbook_April_2011.pdf . London: LSE; 2011.
• 15. HEFCE. Consultation on the second Research Excellence Framework. 2016.
• 18. Merriam-Webster Dictionary 2017. Available from: https://www.merriam-webster.com/dictionary/methodology .
• 19. Oxford Dictionaries—pathway 2016 [cited 2016 19 June]. Available from: http://www.oxforddictionaries.com/definition/english/pathway .
• 20. Oxford Dictionaries—metric 2016 [cited 2016 15 Sep]. Available from: https://en.oxforddictionaries.com/definition/metric .
• 21. WHO. WHO Ethical and Safety Guidelines for Interviewing Trafficked Women 2003 [cited 2016 29 July]. Available from: http://www.who.int/mip/2003/other_documents/en/Ethical_Safety-GWH.pdf .
• 31. Kalucy L, et al. Primary Health Care Research Impact Project: Final Report Stage 1 Adelaide: Primary Health Care Research & Information Service; 2007 [cited 2016 26 Feb]. Available from: http://www.phcris.org.au/phplib/filedownload.php?file=/elib/lib/downloaded_files/publications/pdfs/phcris_pub_3338.pdf .
• 33. Canadian Academy of Health Sciences. Making an impact—A preferred framework and indicators to measure returns on investment in health research 2009 [cited 2016 26 Feb]. Available from: http://www.cahs-acss.ca/wp-content/uploads/2011/09/ROI_FullReport.pdf .
• 39. HEFCE. RAE 2008—Guidance in submissions 2005 [cited 2016 15 Feb]. Available from: http://www.rae.ac.uk/pubs/2005/03/rae0305.pdf .
• 41. Royal Netherlands Academy of Arts and Sciences. The societal impact of applied health research—Towards a quality assessment system 2002 [cited 2016 29 Feb]. Available from: https://www.knaw.nl/en/news/publications/the-societal-impact-of-applied-health-research/@@download/pdf_file/20021098.pdf .
• 48. Weiss CH. Using social research in public policy making: Lexington Books; 1977.
• 50. Kogan M, Henkel M. Government and research: the Rothschild experiment in a government department: Heinemann Educational Books; 1983.
• 51. Thomas P. The Aims and Outcomes of Social Policy Research. Croom Helm; 1985.
• 52. Bulmer M. Social Science Research and Government: Comparative Essays on Britain and the United States: Cambridge University Press; 2010.
• 53. Booth T. Developing Policy Research. Aldershot, Gower1988.
• 55. Kalucy L, et al Exploring the impact of primary health care research Stage 2 Primary Health Care Research Impact Project Adelaide: Primary Health Care Research & Information Service (PHCRIS); 2009 [cited 2016 26 Feb]. Available from: http://www.phcris.org.au/phplib/filedownload.php?file=/elib/lib/downloaded_files/publications/pdfs/phcris_pub_8108.pdf .
• 56. CHSRF. Canadian Health Services Research Foundation 2000. Health Services Research and Evidence-based Decision Making [cited 2016 February]. Available from: http://www.cfhi-fcass.ca/migrated/pdf/mythbusters/EBDM_e.pdf .
• 58. W.K. Kellogg Foundation. Logic Model Development Guide 2004 [cited 2016 19 July]. Available from: http://www.smartgivers.org/uploads/logicmodelguidepdf.pdf .
• 59. United Way of America. Measuring Program Outcomes: A Practical Approach 1996 [cited 2016 19 July]. Available from: https://www.bttop.org/sites/default/files/public/W.K.%20Kellogg%20LogicModel.pdf .
• 60. Nutley S, Percy-Smith J and Solesbury W. Models of research impact: a cross sector review of literature and practice. London: Learning and Skills Research Centre 2003.
• 61. Spaapen J, van Drooge L. SIAMPI final report [cited 2017 Jan]. Available from: http://www.siampi.eu/Content/SIAMPI_Final%20report.pdf .
• 63. LSHTM. The Health Risks and Consequences of Trafficking in Women and Adolescents—Findings from a European Study 2003 [cited 2016 29 July]. Available from: http://www.oas.org/atip/global%20reports/zimmerman%20tip%20health.pdf .
• 70. Russell G. Response to second HEFCE consultation on the Research Excellence Framework 2009 [cited 2016 04 Apr]. Available from: http://russellgroup.ac.uk/media/5262/ref-consultation-response-final-dec09.pdf .

An official website of the Department of Health and Human Services

Browse Topics

Priority populations.

• Children/Adolescents
• Racial/Ethnic Minorities
• Rural/Inner-City Residents
• Special Healthcare Needs
• Clinicians & Providers
• Data & Measures
• Digital Healthcare Research
• Education & Training
• Hospitals & Health Systems
• Prevention & Chronic Care
• Quality & Patient Safety

Publications & Products

• AHRQ Publishing and Communications Guidelines
• Search Publications

Research Findings & Reports

• Evidence-based Practice Center (EPC) Reports
• Fact Sheets
• Grantee Final Reports: Patient Safety
• Making Healthcare Safer Report
• National Healthcare Quality and Disparities Reports
• Technology Assessment Program
• AHRQ Research Studies

National Healthcare Quality and Disparities Report

Latest available findings on quality of and access to health care

Data & Analytics

• Data Infographics
• Data Visualizations
• Data Innovations
• All-Payer Claims Database
• Consumer Assessment of Healthcare Providers and Systems (CAHPS®) Program
• Healthcare Cost and Utilization Project (HCUP)
• Medical Expenditure Panel Survey (MEPS)
• National Healthcare Quality and Disparities Report Data Tools
• AHRQ Quality Indicator Tools for Data Analytics
• United States Health Information Knowledgebase (USHIK)
• Data Sources Available from AHRQ

Funding & Grants

Notice of funding opportunities, research policies.

• Notice of Funding Opportunity Guidance
• AHRQ Grants Policy Notices
• AHRQ Informed Consent & Authorization Toolkit for Minimal Risk Research
• HHS Grants Policy Statement
• Federal Regulations & Authorities
• Federal Register Notices
• AHRQ Public Access Policy
• Protection of Human Subjects

Funding Priorities

• Special Emphasis Notices
• Staff Contacts

Training & Education Funding

Grant application, review & award process.

• Grant Application Basics
• Application Forms
• Application Deadlines & Important Dates
• AHRQ Tips for Grant Applicants
• Grant Mechanisms & Descriptions
• Application Receipt & Review
• Study Sections for Scientific Peer Review
• Award Process

Post-Award Grant Management

• AHRQ Grantee Profiles
• Getting Recognition for Your AHRQ-Funded Study
• Grants by State
• No-Cost Extensions (NCEs)

AHRQ Grants by State

Searchable database of AHRQ Grants

AHRQ Projects funded by the Patient-Centered Outcomes Research Trust Fund.

• Press Releases
• AHRQ Social Media
• Impact Case Studies
• AHRQ News Now
• AHRQ Research Summit on Diagnostic Safety
• AHRQ Research Summit on Learning Health Systems
• National Advisory Council Meetings
• AHRQ Research Conferences
• AHRQ's 35th Anniversary
• Mission and Budget
• AHRQ's Core Competencies
• National Advisory Council
• Careers at AHRQ
• Maps and Directions
• Other AHRQ Web Sites
• Other HHS Agencies
• Testimonials

Organization & Contacts

• Centers and Offices
• Organization Chart
• Key Contacts

• Health Services and Primary Care Research Study

Measuring Climate Resiliency: A Virtual Roundtable Discussion

• Primer: Reducing Healthcare Carbon Emissions

Health Systems Research

Health services research and primary care research.

AHRQ convened an interactive roundtable meeting of leaders in health systems, public health preparedness, and resiliency, facilitated by the Department of Health and Human Services Office of Climate Change and Health Equity interim director. The roundtable discussion laid the groundwork for advancing work on health system resilience, along with the relationship between operational resilience and operational sustainability.

AHRQ Primer for Reducing Healthcare Carbon Emissions

To support healthcare organizations in advancing toward their decarbonization commitments, AHRQ contracted with the Institute for Healthcare Improvement to develop a primer that offers guidance on high-priority measures and strategies for healthcare organizations to reduce their carbon footprint. Read the report . 

AHRQ Releases Independent RAND Research Study

As mandated by Congress in 2018, AHRQ commissioned the RAND Corporation to conduct an independent assessment of federally funded health services research and primary care research to identify gaps and to propose strategies for better coordination across the Federal enterprise. Read the report, recommendations, and next steps .

Funding Researchers

AHRQ is a major national funder of investigator-initiated health systems research. We fund research to improve healthcare. AHRQ works to ensure that the research we fund is addressing the most pressing issues facing our nation. Our funding priorities and our current funding opportunity announcements give you an idea of where there are critical evidence gaps and where our passions lie.

Training Researchers

We have grants to train health systems researchers , especially those who want experience in learning health systems .

Conducting Intramural Research

We publish our own intramural health systems research that examines current topics in healthcare.

• Medical Expenditure Panel Survey publications .
• Healthcare Cost and Utilization Project reports .
• National Scorecard on Hospital-Acquired Conditions .
• National Quality and Disparities Reports .

Publishing in Journals

Our intramural analysts, grantees, and contractors regularly publish articles in top-tier academic journals. Our website publishes monthly summaries of these articles.

Internet Citation: Health Systems Research. Content last reviewed March 2023. Agency for Healthcare Research and Quality, Rockville, MD. https://www.ahrq.gov/healthsystemsresearch/index.html

• Open access
• Published: 29 August 2023

Healthcare predictive analytics using machine learning and deep learning techniques: a survey

• Mohammed Badawy   ORCID: orcid.org/0000-0001-9494-1386 1 ,
• Nagy Ramadan 1 &
• Hesham Ahmed Hefny 2  

Journal of Electrical Systems and Information Technology volume  10 , Article number:  40 ( 2023 ) Cite this article

14k Accesses

18 Citations

Metrics details

Healthcare prediction has been a significant factor in saving lives in recent years. In the domain of health care, there is a rapid development of intelligent systems for analyzing complicated data relationships and transforming them into real information for use in the prediction process. Consequently, artificial intelligence is rapidly transforming the healthcare industry, and thus comes the role of systems depending on machine learning and deep learning in the creation of steps that diagnose and predict diseases, whether from clinical data or based on images, that provide tremendous clinical support by simulating human perception and can even diagnose diseases that are difficult to detect by human intelligence. Predictive analytics for healthcare a critical imperative in the healthcare industry. It can significantly affect the accuracy of disease prediction, which may lead to saving patients' lives in the case of accurate and timely prediction; on the contrary, in the case of an incorrect prediction, it may endanger patients' lives. Therefore, diseases must be accurately predicted and estimated. Hence, reliable and efficient methods for healthcare predictive analysis are essential. Therefore, this paper aims to present a comprehensive survey of existing machine learning and deep learning approaches utilized in healthcare prediction and identify the inherent obstacles to applying these approaches in the healthcare domain.

Introduction

Each day, human existence evolves, yet the health of each generation either improves or deteriorates. There are always uncertainties in life. Occasionally encounter many individuals with fatal health problems due to the late detection of diseases. Concerning the adult population, chronic liver disease would affect more than 50 million individuals worldwide. However, if the sickness is diagnosed early, it can be stopped. Disease prediction based on machine learning can be utilized to identify common diseases at an earlier stage. Currently, health is a secondary concern, which has led to numerous problems. Many patients cannot afford to see a doctor, and others are extremely busy and on a tight schedule, yet ignoring recurring symptoms for an extended length of time can have significant health repercussions [ 1 ].

Diseases are a global issue; thus, medical specialists and researchers are exerting their utmost efforts to reduce disease-related mortality. In recent years, predictive analytic models has played a pivotal role in the medical profession because of the increasing volume of healthcare data from a wide range of disparate and incompatible data sources. Nonetheless, processing, storing, and analyzing the massive amount of historical data and the constant inflow of streaming data created by healthcare services has become an unprecedented challenge utilizing traditional database storage [ 2 , 3 , 4 ]. A medical diagnosis is a form of problem-solving and a crucial and significant issue in the real world. Illness diagnosis is the process of translating observational evidence into disease names. The evidence comprises data received from evaluating a patient and substances generated from the patient; illnesses are conceptual medical entities that detect anomalies in the observed evidence [ 5 ].

Healthcare is the collective effort of society to ensure, provide, finance, and promote health. In the twentieth century, there was a significant shift toward the ideal of wellness and the prevention of sickness and incapacity. The delivery of healthcare services entails organized public or private efforts to aid persons in regaining health and preventing disease and impairment [ 6 ]. Health care can be described as standardized rules that help evaluate actions or situations that affect decision-making [ 7 ]. Healthcare is a multi-dimensional system. The basic goal of health care is to diagnose and treat illnesses or disabilities. A healthcare system’s key components are health experts (physicians or nurses), health facilities (clinics and hospitals that provide medications and other diagnostic services), and a funding institution to support the first two [ 8 ].

With the introduction of systems based on computers, the digitalization of all medical records and the evaluation of clinical data in healthcare systems have become widespread routine practices. The phrase "electronic health records" was chosen by the Institute of Medicine, a division of the National Academies of Sciences, Engineering, and Medicine, in 2003 to define the records that continued to enhance the healthcare sector for the benefit of both patients and physicians. Electronic Health Records (EHR) are "computerized medical records for patients that include all information in an individual's past, present, or future that occurs in an electronic system used to capture, store, retrieve, and link data primarily to offer healthcare and health-related services," according to Murphy, Hanken, and Waters [ 8 ].

Daily, healthcare services produce an enormous amount of data, making it increasingly complicated to analyze and handle it in "conventional ways." Using machine learning and deep learning, this data may be properly analyzed to generate actionable insights. In addition, genomics, medical data, social media data, environmental data, and other data sources can be used to supplement healthcare data. Figure  1 provides a visual picture of these data sources. The four key healthcare applications that can benefit from machine learning are prognosis, diagnosis, therapy, and clinical workflow, as outlined in the following section [ 9 ].

Illustration of heterogeneous sources contributing to healthcare data [ 9 ]

The long-term investment in developing novel technologies based on machine learning as well as deep learning techniques to improve the health of individuals via the prediction of future events reflects the increased interest in predictive analytics techniques to enhance healthcare. Clinical predictive models, as they have been formerly referred to, assisted in the diagnosis of people with an increased probability of disease. These prediction algorithms are utilized to make clinical treatment decisions and counsel patients based on some patient characteristics [ 10 ].

The concept of medical care is used to stress the organization and administration of curative care, which is a subset of health care. The ecology of medical care was first introduced by White in 1961. White also proposed a framework for perceiving patterns of health concerning symptoms experienced by populations of interest, along with individuals’ choices in getting medical treatment. In this framework, it is possible to calculate the proportion of the population that used medical services over a specific period of time. The "ecology of medical care" theory has become widely accepted in academic circles over the past few decades [ 6 ].

Medical personnel usually face new problems, changing tasks, and frequent interruptions because of the system's dynamism and scalability. This variability often makes disease recognition a secondary concern for medical experts. Moreover, the clinical interpretation of medical data is a challenging task from an epistemological point of view. This not only applies to professionals with extensive experience but also to representatives, such as young physician assistants, with varied or little experience [ 11 ]. The limited time available to medical personnel, the speedy progression of diseases, and the fluctuating patient dynamics make diagnosis a particularly complex process. However, a precise method of diagnosis is critical to ensuring speedy treatment and, thus, patient safety [ 12 ].

Predictive analytics for health care are critical industry requirements. It can have a significant impact on the accuracy of disease prediction, which can save patients' lives in the case of an accurate and timely prediction but can also endanger patients' lives in the case of an incorrect prediction. Diseases must therefore be accurately predicted and estimated. As a result, dependable and efficient methods for healthcare predictive analysis are required.

The purpose of this paper is to present a comprehensive review of common machine learning and deep learning techniques that are utilized in healthcare prediction, in addition to identifying the inherent obstacles that are associated with applying these approaches in the healthcare domain.

The rest of the paper is organized as follows: Section  " Background " gives a theoretical background on artificial intelligence, machine learning, and deep learning techniques. Section  " Disease prediction with analytics " outlines the survey methodology and presents a literature review of machine learning as well as deep learning approaches employed in healthcare prediction. Section  " Results and Discussion " gives a discussion of the results of previous works related to healthcare prediction. Section  " Challenges " covers the existing challenges related to the topic of this survey. Finally, Section  " Conclusion " concludes the paper.

The extensive research and development of cutting-edge tools based on machine learning and deep learning for predicting individual health outcomes demonstrate the increased interest in predictive analytics techniques to improve health care. Clinical predictive models assisted physicians in better identifying and treating patients who were at a higher risk of developing a serious illness. Based on a variety of factors unique to each individual patient, these prediction algorithms are used to advise patients and guide clinical practice.

Artificial intelligence (AI) is the ability of a system to interpret data, and it makes use of computers and machines to improve humans' capacity for decision-making, problem-solving, and technological innovation [ 13 ]. Figure  2 depicts machine learning and deep learning as subsets of AI.

AI, ML, and DL

Machine learning

Machine learning (ML) is a subfield of AI that aims to develop predictive algorithms based on the idea that machines should have the capability to access data and learn on their own [ 14 ]. ML utilizes algorithms, methods, and processes to detect basic correlations within data and create descriptive and predictive tools that process those correlations. ML is usually associated with data mining, pattern recognition, and deep learning. Although there are no clear boundaries between these areas and they often overlap, it is generally accepted that deep learning is a relatively new subfield of ML that uses extensive computational algorithms and large amounts of data to define complex relationships within data. As shown in Fig.  3 , ML algorithms can be divided into three categories: supervised learning, unsupervised learning, and reinforcement learning [ 15 ].

Different types of machine learning algorithms

Supervised learning

Supervised learning is an ML model for investigating the input–output correlation information of a system depending on a given set of training examples that are paired between the inputs and the outputs [ 16 ]. The model is trained with a labeled dataset. It matches how a student learns fundamental math from a teacher. This kind of learning requires labeled data with predicted correct answers based on algorithm output [ 17 ]. The most widely used supervised learning-based techniques include linear regression, logistic regression, decision trees, random forests, support vector machines, K-nearest neighbor, and naive Bayes.

A. Linear regression

Linear regression is a statistical method commonly used in predictive investigations. It succeeds in forecasting the dependent, output, variable (Y) based on the independent, input, variable (X). The connection between X and Y is represented as shown in Eq.  1 assuming continuous, real, and numeric parameters.

where m indicates the slope and c indicates the intercept. According to Eq.  1 , the association between the independent parameters (X) and the dependent parameters (Y) can be inferred [ 18 ].

The advantage of linear regression is that it is straightforward to learn and easy to-eliminate overfitting through regularization. One drawback of linear regression is that it is not convenient when applied to nonlinear relationships. However, it is not recommended for most practical applications as it greatly simplifies real-world problems [ 19 ]. The implementation tools utilized in linear regression are Python, R, MATLAB, and Excel.

As shown in Fig.  4 , observations are highlighted in red, and random deviations' result (shown in green) from the basic relationship (shown in yellow) between the independent variable (x) and the dependent variable (y) [ 20 ].

Linear regression model

B. Logistic regression

Logistic regression, also known as the logistic model, investigates the correlation between many independent variables and a categorical dependent variable and calculates the probability of an event by fitting the data to a logistic curve [ 21 ]. Discrete mean values must be binary, i.e., have only two outcomes: true or false, 0 or 1, yes or no, or either superscript or subscript. In logistic regression, categorical variables need to be predicted and classification problems should be solved. Logistic regression can be implemented using various tools such as R, Python, Java, and MATLAB [ 18 ]. Logistic regression has many benefits; for example, it shows the linear relationship between dependent and independent variables with the best results. It is also simple to understand. On the other hand, it can only predict numerical output, is not relevant to nonlinear data, and is sensitive to outliers [ 22 ].

C. Decision tree

The decision tree (DT) is the supervised learning technique used for classification. It combines the values of attributes based on their order, either ascending or descending [ 23 ]. As a tree-based strategy, DT defines each path starting from the root using a data-separating sequence until a Boolean conclusion is attained at the leaf node [ 24 , 25 ]. DT is a hierarchical representation of knowledge interactions that contains nodes and links. When relations are employed to classify, nodes reflect purposes [ 26 , 27 ]. An example of DT is presented in Fig.  5 .

Example of a DT

DTs have various drawbacks, such as increased complexity with increasing nomenclature, small modifications that may lead to a different architecture, and more processing time to train data [ 18 ]. The implementation tools used in DT are Python (Scikit-Learn), RStudio, Orange, KNIME, and Weka [ 22 ].

D. Random forest

Random forest (RF) is a basic technique that produces correct results most of the time. It may be utilized for classification and regression. The program produces an ensemble of DTs and blends them [ 28 ].

In the RF classifier, the higher the number of trees in the forest, the more accurate the results. So, the RF has generated a collection of DTs called the forest and combined them to achieve more accurate prediction results. In RF, each DT is built only on a part of the given dataset and trained on approximations. The RF brings together several DTs to reach the optimal decision [ 18 ].

As indicated in Fig.  6 , RF randomly selects a subset of features from the data, and from each subset it generates n random trees [ 20 ]. RF will combine the results from all DTs and provide them in the final output.

Random forest architecture

Two parameters are being used for tuning RF models: mtry —the count of randomly selected features to be considered in each division; and ntree —the model trees count. The mtry parameter has a trade-off: Large values raise the correlation between trees, but enhance the per-tree accuracy [ 29 ].

The RF works with a labeled dataset to do predictions and build a model. The final model is utilized to classify unlabeled data. The model integrates the concept of bagging with a random selection of traits to build variance-controlled DTs [ 30 ].

RF offers significant benefits. First, it can be utilized for determining the relevance of the variables in a regression and classification task [ 31 , 32 ]. This relevance is measured on a scale, based on the impurity drop at each node used for data segmentation [ 33 ]. Second, it automates missing values contained in the data and resolves the overfitting problem of DT. Finally, RF can efficiently handle huge datasets. On the other side, RF suffers from drawbacks; for example, it needs more computing and resources to generate the output results and it requires training effort due to the multiple DTs involved in it. The implementation tools used in RF are Python Scikit-Learn and R [ 18 ].

E. Support vector machine

The supervised ML technique for classification issues and regression models is called the support vector machine (SVM). SVM is a linear model that offers solutions to issues that are both linear and nonlinear. as shown in Fig.  7 . Its foundation is the idea of margin calculation. The dataset is divided into several groups to build relations between them [ 18 ].

Support vector machine

SVM is a statistics-based learning method that follows the principle of structural risk minimization and aims to locate decision bounds, also known as hyperplanes, that can optimally separate classes by finding a hyperplane in a usable N-dimensional space that explicitly classifies data points [ 34 , 35 , 36 ]. SVM indicates the decision boundary between two classes by defining the value of each data point, in particular the support vector points placed on the boundary between the respective classes [ 37 ].

SVM has several advantages; for example, it works perfectly with both semi-structured and unstructured data. The kernel trick is a strong point of SVM. Moreover, it can handle any complex problem with the right functionality and can also handle high-dimensional data. Furthermore, SVM generalization has less allocation risk. On the other hand, SVM has many downsides. The model training time is increased on a large dataset. Choosing the right kernel function is also a difficult process. In addition, it is not working well with noisy data. Implementation tools used in SVM include SVMlight with C, LibSVM with Python, MATLAB or Ruby, SAS, Kernlab, Scikit-Learn, and Weka [ 22 ].

F. K-nearest neighbor

K-nearest neighbor (KNN) is an "instance-based learning" or non-generalized learning algorithm, which is often known as a “lazy learning” algorithm [ 38 ]. KNN is used for solving classification problems. To anticipate the target label of the novel test data, KNN determines the distance of the nearest training data class labels with a new test data point in the existence of a K value, as shown in Fig.  8 . It then calculates the number of nearest data points using the K value and terminates the label of the new test data class. To determine the number of nearest-distance training data points, KNN usually sets the value of K according to (1): k  =  n ^(1/2), where n is the size of the dataset [ 22 ].

K-nearest neighbor

KNN has many benefits; for example, it is sufficiently powerful if the size of the training data is large. It is also simple and flexible, with attributes and distance functions. Moreover, it can handle multi-class datasets. KNN has many drawbacks, such as the difficulty of choosing the appropriate K value, it being very tedious to choose the distance function type for a particular dataset, and the computation cost being a little high due to the distance between all the training data points, the implementation tools used in KNN are Python (Scikit-Learn), WEKA, R, KNIME, and Orange [ 22 ].

G. Naive Bayes

Naive Bayes (NB) focuses on the probabilistic model of Bayes' theorem and is simple to set up as the complex recursive parameter estimation is basically none, making it suitable for huge datasets [ 39 ]. NB determines the class membership degree based on a given class designation [ 40 ]. It scans the data once, and thus, classification is easy [ 41 ]. Simply, the NB classifier assumes that there is no relation between the presence of a particular feature in a class and the presence of any other characteristic. It is mainly targeted at the text classification industry [ 42 ].

NB has great benefits such as ease of implementation, can provide a good result even using fewer training data, can manage both continuous and discrete data, and is ideal to solve the prediction of multi-class problems, and the irrelevant feature does not affect the prediction. NB, on the other hand, has the following drawbacks: It assumes that all features are independent which is not always viable in real-world problems, suffers from zero frequency problems, and the prediction of NB is not usually accurate. Implementation tools are WEKA, Python, RStudio, and Mahout [ 22 ].

To summarize the previously discussed models, Table 1 demonstrates the advantages and disadvantages of each model.

Unsupervised learning

Unlike supervised learning, there are no correct answers and no teachers in unsupervised learning [ 42 ]. It follows the concept that a machine can learn to understand complex processes and patterns on its own without external guidance. This approach is particularly useful in cases where experts have no knowledge of what to look for in the data and the data itself do not include the objectives. The machine predicts the outcome based on past experiences and learns to predict the real-valued outcome from the information previously provided, as shown in Fig.  9 .

Workflow of unsupervised learning [ 23 ]

Unsupervised learning is widely used in the processing of multimedia content, as clustering and partitioning of data in the lack of class labels is often a requirement [ 43 ]. Some of the most popular unsupervised learning-based approaches are k-means, principal component analysis (PCA), and apriori algorithm.

The k-means algorithm is the common portioning method [ 44 ] and one of the most popular unsupervised learning algorithms that deal with the well-known clustering problem. The procedure classifies a particular dataset by a certain number of preselected (assuming k -sets) clusters [ 45 ]. The pseudocode of the K-means algorithm is shown in Pseudocode 1.

K means has several benefits such as being more computationally efficient than hierarchical grouping in case of large variables. It provides more compact clusters than hierarchical ones when a small k is used. Also, the ease of implementation and comprehension of assembly results is another benefit. However, K -means have disadvantages such as the difficulty of predicting the value of K . Also, as different starting sections lead to various final combinations, the performance is affected. It is accurate for raw points and local optimization, and there is no single solution for a given K value—so the average of the K value must be run multiple times (20–100 times) and then pick the results with the minimum J [ 19 ].

B. Principal component analysis

In modern data analysis, principal component analysis (PCA) is an essential tool as it provides a guide for extracting the most important information from a dataset, compressing the data size by keeping only those important features without losing much information, and simplifying the description of a dataset [ 46 , 47 ].

PCA is frequently used to reduce data dimensions before applying classification models. Moreover, unsupervised methods, such as dimensionality reduction or clustering algorithms, are commonly used for data visualizations, detection of common trends or behaviors, and decreasing the data quantity to name a few only [ 48 ].

PCA converts the 2D data into 1D data. This is done by changing the set of variables into new variables known as principal components (PC) which are orthogonal [ 23 ]. In PCA, data dimensions are reduced to make calculations faster and easier. To illustrate how PCA works, let us consider an example of 2D data. When these data are plotted on a graph, it will take two axes. Applying PCA, the data turn into 1D. This process is illustrated in Fig.  10 [ 49 ].

Visualization of data before and after applying PCA [ 49 ]

Apriori algorithm is considered an important algorithm, which was first introduced by R. Agrawal and R. Srikant, and published in [ 50 , 51 ].

The principle of the apriori algorithm is to represent the filter generation strategy. It creates a filter element set ( k  + 1) based on the repeated k element groups. Apriori uses an iterative strategy called planar search, where k item sets are employed to explore ( k  + 1) item sets. First, the set of repeating 1 item is produced by scanning the dataset to collect the number for each item and then collecting items that meet the minimum support. The resulting group is called L1. Then L1 is used to find L2, the recursive set of two elements is used to find L3, and so on until no repeated k element groups are found. Finding every Lk needs a full dataset scan. To improve production efficiency at the level-wise of repeated element groups, a key property called the apriori property is used to reduce the search space. Apriori property states that all non-empty subsets of a recursive element group must be iterative. A two-step technique is used to identify groups of common elements: join and prune activities [ 52 ].

Although it is simple, the apriori algorithm suffers from several drawbacks. The main limitation is the costly wasted time to contain many candidates sets with a lot of redundant item sets. It also suffers from low minimum support or large item sets, and multiple rounds of data are needed for data mining which usually results in irrelevant items, in addition to difficulties in discovering individual elements of events [ 53 , 54 ].

To summarize the previously discussed models, Table 2 demonstrates the advantages and disadvantages of each model.

Reinforcement learning

Reinforcement learning (RL) is different from supervised learning and unsupervised learning. It is a goal-oriented learning approach. RL is closely related to an agent (controller) that takes responsibility for the learning process to achieve a goal. The agent chooses actions, and as a result, the environment changes its state and returns rewards. Positive or negative numerical values are used as rewards. An agent's goal is to maximize the rewards accumulated over time. A job is a complete environment specification that identifies how to generate rewards [ 55 ]. Some of the most popular reinforcement learning-based algorithms are the Q-learning algorithm and the Monte Carlo tree search (MCTS).

A. Q-learning

Q-learning is a type of model-free RL. It can be considered an asynchronous dynamic programming approach. It enables agents to learn how to operate optimally in Markovian domains by exploring the effects of actions, without the need to generate domain maps [ 56 ]. It represented an incremental method of dynamic programming that imposed low computing requirements. It works through the successive improvement of the assessment of individual activity quality in particular states [ 57 ].

In information theory, Q-learning is strongly employed, and other related investigations are underway. Recently, Q-learning combined with information theory has been employed in different disciplines such as natural language processing (NLP), pattern recognition, anomaly detection, and image classification [ 58 , 59 , 60 , 60 ]. Moreover, a framework has been created to provide a satisfying response based on the user’s utterance using RL in a voice interaction system [ 61 ]. Furthermore, a high-resolution deep learning-based prediction system for local rainfall has been constructed [ 62 ].

The advantage of developmental Q-learning is that it is possible to identify the reward value effectively on a given multi-agent environment method as agents in ant Q-learning are interacting with each other. The problem with Q-learning is that its output can be stuck in the local minimum as agents just take the shortest path [ 63 ].

B. Monte Carlo tree search

Monte Carlo tree search (MCTS) is an effective technique for solving sequential selection problems. Its strategy is based on a smart tree search that balances exploration and exploitation. MCTS presents random samples in the form of simulations and keeps activity statistics for better educated choices in each future iteration. MCTS is a decision-making algorithm that is employed in searching tree-like huge complex regions. In such trees, each node refers to a state, which is also referred to as problem configuration, while edges represent transitions from one state to another [ 64 ].

The MCTS is related directly to cases that can be represented by a Markov decision process (MDP), which is a type of discrete-time random control process. Some modifications of the MCTS make it possible to apply it to partially observable Markov decision processes (POMDP) [ 65 ]. Recently, MCTS coupled with deep RL became the base of AlphaGo developed by Google DeepMind and documented in [ 66 ]. The basic MCTS method is conceptually simple, as shown in Fig.  11 .

Basic MCTS process

Tree 1 is constructed progressively and unevenly. The tree policy is utilized to get the critical node of the current tree for each iteration of the method. The tree strategy seeks to strike a balance between exploration and exploitation concerns. Then, from the specified node, simulation 2 is run, and the search tree is then updated according to the obtained results. This comprises adding a child node that matches the specified node's activity and updating its ancestor's statistics. During this simulation, movements are performed based on some default policy, which in its simplest case is to make uniform random movements. The benefit of MCTS is that there is no need to evaluate the values of the intermediate state, which significantly minimizes the amount of required knowledge in the field [ 67 ].

To summarize the previously discussed models, Table 3 demonstrates the advantages and disadvantages of each model.

Deep learning

Over the past decades, ML has had a significant impact on our daily lives with examples including efficient computer vision, web search, and recognition of optical characters. In addition, by applying ML approaches, AI at the human level has also been improved [ 68 , 69 , 70 ]. However, when it comes to the mechanisms of human information processing (such as sound and vision), the performance of traditional ML algorithms is far from satisfactory. The idea of deep learning (DL) was formed in the late 20th inspired by the deep hierarchical structures of human voice recognition and production systems. DL breaks have been introduced in 2006 when Hinton built a deep-structured learning architecture called deep belief network (DBN) [ 71 ].

The performance of classifiers using DL has been extensively improved with the increased complexity of data compared to classical learning methods. Figure  12 shows the performance of classic ML algorithms and DL methods [ 72 ]. The performance of typical ML algorithms becomes stable when they reach the training data threshold, but DL improves its performance as the complexity of data increases [ 73 ].

Performance of deep learning concerning the complexity of data

DL (deep ML, or deep-structured learning) is a subset of ML that involves a collection of algorithms attempting to represent high-level abstractions for data through a model that has complicated structures or is otherwise, composed of numerous nonlinear transformations. The most important characteristic of DL is the depth of the network. Another essential aspect of DL is the ability to replace handcrafted features generated by efficient algorithms for unsupervised or semi-supervised feature learning and hierarchical feature extraction [ 74 ].

DL has significantly advanced the latest technologies in a variety of applications, including machine translation, speech, and visual object recognition, NLP, and text automation, using multilayer artificial neural networks (ANNs) [ 15 ].

Different DL designs in the past two decades give enormous potential for employment in various sectors such as automatic voice recognition, computer vision, NLP, and bioinformatics. This section discusses the most common architectures of DL such as convolutional neural networks (CNNs), long short-term memory (LSTM), and recurrent convolution neural networks (RCNNs) [ 75 ].

A. Convolutional neural network

CNNs are special types of neural networks inspired by the human visual cortex and used in computer vision. It is an automatic feed-forward neural network in which information transfers exclusively in the forward direction [ 76 ]. CNN is frequently applied in face recognition, human organ localization, text analysis, and biological image recognition [ 77 ].

Since CNN was first created in 1989, it has done well in disease diagnosis over the past three decades [ 78 ]. Figure  13 depicts the general architecture of a CNN composed of feature extractors and a classifier. Each layer of the network accepts the output of the previous layer as input and passes it on to the next layer in feature extraction layers. A typical CNN architecture consists of three types of layers: convolution, pooling, and classification. There are two types of layers at the network's low and middle levels: convolutional layers and pooling layers. Even-numbered layers are used for convolutions, while odd-numbered layers are used for pooling operations. The convolution and pooling layers' output nodes are categorized in a two-dimensional plane called feature mapping. Each layer level is typically generated by combining one or more previous layers [ 79 ].

Architecture of CNN [ 79 ]

CNN has a lot of benefits, including a human optical processing system, greatly improved 2D and 3D image processing structure, and is effective in learning and extracting abstract information from 2D information. The max-pooling layer in CNN is efficient in absorbing shape anisotropy. Furthermore, they are constructed from sparse connections with paired weights and contain far fewer parameters than a fully connected network of equal size. CNNs are trained using a gradient-based learning algorithm and are less susceptible to the diminishing gradient problem because the gradient-based approach trains the entire network to directly reduce the error criterion, allowing CNNs to provide highly optimized weights [ 79 ].

B. Long short-term memory

LSTM is a special type of recurrent neural network (RNN) with internal memory and multiplicative gates. Since the original LSTM introduction in 1997 by Sepp Hochrieiter and Jürgen Schmidhuber, a variety of LSTM cell configurations have been described [ 80 ].

LSTM has contributed to the development of well-known software such as Alexa, Siri, Cortana, Google Translate, and Google voice assistant [ 81 ]. LSTM is an implementation of RNN with a special connection between nodes. The special components within the LSTM unit include the input, output, and forget gates. Figure  14 depicts a single LSTM cell.

LSTM unit [ 82 ]

x t  = Input vector at the time t.

h t-1  = Previous hidden state.

c t-1  = Previous memory state.

h t  = Current hidden state.

c t  = Current memory state.

[ x ] = Multiplication operation.

[+] = Addition operation.

LSTM is an RNN module that handles gradient loss problems. In general, RNN uses LSTM to eliminate propagation errors. This allows the RNN to learn over multiple time steps. LSTM is characterized by cells that hold information outside the recurring network. This cell enables the RNN to learn over many time steps. The basic principle of LSTMs is the state of the cell, which contains information outside the recurrent network. A cell is like a memory in a computer, which decides when data should be stored, written, read, or erased via the LSTM gateway [ 82 ]. Many network architectures use LSTM such as bidirectional LSTM, hierarchical and attention-based LSTM, convolutional LSTM, autoencoder LSTM, grid LSTM, cross-modal, and associative LSTM [ 83 ].

Bidirectional LSTM networks move the state vector forward and backward in both directions. This implies that dependencies must be considered in both temporal directions. As a result of inverse state propagation, the expected future correlations can be included in the network's current output [ 84 ]. Bidirectional LSTM investigates and analyzes this because it encapsulates spatially and temporally scattered information and can tolerate incomplete inputs via a flexible cell state vector propagation communication mechanism. Based on the detected gaps in data, this filtering mechanism reidentifies the connections between cells for each data sequence. Figure  15 depicts the architecture. A bidirectional network is used in this study to process properties from multiple dimensions into a parallel and integrated architecture [ 83 ].

(left) Bidirectional LSTM and (right) filter mechanism for processing incomplete data [ 84 ]

Hierarchical LSTM networks solve multi-dimensional problems by breaking them down into subproblems and organizing them in a hierarchical structure. This has the advantage of focusing on a single or multiple subproblems. This is accomplished by adjusting the weights within the network to generate a certain level of interest [ 83 ]. A weighting-based attention mechanism that analyzes and filters input sequences is also used in hierarchical LSTM networks for long-term dependency prediction [ 85 ].

Convolutional LSTM reduces and filters input data collected over a longer period using convolutional operations applied in LSTM networks or the LSTM cell architecture directly. Furthermore, due to their distinct characteristics, convolutional LSTM networks are useful for modeling many quantities such as spatially and temporally distributed relationships. However, many quantities can be expected collectively in terms of reduced feature representation. Decoding or decoherence layers are required to predict different output quantities not as features but based on their parent units [ 83 ].

The LSTM autoencoder solves the problem of predicting high-dimensional parameters by shrinking and expanding the network [ 86 ]. The autoencoder architecture is separately trained with the aim of accurate reconstruction of the input data as reported in [ 87 ]. Only the encoder is used during testing and commissioning to extract the low-dimensional properties that are transmitted to the LSTM. The LSTM was extended to multimodal prediction using this strategy. To compress the input data and cell states, the encoder and decoder are directly integrated into the LSTM cell architecture. This combined reduction improves the flow of information in the cell and results in an improved cell state update mechanism for both short-term and long-term dependency [ 83 ].

Grid long short-term memory is a network of LSTM cells organized into a multi-dimensional grid that can be applied to sequences, vectors, or higher-dimensional data like images [ 88 ]. Grid LSTM has connections to the spatial or temporal dimensions of input sequences. Thus, connections of different dimensions within cells extend the normal flow of information. As a result, grid LSTM is appropriate for the parallel prediction of several output quantities that may be independent, linear, or nonlinear. The network's dimensions and structure are influenced by the nature of the input data and the goal of the prediction [ 89 ].

A novel method for the collaborative prediction of numerous quantities is the cross-modal and associative LSTM. It uses several standard LSTMs to separately model different quantities. To calculate the dependencies of the quantities, these LSTM streams communicate with one another via recursive connections. The chosen layers' outputs are added as new inputs to the layers before and after them in other streams. Consequently, a multimodal forecast can be made. The benefit of this approach is that the correlation vectors that are produced have the same dimensions as the input vectors. As a result, neither the parameter space nor the computation time increases [ 90 ].

C. Recurrent convolution neural network

CNN is a key method for handling various computer vision challenges. In recent years, a new generation of CNNs has been developed, the recurrent convolution neural network (RCNN), which is inspired by large-scale recurrent connections in the visual systems of animals. The recurrent convolutional layer (RCL) is the main feature of RCNN, which integrates repetitive connections among neurons in the normal convolutional layer. With the increase in the number of repetitive computations, the receptive domains (RFs) of neurons in the RCL expand infinitely, which is contrary to biological facts [ 91 ].

The RCNN prototype was proposed by Ming Liang and Xiaolin Hu [ 92 , 93 ], and the structure is illustrated in Fig.  16 , in which both forward and redundant connections have local connectivity and weights shared between distinct sites. This design is quite like the recurrent multilayer perceptron (RMLP) concept which is often used for dynamic control [ 94 , 95 ] (Fig.  17 , middle). Like the distinction between MLP and CNN, the primary distinction is that in RMLP, common local connections are used in place of full connections. For this reason, the proposed model is known as RCNN [ 96 ].

Illustration of the architectures of CNN, RMLP, and RCNN [ 85 ]

Illustration of the total number of reviewed papers

The main unit of RCNN is the RCL. RCLs develop through discrete time steps. RCNN offers three basic advantages. First, it allows each unit to accommodate background information in an arbitrarily wide area in the current layer. Second, recursive connections improve the depth of the network while keeping the number of mutable parameters constant through weight sharing. This is consistent with the trend of modern CNN architecture to grow deeper with a relatively limited number of parameters. The third aspect of RCNN is the time exposed in RCNN which is a CNN with many paths between the input layer and the output layer, which makes learning simple. On one hand, having longer paths makes it possible for the model to learn very complex features. On the other hand, having shorter paths may improve the inverse gradient during training [ 91 ].

To summarize the previously discussed models, Table 4 demonstrates the advantages and disadvantages of each model.

Disease prediction with analytics

The studies discussed in this paper have been presented and published in high-quality journals and international conferences published by IEEE, Springer, and Elsevier, and other major scientific publishers such as Hindawi, Frontiers, Taylor, and MDPI. The search engines used are Google Scholar, Scopus, and Science Direct. All papers selected covered the period from 2019 to 2022. Machine learning, deep learning, health care, surgery, cardiology, radiology, hepatology, and nephrology are some of the terms used to search for these studies. The studies chosen for this survey are concerned with the use of machine learning as well as deep learning algorithms in healthcare prediction. For this survey, empirical and review articles on the topics were considered. This section discusses existing research efforts that healthcare prediction using various techniques in ML and DL. This survey gives a detailed discussion about the methods and algorithms which are used for predictions, performance metrics, and tools of their model.

ML-based healthcare prediction

To predict diabetes patients, the authors of [ 97 ] utilized a framework to develop and evaluate ML classification models like logistic regression, KNN, SVM, and RF. ML method was implemented on the Pima Indian Diabetes Database (PIDD) which has 768 rows and 9 columns. The forecast accuracy delivers 83%. Results of the implementation approach indicate how the logistic regression outperformed other algorithms of ML, in addition only a structured dataset was selected but unstructured data are not considered, also model should be implemented in other healthcare domains like heart disease, and COVID-19, finally other factors should be considered for diabetes prediction, like family history of diabetes, smoking habits, and physical inactivity.

The authors created a diagnosis system in [ 98 ] that uses two different datasets (Frankfurt Hospital in Germany and PIDD provided by the UCI ML repository) and four prediction models (RF, SVM, NB, and DT) to predict diabetes. the SVM algorithm performed with an accuracy of 83.1 percent. There are some aspects of this study that need to be improved; such as, using a DL approach to predict diabetes may lead to achieving better results; furthermore, the model should be tested in other healthcare domains such as heart disease and COVID-19 prediction datasets.

In [ 99 ], the authors proposed three ML methods (logistic regression, DT, and boosted RF) to assess COVID-19 using OpenData Resources from Mexico and Brazil. To predict rescue and death, the proposed model incorporates just the COVID-19 patient's geographical, social, and economic conditions, as well as clinical risk factors, medical reports, and demographic data. On the dataset utilized, the model for Mexico has a 93 percent accuracy, and an F1 score is 0.79. On the other hand, on the used dataset, the Brazil model has a 69 percent accuracy and an F1 score is 0.75. The three ML algorithms have been examined and the acquired results showed that logistic regression is the best way of processing data. The authors should be concerned about the usage of authentication and privacy management of the created data.

A new model for predicting type 2 diabetes using a network approach and ML techniques was presented by the authors in [ 100 ] (logistic regression, SVM, NB, KNN, decision tree, RF, XGBoost, and ANN). To predict the risk of type 2 diabetes, the healthcare data of 1,028 type 2 diabetes patients and 1,028 non-type 2 diabetes patients were extracted from de-identified data. The experimental findings reveal the models’ effectiveness with an area under curve (AUC) varied from 0.79 to 0.91. The RF model achieved higher accuracy than others. This study relies only on the dataset providing hospital admission and discharge summaries from one insurance company. External hospital visits and information from other insurance companies are missing for people with many insurance providers.

The authors of [ 101 ] proposed a healthcare management system that can be used by patients to schedule appointments with doctors and verify prescriptions. It gives support for ML to detect ailments and determine medicines. ML models including DT, RF, logistic regression, and NB classifiers are applied to the datasets of diabetes, heart disease, chronic kidney disease, and liver. The results showed that among all the other models, logistic regression had the highest accuracy of 98.5 percent in the heart dataset. while the least accuracy is of the DT classifier which came out to be 92 percent. In the liver dataset the logistic regression with maximum accuracy of 75.17% among all others. In the chronic renal disease dataset, the logistic regression, RF, and Gaussian NB, all performed well with an accuracy of 1, the accuracy of 100% should be verified by using k-fold cross-validation to test the reliability of the models. In the diabetes dataset random forest with maximum accuracy of 83.67 percent. The authors should include a hospital directory as then various hospitals and clinics can be accessed through a single portal. Additionally, image datasets could be included to allow image processing of reports and the deployment of DL to detect diseases.

In [ 102 ], the authors developed an ML model to predict the occurrence of Type 2 Diabetes in the following year (Y + 1) using factors in the present year (Y). Between 2013 and 2018, the dataset was obtained as an electronic health record from a private medical institute. The authors applied logistic regression, RF, SVM, XGBoost, and ensemble ML algorithms to predict the outcome of non-diabetic, prediabetes, and diabetes. Feature selection was applied to choose the three classes efficiently. FPG, HbA1c, triglycerides, BMI, gamma-GTP, gender, age, uric acid, smoking, drinking, physical activity, and family history were among the features selected. According to the experimental results, the maximum accuracy was 73% from RF, while the lowest was 71% from the logistic regression model. The authors presented a model that used only one dataset. As a result, additional data sources should be applied to verify the models developed in this study.

The authors of [ 103 ] classified the diabetes dataset using SVM and NB algorithms with feature selection to improve the model's accuracy. PIDD is taken from the UCI Repository for analysis. For training and testing purposes the authors employed the k-fold cross-validation model, the SVM classifier was performing better than the NB method it offers around 91% correct predictions; however, the authors acknowledge that they need to extend to the latest dataset that will contain additional attributes and rows.

K-means clustering is an unsupervised ML algorithm that was introduced by the authors of [ 104 ] for the purpose of detecting heart disease in its earliest stages using the UCI heart disease dataset. PCA is used for dimensionality reduction. The outcome of the method demonstrates early cardiac disease prediction with 94.06% accuracy. The authors should apply the proposed technique using more than one algorithm and use more than one dataset.

In [ 105 ], the authors constructed a predictive model for the classification of diabetes data using the logistic regression classification technique. The dataset includes 459 patients for training data and 128 cases for testing data. The prediction accuracy using logistic regression was obtained at 92%. The main limitation of this research is that the authors have not compared the model with other diabetes prediction algorithms, so it cannot be confirmed.

The authors of [ 106 ] developed a prediction model that analyzes the user's symptoms and predicts the disease using ML algorithms (DT classifier, RF classifier, and NB classifier). The purpose of this study was to solve health-related problems by allowing medical professionals to predict diseases at an early stage. The dataset is a sample of 4920 patient records with 41 illnesses diagnosed. A total of 41 disorders were included as a dependent variable. All algorithms achieved the same accuracy score of 95.12%. The authors noticed that overfitting occurred when all 132 symptoms from the original dataset were assessed instead of 95 symptoms. That is, the tree appears to remember the dataset provided and thus fails to classify new data. As a result, just 95 symptoms were assessed during the data-cleansing process, with the best ones being chosen.

In [ 107 ], the authors built a decision-making system that assists practitioners to anticipate cardiac problems in exact classification through a simpler method and will deliver automated predictions about the condition of the patient’s heart. implemented 4 algorithms (KNN, RF, DT, and NB), all these algorithms were used in the Cleveland Heart Disease dataset. The accuracy varies for different classification methods. The maximum accuracy is given when they utilized the KNN algorithm with the Correlation factor which is almost 94 percent. The authors should extend the presented technique to leverage more than one dataset and forecast different diseases.

The authors of [ 108 ] used the Cleveland dataset, which included 303 cases and 76 attributes, to test out three different classification strategies: NB, SVM, and DT in addition to KNN. Only 14 of these 76 characteristics are going to be put through the testing process. The authors performed data preprocessing to remove noisy data. The KNN obtained the greatest accuracy with 90.79 percent. The authors need to use more sophisticated models to improve the accuracy of early heart disease prediction.

The authors of [ 109 ] proposed a model to predict heart disease by making use of a cardiovascular dataset, which was then classified through the application of supervised machine learning algorithms (DT, NB, logistic regression, RF, SVM, and KNN). The results reveal that the DT classification model predicted cardiovascular disorders better than other algorithms with an accuracy of 73 percent. The authors highlighted that the ensemble ML techniques employing the CVD dataset can generate a better illness prediction model.

In [ 110 ], the authors attempted to increase the accuracy of heart disease prediction by applying a logistic regression using a healthcare dataset to determine whether patients have heart illness problems or not. The dataset was acquired from an ongoing cardiovascular study on people of the town of Framingham, Massachusetts. The model reached an accuracy prediction of 87 percent. The authors acknowledge the model could be improved with more data and the use of more ML models.

Because breast cancer affects one in every 28 women in India, the author of [ 111 ] presented an accurate classification technique to examine the breast cancer dataset containing 569 rows and 32 columns. Similarly employing a heart disease dataset and Lung cancer dataset, this research offered A novel way to function selection. This method of selection is based on genetic algorithms mixed with the SVM classification. The classifier results are Lung cancer 81.8182, Diabetes 78.9272. noticed that the size, kind, and source of data used are not indicated.

In [ 112 ], the authors predicted the risk factors that cause heart disease using the K-means clustering algorithm and analyzed with a visualization tool using a Cleveland heart disease dataset with 76 features of 303 patients, holds 209 records with 8 attributes such as age, chest pain type, blood pressure, blood glucose level, ECG in rest, heart rate as well as four types of chest pain. The authors forecast cardiac diseases by taking into consideration the primary characteristics of four types of chest discomfort solely and K-means clustering is a common unsupervised ML technique.

The aim of the article [ 113 ] was to report the advantages of using a variety of data mining (DM) methods and validated heart disease survival prediction models. From the observations, the authors proposed that logistic regression and NB achieved the highest accuracy when performed on a high-dimensional dataset on the Cleveland hospital dataset and DT and RF produce better results on low-dimensional datasets. RF delivers more accuracy than the DT classifier as the algorithm is an optimized learning algorithm. The author mentioned that this work can be extended to other ML algorithms, the model could be developed in a distributed environment such as Map–Reduce, Apache Mahout, and HBase.

In [ 114 ], the authors proposed a single algorithm named hybridization to predict heart disease that combines used techniques into one single algorithm. The presented method has three phases. Preprocessing phase, classification phase, and diagnosis phase. They employed the Cleveland database and algorithms NB, SVM, KNN, NN, J4.8, RF, and GA. NB and SVM always perform better than others, whereas others depend on the specified features. results attained an accuracy of 89.2 percent. The authors need to is the key goal. Notice that the dataset is little; hence, the system was not able to train adequately, so the accuracy of the method was bad.

Using six algorithms (logistic regression, KNN, DT, SVM, NB, and RF), the authors of [ 115 ] explored different data representations to better understand how to use clinical data for predicting liver disease. The original dataset was taken from the northeast of Andhra Pradesh, India. includes 583 liver patient data, whereas 75.64 percent are male, and 24.36 percent are female. The analysis result indicated that the logistic regression classifier delivers the most increased order exactness of 75 percent depending on the f1 measure to forecast the liver illness and NB gives the least precision of 53 percent. The authors merely studied a few prominent supervised ML algorithms; more algorithms can be picked to create an increasingly exact model of liver disease prediction and performance can be steadily improved.

In [ 116 ], the authors aimed to predict coronary heart disease (CHD) based on historical medical data using ML technology. The goal of this study is to use three supervised learning approaches, NB, SVM, and DT, to find correlations in CHD data that could aid improve prediction rates. The dataset contains a retrospective sample of males from KEEL, a high-risk heart disease location in the Western Cape of South Africa. The model utilized NB, SVM, and DT. NB achieved the most accurate among the three models. SVM and DT J48 outperformed NB with a specificity rate of 82 percent but showed an inadequate sensitivity rate of less than 50 percent.

With the help of DM and network analysis methods, the authors of [ 117 ] created a chronic disease risk prediction framework that was created and evaluated in the Australian healthcare system to predict type 2 diabetes risk. Using a private healthcare funds dataset from Australia that spans six years and three different predictive algorithms (regression, parameter optimization, and DT). The accuracy of the prediction ranges from 82 to 87 percent. The hospital admission and discharge summary are the dataset's source. As a result, it does not provide information about general physician visits or future diagnoses.

DL-based healthcare prediction

With the help of DL algorithms such as CNN for autofeature extraction and illness prediction, the authors of [ 118 ] proposed a system for predicting patients with the more common inveterate diseases, and they used KNN for distance calculation to locate the exact matching in the dataset and the outcome of the final sickness prediction. A combination of disease symptoms was made for the structure of the dataset, the living habits of a person, and the specific attaches to doctor consultations which are acceptable in this general disease prediction. In this study, the Indian chronic kidney disease dataset was utilized that comprises 400 occurrences, 24 characteristics, and 2 classes were restored from the UCI ML store. Finally, a comparative study of the proposed system with other algorithms such as NB, DT, and logistic regression has been demonstrated in this study. The findings showed that the proposed system gives an accuracy of 95% which is higher than the other two methods. So, the proposed technique should be applied using more than one dataset.

In [ 119 ], the authors developed a DL approach that uses chest radiography images to differentiate between patients with mild, pneumonia, and COVID-19 infections, providing a valid mechanism for COVID-19 diagnosis. To increase the intensity of the chest X-ray image and eliminate noise, image-enhancing techniques were used in the proposed system. Two distinct DL approaches based on a pertained neural network model (ResNet-50) for COVID-19 identification utilizing chest X-ray (CXR) pictures are proposed in this work to minimize overfitting and increase the overall capabilities of the suggested DL systems. The authors emphasized that tests using a vast and hard dataset encompassing several COVID-19 cases are necessary to establish the efficacy of the suggested system.

Diabetes disease prediction was the topic of the article [ 120 ], in which the authors presented a cuckoo search-based deep LSTM classifier for prediction. The deep convLSTM classifier is used in cuckoo search optimization, which is a nature-inspired method for accurately predicting disease by transferring information and therefore reducing time consumption. The PIMA dataset is used to predict the onset of diabetes. The National Institute of Diabetes and Digestive and Kidney Diseases provided the data. The dataset is made up of independent variables including insulin level, age, and BMI index, as well as one dependent variable. The new technique was compared to traditional methods, and the results showed that the proposed method achieved 97.591 percent accuracy, 95.874 percent sensitivity, and 97.094 percent specificity, respectively. The authors noticed more datasets are needed, as well as new approaches to improve the classifier's effectiveness.

In [ 121 ], the authors presented a wavelet-based convolutional neural network to handle data limitations in this time of COVID-19 fast emergence. By investigating the influence of discrete wavelet transform decomposition up to 4 levels, the model demonstrated the capability of multi-resolution analysis for detecting COVID-19 chest X-rays. The wavelet sub-bands are CNN’s inputs at each decomposition level. COVID-19 chest X-ray-12 is a collection of 1,944 chest X-ray pictures divided into 12 groups that were compiled from two open-source datasets (National Institute Health containing several X-rays of pneumonia-related diseases, whereas the COVID-19 dataset is collected from Radiology Society North America). COVID-Neuro wavelet, a suggested model, was trained alongside other well-known ImageNet pre-trained models on COVID-CXR-12. The authors acknowledge they hope to investigate the effects of other wavelet functions besides the Haar wavelet.

A CNN framework for COVID-19 identification was suggested in [ 122 ] it made use of computed tomography images that was developed by the authors. The proposed framework employs a public CT dataset of 2482 CT images from patients of both classifications. the system attained an accuracy of 96.16 percent and recall of 95.41 percent after training using only 20 percent of the dataset. The authors stated that the use of the framework should be extended to multimodal medical pictures in the future.

Using an LSTM network enhanced by two processes to perform multi-label classification based on patients' clinical visit records, the authors of [ 123 ] performed multi-disease prediction for intelligent clinical decision support. A massive dataset of electronic health records was collected from a prominent hospital in southeast China. The suggested LSTM approach outperforms several standard and DL models in predicting future disease diagnoses, according to model evaluation results. The F1 score rises from 78.9 to 86.4 percent, respectively, with the state-of-the-art conventional and DL models, to 88.0 percent with the suggested technique. The authors stated that the model prediction performance may be enhanced further by including new input variables and that to reduce computational complexity, the method only uses one data source.

In [ 124 ], the authors introduced an approach to creating a supervised ANN structure based on the subnets (the group of neurons) instead of layers, in the cases of low datasets, this effectively predicted the disease. The model was evaluated using textual data and compared to multilayer perceptrons (MLPs) as well as LSTM recurrent neural network models using three small-scale publicly accessible benchmark datasets. On the Iris dataset, the experimental findings for classification reached 97% accuracy, compared to 92% for RNN (LSTM) with three layers, and the model had a lower error rate, 81, than RNN (LSTM) and MLP on the diabetic dataset, while RNN (LSTM) has a high error rate of 84. For larger datasets, however, this method is useless. This model is useless because it has not been implemented on large textual and image datasets.

The authors of [ 125 ] presented a novel AI and Internet of Things (IoT) convergence-based disease detection model for a smart healthcare system. Data collection, reprocessing, categorization, and parameter optimization are all stages of the proposed model. IoT devices, such as wearables and sensors, collect data, which AI algorithms then use to diagnose diseases. The forest technique is then used to remove any outliers found in the patient data. Healthcare data were used to assess the performance of the CSO-LSTM model. During the study, the CSO-LSTM model had a maximum accuracy of 96.16% on heart disease diagnoses and 97.26% on diabetes diagnoses. This method offered a greater prediction accuracy for heart disease and diabetes diagnosis, but there was no feature selection mechanism; hence, it requires extensive computations.

The global health crisis posed by coronaviruses was a subject of [ 126 ]. The authors aimed at detecting disease in people whose X-ray had been selected as potential COVID-19 candidates. Chest X-rays of people with COVID-19, viral pneumonia, and healthy people are included in the dataset. The study compared the performance of two DL algorithms, namely CNN and RNN. DL techniques were used to evaluate a total of 657 chest X-ray images for the diagnosis of COVID-19. VGG19 is the most successful model, with a 95% accuracy rate. The VGG19 model successfully categorizes COVID-19 patients, healthy individuals, and viral pneumonia cases. The dataset's most failing approach is InceptionV3. The success percentage can be improved, according to the authors, by improving data collection. In addition to chest radiography, lung tomography can be used. The success ratio and performance can be enhanced by creating numerous DL models.

In [ 127 ], the authors developed a method based on the RNN algorithm for predicting blood glucose levels for diabetics a maximum of one hour in the future, which required the patient's glucose level history. The Ohio T1DM dataset for blood glucose level prediction, which included blood glucose level values for six people with type 1 diabetes, was used to train and assess the approach. The distribution features were further honed with the use of studies that revealed the procedure's certainty estimate nature. The authors point out that they can only evaluate prediction goals with enough glucose level history; thus, they cannot anticipate the beginning levels after a gap, which does not improve the prediction's quality.

To build a new deep anomaly detection model for fast, reliable screening, the authors of [ 128 ] used an 18-layer residual CNN pre-trained on ImageNet with a different anomaly detection mechanism for the classification of COVID-19. On the X-ray dataset, which contains 100 images from 70 COVID-19 persons and 1431 images from 1008 non-COVID-19 pneumonia subjects, the model obtains a sensitivity of 90.00 percent specificity of 87.84 percent or sensitivity of 96.00 percent specificity of 70.65 percent. The authors noted that the model still has certain flaws, such as missing 4% of COVID-19 cases and having a 30% false positive rate. In addition, more clinical data are required to confirm and improve the model's usefulness.

In [ 129 ], the authors developed COVIDX-Net, a novel DL framework that allows radiologists to diagnose COVID-19 in X-ray images automatically. Seven algorithms (MobileNetV2, ResNetV2, VGG19, DenseNet201, InceptionV3, Inception, and Xception) were evaluated using a small dataset of 50 photographs (MobileNetV2, ResNetV2, VGG19, DenseNet201, InceptionV3, Inception, and Xception). Each deep neural network model can classify the patient's status as a negative or positive COVID-19 case based on the normalized intensities of the X-ray image. The f1-scores for the VGG19 and dense convolutional network (DenseNet) models were 0.89 and 0.91, respectively. With f1-scores of 0.67, the InceptionV3 model has the weakest classification performance.

The authors of [ 130 ] designed a DL approach for delivering 30-min predictions about future glucose levels based on a Dilated RNN (DRNN). The performance of the DRNN models was evaluated using data from two electronic health records datasets: OhioT1DM from clinical trials and the in silicon dataset from the UVA-Padova simulator. It outperformed established glucose prediction approaches such as neural networks (NNs), support vector regression (SVR), and autoregressive models (ARX). The results demonstrated that it significantly improved glucose prediction performance, although there are still some limits, such as the authors' creation of a data-driven model that heavily relies on past EHR. The quality of the data has a significant impact on the accuracy of the prediction. The number of clinical datasets is limited and , however, often restricted. Because certain data fields are manually entered, they are occasionally incorrect.

In [ 131 ], the authors utilized a deep neural network (DNN) to discover 15,099 stroke patients, researchers were able to predict stroke death based on medical history and human behaviors utilizing large-scale electronic health information. The Korea Centers for Disease Control and Prevention collected data from 2013 to 2016 and found that there are around 150 hospitals in the country, all having more than 100 beds. Gender, age, type of insurance, mode of admission, necessary brain surgery, area, length of hospital stays, hospital location, number of hospital beds, stroke kind, and CCI were among the 11 variables in the DL model. To automatically create features from the data and identify risk factors for stroke, researchers used a DNN/scaled principal component analysis (PCA). 15,099 people with a history of stroke were enrolled in the study. The data were divided into a training set (66%) and a testing set (34%), with 30 percent of the samples used for validation in the training set. DNN is used to examine the variables of interest, while scaled PCA is utilized to improve the DNN's continuous inputs. The sensitivity, specificity, and AUC values were 64.32%, 85.56%, and 83.48%, respectively.

The authors of [ 132 ] proposed (GluNet), an approach to glucose forecasting. This method made use of a personalized DNN to forecast the probabilistic distribution of short-term measurements for people with Type 1 diabetes based on their historical data. These data included insulin doses, meal information, glucose measurements, and a variety of other factors. It utilized the newest DL techniques consisting of four components: post-processing, dilated CNN, label recovery/ transform, and data preprocessing. The authors run the models on the subjects from the OhioT1DM datasets. The outcomes revealed significant enhancements over the previous procedures via a comprehensive comparison concerning the and root mean square error (RMSE) having a time lag of 60 min prediction horizons (PH) and RMSE having a small-time lag for the case of prediction horizons in the virtual adult participants. If the PH is properly matched to the lag between input and output, the user may learn the control of the system more frequently and it achieves good performance. Additionally, GluNet was validated on two clinical datasets. It attained an RMSE with a time lag of 60 min PH and RMSE with a time lag of 30-min PH. The authors point out that the model does not consider physiological knowledge, and that they need to test GluNet with larger prediction horizons and use it to predict overnight hypoglycemia.

The authors of [ 133 ] proposed the short-term blood glucose prediction model (VMD-IPSO-LSTM), which is a short-term strategy for predicting blood glucose (VMD-IPSO-LSTM). Initially, the intrinsic modal functions (IMF) in various frequency bands were obtained using the variational modal decomposition (VMD) technique, which deconstructed the blood glucose content. The short- and long-term memory networks then constructed a prediction mechanism for each blood glucose component’s intrinsic modal functions (IMF). Because the time window length, learning rate, and neuron count are difficult to set, the upgraded PSO approach optimized these parameters. The improved LSTM network anticipated each IMF, and the projected subsequence was superimposed in the final step to arrive at the ultimate prediction result. The data of 56 participants were chosen as experimental data among 451 diabetic Mellitus patients. The experiments revealed that it improved prediction accuracy at "30 min, 45 min, and 60 min." The RMSE and MAPE were lower than the "VMD-PSO-LSTM, VMD-LSTM, and LSTM," indicating that the suggested model is effective. The longer time it took to anticipate blood glucose levels and the higher accuracy of the predictions gave patients and doctors more time to improve the effectiveness of diabetes therapy and manage blood glucose levels. The authors noted that they still faced challenges, such as an increase in calculation volume and operation time. The time it takes to estimate glucose levels in the short term will be reduced.

To speed up diagnosis and cut down on mistakes, the authors of [ 134 ] proposed a new paradigm for primary COVID-19 detection based on a radiology review of chest radiography or chest X-ray. The authors used a dataset of chest X-rays from verified COVID-19 patients (408 photographs), confirmed pneumonia patients (4273 images), and healthy people (1590 images) to perform a three-class image classification (1590 images). There are 6271 people in total in the dataset. To fulfill this image categorization problem, the authors plan to use CNN and transfer learning. For all the folds of data, the model's accuracy ranged from 93.90 percent to 98.37 percent. Even the lowest level of accuracy, 93.90 percent, is still quite good. The authors will face a restriction, particularly when it comes to adopting such a model on a large scale for practical usage.

In [ 135 ], the authors proposed DL models for predicting the number of COVID-19-positive cases in Indian states. The Ministry of Health and Family Welfare dataset contains time series data for 32 individual confirmed COVID-19 cases in each of the states (28) and union territories (4) since March 14, 2020. This dataset was used to conduct an exploratory analysis of the increase in the number of positive cases in India. As prediction models, RNN-based LSTMs are used. Deep LSTM, convolutional LSTM, and bidirectional LSTM models were tested on 32 states/union territories, and the model with the best accuracy was chosen based on absolute error. Bidirectional LSTM produced the best performance in terms of prediction errors, while convolutional LSTM produced the worst performance. For all states, daily and weekly forecasts were calculated, and bi-LSTM produced accurate results (error less than 3%) for short-term prediction (1–3 days).

With the goal of increasing the reliability and precision of type 1 diabetes predictions, the authors of [ 136 ] proposed a new method based on CNNs and DL. It was about figuring out how to extract the behavioral pattern. Numerous observations of identical behaviors were used to fill in the gaps in the data. The suggested model was trained and verified using data from 759 people with type 1 diabetes who visited Sheffield Teaching Hospitals between 2013 and 2015. A subject's type 1 diabetes test, demographic data (age, gender, years with diabetes), and the final 84 days (12 weeks) of self-monitored blood glucose (SMBG) measurements preceding the test formed each item in the training set. In the presence of insufficient data and certain physiological specificities, prediction accuracy deteriorates, according to the authors.

The authors of [ 137 ] constructed a framework using the PIDD. PID's participants are all female and at least 21 years old. PID comprises 768 incidences, with 268 samples diagnosed as diabetic and 500 samples not diagnosed as diabetic. The eight most important characteristics that led to diabetes prediction. The accuracy of functional classifiers such as ANN, NB, DT, and DL is between 90 and 98 percent. On the PIMA dataset, DL had the best results for diabetes onset among the four, with an accuracy rate of 98.07 percent. The technique uses a variety of classifiers to accurately predict the disease, but it failed to diagnose it at an early stage.

To summarize all previous works discussed in this section, we will categorize them according to the diseases along with the techniques used to predict each disease, the datasets used, and the main findings, as shown in Table 5 .

Results and discussion

This study conducted a systematic review to examine the latest developments in ML and DL for healthcare prediction. It focused on healthcare forecasting and how the use of ML and DL can be relevant and robust. A total of 41 papers were reviewed, 21 in ML and 20 in DL as depicted in Fig.  17 .

In this study, the reviewed paper were classified by diseases predicted; as a result, 5 diseases were discussed including diabetes, COVID-19, heart, liver, and chronic kidney). Table 6 illustrates the number of reviewed papers for each disease in addition to the adopted prediction techniques in each disease.

Table 6 provides a comprehensive summary of the various ML and DL models used for disease prediction. It indicates the number of studies conducted on each disease, the techniques employed, and the highest level of accuracy attained. As shown in Table 6 , the optimal diagnostic accuracy for each disease varies. For diabetes, the DL model achieved a 98.07% accuracy rate. For COVID-19, the accuracy of the logistic regression model was 98.5%. The CSO-LSTM model achieved an accuracy of 96.16 percent for heart disease. For liver disease, the accuracy of the logistic regression model was 75%. The accuracy of the logistic regression model for predicting multiple diseases was 98.5%. It is essential to note that these are merely the best accuracy included in this survey. In addition, it is essential to consider the size and quality of the datasets used to train and validate the models. It is more likely that models trained on larger and more diverse datasets will generalize well to new data. Overall, the results presented in Table 6 indicate that ML and DL models can be used to accurately predict disease. When selecting a model for a specific disease, it is essential to carefully consider the various models and techniques.

Although ML and DL have made incredible strides in recent years, they still have a long way to go before they can effectively be used to solve the fundamental problems plaguing the healthcare systems. Some of the challenges associated with implementing ML and DL approaches in healthcare prediction are discussed here.

The Biomedical Data Stream is the primary challenge that needs to be handled. Significant amounts of new medical data are being generated rapidly, and the healthcare industry as a whole is evolving rapidly. Some examples of such real-time biological signals include measurements of blood pressure, oxygen saturation, and glucose levels. While some variants of DL architecture have attempted to address this problem, many challenges remain before effective analyses of rapidly evolving, massive amounts of streaming data can be conducted. These include problems with memory consumption, feature selection, missing data, and computational complexity. Another challenge for ML and DL is tackling the complexity of the healthcare domain.

Healthcare and biomedical research present more intricate challenges than other fields. There is still a lot we do not know about the origins, transmission, and cures for many of these incredibly diverse diseases. It is hard to collect sufficient data because there are not always enough patients. A solution to this issue may be found, however. The small number of patients necessitates exhaustive patient profiling, innovative data processing, and the incorporation of additional datasets. Researchers can process each dataset independently using the appropriate DL technique and then represent the results in a unified model to extract patient data.

The use of ML and DL techniques for healthcare prediction has the potential to change the way traditional healthcare services are delivered. In the case of ML and DL applications, healthcare data is deemed the most significant component that contributes to medical care systems. This paper aims to present a comprehensive review of the most significant ML and DL techniques employed in healthcare predictive analytics. In addition, it discussed the obstacles and challenges of applying ML and DL Techniques in the healthcare domain. As a result of this survey, a total of 41 papers covering the period from 2019 to 2022 were selected and thoroughly reviewed. In addition, the methodology for each paper was discussed in detail. The reviewed studies have shown that AI techniques (ML and DL) play a significant role in accurately diagnosing diseases and helping to anticipate and analyze healthcare data by linking hundreds of clinical records and rebuilding a patient's history using these data. This work advances research in the field of healthcare predictive analytics using ML and DL approaches and contributes to the literature and future studies by serving as a resource for other academics and researchers.

Availability of data and materials

Not applicable.

Abbreviations

Artificial Intelligence

Machine Learning

Decision Tree

Electronic Health Records

Random Forest

Support Vector Machine

K-Nearest Neighbor

Naive Bayes

Reinforcement Learning

Natural Language Processing

Monte Carlo Tree Search

Partially Observable Markov Decision Processes

Deep Learning

Deep Belief Network

Artificial Neural Networks

Convolutional Neural Networks

Long Short-Term Memory

Recurrent Convolution Neural Networks

Recurrent Neural Networks

Recurrent Convolutional Layer

Receptive Domains

Recurrent Multilayer Perceptron

Pima Indian Diabetes Database

Coronary Heart Disease

Chest X-Ray

Multilayer Perceptrons

Internet of Things

Dilated RNN

Neural Networks

Support Vector Regression

Principal Component Analysis

Deep Neural Network

Prediction Horizons

Root Mean Square Error

Intrinsic Modal Functions

Variational Modal Decomposition

Self-Monitored Blood Glucose

Latha MH, Ramakrishna A, Reddy BSC, Venkateswarlu C, Saraswathi SY (2022) Disease prediction by stacking algorithms over big data from healthcare communities. Intell Manuf Energy Sustain: Proc ICIMES 2021(265):355

Google Scholar  

Van Calster B, Wynants L, Timmerman D, Steyerberg EW, Collins GS (2019) Predictive analytics in health care: how can we know it works? J Am Med Inform Assoc 26(12):1651–1654

Sahoo PK, Mohapatra SK, Wu SL (2018) SLA based healthcare big data analysis and computing in cloud network. J Parallel Distrib Comput 119:121–135

Thanigaivasan V, Narayanan SJ, Iyengar SN, Ch N (2018) Analysis of parallel SVM based classification technique on healthcare using big data management in cloud storage. Recent Patents Comput Sci 11(3):169–178

Elmahdy HN (2014) Medical diagnosis enhancements through artificial intelligence

Xiong X, Cao X, Luo L (2021) The ecology of medical care in Shanghai. BMC Health Serv Res 21:1–9

Donev D, Kovacic L, Laaser U (2013) The role and organization of health care systems. Health: systems, lifestyles, policies, 2nd edn. Jacobs Verlag, Lage, pp 3–144

Murphy G F, Hanken M A, & Waters K A (1999) Electronic health records: changing the vision

Qayyum A, Qadir J, Bilal M, Al-Fuqaha A (2020) Secure and robust machine learning for healthcare: a survey. IEEE Rev Biomed Eng 14:156–180

El Seddawy AB, Moawad R, Hana MA (2018) Applying data mining techniques in CRM

Wang Y, Kung L, Wang WYC, Cegielski CG (2018) An integrated big data analytics-enabled transformation model: application to health care. Inform Manag 55(1):64–79

Mirbabaie M, Stieglitz S, Frick NR (2021) Artificial intelligence in disease diagnostics: a critical review and classification on the current state of research guiding future direction. Heal Technol 11(4):693–731

Tang R, De Donato L, Besinović N, Flammini F, Goverde RM, Lin Z, Wang Z (2022) A literature review of artificial intelligence applications in railway systems. Transp Res Part C: Emerg Technol 140:103679

Singh G, Al’Aref SJ, Van Assen M, Kim TS, van Rosendael A, Kolli KK, Dwivedi A, Maliakal G, Pandey M, Wang J, Do V (2018) Machine learning in cardiac CT: basic concepts and contemporary data. J Cardiovasc Comput Tomograph 12(3):192–201

Kim KJ, Tagkopoulos I (2019) Application of machine learning in rheumatic disease research. Korean J Intern Med 34(4):708

Liu B (2011) Web data mining: exploring hyperlinks, contents, and usage data. Spriger, Berlin

MATH   Google Scholar  

Haykin S, Lippmann R (1994) Neural networks, a comprehensive foundation. Int J Neural Syst 5(4):363–364

Gupta M, Pandya SD (2022) A comparative study on supervised machine learning algorithm. Int J Res Appl Sci Eng Technol (IJRASET) 10(1):1023–1028

Ray S (2019) A quick review of machine learning algorithms. In: 2019 international conference on machine learning, big data, cloud and parallel computing (COMITCon) (pp 35–39). IEEE

Srivastava A, Saini S, & Gupta D (2019) Comparison of various machine learning techniques and its uses in different fields. In: 2019 3rd international conference on electronics, communication and aerospace technology (ICECA) (pp 81–86). IEEE

Park HA (2013) An introduction to logistic regression: from basic concepts to interpretation with particular attention to nursing domain. J Korean Acad Nurs 43(2):154–164

Obulesu O, Mahendra M, & Thrilok Reddy M (2018) Machine learning techniques and tools: a survey. In: 2018 international conference on inventive research in computing applications (ICIRCA) (pp 605–611). IEEE

Dhall D, Kaur R, & Juneja M (2020) Machine learning: a review of the algorithms and its applications. Proceedings of ICRIC 2019: recent innovations in computing 47–63

Yang F J (2019) An extended idea about Decision Trees. In: 2019 international conference on computational science and computational intelligence (CSCI) (pp 349–354). IEEE

Eesa AS, Orman Z, Brifcani AMA (2015) A novel feature-selection approach based on the cuttlefish optimization algorithm for intrusion detection systems. Expert Syst Appl 42(5):2670–2679

Shamim A, Hussain H, & Shaikh M U (2010) A framework for generation of rules from Decision Tree and decision table. In: 2010 international conference on information and emerging technologies (pp 1–6). IEEE

Eesa AS, Abdulazeez AM, Orman Z (2017) A dids based on the combination of cuttlefish algorithm and Decision Tree. Sci J Univ Zakho 5(4):313–318

Bakyarani ES, Srimathi H, Bagavandas M (2019) A survey of machine learning algorithms in health care. Int J Sci Technol Res 8(11):223

Resende PAA, Drummond AC (2018) A survey of random forest based methods for intrusion detection systems. ACM Comput Surv (CSUR) 51(3):1–36

Breiman L (2001) Random forests. Mach learn 45:5–32

Ho TK (1998) The random subspace method for constructing decision forests. IEEE Trans Pattern Anal Mach Intell 20(8):832–844

Hofmann M, & Klinkenberg R (2016) RapidMiner: data mining use cases and business analytics applications. CRC Press

Chow CKCN, Liu C (1968) Approximating discrete probability distributions with dependence trees. IEEE Trans Inf Theory 14(3):462–467

Burges CJ (1998) A tutorial on support vector machines for pattern recognition. Data Min Knowl Disc 2(2):121–167

Han J, Pei J, Kamber M (1999) Data mining: concepts and techniques. 2011

Cortes C, Vapnik V (1995) Support-vector networks. Mach learn 20:273–297

Aldahiri A, Alrashed B, Hussain W (2021) Trends in using IoT with machine learning in health prediction system. Forecasting 3(1):181–206

Sarker IH (2021) Machine learning: Algorithms, real-world applications and research directions. SN Comput Sci 2(3):160

Ting K M, & Zheng Z (1999) Improving the performance of boosting for naive Bayesian classification. In: Methodologies for knowledge discovery and data mining: third Pacific-Asia conference, PAKDD-99 Beijing, China, Apr 26–28, 1999 proceedings 3 (pp 296–305). Springer Berlin Heidelberg

Oladipo ID, AbdulRaheem M, Awotunde JB, Bhoi AK, Adeniyi EA, Abiodun MK (2022) Machine learning and deep learning algorithms for smart cities: a start-of-the-art review. In: IoT and IoE driven smart cities, pp 143–162

Shailaja K, Seetharamulu B, & Jabbar M A Machine learning in healthcare: a review. In: 2018 second international conference on electronics, communication and aerospace technology (ICECA) 2018 Mar 29 (pp 910–914)

Mahesh B (2020) Machine learning algorithms-a review. Int J Sci Res (IJSR) 9:381–386

Greene D, Cunningham P, & Mayer R (2008) Unsupervised learning and clustering. Mach learn Techn Multimed: Case Stud Organ Retriev 51–90

Jain AK, Dubes RC (1988) Algorithms for clustering data. Prentice-Hall, Inc, USA

Kodinariya TM, Makwana PR (2013) Review on determining number of cluster in K-means clustering. Int J 1(6):90–95

Smith LI (2002) A tutorial on principal components analysis

Mishra SP, Sarkar U, Taraphder S, Datta S, Swain D, Saikhom R, Laishram M (2017) Multivariate statistical data analysis-principal component analysis (PCA). Int J Livestock Res 7(5):60–78

Kamani M, Farzin Haddadpour M, Forsati R, and Mahdavi M (2019) "Efficient Fair Principal Component Analysis." arXiv e-prints: arXiv-1911.

Dey A (2016) Machine learning algorithms: a review. Int J Comput Sci Inf Technol 7(3):1174–1179

Agrawal R, Imieliński T, & Swami A (1993) Mining association rules between sets of items in large databases. In: proceedings of the 1993 ACM SIGMOD international conference on Management of data (pp 207–216)

Agrawal R, & Srikant R (1994) Fast algorithms for mining association rules. In: Proceeding of 20th international conference very large data bases, VLDB (Vol 1215, pp 487-499)

Singh J, Ram H, Sodhi DJ (2013) Improving efficiency of apriori algorithm using transaction reduction. Int J Sci Res Publ 3(1):1–4

Al-Maolegi M, & Arkok B (2014) An improved Apriori algorithm for association rules. arXiv preprint arXiv:1403.3948

Abaya SA (2012) Association rule mining based on Apriori algorithm in minimizing candidate generation. Int J Sci Eng Res 3(7):1–4

Coronato A, Naeem M, De Pietro G, Paragliola G (2020) Reinforcement learning for intelligent healthcare applications: a survey. Artif Intell Med 109:101964

Watkins CJ, Dayan P (1992) Q-learning. Mach Learn 8:279–292

Jang B, Kim M, Harerimana G, Kim JW (2019) Q-learning algorithms: a comprehensive classification and applications. IEEE access 7:133653–133667

Achille A, Soatto S (2018) Information dropout: Learning optimal representations through noisy computation. IEEE Trans Pattern Anal Mach Intell 40(12):2897–2905

Williams G, Wagener N, Goldfain B, Drews P, Rehg J M, Boots B, & Theodorou E A (2017) Information theoretic MPC for model-based reinforcement learning. In: 2017 IEEE international conference on robotics and automation (ICRA) (pp 1714–1721). IEEE

Wilkes JT, Gallistel CR (2017) Information theory, memory, prediction, and timing in associative learning. Comput Models Brain Behav 29:481–492

Ning Y, Jia J, Wu Z, Li R, An Y, Wang Y, Meng H (2017) Multi-task deep learning for user intention understanding in speech interaction systems. In: Proceedings of the AAAI conference on artificial intelligence (Vol 31, No. 1)

Shi X, Gao Z, Lausen L, Wang H, Yeung DY, Wong WK, Woo WC (2017) Deep learning for precipitation nowcasting: a benchmark and a new model. In: Guyon I, Von Luxburg U, Bengio S, Wallach H, Fergus R, Vishwanathan S, Garnett R (Eds) Advances in neural information processing systems, vol 30. Curran Associates, Inc.,. https://proceedings.neurips.cc/paper_files/paper/2017/file/a6db4ed04f1621a119799fd3d7545d3d-Paper.pdf

Juang CF, Lu CM (2009) Ant colony optimization incorporated with fuzzy Q-learning for reinforcement fuzzy control. IEEE Trans Syst, Man, Cybernet-Part A: Syst Humans 39(3):597–608

Świechowski M, Godlewski K, Sawicki B, Mańdziuk J (2022) Monte Carlo tree search: a review of recent modifications and applications. Artif Intell Rev 56:1–66

Lizotte DJ, Laber EB (2016) Multi-objective Markov decision processes for data-driven decision support. J Mach Learn Res 17(1):7378–7405

MathSciNet   MATH   Google Scholar  

Silver D, Huang A, Maddison CJ, Guez A, Sifre L, Van Den Driessche G, Hassabis D (2016) Mastering the game of go with deep neural networks and tree search. Nature 529(7587):484–489

Browne CB, Powley E, Whitehouse D, Lucas SM, Cowling PI, Rohlfshagen P, Colton S (2012) A survey of monte carlo tree search methods. IEEE Trans Comput Intell AI Games 4(1):1–43

Ling ZH, Kang SY, Zen H, Senior A, Schuster M, Qian XJ, Deng L (2015) Deep learning for acoustic modeling in parametric speech generation: a systematic review of existing techniques and future trends. IEEE Signal Process Magaz 32(3):35–52

Schmidhuber J (2015) Deep learning in neural networks: an overview. Neural Netw 61:85–117

Yu D, Deng L (2010) Deep learning and its applications to signal and information processing [exploratory dsp]. IEEE Signal Process Mag 28(1):145–154

Hinton GE, Osindero S, Teh YW (2006) A fast learning algorithm for deep belief nets. Neural Comput 18(7):1527–1554

Goyal P, Pandey S, Jain K, Goyal P, Pandey S, Jain K (2018) Introduction to natural language processing and deep learning. Deep Learn Nat Language Process: Creat Neural Netw Python 1–74. https://doi.org/10.1007/978-1-4842-3685-7

Mathew A, Amudha P, Sivakumari S (2021) Deep learning techniques: an overview. Adv Mach Learn Technol Appl: Proc AMLTA 2020:599–608

Goodfellow I, Bengio Y, Courville A (2016) Deep learning. MIT press, USA

Gomes L (2014) Machine-learning maestro Michael Jordan on the delusions of big data and other huge engineering efforts. IEEE Spectrum 20. https://spectrum.ieee.org/machinelearning-maestro-michael-jordan-on-the-delusions-of-big-data-and-other-huge-engineering-efforts

Huang G, Liu Z, Van Der Maaten L, & Weinberger K Q (2017) Densely connected convolutional networks. In: proceedings of the IEEE conference on computer vision and pattern recognition (pp 4700–4708)

Yap MH, Pons G, Marti J, Ganau S, Sentis M, Zwiggelaar R, Marti R (2017) Automated breast ultrasound lesions detection using convolutional neural networks. IEEE J Biomed Health Inform 22(4):1218–1226

Hayashi Y (2019) The right direction needed to develop white-box deep learning in radiology, pathology, and ophthalmology: a short review. Front Robot AI 6:24

Alom MZ, Taha TM, Yakopcic C, Westberg S, Sidike P, Nasrin MS, Asari VK (2019) A state-of-the-art survey on deep learning theory and architectures. Electronics 8(3):292

Schmidhuber J, Hochreiter S (1997) Long short-term memory. Neural Comput 9(8):1735–1780

Smagulova K, James AP (2019) A survey on LSTM memristive neural network architectures and applications. Eur Phys J Spec Topics 228(10):2313–2324

Setyanto A, Laksito A, Alarfaj F, Alreshoodi M, Oyong I, Hayaty M, Kurniasari L (2022) Arabic language opinion mining based on long short-term memory (LSTM). Appl Sci 12(9):4140

Lindemann B, Müller T, Vietz H, Jazdi N, Weyrich M (2021) A survey on long short-term memory networks for time series prediction. Procedia CIRP 99:650–655

Cui Z, Ke R, Pu Z, & Wang Y (2018) Deep bidirectional and unidirectional LSTM recurrent neural network for network-wide traffic speed prediction. arXiv preprint arXiv:1801.02143

Villegas R, Yang J, Zou Y, Sohn S, Lin X, & Lee H (2017) Learning to generate long-term future via hierarchical prediction. In: international conference on machine learning (pp 3560–3569). PMLR

Gensler A, Henze J, Sick B, & Raabe N (2016) Deep learning for solar power forecasting—an approach using autoencoder and LSTM neural networks. In: 2016 IEEE international conference on systems, man, and cybernetics (SMC) (pp 002858–002865). IEEE

Lindemann B, Fesenmayr F, Jazdi N, Weyrich M (2019) Anomaly detection in discrete manufacturing using self-learning approaches. Procedia CIRP 79:313–318

Kalchbrenner N, Danihelka I, & Graves A (2015) Grid long short-term memory. arXiv preprint arXiv:1507.01526

Cheng B, Xu X, Zeng Y, Ren J, Jung S (2018) Pedestrian trajectory prediction via the social-grid LSTM model. J Eng 2018(16):1468–1474

Veličković P, Karazija L, Lane N D, Bhattacharya S, Liberis E, Liò P & Vegreville M (2018) Cross-modal recurrent models for weight objective prediction from multimodal time-series data. In: proceedings of the 12th EAI international conference on pervasive computing technologies for healthcare (pp 178–186)

Wang J, Hu X (2021) Convolutional neural networks with gated recurrent connections. IEEE Trans Pattern Anal Mach Intell 44(7):3421–3435

Liang M, & Hu X (2015) Recurrent convolutional neural network for object recognition. In: proceedings of the IEEE conference on computer vision and pattern recognition (pp 3367–3375)

Liang M, Hu X, Zhang B (2015) Convolutional neural networks with intra-layer recurrent connections for scene labeling. In: Cortes C, Lawrence N, Lee D, Sugiyama M, Garnett R (Eds) Advances in Neural Information Processing Systems, vol 28. Curran Associates, Inc. https://proceedings.neurips.cc/paper_files/paper/2015/file/9cf81d8026a9018052c429cc4e56739b-Paper.pdf

Fernandez B, Parlos A G, & Tsai W K (1990) Nonlinear dynamic system identification using artificial neural networks (ANNs). In: 1990 IJCNN international joint conference on neural networks (pp 133–141). IEEE

Puskorius GV, Feldkamp LA (1994) Neurocontrol of nonlinear dynamical systems with Kalman filter trained recurrent networks. IEEE Trans Neural Netw 5(2):279–297

Rumelhart DE (1986) Learning representations by error propagation. In: DE Rumelhart and JL McClelland & PDP Research Group, eds, Parallel distributed processing: explorations in the microstructure of cognition. Bradford Books MITPress, Cambridge, Mass

Krishnamoorthi R, Joshi S, Almarzouki H Z, Shukla P K, Rizwan A, Kalpana C, & Tiwari B (2022) A novel diabetes healthcare disease prediction framework using machine learning techniques. J Healthcare Eng. https://doi.org/10.1155/2022/1684017

Edeh MO, Khalaf OI, Tavera CA, Tayeb S, Ghouali S, Abdulsahib GM, Louni A (2022) A classification algorithm-based hybrid diabetes prediction model. Front Publ Health 10:829510

Iwendi C, Huescas C G Y, Chakraborty C, & Mohan S (2022) COVID-19 health analysis and prediction using machine learning algorithms for Mexico and Brazil patients. J Experiment Theor Artif Intell 1–21. https://doi.org/10.1080/0952813X.2022.2058097

Lu H, Uddin S, Hajati F, Moni MA, Khushi M (2022) A patient network-based machine learning model for disease prediction: the case of type 2 diabetes mellitus. Appl Intell 52(3):2411–2422

Chugh M, Johari R, & Goel A (2022) MATHS: machine learning techniques in healthcare system. In: international conference on innovative computing and communications: proceedings of ICICC 2021, Volume 3 (pp 693–702). Springer Singapore

Deberneh HM, Kim I (2021) Prediction of type 2 diabetes based on machine learning algorithm. Int J Environ Res Public Health 18(6):3317

Gupta S, Verma H K, & Bhardwaj D (2021) Classification of diabetes using Naive Bayes and support vector machine as a technique. In: operations management and systems engineering: select proceedings of CPIE 2019 (pp 365–376). Springer Singapore

Islam M T, Rafa S R, & Kibria M G (2020) Early prediction of heart disease using PCA and hybrid genetic algorithm with k-means. In: 2020 23rd international conference on computer and information technology (ICCIT) (pp 1–6). IEEE

Qawqzeh Y K, Bajahzar A S, Jemmali M, Otoom M M, Thaljaoui A (2020) Classification of diabetes using photoplethysmogram (PPG) waveform analysis: logistic regression modeling. BioMed Res Int. https://doi.org/10.1155/2020/3764653

Grampurohit S, Sagarnal C (2020) Disease prediction using machine learning algorithms. In: 2020 international conference for emerging technology (INCET) (pp 1–7). IEEE

Moturi S, Srikanth Vemuru DS (2020) Classification model for prediction of heart disease using correlation coefficient technique. Int J 9(2). https://doi.org/10.30534/ijatcse/2020/185922020

Barik S, Mohanty S, Rout D, Mohanty S, Patra A K, & Mishra A K (2020) Heart disease prediction using machine learning techniques. In: advances in electrical control and signal systems: select proceedings of AECSS 2019 (pp 879–888). Springer, Singapore

Princy R J P, Parthasarathy S, Jose P S H, Lakshminarayanan A R, & Jeganathan S (2020) Prediction of cardiac disease using supervised machine learning algorithms. In: 2020 4th international conference on intelligent computing and control systems (ICICCS) (pp 570–575). IEEE

Saw M, Saxena T, Kaithwas S, Yadav R, & Lal N (2020) Estimation of prediction for getting heart disease using logistic regression model of machine learning. In: 2020 international conference on computer communication and informatics (ICCCI) (pp 1–6). IEEE

Soni VD (2020) Chronic disease detection model using machine learning techniques. Int J Sci Technol Res 9(9):262–266

Indrakumari R, Poongodi T, Jena SR (2020) Heart disease prediction using exploratory data analysis. Procedia Comput Sci 173:130–139

Wu C S M, Badshah M, & Bhagwat V (2019) Heart disease prediction using data mining techniques. In: proceedings of the 2019 2nd international conference on data science and information technology (pp 7–11)

Tarawneh M, & Embarak O (2019) Hybrid approach for heart disease prediction using data mining techniques. In: advances in internet, data and web technologies: the 7th international conference on emerging internet, data and web technologies (EIDWT-2019) (pp 447–454). Springer International Publishing

Rahman AS, Shamrat FJM, Tasnim Z, Roy J, Hossain SA (2019) A comparative study on liver disease prediction using supervised machine learning algorithms. Int J Sci Technol Res 8(11):419–422

Gonsalves A H, Thabtah F, Mohammad R M A, & Singh G (2019) Prediction of coronary heart disease using machine learning: an experimental analysis. In: proceedings of the 2019 3rd international conference on deep learning technologies (pp 51–56)

Khan A, Uddin S, Srinivasan U (2019) Chronic disease prediction using administrative data and graph theory: the case of type 2 diabetes. Expert Syst Appl 136:230–241

Alanazi R (2022) Identification and prediction of chronic diseases using machine learning approach. J Healthcare Eng. https://doi.org/10.1155/2022/2826127

Gouda W, Almurafeh M, Humayun M, Jhanjhi NZ (2022) Detection of COVID-19 based on chest X-rays using deep learning. Healthcare 10(2):343

Kumar A, Satyanarayana Reddy S S, Mahommad G B, Khan B, & Sharma R (2022) Smart healthcare: disease prediction using the cuckoo-enabled deep classifier in IoT framework. Sci Progr. https://doi.org/10.1155/2022/2090681

Monday H N, Li J P, Nneji G U, James E C, Chikwendu I A, Ejiyi C J, & Mgbejime G T (2021) The capability of multi resolution analysis: a case study of COVID-19 diagnosis. In: 2021 4th international conference on pattern recognition and artificial intelligence (PRAI) (pp 236–242). IEEE

Al Rahhal MM, Bazi Y, Jomaa RM, Zuair M, Al Ajlan N (2021) Deep learning approach for COVID-19 detection in computed tomography images. Cmc-Comput Mater Continua 67(2):2093–2110

Men L, Ilk N, Tang X, Liu Y (2021) Multi-disease prediction using LSTM recurrent neural networks. Expert Syst Appl 177:114905

Ahmad U, Song H, Bilal A, Mahmood S, Alazab M, Jolfaei A & Saeed U (2021) A novel deep learning model to secure internet of things in healthcare. Mach Intell Big Data Anal Cybersec Appl 341–353

Mansour RF, El Amraoui A, Nouaouri I, Díaz VG, Gupta D, Kumar S (2021) Artificial intelligence and internet of things enabled disease diagnosis model for smart healthcare systems. IEEE Access 9:45137–45146

Sevi M, & Aydin İ (2020) COVID-19 detection using deep learning methods. In: 2020 international conference on data analytics for business and industry: way towards a sustainable economy (ICDABI) (pp 1–6). IEEE

Martinsson J, Schliep A, Eliasson B, Mogren O (2020) Blood glucose prediction with variance estimation using recurrent neural networks. J Healthc Inform Res 4:1–18

Zhang J, Xie Y, Pang G, Liao Z, Verjans J, Li W, Xia Y (2020) Viral pneumonia screening on chest X-rays using confidence-aware anomaly detection. IEEE Trans Med Imaging 40(3):879–890

Hemdan E E D, Shouman M A, & Karar M E (2020) Covidx-net: a framework of deep learning classifiers to diagnose covid-19 in x-ray images. arXiv preprint arXiv:2003.11055

Zhu T, Li K, Chen J, Herrero P, Georgiou P (2020) Dilated recurrent neural networks for glucose forecasting in type 1 diabetes. J Healthc Inform Res 4:308–324

Cheon S, Kim J, Lim J (2019) The use of deep learning to predict stroke patient mortality. Int J Environ Res Public Health 16(11):1876

Li K, Liu C, Zhu T, Herrero P, Georgiou P (2019) GluNet: a deep learning framework for accurate glucose forecasting. IEEE J Biomed Health Inform 24(2):414–423

Wang W, Tong M, Yu M (2020) Blood glucose prediction with VMD and LSTM optimized by improved particle swarm optimization. IEEE Access 8:217908–217916

Rashid N, Hossain M A F, Ali M, Sukanya M I, Mahmud T, & Fattah S A (2020) Transfer learning based method for COVID-19 detection from chest X-ray images. In: 2020 IEEE region 10 conference (TENCON) (pp 585–590). IEEE

Arora P, Kumar H, Panigrahi BK (2020) Prediction and analysis of COVID-19 positive cases using deep learning models: a descriptive case study of India. Chaos, Solitons Fractals 139:110017

MathSciNet   Google Scholar  

Zaitcev A, Eissa MR, Hui Z, Good T, Elliott J, Benaissa M (2020) A deep neural network application for improved prediction of in type 1 diabetes. IEEE J Biomed Health Inform 24(10):2932–2941

Naz H, Ahuja S (2020) Deep learning approach for diabetes prediction using PIMA Indian dataset. J Diabetes Metab Disord 19:391–403

Download references

Acknowledgements

Author information, authors and affiliations.

Department of Information Systems and Technology, Faculty of Graduate Studies for Statistical Research, Cairo University, Giza, Egypt

Mohammed Badawy & Nagy Ramadan

Department of Computer Sciences, Faculty of Graduate Studies for Statistical Research, Cairo University, Giza, Egypt

Hesham Ahmed Hefny

You can also search for this author in PubMed   Google Scholar

Contributions

MB wrote the main text of the manuscript; NR and HAH revised the manuscript. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Mohammed Badawy .

Ethics declarations

Competing interests.

The authors declare that they have no competing interests. All authors approved the final manuscript.

Additional information

Publisher's note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Badawy, M., Ramadan, N. & Hefny, H.A. Healthcare predictive analytics using machine learning and deep learning techniques: a survey. Journal of Electrical Systems and Inf Technol 10 , 40 (2023). https://doi.org/10.1186/s43067-023-00108-y

Download citation

Received : 27 December 2022

Accepted : 31 July 2023

Published : 29 August 2023

DOI : https://doi.org/10.1186/s43067-023-00108-y

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Healthcare prediction
• Artificial intelligence (AI)
• Machine learning (ML)
• Deep learning (DL)
• Medical diagnosis

• Open access
• Published: 27 July 2006

The importance of human resources management in health care: a global context

• Stefane M Kabene 1 , 3 ,
• Carole Orchard 3 ,
• John M Howard 2 ,
• Mark A Soriano 1 &
• Raymond Leduc 1  

Human Resources for Health volume  4 , Article number:  20 ( 2006 ) Cite this article

806k Accesses

173 Citations

12 Altmetric

Metrics details

This paper addresses the health care system from a global perspective and the importance of human resources management (HRM) in improving overall patient health outcomes and delivery of health care services.

We explored the published literature and collected data through secondary sources.

Various key success factors emerge that clearly affect health care practices and human resources management. This paper will reveal how human resources management is essential to any health care system and how it can improve health care models. Challenges in the health care systems in Canada, the United States of America and various developing countries are examined, with suggestions for ways to overcome these problems through the proper implementation of human resources management practices. Comparing and contrasting selected countries allowed a deeper understanding of the practical and crucial role of human resources management in health care.

Proper management of human resources is critical in providing a high quality of health care. A refocus on human resources management in health care and more research are needed to develop new policies. Effective human resources management strategies are greatly needed to achieve better outcomes from and access to health care around the world.

Peer Review reports

Defining human resources in health care

Within many health care systems worldwide, increased attention is being focused on human resources management (HRM). Specifically, human resources are one of three principle health system inputs, with the other two major inputs being physical capital and consumables [ 1 ]. Figure 1 depicts the relationship between health system inputs, budget elements and expenditure categories.

Relationship between health system inputs, budget elements and expenditure categories . Source: World Health Report 2000 Figure 4.1 pg.75. http://www.who.int.proxy.lib.uwo.ca:2048/whr/2000/en/whr00_ch4_en.pdf Figure 1 identifies three principal health system inputs: human resources, physical capital and consumables. It also shows how the financial resources to purchase these inputs are of both a capital investment and a recurrent character. As in other industries, investment decisions in health are critical because they are generally irreversible: they commit large amounts of money to places and activities that are difficult, even impossible, to cancel, close or scale down [1].

Human resources, when pertaining to health care, can be defined as the different kinds of clinical and non-clinical staff responsible for public and individual health intervention [ 1 ]. As arguably the most important of the health system inputs, the performance and the benefits the system can deliver depend largely upon the knowledge, skills and motivation of those individuals responsible for delivering health services [ 1 ].

As well as the balance between the human and physical resources, it is also essential to maintain an appropriate mix between the different types of health promoters and caregivers to ensure the system's success [ 1 ]. Due to their obvious and important differences, it is imperative that human capital is handled and managed very differently from physical capital [ 1 ]. The relationship between human resources and health care is very complex, and it merits further examination and study.

Both the number and cost of health care consumables (drugs, prostheses and disposable equipment) are rising astronomically, which in turn can drastically increase the costs of health care. In publicly-funded systems, expenditures in this area can affect the ability to hire and sustain effective practitioners. In both government-funded and employer-paid systems, HRM practices must be developed in order to find the appropriate balance of workforce supply and the ability of those practitioners to practise effectively and efficiently. A practitioner without adequate tools is as inefficient as having the tools without the practitioner.

Key questions and issues pertaining to human resources in health care

When examining health care systems in a global context, many general human resources issues and questions arise. Some of the issues of greatest relevance that will be discussed in further detail include the size, composition and distribution of the health care workforce, workforce training issues, the migration of health workers, the level of economic development in a particular country and sociodemographic, geographical and cultural factors.

The variation of size, distribution and composition within a county's health care workforce is of great concern. For example, the number of health workers available in a country is a key indicator of that country's capacity to provide delivery and interventions [ 2 ]. Factors to consider when determining the demand for health services in a particular country include cultural characteristics, sociodemographic characteristics and economic factors [ 3 ].

Workforce training is another important issue. It is essential that human resources personnel consider the composition of the health workforce in terms of both skill categories and training levels [ 2 ]. New options for the education and in-service training of health care workers are required to ensure that the workforce is aware of and prepared to meet a particular country's present and future needs [ 2 ]. A properly trained and competent workforce is essential to any successful health care system.

The migration of health care workers is an issue that arises when examining global health care systems. Research suggests that the movement of health care professionals closely follows the migration pattern of all professionals in that the internal movement of the workforce to urban areas is common to all countries [ 2 ]. Workforce mobility can create additional imbalances that require better workforce planning, attention to issues of pay and other rewards and improved overall management of the workforce [ 2 ]. In addition to salary incentives, developing countries use other strategies such as housing, infrastructure and opportunities for job rotation to recruit and retain health professionals [ 2 ], since many health workers in developing countries are underpaid, poorly motivated and very dissatisfied [ 3 ]. The migration of health workers is an important human resources issue that must be carefully measured and monitored.

Another issue that arises when examining global health care systems is a country's level of economic development. There is evidence of a significant positive correlation between the level of economic development in a country and its number of human resources for health [ 3 ]. Countries with higher gross domestic product (GDP) per capita spend more on health care than countries with lower GDP and they tend to have larger health workforces [ 3 ]. This is an important factor to consider when examining and attempting to implement solutions to problems in health care systems in developing countries.

Socio-demographic elements such as age distribution of the population also play a key role in a country's health care system. An ageing population leads to an increase in demand for health services and health personnel [ 3 ]. An ageing population within the health care system itself also has important implications: additional training of younger workers will be required to fill the positions of the large number of health care workers that will be retiring.

It is also essential that cultural and geographical factors be considered when examining global health care systems. Geographical factors such as climate or topography influence the ability to deliver health services; the cultural and political values of a particular nation can also affect the demand and supply of human resources for health [ 3 ]. The above are just some of the many issues that must be addressed when examining global health care and human resources that merit further consideration and study.

The impact of human resources on health sector reform

When examining global health care systems, it is both useful and important to explore the impact of human resources on health sector reform. While the specific health care reform process varies by country, some trends can be identified. Three of the main trends include efficiency, equity and quality objectives [ 3 ].

Various human resources initiatives have been employed in an attempt to increase efficiency. Outsourcing of services has been used to convert fixed labor expenditures into variable costs as a means of improving efficiency. Contracting-out, performance contracts and internal contracting are also examples of measures employed [ 3 ].

Many human resources initiatives for health sector reform also include attempts to increase equity or fairness. Strategies aimed at promoting equity in relation to needs require more systematic planning of health services [ 3 ]. Some of these strategies include the introduction of financial protection mechanisms, the targeting of specific needs and groups, and re-deployment services [ 3 ]. One of the goals of human resource professionals must be to use these and other measures to increase equity in their countries.

Human resources in health sector reform also seek to improve the quality of services and patients' satisfaction. Health care quality is generally defined in two ways: technical quality and sociocultural quality. Technical quality refers to the impact that the health services available can have on the health conditions of a population [ 3 ]. Sociocultural quality measures the degree of acceptability of services and the ability to satisfy patients' expectations [ 3 ].

Human resource professionals face many obstacles in their attempt to deliver high-quality health care to citizens. Some of these constraints include budgets, lack of congruence between different stakeholders' values, absenteeism rates, high rates of turnover and low morale of health personnel [ 3 ].

Better use of the spectrum of health care providers and better coordination of patient services through interdisciplinary teamwork have been recommended as part of health sector reform [ 4 ]. Since all health care is ultimately delivered by people, effective human resources management will play a vital role in the success of health sector reform.

In order to have a more global context, we examined the health care systems of Canada, the United States of America, Germany and various developing countries. The data collection was achieved through secondary sources such as the Canadian Health Coalition, the National Coalition on Health Care and the World Health Organization Regional Office for Europe. We were able to examine the main human resources issues and questions, along with the analysis of the impact of human resources on the health care system, as well as the identification of the trends in health sector reform. These trends include efficiency, equity and quality objectives.

Health care systems

The Canadian health care system is publicly funded and consists of five general groups: the provincial and territorial governments, the federal government, physicians, nurses and allied health care professionals. The roles of these groups differ in numerous aspects. See Figure 2 for an overview of the major stakeholders in the Canadian health care system.

Overview of the major stakeholders in the Canadian health care system . Figure 2 depicts the major stakeholders in the Canadian health care system and how they relate.

Provincial and territorial governments are responsible for managing and delivering health services, including some aspects of prescription care, as well as planning, financing, and evaluating hospital care provision and health care services [ 5 ]. For example, British Columbia has shown its commitment to its health care program by implementing an increase in funding of CAD 6.7 million in September 2003, in order to strengthen recruitment, retention and education of nurses province-wide [ 6 ]. In May 2003, it was also announced that 30 new seats would be funded to prepared nurse practitioners at the University of British Columbia and at the University of Victoria [ 6 ]. Recently the Ontario Ministry of Health and Long Term Care announced funding for additional nurse practitioner positions within communities. Furthermore, most provinces and territories in Canada have moved the academic entry requirement for registered nurses to the baccalaureate level, while increasing the length of programmes for Licensed Practice Nurses to meet the increasing complexity of patient-care needs. Several provinces and territories have also increased seats in medical schools aimed towards those students wishing to become family physicians [ 7 ].

The federal government has other responsibilities, including setting national health care standards and ensuring that standards are enforced by legislative acts such as the Canada Health Act (CHA) [ 5 ]. Constitutionally the provinces are responsible for the delivery of health care under the British North America (BNA) Act; the provinces and territories must abide by these standards if they wish to receive federal funding for their health care programs [ 8 ]. The federal government also provides direct care to certain groups, including veterans and First Nation's peoples, through the First Nationals and Inuit Health Branch (FNIHB). Another role of the federal government is to ensure disease protection and to promote health issues [ 5 ].

The federal government demonstrates its financial commitment to Canada's human resources in health care by pledging transfer funds to the provinces and direct funding for various areas. For example, in the 2003 Health Care Renewal Accord, the federal government provided provinces and territories with a three-year CAD 1.5 billion Diagnostic/Medical Equipment Fund. This was used to support specialized staff training and equipment that improved access to publicly funded services [ 6 ].

The third group – private physicians – is generally not employed by the government, but rather is self-employed and works in a private practice. They deliver publicly-funded care to Canadian citizens. Physicians will negotiate fee schedules for their services with their provincial governments and then submit their claims to the provincial health insurance plan in order to receive their reimbursement [ 5 ].

The roles of nurses consist of providing care to individuals, groups, families, communities and populations in a variety of settings. Their roles require strong, consistent and knowledgeable leaders, who inspire others and support professional nursing practice. Leadership is an essential element for high-quality professional practice environments in which nurses can provide high-quality nursing care [ 9 ].

In most Canadian health care organizations, nurses manage both patient care and patient care units within the organization. Nurses have long been recognized as the mediators between the patient and the health care organization [ 10 ]. In care situations, they generally perform a coordinating role for all services needed by patients. They must be able to manage and process nursing data, information and knowledge to support patient care delivery in diverse care-delivery settings [ 10 ]. Workplace factors most valued by nurses include autonomy and control over the work environment, ability to initiate and sustain a therapeutic relationship with patients and a collaborative relationship with physicians at the unit level [ 11 ].

In addition to doctors and nurses, there are many more professionals involved in the health care process. Allied health care professionals can consist of pharmacists, dietitians, social workers and case managers, just to name a few. While much of the focus is on doctors and nurses, there are numerous issues that affect other health care providers as well, including workplace issues, scopes of practice and the impact of changing ways of delivering services [ 12 ]. Furthermore, with health care becoming so technologically advanced, the health care system needs an increasing supply of highly specialized and skilled technicians [ 12 ]. Thus we can see the various roles played by these five groups and how they work together to form the Canadian health care system.

Canada differs from other nations such as the United States of America for numerous reasons, one of the most important being the CHA. As previously mentioned, the CHA sets national standards for health care in Canada. The CHA ensures that all Canadian citizens, regardless of their ability to pay, will have access to health care services in Canada. "The aim of the CHA is to ensure that all eligible residents of Canada have reasonable access to insured health services on a prepaid basis, without direct charges at the point of service" [ 6 ].

Two of the most significant stipulations of the CHA read: "reasonable access to medically necessary hospital and physician services by insured persons must be unimpeded by financial or other barriers" and "health services may not be withheld on the basis of income, age, health status, or gender" [ 5 ]. These two statements identify the notable differences between the Canadian and American health care systems. That is, coverage for the Canadian population is much more extensive.

Furthermore in Canada, there has been a push towards a more collaborative, interdisciplinary team approach to delivering health care; this raises many new issues, one of which will involve successful knowledge transfer within these teams [ 13 ]. Effective knowledge management, which includes knowledge transfer, is increasingly being recognized as a crucial aspect of an organization's basis for long-term, sustainable, competitive advantage [ 34 ]. Even though health care in Canada is largely not for profit, there will still be the need for effective knowledge management practices to be developed and instituted. The introduction of interdisciplinary health teams in Canadian hospitals is a relatively new phenomenon and their connection to the knowledge management policies and agendas of governments and hospital administrations raises important questions about how such teams will work and to what extent they can succeed in dealing with the more difficult aspects of knowledge management, such as the transfer of tacit knowledge.

The multidisciplinary approach tends to be focused around specific professional disciplines, with health care planning being mainly top-down and dominated by medical professionals. Typically there is a lead professional (usually a physician) who determines the care and, if necessary, directs the patient to other health care specialists and allied professionals (aides, support workers). There is generally little involvement by the patient in the direction and nature of the care. Interdisciplinary health care is a patient-centred approach in which all those involved, including the patient, have input into the decisions being made.

The literature on teamwork and research on the practices in hospitals relating to multidisciplinary teams suggests that interdisciplinary teams face enormous challenges [ 13 ], therefore multidisciplinary teamwork will continue to be a vital part of the health care system. However, the goal of this teamwork should not be to displace one health care provider with another, but rather to look at the unique skills each one brings to the team and to coordinate the deployment of these skills. Clients need to see the health worker most appropriate to deal with their problem [ 14 ].

Some of the issues regarding the Canadian public system of health have been identified in the Mazankowski Report, which was initiated by Alberta's Premier Ralph Klein in 2000. Many issues have arisen since this time and have been debated among Canadians. One of the most contentious, for example, is the possibility of introducing a two-tier medical system. One tier of the proposed new system would be entirely government-funded through tax dollars and would serve the same purpose as the current publicly-funded system. The second tier would be a private system and funded by consumers [ 5 ].

However, the CHA and the Canadian Nurses Association (CNA) are critical of any reforms that pose a threat to the public health care system. It should be noted that although Canada purports to have a one-tier system, the close proximity of private, fee-for-service health care in the United States really creates a pay-as-you-go second tier for wealthy Canadians. In addition, many health care services such as most prescriptions and dental work are largely funded by individuals and/or private or employer paid insurance plans.

It is important to realize the differences between the proposed two-tier system and the current health care system. Presently, the public health care system covers all medically necessary procedures and the private sector provides 30% for areas such as dental care. With the new system, both public and private care would offer all services and Canadians would have the option of choosing between the two.

The proposal of the two-tier system is important because it highlights several important issues that concern many Canadians, mainly access to the system and cost reduction. Many Canadians believe the current public system is not sustainable and that a two-tiered system would force the public system to become more efficient and effective, given the competition of the private sector. However, the two-tiered system is not within the realm of consideration, since the majority of Canadians are opposed to the idea of a privatized system [ 5 ]. No proposals have come forward that show how a privately funded system would provide an equal quality of services for the same cost as the current publicly funded system.

United States of America

The health care system in the United States is currently plagued by three major challenges. These include: rapidly escalating health care costs, a large and growing number of Americans without health coverage and an epidemic of substandard care [ 15 ].

Health insurance premiums in the United States have been rising at accelerating rates. The premiums themselves, as well as the rate of increase in premiums, have increased every year since 1998; independent studies and surveys indicate that this trend is likely to continue over the next several years [ 15 ]. As a result of these increases, it is more difficult for businesses to provide health coverage to employees, with individuals and families finding it more difficult to pay their share of the cost of employer-sponsored coverage [ 15 ]. The rising trend in the cost of employer-sponsored family health coverage is illustrated in Figure 3 .

The trend of the cost of employer-sponsored family health care coverage in the United States . Source: National Coalition on Health Care 2004 pg.9. http://www.nchc.org/materials/studies/reform.pdf . Figure 3 illustrates the increase in health insurance premiums since 2001. These increases are making it more difficult for businesses to continue to provide health coverage for their employees and retirees [15].

To help resolve this problem, health maintenance organizations (HMO) have been introduced, with the goal of focusing on keeping people well and out of hospitals in the hope of decreasing employer costs. HMOs are popular alternatives to traditional health care plans offered by insurance companies because they can cover a wide variety of services, usually at a significantly lower cost [ 16 ]. HMOs use "networks" of selected doctors, hospitals, clinics and other health care providers that together provide comprehensive health services to the HMOs members [ 16 ]. The overall trade-off with an HMO is reduced choice in exchange for increased affordability.

Another problem to address regarding the American health care system is the considerable and increasing number of Americans without health coverage. Health care coverage programs such as Medicare offer a fee-for-service plan that covers many health care services and certain drugs. It also provides access to any doctor or hospital that accepts Medicare [ 17 ]. Patients with limited income and resources may qualify for Medicaid, which provide extra help paying for prescription drug costs [ 17 ]. However, according to figures from the United States Census Bureau, the number of Americans without health coverage grew to 43.6 million in 2002; it is predicted that the number of uninsured Americans will increase to between 51.2 and 53.7 million in 2006 [ 15 ].

Those Americans without health care insurance receive less care, receive care later and are, on average, less healthy and less able to function in their daily lives than those who have health care insurance. Additionally, the risk of mortality is 25% higher for the uninsured than for the insured [ 15 ].

Despite excellent care in some areas, the American health care system is experiencing an epidemic of substandard care; the system is not consistently providing high-quality care to its patients [ 15 ]. There appears to be a large discrepancy between the care patients should be receiving and the care they are actually getting. The Institute of Medicine has estimated that between 44 000 and 98 000 Americans die each year from preventable medical errors in hospitals [ 15 ].

It is also useful to examine the demographic characteristics of those Americans more likely to receive substandard care. Research shows that those Americans with little education and low income receive a lower standard of care [ 18 ]. This finding may be explained by the fact that patients who have lower education levels tend to have more difficulty explaining their concerns to physicians, as well as eliciting a response for those concerns because health professionals often do not value their opinions [ 18 ].

Case studies

As shown by the extensive literature, statistics and public opinion, there is a growing need for health care reform in the United States of America. There is a duty and responsibility of human resources professionals to attempt to elicit change and implement policies that will improve the health care system.

It is informative to examine case studies in which human resources professionals have enacted positive change in a health care setting. One such case from 1995 is that of a mid-sized, private hospital in the New York metropolitan area. This case presents a model of how human resources can be an agent for change and can partner with management to build an adaptive culture to maintain strong organizational growth [ 19 ].

One of the initiatives made by human resources professionals in an attempt to improve the overall standard of care in the hospital was to examine and shape the organization's corporate culture. Steps were taken to define the values, behaviors and competences that characterized the current culture, and analyze these against the desired culture [ 19 ]. A climate survey was conducted in the organization; it became the goal of the human resources professionals to empower employees to be more creative and innovative [ 19 ]. To achieve this, a new model of care was designed that emphasized a decentralized nursing staff and a team-based approach to patient care. Nursing stations were redesigned to make them more accessible and approachable [ 19 ].

Human resources management also played an important role in investing in employee development. This was achieved by assisting employees to prepare and market themselves for internal positions and if desired, helping them pursue employment opportunities outside the organization [ 19 ]. This case makes obvious the important roles that human resources management can play in orchestrating organizational change.

Another case study that illustrates the importance of human resources management to the health care system is that of The University of Nebraska Medical Center in 1995. During this period, the hospital administrative staff recognized a variety of new challenges that were necessitating organizational change. Some of these challenges included intense price competition and payment reform in health care, reduced state and federal funding for education and research, and changing workforce and population demographics [ 20 ]. The organizational administrators recognized that a cultural reformation was needed to meet these new challenges. A repositioning process was enacted, resulting in a human resources strategy that supported the organization's continued success [ 20 ]. This strategy consisted of five major objectives, each with a vision statement and series of action steps.

Staffing: Here, the vision was to integrate a series of organization-wide staffing strategies that would anticipate and meet changing workforce requirements pertaining to staff, faculty and students. To achieve this vision, corporate profiles were developed for each position to articulate the core competences and skills required [ 20 ].

Performance management: The vision was to hold all faculty and staff accountable and to reward individual and team performance. With this strategy, managers would be able to provide feedback and coaching to employees in a more effective and timely manner [ 20 ].

Development and learning: The vision was to have all individuals actively engaged in the learning process and responsible for their own development. Various unit-based training functions were merged into a single unit, which defined critical technical and behavioral competencies [ 20 ].

Valuing people: The vision was to have the hospital considered as a favored employer and to be able to attract and retain the best talent. To facilitate this vision, employee services such as child care and wellness were expanded [ 20 ].

Organizational effectiveness. The vision was to create an organization that is flexible, innovative and responsive [ 20 ]. The developments of these human resources strategies were essential to the effectiveness of the organization and to demonstrate the importance of human resources in the health care industry.

Both these case studies illustrate that effective human resources management is crucial to health care in a practical setting and that additional human resources initiatives are required if solutions are to be found for the major problems in the United States health care system.

Approximately 92% of Germany's population receives health care through the country's statutory health care insurance program, Gesetzliche Krankenversicherung (GKV). GKV designed an organizational framework for health care in Germany and has identified and constructed the roles of payers, providers and hospitals. Private, for-profit companies cover slightly less than 8% of the population. This group would include, for example, civil servants and the self-employed. It is estimated that approximately 0.2% of the population does not have health care insurance [ 21 ]. This small fragment may be divided into two categories: either the very rich, who do not require it, or the very poor, who obtain their coverage through social insurance. All Germans, regardless of their coverage, use the same health care facilities. With these policies nearly all citizens are guaranteed access to high-quality medical care [ 22 ].

While the federal government plays a major part in setting the standards for national health care policies, the system is actually run by national and regional autonomous organizations. Rather than being financed solely through taxes, the system is covered mostly by health care premiums [ 22 ]. In 2003, about 11.1% of Germany's gross domestic product (GDP) went into the health care system [ 23 ] versus the United States, with 15% [ 24 ] and Canada at 9.9% [ 25 ]. However, Germany still put about one third of its social budget towards health care [ 22 ].

The supply of physicians in Germany is high, especially compared to the United States, and this is attributed largely to the education system. If one meets the academic requirements in Germany, the possibility to study medicine is legally guaranteed [ 26 ]. This has led to a surplus of physicians and unemployment for physicians has become a serious problem. In 2001, the unemployment rate for German physicians of 2.1% led many German doctors to leave for countries such as Norway, Sweden and the United Kingdom, all of which actively recruit from Germany [ 27 ].

Germany's strong and inexpensive academic system has led the country to educate far more physicians than the United States and Canada. In 2003, Germany had 3.4 practicing physicians per 1000 inhabitants [ 23 ], versus the United States, which had 2.3 practicing physicians per 1000 inhabitants in 2002 [ 24 ] and Canada, which had 2.1 practicing physicians per 1000 inhabitants in 2003 [ 25 ]. It is also remarkable that health spending per capita in Germany (USD 2996) [ 23 ] amounted to about half of health spending per capita in the United States (USD 5635) [ 24 ], and slightly less than Canada's health spending (USD 3003) [ 25 ]. This clearly demonstrates the Germans' strength regarding cost containment.

There are several issues that physicians face in the German health care system. In a 1999 poll, 49.9% of respondents said they were very or fairly satisfied with their health care system, while 47.7% replied they were very or fairly dissatisfied with it [ 28 ]. Furthermore, the degree of competition between physicians is very high in Germany and this could lead to a reduction in physician earnings. Due to this competition, many younger physicians currently face unemployment. The German law also limits the number of specialists in certain geographical areas where there are issues of overrepresentation [ 22 ]. Thus, the oversupply of physicians in Germany leads to many challenges, including human resources management in the health care system.

In Germany a distinction is made between office-based physicians and hospital-based physicians. The income of office-based physicians is based on the number and types of services they provide, while hospital-based physicians are compensated on a salary basis. This division has created a separated workforce that German legislation is now working to eliminate by encouraging the two parties to work together, with the aim of reducing overall medical costs [ 22 ].

Developing countries

Accessing good-quality health care services can be incredibly arduous for those living in developing countries, and more specifically, for those residing in rural areas. For many reasons, medical personnel and resources may not be available or accessible for such residents. As well, the issue of migrant health care workers is critical. Migrant health workers can be defined as professionals who have a desire and the ability to leave the country in which they were educated and migrate to another country. The workers are generally enticed to leave their birth country by generous incentive offers from the recruiting countries [ 29 ].

Developing countries struggle to find means to improve living conditions for their residents; countries such as Ghana, Kenya, South Africa and Zimbabwe are seeking human resources solutions to address their lack of medically trained professionals. Shortages in these countries are prevalent due to the migration of their highly educated and medically trained personnel.

Professionals tend to migrate to areas where they believe their work will be more thoroughly rewarded. The International Journal for Equity in Health (2003) suggested that those who work in the health care profession tend to migrate to areas that are more densely populated and where their services may be better compensated. Health care professionals look to areas that will provide their families with an abundance of amenities, including schools for their children, safe neighborhoods and relatives in close proximity. For medical professionals, the appeal of promotions also serves as an incentive for educating oneself further [ 30 ]. As one becomes more educated, the ability and opportunity to migrate increases and this can lead to a further exodus of needed health care professionals.

These compelling reasons tend to cause medical professionals to leave their less-affluent and less-developed areas and migrate to areas that can provide them with better opportunities. This has caused a surplus in some areas and a huge deficit in others. This epidemic can be seen in nations such as Nicaragua. Its capital city, Managua, holds only one fifth of the country's population, yet it employs almost 50% of the medically trained health care workers. The same situation can be found in other countries, such as Bangladesh, where almost one third of the available health personnel are employed "in four metropolitan districts where less than 15% of the population lives" [ 30 ]. Clearly this presents a problem for those living outside these metropolitan districts.

Other possible explanations put forth by Dussault and Franceschini, both of the Human Development Division of the World Bank Institute, include "management style, incentive and career structures, salary scales, recruitment, posting and retention practices" [ 31 ]. Salary scales can differ quite drastically between originating and destination countries, which are shown in Figures 4 and 5 . They also state that in developing countries the earning potential one would see in more affluent or populated urban areas is much higher than one would expect to earn in rural areas.

Ratio of nurse wages (PPP USD), destination country to source country . Source: Vujicic M, Zurn P, Diallo K, Orvill A, Dal Poz MR 2004. http://www.human-resources-health.com/content/2/1/3 . Figure 4 shows the difference between the wage in the source country and destination country for nurses. This difference is also known as the "wage premium" [29].

Ratio of physician wages (PPP USD), destination country to source country . Source: Vujicic M, Zurn P, Diallo K, Orvill A, Dal Poz MR 2004. http://www.human-resources-health.com/content/2/1/3 . Figure 5 shows the difference between the wage in the source country and destination country for physicians [29].

As more health professionals emigrate to urban areas, the workloads for those in the rural areas greatly increase. This leads to a domino effect, in that those in such dire situations look for areas where they may be able to find more satisfactory and less demanding working conditions [ 31 ]. Vujicic et al. (2004) summarizes numerous variables that influence the migration pattern and has created a formula to express their impact. It is possible to quantify the factors, and human resources professionals need to look at the costs and benefits of altering the factors so that the migration pattern is more favorable. This formula is expressed as the results shown in Table 1 , which shows the different reasons for one to migrate in terms of the popularity of a given reason.

There is a tendency for developed countries faced with decreasing numbers of nationally trained medical personnel to recruit already-trained individuals from other nations by enticing them with incentives. Zimbabwe has been particularly affected by this problem. In 2001, out of approximately 730 nursing graduates, more than one third (237) of them relocated to the United Kingdom [ 29 ]. This was a dramatic increase from 1997, when only 26 (approximately 6.2%) of the 422 nursing program graduates migrated to the United Kingdom [ 29 ]. This leads to the loss of skilled workers in developing countries and can be very damaging, since the education systems in developing countries are training individuals for occupations in the medical profession, yet are not able to retain them [ 29 ].

Countries that have the capacity to educate more people than necessary in order to meet their domestic demand have tried to counterbalance this problem by increasing their training quota. Vujicic et al. (2004) identify that "the Philippines has for many years trained more nurses than are required to replenish the domestic stock, in an effort to encourage migration and increase the level of remittance flowing back into the country" [ 29 ].

Developed countries attract internationally trained medical professionals for many reasons. To begin with, "political factors, concerns for security, domestic birth rates, the state of the economy and war (both at home and abroad)" [ 26 ] influence the number of people that will be allowed or recruited into a country. Also, due to the conditions of the labor market compared to the demand in developed countries, governments may make allowances to their strict policies regarding the type of and number of professionals they will allow into their country [ 29 ]. This can be seen in a Canadian example:

Canada maintains] a list of occupations within which employment vacancies [are] evident. Potential immigrants working in one of these [listed] occupations would have a much higher chance of being granted entry than if they worked in a non-listed occupation [ 29 ].

Though Canada attracts internationally trained medical professionals, those employment vacancies may not always be open. Although there may be up to 10 000 international medical graduates (IMG) in Canada, many are not legally allowed to practice. Many immigrants cannot afford the costs of retraining and may be forced to find a new job in a completely unrelated field, leaving their skills to go to waste [ 32 ]. In 2004, Ontario had between 2000 and 4000 IMGs looking for work in medical fields related to their training and background [ 33 ]. That year, IMG Ontario accepted 165 IMGs into assessment and training positions, which was a 50% increase over the last year, and a 600% increase from the 24 positions in 1999 [ 33 ].

Another appeal for developed countries with regard to foreign trained health care professionals is that they may be less of a financial burden to the host country than those trained domestically. This is because educational costs and the resources necessary for training are already taken care of by the international medical schools and governments [ 29 ]. Though these reasons may make recruiting foreign medical professionals seem appealing, there are still ongoing debates as to whether those trained outside the host country are equally qualified and culturally sensitive to the country to which they migrate. Developing countries are addressing these concerns by establishing health professional training programs similar to those in developed countries [ 29 ]. These practices can be seen in, "the majority of nursing programs in Bangladesh, the Philippines and South Africa [which] are based on curricula from United Kingdom or USA nursing schools" [ 29 ]. Because of these actions, those who are trained may be more likely to leave and use their skills where they will be recognized and more highly rewarded.

There are also ethical considerations when examining the practice of recruiting health care professionals, particularly if they are recruited from regions or countries where health care shortages already exist. The rights of individuals to move as they see fit may need to be balanced against the idea of the greater good of those left behind.

Due to the shortages, it has been found the level of health service in rural or poor areas has decreased, leading to lower quality and productivity of health services, closure of hospital wards, increased waiting times, reduced numbers of available beds for inpatients, diversion of emergency department patients and underuse of remaining personnel or substitution with persons lacking the required skills for performing critical interventions [ 30 ].

The article "Not enough here, too many there: understanding geographical imbalances in the distribution of the health workforce" (2003), states that a reduced number of health care workers in a given area has a direct effect on the life expectancy of its residents. For example, in the rural areas of Mexico, life expectancy is 55 years, compared to 71 years in the urban areas. Additionally, in "the wealthier, northern part of the country, infant mortality is 20/1000 as compared to more than 50/1000 in the poorer southern states" [ 31 ].

Globalization – a common thread

While the issues raised in this article are common to many countries, the approaches taken to address them may not be the same in each country. Factors affecting the approaches that can be taken, some of which have been raised, include demographics, resources and philosophical and political perspectives. However, an overarching issue that affects not only health care but many other areas is that of globalization itself.

Different countries have traditionally had different perspectives on health care that have influenced their approaches to health care delivery. In Canada for example, health care is considered a right; its delivery is defined by the five main principles of the Canada Health Act, which officially precludes a significant role for private delivery of essential services. In the United States, health care is treated more as another service that, while it should be accessible, is not considered a right. Therefore there is a much larger private presence in health care delivery the United States than there is in Canada. In other parts of the world, the approach to health care falls between these perspectives.

As the move towards globalization for many goods and services increases, countries will have to consider how this will affect their approaches to health care delivery. As mentioned earlier, there is already a degree of labor mobility within a country that affects the quality and availability of health care services. There is also already a degree of international mobility of health care workers, as shown by the number of workers recruited developed countries.

While the international mobility of labor is generally not as unencumbered as that for goods and capital, that may be changing as more and more regional free trade agreements are considered. Canada, the United States and Mexico have NAFTA (North American Free Trade Agreement), Europe has the EU (European Union) and talks are under way to consider expanding the NAFTA agreement to include Central and South America, to expand EU membership and to consider an Asian trading bloc including China and India.

If health care becomes a part of these new trade agreements, countries will be obliged to treat health care delivery according to the rules of the agreement. Using the NAFTA as an example, if health care is included, governments could not treat domestic providers more favorably than foreign firms wanting to deliver services. In Canada the concern is that it would mean the end of the Canada Health Act, since NAFTA would allow private, for-profit American or Mexican firms to open.

All five issues raised in this research would be affected by the increase in international trade agreements that included health care. Therefore, governments, health care providers and human resources professionals cannot ignore this important consideration and trend when examining solutions to the issues. Depending upon their relative negotiation strengths and positions, some countries may not benefit as much as others with these agreements.

For example, it is more likely that countries with well-developed private, for-profit, health care expertise, such as the United States, would expand into developing countries rather than the other way around. If there is an increased ability for labor mobility, then it is likely that health care professionals in the poorer, developing countries would move to where the opportunities are better. We already see this internally in the move from rural to urban centers; this would likely continue if the health care professionals had the opportunity to move out of country to where they could have greater financial rewards for their expertise.

When considering the countries examined in this paper, it is likely that Canada and the United States would initially be the two most likely to move towards a more integrated approach to health care delivery. There is already a trade agreement in place, many of the factors influencing health care are similar (demographics, training, level of economic development, geography, cultural factors) and they are currently each other's largest trading partners. While the current agreement, which includes Mexico, does not cover health care, there is pressure to broaden the agreement to include areas not currently covered. If this happens, human resources professionals will have to increase their understanding of what the new health care delivery realities could be. For example, if the move is more towards the Canadian example of a largely not-for-profit, mainly publicly-funded health care delivery system, then it will be more of an adjustment for the American professionals.

However, the likelihood of the Canadian approach to health care's being adopted in the United States is very slim. During the presidency of Bill Clinton, the government attempted to introduce a more universal health care delivery system, which failed completely. Even though there are over 40 million Americans with no health care coverage, the idea of a universal, publicly-funded system went nowhere. Also, within Canada there is increasing pressure to consider a more active role for private health care delivery. Therefore, it is more likely that Canadian health care and human resource professionals will have to adapt to a style more like the American, privately delivered, for-profit approach.

If this is the direction of change, human resources professionals in Canada will need to adjust how they approach the challenges and new realities. For instance, there would likely be an increased role for insurance companies and health maintenance organizations (HMO) as they move towards the managed care model of the United States. With an HMO approach, financial as well as health needs of the patients are considered when making medical decisions. An insured patient would select from the range of services and providers that his/her policy covers and approves. Human resources professionals would need to work with a new level of administration, the HMO, which currently does not exist to any significant degree in Canada.

As mentioned earlier, it is likely that developing countries would be receiving health care models and approaches from developed countries rather than the other way around. In particular, a country such as the United States that has a strong, private, for-profit approach already in place would likely be the source from which the health care models would be drawn. Therefore, health care, as well as human resources professionals in those countries, would also need to adapt to these new realities.

In Germany, where there is currently an oversupply of physicians, a move towards a more global approach to health care delivery, through increased trade agreements, could result in even more German health care professionals' leaving the country. The challenge to be addressed by human resource professionals within the German health care system in this situation would be to prevent, or slow, the loss of the best professionals to other countries. Spending public resources in educating professionals only to have significant numbers of them leave the country is not a financially desirable or sustainable situation for a country.

While examining health care systems in various countries, we have found significant differences pertaining to human resources management and health care practices. It is evident that in Canada, CHA legislation influences human resources management within the health care sector. Furthermore, the result of the debate on Canada's one-tier versus two-tier system may have drastic impacts on the management of human resources in health care. Additionally, due to a lack of Canadian trained health professionals, we have found that Canada and the United States have a tendency to recruit from developing countries such as South Africa and Ghana, in order to meet demand.

Examination of the relationship between health care in the United States and human resources management reveals three major problems: rapidly escalating health care costs, a growing number of Americans without health care coverage and an epidemic regarding the standard of care. These problems each have significant consequences for the well-being of individual Americans and will have devastating affects on the physical and psychological health and well-being of the nation as a whole.

The physical health of many Americans is compromised because these factors make it difficult for individuals to receive proper consultation and treatment from physicians. This can have detrimental effects on the mental state of the patient and can lead to large amounts of undue stress, which may further aggravate the physical situation.

Examining case studies makes it evident that human resources management can and does play an essential role in the health care system. The practices, policies and philosophies of human resources professionals are imperative in developing and improving American health care. The implication is that further research and studies must be conducted in order to determine additional resource practices that can be beneficial to all organizations and patients.

Compared to the United States, Canada and developing countries, Germany is in a special situation, given its surplus of trained physicians. Due to this surplus, the nation has found itself with a high unemployment rate in the physician population group. This is a human resources issue that can be resolved through legislation. Through imposing greater restrictive admissions criteria for medical schools in Germany, they can reduce the number of physicians trained. Accompanying the surplus problem is the legislative restriction limiting the number of specialists allowed to practice in geographical areas. These are two issues that are pushing German-trained physicians out of the country and thus not allowing the country to take full advantage of its national investment in training these professionals.

Developing countries also face the problem of investing in the training of health care professionals, thus using precious national resources, but losing many of their trained professionals to other areas of the world that are able to provide them with more opportunities and benefits. Human resources professionals face the task of attempting to find and/or retain workers in areas that are most severely affected by the loss of valuable workers.

Human resources management plays a significant role in the distribution of health care workers. With those in more developed countries offering amenities otherwise unavailable, chances are that professionals will be more enticed to relocate, thus increasing shortages in all areas of health care. Due to an increase in globalization, resources are now being shared more than ever, though not always distributed equally.

Human resources implications of the factors

While collectively the five main areas addressed in the article represent health care issues affecting and affected by human resources practices, they are not all equal in terms of their influence in each country. For instance, in Canada there are fewer health care issues surrounding the level of economic development or migration of health workers, whereas these issues are much more significant in developing countries. In the United States, the level of economic development is not a significant issue, but the accessibility of health care based upon an individual's financial situation certainly is, as evidenced by the more than 40 million Americans who have no health care coverage. Germany's issues with the size of its health care worker base have to do with too many physicians, whereas in Canada one of the issues is having too few physicians. Table 2 summarizes some of the implications for health care professionals with regard to the five main issues raised in the article. One of the main implications of this paper, as shown in Table 2 , is that HRP will have a vital role in addressing all the factors identified. Solutions to health care issues are not just medical in nature.

Policy approaches in a global approach to health care delivery

As mentioned at the start of this paper, there are three main health system inputs: human resources, physical capital and consumables. Given that with sufficient resources any country can obtain the same physical capital and consumables, it is clear that the main differentiating input is the human resources. This is the input that is the most difficult to develop, manage, motivate, maintain and retain, and this is why the role of the human resources professional is so critical.

The case studies described earlier showed how human resources initiatives aimed at improving organizational culture had a significant and positive effect on the efficiency and effectiveness of the hospitals studied. Ultimately all health care is delivered by people, so health care management can really be considered people management; this is where human resources professionals must make a positive contribution.

Human resource professionals understand the importance of developing a culture that can enable an organization to meet its challenges. They understand how communities of practice can form around common goals and interests, and the importance of aligning these to the goals and interests of the organization.

Given the significant changes that globalization of health care can introduce, it is important that human resources professionals be involved at the highest level of strategic planning, and not merely be positioned at the more functional, managerial levels. By being actively involved at the strategic levels, they can ensure that the HR issues are raised, considered and properly addressed.

Therefore, human resources professionals will also need to have an understanding not only of the HR area, but of all areas of an organization, including strategy, finance, operations, etc. This need will have an impact on the educational preparation as well as the possible need to have work experience in these other functional areas.

We have found that the relationship between human resources management and health care is extremely complex, particularly when examined from a global perspective. Our research and analysis have indicated that several key questions must be addressed and that human resources management can and must play an essential role in health care sector reform.

The various functions of human resources management in health care systems of Canada, the United States of America, Germany and various developing countries have been briefly examined. The goals and motivations of the main stakeholders in the Canadian health care system, including provincial governments, the federal government, physicians, nurses and allied health care professionals, have been reviewed. The possibility of a major change in the structure of Canadian health care was also explored, specifically with regard to the creation of a two-tier system. The American health care system is currently challenged by several issues; various American case studies were examined that displayed the role of human resources management in a practical setting. In Germany, the health care situation also has issues due to a surplus of physicians; some of the human resources implications of this issue were addressed. In developing countries, the migration of health workers to more affluent regions and/or countries is a major problem, resulting in citizens in rural areas of developing countries experiencing difficulties receiving adequate medical care.

Since all health care is ultimately delivered by and to people, a strong understanding of the human resources management issues is required to ensure the success of any health care program. Further human resources initiatives are required in many health care systems, and more extensive research must be conducted to bring about new human resources policies and practices that will benefit individuals around the world.

World Health Organization: World Health Report 2000. Health Systems: Improving Performance. Geneva. 2000, [ http://www.who.int.proxy.lib.uwo.ca:2048/whr/2000/en/whr00_ch4_en.pdf ]

Google Scholar  

World Health Organization: World Health Report 2003: Shaping the Future. Geneva. 2003, [ http://www.who.int.proxy.lib.uwo.ca:2048/whr/2003/en/Chapter7-en.pdf ]

Book   Google Scholar  

Zurn P, Dal Poz MR, Stilwell B, Adams O: Imbalance in the health workforce. Human Resources for Health. 2004, 2: 13-10.1186/1478-4491-2-13.

Article   PubMed   PubMed Central   Google Scholar  

Kirby MJL: The health of Canadians – the federal role. The Senate of the Government of Canada. 2002, Ottawa, ON: Government of Canada, 6: 78-

Makarenko J: The Mazankowski Report: A Diagnosis of Health Care in Canada. 2002, Edmonton, AB: Government of the Province of Alberta, [ http://www.mapleleafweb.com/features/medicare/mazankowski/index.html .]

Dosanjh U: Canada Health Act Report 2003–2004. 2004, Ottawa: Government of Canada, [ http://www.hcsc.gc.ca/medicare/Documents/CHAAR%202003-04.pdf ]

Ministry of Health and Long term Care: Report on the Integration of Primary and Health Care Nurse Practitioners into the Province of Ontario. 2005, Toronto, ON, [ http://www.health.gov.on.ca/english/public/pub/ministry_reports/nurseprac03/exec_summ.pdf ]

Canadian Health Coalition: The History of Medicare Shows that Canadians Can Do It. Ottawa, ON, [ http://www.healthcoalition.ca/history.html ]

Canadian Nurses Association: Succession Planning for Nursing Leadership. 2006, Ottawa, ON, [ http://www.cna-nurses.ca/CNA/practice/leadership/default_e.aspx ]

Hannah KJ: Health informatics and nursing in Canada. Healthcare Information Management and Communications Canada. 2005, 14: 3-[ http://hcccinc.qualitygroup.com/hcccinc2/pdf/Vol_XIX_No_3/Vol_XIX_No_3_7.pdf ]

Manojlovich M, Ketefian S: The effects of organizational culture on nursing professionalism: Implications for health resource planning. The Canadian Journal of Nursing Research. 2002, 33: 15-34.

CAS   PubMed   Google Scholar  

Health Canada: Investing in Health Care Providers. 2003, Ottawa, ON, [ http://www.hc-sc.gc.ca/english/pdf/romanow/pdfs/HCC_Chapter_4.pdf ]

Romanow RJ: Building on Values: The Future of Health Care in Canada. 2002, Commission on the Future of Health Care in Canada. Ottawa, ON

Kirby MJL: The Health of Canadians – The Federal Role. 2002, The Senate of the Government of Canada. Ottawa, ON: Government of Canada, 4: 111-

National Coalition on Health Care: Building a Better Health Care System: Specifications for Reform. Report from the National Coalition on Health Care. Washington, DC. 2004, 5-12. [ http://www.nchc.org/materials/studies/reform.pdf ]

The Texas Department of Insurance: Health Maintenance Organizations. 2005, Austin, TX, [ http://www.tdi.state.tx.us/consumer/cbo69.html ]

Centers for Medicare and Medicaid Services: Medicare and You. 2006, Baltimore, Maryland, [ http://www.medicare.gov/publications/pubs/pdf/10050.pdf ]

Malat J: Social distance and patient's ratings of health care providers. Journal of Health and Social Behavior. 2001, 42: 360-72. 10.2307/3090184.

Article   CAS   PubMed   Google Scholar  

Anson BR: Taking charge in a volatile health care marketplace. Human Resource Planning. 2003, 23 (4): 21-34.

Jones DA: Repositioning human resources: a case study. Human Resources Planning. 1996, 19 (1): 51-54.

World Health Organization Regional Office for Europe: Highlights on Health. Germany. Copenhagen. 2004, [ http://www.euro.who.int/eprise/main/who/progs/chhdeu/system/20050311_1 ]

The Library of Congress: A Country Study: Germany. Washington, DC. 1995, [ http://lcweb2.loc.gov/frd/cs/detoc.html ]

Organization for Economic Co-operation and Development: OECD Health Data 2005. How Does Germany Compare. Paris. 2005, [ http://www.oecd.org/dataoecd/16/6/34970073.pdf ]

Organization for Economic Co-operation and Development: OECD Health Data 2005. How Does the United States Compare. Paris. 2005, [ http://www.oecd.org/dataoecd/15/23/34970246.pdf ]

Organization for Economic Co-operation and Development: OECD Health Data 2005. How Does Canada Compare. Paris. 2005, [ http://www.oecd.org/dataoecd/16/9/34969633.pdf ]

Bundesministerium für Gesundheit: Information on Medical Training in the Federal Republic of Germany. 2005, Kohn, GDR, [ http://www.bmg.bund.de/cln_041/nn_617014/EN/Health/health-node,param=.html_nnn=true ]

Medknowledge: Working Formalities for Foreign Physicians in Germany. Munster. 2000, [ http://www.medknowledge.de/germany/ ]

National Coalition on Health Care: Health Care in Germany. Washington, DC. 1999, [ http://www.nchc.org/facts/Germany.pdf ]

Vujicic M, Zurn P, Diallo K, Orvill A, Dal Poz MR: The role of wages in the migration of health care professionals from developing countries. Human Resources for Health. 2004, 2: 3-10.1186/1478-4491-2-3. [ http://www.human-resources-health.com/content/2/1/3 ]

Gupta N, Zurn P, Diallo K, Dal Poz MR: Uses of population census data for monitoring geographical imbalance in the health workforce: snapshots from three developing countries. International Journal for Equity in Health. 2003, 2: 11-10.1186/1475-9276-2-11. [ http://www.equityhealthj.com/content/2/1/11 ]

Dussault G, Franceschini M: Not enough here, too many there: understanding geographical imbalances in the distribution of the health workforce. Washington, DC: The World Bank Institute. 2003, [ http://www.lachsr.org/observatorio/eng/pdfs/Geographical Imbalances05-13-03.pdf ]

Findlay J: Doctors with Borders: Struggles Facing Foreign Physicians in Canada. 2005, New Media Journalism. University of Western Ontario. London, ON, [ http://www.fims.uwo.ca/newmedia2005/default.asp?id=166 ]

Findlay J: Facts on Foreign Doctors. 2005, New Media Journalism, University of Western Ontario. London, ON, [ http://www.fims.uwo.ca/newmedia2005/default.asp?id=175 ]

Barney J: Gaining and Sustaining Competitive Advantage. 1997, Reading, MASS: Addison-Wesley Publishing Co.

Download references

Acknowledgements

The authors are grateful to Valerie Sloby from PCHealthcare for her editorial assistance and helping in reviewing the manuscript.

Author information

Authors and affiliations.

Management and Organizational Studies, The University of Western Ontario, London, Ontario, Canada

Stefane M Kabene, Mark A Soriano & Raymond Leduc

Schulich School of Medicine, The University of Western Ontario, London, Ontario, Canada

John M Howard

School of Nursing, The University of Western Ontario, London, Ontario, Canada

Stefane M Kabene & Carole Orchard

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Stefane M Kabene .

Additional information

Competing interests.

The author(s) declare that they have no competing interests.

Authors' contributions

SK conceived the paper, worked on research design, did data analysis and led the writing of the paper. CO, JH, MS and RL all actively participated in data analysis, manuscript writing and review. All authors read and approved the final manuscript.

Authors’ original submitted files for images

Below are the links to the authors’ original submitted files for images.

Authors’ original file for figure 1

Authors’ original file for figure 2, authors’ original file for figure 3, authors’ original file for figure 4, authors’ original file for figure 5, rights and permissions.

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( http://creativecommons.org/licenses/by/2.0 ), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Reprints and permissions

About this article

Cite this article.

Kabene, S.M., Orchard, C., Howard, J.M. et al. The importance of human resources management in health care: a global context. Hum Resour Health 4 , 20 (2006). https://doi.org/10.1186/1478-4491-4-20

Download citation

Received : 13 April 2006

Accepted : 27 July 2006

Published : 27 July 2006

DOI : https://doi.org/10.1186/1478-4491-4-20

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Human Resource
• Health Care System
• Gross Domestic Product
• Human Resource Management
• Health Care Delivery

Human Resources for Health

ISSN: 1478-4491

• Submission enquiries: Access here and click Contact Us
• General enquiries: [email protected]

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List
• v.15(9); 2023 Sep
• PMC10548490

Pakistan’s Healthcare System: A Review of Major Challenges and the First Comprehensive Universal Health Coverage Initiative

Salman j khan.

1 Hematology and Oncology, Mayo Clinic, Jacksonville, USA

2 Public Health, University of Massachusetts, Amherst, USA

Muhammad Asif

3 Internal Medicine, Avera McKennan Hospital and University Health Center, Sioux Falls, USA

Sadia Aslam

Wahab j khan.

4 Internal Medicine, University of South Dakota Sanford School of Medicine, Sioux Falls, USA

Syed A Hamza

5 Cardiology, Punjab Institute of Cardiology, Lahore, PAK

Each country's healthcare system has a different structure and functioning designed to meet the needs of its people utilizing the available resources. Due to ever-growing population needs and constantly emerging public health problems, it is vital for any healthcare system to be ready to adapt, recognize its limitations, and improve its flaws by learning from other healthcare models across the globe. In this article, we analyzed the significant challenges faced by Pakistan's healthcare system (PHS) and the first comprehensive initiative taken for universal health coverage in Pakistan. Inequitable distribution of resources, inadequate healthcare spending, non-adherence to preventative healthcare and brain drain are the major problems in the PHS. On the other hand, the recently introduced universal health coverage initiative, the Sehat Sahulat Program (SSP), can be considered one of the biggest achievements of the country’s healthcare system.

Introduction

No healthcare system can be labeled as perfect because of the growing needs of people, constantly emerging new public health challenges, and the diversity of population demographics around the globe. Every system needs continuous pruning to fulfill the needs of its people through analysis of its shortcomings and strengths. Pakistan's healthcare system (PHS) is not an exception to this principle. PHS comprises private and public sectors, catering to a huge population of more than 220 million [ 1 ]. There are many challenges faced by PHS including inadequate funding, infrastructural limitations, brain drain of health professionals, limited focus on preventive healthcare (PHC), and inequitable resource allocation. Among these issues, Pakistan's first comprehensive universal health coverage (UHC) initiative, Sehat Sahulat Program (SSP), can be considered the most outstanding achievement of the PHS.

PHS faces many challenges that hinder its ability to provide adequate and efficient healthcare services to its citizens. One of the significant challenges is insufficient funding. Pakistan spends around 38 US Dollars (USD) per capita on healthcare, which is much lower than other developing countries [ 2 ]. As compared to Pakistan, India, the Philippines, and Ghana spend 57, 165, and 85 USD per capita on healthcare, respectively [ 2 ]. Pakistan spent 1.2% of its gross domestic product (GDP) on the public health sector in 2020-2021 as compared to 1.1 in 2019-2020, which is not a significant increase when viewed in terms of GDP percentage [ 3 ]. The lack of sufficient investment in the PHS has led to another challenge which is a shortage of health infrastructure, medicines, medical equipment, and qualified healthcare professionals. Although there is an increase in human resources from 2014 to 2021, this growth is not enough to cater to the needs of the population growing at 2% per annum (Table ​ (Table1) 1 ) [ 3 ]. Around 32,879 physicians graduate every year in Pakistan and 40% of them go abroad for better opportunities citing low income, long hours of job, and inequality as the main reasons [ 4 ]. According to a study conducted at two different medical colleges, 33% of medical students plan to leave the country to practice healthcare abroad. This brain drain puts undue pressure on the PHS resulting in inadequate provision of health facilities to people.

Open access source: Pakistan Economic Survey 2021-22 [ 3 ]

Limited focus on PHC is another significant issue PHS faces. PHC includes measures to prevent diseases and promote health, such as immunizations, screenings, and health education. Pakistan's government has taken several steps over the years to promote PHC which include the Lady Health Workers (LHW) programme, the expanded programme on immunization (EPI), the Polio Eradication Initiative (PEI) Programme, the Malaria Control Programme (MCP), Tuberculosis (TB), Control Programme, and establishment of basic health units (BHUs) and rural health units and rural health units (RHUs). In 2021, there were 1,276 hospitals, 5,558 BHUs, 736 RHCs, 5,802 Dispensaries, 780 Maternity and Child Health Centers, and 416 TB centers in Pakistan [ 3 ]. However, all these initiatives have not been able to drastically improve the health indicators of Pakistan, which are much worse than its peers (Table ​ (Table2). 2 ). These initiatives are not enough for a population of more than 220 million [ 1 ]. There is still a scarcity of resources in the PHC realm and the people do not have access to these services because of less developed PHC centers or even the absence of these centers nearby. Even with access to these facilities, the population does not get involved in preventive health because of a lack of awareness and education regarding its importance to their own health. Many people in Pakistan lack basic health literacy, which means they do not have the knowledge and skills to access and use healthcare services effectively.

Open access source: Economic Survey of Pakistan [ 3 ]

The inequitable distribution of healthcare resources is a serious threat to the PHS. The healthcare resources, including hospitals, clinics, and healthcare professionals, are concentrated in the urban areas, leaving rural areas with inadequate healthcare facilities. It leads to a significant disparity in healthcare access and outcomes between urban and rural populations [ 5 ]. The Community Health Index (CHI) reflects the unequal distribution of healthcare resources. CHI measures the disparities between different regions based on health and well-being. Pakistan scored an inequality ratio of 16.59 CHI, which means that the upper-tier districts are 16.59 times healthier than the lower-tier districts. The disparity ratio differs by approximately 10 points between the urban and rural areas (7.78 and 17.54, respectively) showing a huge disparity in resources. This data reveals the inequitable distribution of resources in the healthcare domain in Pakistan [ 5 ]. Consequently, the rural healthcare system lacks basic medical equipment, diagnostic facilities, and medications, leading to a lack of proper patient diagnosis and treatment. All these shortages increase the burden on the infrastructure in cities and, in turn, lead to inadequate provision of health facilities, physician shortages, and dissatisfaction among patients.

Sehat Sahulat Program: UHC Initiative

UHC is a concept coined by the WHO that aims to ensure essential health services to everyone without any financial hardship. UHC is a part of the Sustainable Development Goals (SDGs) adopted by the United Nations (UN) in 2015. Pakistan is a signatory of the SDGs. The goal of UHC is expressed in the UN 2030 agenda as part of the SDGs in Goal 3, which focuses on health (target 3.8). UHC is the primary step toward providing health as a fundamental right of citizens [ 2 ]. The biggest achievement of the PHS is the UHC initiative in the form of the SSP.

SSP is a public sector-funded health insurance initiative of the federal and provincial governments working to provide financial health protection to all citizens against extraordinary healthcare expenditure. SSP is a landmark healthcare initiative that is considered an important step toward UHC. SSP was implemented first by the Khyber Pakhtunkhwa (KPK) provincial government in 2015 to provide free health insurance coverage to the poor and vulnerable populations only. Then, the federal government of Pakistan in cooperation with the provincial governments rolled out the SSP in other provinces in 2019. The program is funded by the government of Pakistan and is managed by the Ministry of National Health Services, Regulations, and Coordination. The program has two main components: (i) free health insurance coverage for eligible households and (ii) a network of participating hospitals and clinics where eligible households can access healthcare services. The SSP initially provided social insurance only to families living below the poverty line but is now gradually moving toward every citizen. As of 2022, the SSP has been implemented in 36 districts of Punjab, 35 districts of Khyber Pakhtunkhwa, 10 districts of Azad Jammu and Kashmir (AJK), 10 districts of Gilgit Baltistan (GB), Islamabad Capital Territory (ICT) and Hardaker district of Sindh, reaching approximately 44.6 million households. The Public Sector Development Program (PSDP) is responsible for contributing premiums from ICT, AJK, GB, Federally Administered Tribal Areas (FATA), and Thar Parker districts. However, Punjab and KP fund 100% premium contributions from various sources [ 3 ].

Under the SSP, households receive health insurance cards, which can be used to access healthcare services up to one million rupees per year at participating hospitals and clinics. The program covers a wide range of inpatient services, including cardiac procedures, cancer management, burn management, dialysis, complications of diabetes mellitus, trauma management, neurosurgical procedures, abdominal surgeries, fracture management, and other medical and surgical interventions [ 3 ]. The program has a tiered benefit structure with higher benefits for households with more vulnerable members, such as women, children, and older people. The SSP has a vast network of more than 1030 paneled hospitals across Pakistan. Beneficiaries from any district can get treatment from any of these paneled hospitals. The program has also positively impacted the financial protection of marginalized communities. Transgender people and persons with disabilities registered with the National Database Regulatory Authority (NADRA) were also enrolled in this program. They have given access to UHC, a giant leap in the inclusion of the ignored community [ 3 ]. In Pakistan, out-of-pocket (OOP) expenditures on health are more than 60% of the total health expenditure [ 3 ]. The SSP has shared this cost at every level of healthcare. Moreover, it serves 154 million people in Pakistan, which is the first-ever health insurance initiative in the history of Pakistan [ 3 ]. Till March 8, 2022, over 3.2 million hospital visits have been recorded under the SSP's health cards [ 1 ]. The shared health expenditure has also facilitated people's access to medical services which they used to avoid in the past due to high healthcare costs, thus promoting health and wellness.

There are a few limitations to this program. Many families have complained about the incompatibility between the cost of treatment in private-sector hospitals and the limits set by the program. Patients are expected to pay the difference. In some instances, patients were turned away without any medical services due to the inability to pay [ 1 ]. Another issue is the interrupted continuity of the SSP due to recent political and economic instability in Pakistan. It is still functional in some parts of the country while being suspended in others.

As Pakistan is a developing country, its healthcare system must make many improvements to meet the needs of its population. The challenges faced by Pakistan's healthcare system include insufficient funding, inadequate healthcare workforce and infrastructure, less focus on preventive health, and inequitable distribution of resources. These challenges need comprehensive policy formulation focused on increases in healthcare funding and allocation of equity-based resources. The most significant achievement of PHS is the initiative toward UHC through the SSP. This initiative has decreased the burden of healthcare expenses and increased access to healthcare services for people, including marginalized communities.

The authors have declared that no competing interests exist.