The Impact of Health Information Technology on Hospital Performance: A Systematic Integrative Literature Review ()
1. Introduction
To maximise the effectiveness and efficiency of clinical care delivery, hospitals improve their performance by using digital technologies, referred to as health information technology (HIT). HIT includes different types of functionalities, such as electronic clinical documentation, results viewing, computerised provider order entry, and decision support [1] - [7] . These functionalities may be integrated in one application, e.g., in electronic health records or electronic medical records (EHR), or they are supported by separate applications with interfaces for data exchange. HIT applications recognise different types of users, such as medical doctors, nurses, pharmacists, and patients [8] .
Yet, despite their importance, we still have a limited understanding of how HIT affects hospital performance, as well as an insight in what this impact of HIT functionalities on hospital performance is. There are two reasons for this. First, the current literature does not provide a conclusive answer whether HIT contributes to hospital performance, despite many studies on the impact of HIT [9] . Second, HIT is by nature a multidisciplinary research field, and it only has been studied separately within the medical, information system or information management research streams, leaving us with only a fragmented understanding of the effect of HIT on hospital performance.
Given our limited understanding and the amount of time and money hospitals spent on implementing HIT, there is a need for a cross-disciplinary synthesis of the HIT studies by making a connection between divergent literature streams. Therefore, we systematically synthesise the quantitative and qualitative studies of HIT as well as provide research directions for researchers studying HIT. By doing so, we provide an overview of what is known, and we develop an integrative understanding of what and how specific types of HIT impacts specific hospital performance indicators. We use a three step approach. First, we organise our research in a framework that encompasses the various aspects of HIT, using an affordance actualization lens [10] [11] . Second, we use this framework to identify what is already known and what remains unknown. Third, we identify future research opportunities.
Our research makes several contributions. First, it provides a framework to organise and categorise the existing literature on HIT, HIT use and hospital performance. The research framework enables us to give an integrative overview of the current status of HIT studies in hospitals and supports us in identifying research gaps and research opportunities. Second, using our framework, we suggest a distinction in types of HIT functionalities and specific dimensions of hospital performance indicators. This categorization helps us to understand mixed results. The proposed framework could help hospitals and researchers to make decisions regarding HIT functionalities and the effects of HIT use in hospitals.
Given our research outcomes, we suggest three overarching future research opportunities to further improve our insight on the impact of HIT on hospital performance. First, future studies should use a reference to types of HIT functionalities to research various aspects of HIT implementation and use. Second, there is a need to study use of HIT. Third, research should examine multiple hospital performance indicators to elucidate trade-offs and interactions in hospital-level outcomes, while differentiating between hospital characteristics.
2. Materials and Methods
2.1. Design and Search
We aimed to systematically review the quantitative and qualitative studies in HIT across multiple disciplines. We therefore mapped existing research to our theoretical research framework, to create an overview of what has been studied and to identify gaps and propose directions for future research. We followed an integrative literature review for searching, screening and synthesis of literature [12] [13] .
We used the Discover! Search engine. Discover! includes many databases, such as EBSCO, Science Direct, Emerald, Springer, Sage, NARCIS and Wiley-Blackwell. To capture as many relevant studies as possible, we developed a broad search string. The search string consists of three parts, roughly “Health Information technology” and “performance indicators” and “hospitals”. Each part contains several keywords. For our search string, see Appendix A. The searches were conducted on August 13th 2022, by searching the abstract of the studies, published in English and Dutch from 2010 to 2023. We choose 2010 as a ‘base line’ given the impact of Agarwal, Gao, DesRoches et al.’s (2010) [9] research. The studies were then uploaded to Mendeley Software and we removed duplicates. Given the broad scope of our research, we could not further tighten the search string. Hence, we included the top 10 journals from multiple research streams as suggested by Webster and Watson (2002) [12] ; namely information systems research, healthcare research, medical research and management and accounting research. In order to obtain the most comprehensive understanding, we included nine different journal guides: Academic Journal Guide 2021 Information management, Academic Journal Guide 2021 Operations and Technology Management, SJR Information systems and management, SJR Management information systems, Academic Journal Guide 2021 Public sector and Health Care, SJR Health professions, SJR Medicine, SJR Pharmacology, Toxicology and Pharmaceutics and SJR Business, management and Accounting. We also included via snowballing “Journal of the American Medical Informatics Association” and “Health Policy and Technology”.
In our first screening step, we screened the titles and abstracts. We included studies that reported on HIT and at least one of our outcome variables. We excluded studies that focused on medical research without HIT use, healthcare system research without hospital performance, research that focused on HIT or outcome variables not both, or research focused on specific HIT applications such as telemedicine, electronic prescription and big data analytics. In our second screening step, we read the selected studies in full text. We excluded one study because the full text was not available and we excluded other studies because, on closer inspection, they were about specific HIT sub-applications, such as supply chain logistics, Internet of Things, revenue cycle management and electronic drug prescription systems only.
2.2. Data Collection and Synthesis
We followed Jiang and Cameron (2020) [11] to categorise and synthesise our literature review by adapting Strong, Volkoff, Johnson et al.’s (2014) [10] affordance actualization theory. Affordance actualization theory explains how HIT functionalities influence hospital goals through the use of HIT. An IT affordance is’ the potential for behaviours associated with achieving an immediate concrete outcome and arising from the relation between an artefact and a goal-oriented actor or actors [10] . To align literature to our research framework, we made a general profile of the included studies by using four categories: 1) the functionalities of HIT and how these functionalities are measured, 2) use and immediate outcomes of HIT functionalities 3) different performance indicators and how HIT functionalities affect them and 4) what hospital characteristics influence the outcome of hospital performance.
3. Results
Our primary search yielded 62,658 references (see Figure 1). After uploading
the references to Mendely Reference Manager and removing duplicates, 49,758 unique studies remained. After selection of journals based on the nine included journal guides, our review included 1152 studies from 81 unique journals. After our screening of these studies based on our exclusion criteria, we included 85 studies that reported on HIT and at least one of our selected outcome variables or on HIT and use. After our second screening based on our exclusion criteria, 52 studies were included from 15 unique journals. From each study, we extracted the study identification information such as author name(s), title, journal name and year of publication. We also extracted study characteristics such as study setting, type of HIT, use of HIT, performance indicator measures, and HIT data source(s). For a complete overview of the results, see Appendix B.
3.1. Characteristics of the Included Studies
Most studies appear in the information system and information management research stream. In the medical research stream, based on journals selected from the journal guides in this discipline, we did not find any relevant studies including HIT and the impact of HIT. The US is the country in which the impact of HIT on hospital performance has been studied the most, with 32 out of 52 studies. Only eight studies focused on countries outside North America and Europe. The level of analysis of the studies within our literature review varies, differing from hospital level studies (71%), medical department level (10%), disease specific level (10%) and a combination of levels (9%). Most study designs (81%) used quantitative research to analyze the impact of HIT, as opposed to qualitative research (13%). Some authors use a combination of methods (6%).
3.2. HIT Functionalities and Their Measurement
In the literature, different authors use a range of definitions referring to HIT and categorise HIT into different functionalities and their affordances [14] [15] . Our analysis of the literature revealed four types of HIT: clinical HIT, decision support HIT, administrative HIT and patient engagement HIT. Clinical HIT describes basic functionalities like record keeping and results viewing and are referred to by names such as clinical information systems, EHR or health information
systems [1] [2] [3] [4] . Decision support HIT (or advanced clinical HIT) describes enhanced features to bolster decision-making capabilities [4] [5] [6] [7] . Patient engagement HIT describes functionalities such as patient monitoring or telehealth [16] . Administrative HIT describes functionalities such as ERP systems that integrate and manage various administrative and financial processes within hospitals [1] [3] [17] . For an overview of types of HIT functionalities in hospitals, see Figure 2.
The lack of a standardised HIT definition also affects the way HIT functionalities are measured. We found that HIT functionalities are measured roughly in four ways: 1) seven studies used the American Hospital Association Annual Information Technology Survey1, 2) twelve studies used the Healthcare Information and Management Systems Society2 Analytics Database, 3) four studies used a combination of AHA and HIMSS data and 4) twenty-six studies used other (self-developed) questionnaires, secondary data or meta-analyses.
3.3. HIT Use and Immediate Outcomes
The equations are an exception to the prescribed specifications of this template. You will need to determine whether or not your equation should be typed using either the Times New Roman or the Symbol font (please no other font). Equations should be edited by Mathtype, not in text or graphic versions. You are suggested to use Mathtype 6.0 (or above version).
According to affordance theory, the actual use of HIT functions and their affordances enable medical professionals to achieve their goals and tasks [18] . Therefore, HIT use is an important variable to consider [5] [6] [15] [19] [20] [21] [22] . However, in our review only ten studies (19.2%) define the use of HIT by explicitly measuring the use rate of HIT, for example by using the technology acceptance model (TAM) or the unified theory of acceptance and use of technology (UTAUT) [23] . These studies show there are factors that influence the use rate of HIT, for example user characteristics, the existence of technical and organisational infrastructure to facilitate the use of a system and the culture of a country [24] [25] . Thirteen (25%) studies refer to the use of HIT, but only measure parts of the HIT functionality. For example, measuring “meaningful use” based on the CMS programme data. Fourteen (27%) studies in our literature review implementation and adoption are used interchangeably but are not separately measured. Therefore saying little about actual use [1] [7] . Fifteen (29%) studies in our literature review do not mention the use of HIT at all.
Only two studies explicitly measure the use of HIT in relation to hospital performance indicators. These studies show a positive impact of HIT use on medical professional satisfaction and HIS use on patient satisfaction [16] [26] .
To measure usage, process quality indicators might be useful. Process quality indicators provide insight if the process of providing care is delivered as intended. For example, whether aspirin is given on time or whether certain actions are performed in a timely manner. Process quality indicators thus say something about the assimilation of HIT with work processes, and this assimilation is necessary to increase hospital performance like mortality and patient satisfaction [1] [27] .
3.4. Hospital Performance and HIT
3.4.1. Performance Indicator Dimensions
In our literature review, 35 studies apply hospital performance indicators, varying in dimensions, such as quality of care, efficiency (costs), medical professional’s satisfaction and patient satisfaction. Of the 35 studies, 19 studies address only one dimension such as quality of care or efficiency, while another 13 studies address two dimensions. Two studies, which were conducted outside the US before 2014, encompass three dimensions. Only one study was found covering all dimensions.
3.4.2. Quality of Care
In general, studies on quality of care indicate that HIT lowers admissions, readmissions or mortality [1] [15] [21] [28] [29] [30] . Others suggest that HIT has no effect on readmissions or mortality [22] [28] [30] . Studies also find negative effects
![]()
Table 1. HIT functionalities and effects on quality of care in hospitals. Explanation of symbols and colours: ↑ higher, ↓ lower, colour green positive, colour red negative.
of HIT on complications and disease specific measures [29] [31] or found mixed results on safety, disease specific measures and the IQI, a general measure of quality of care [5] [32] [33] . Sometimes, inconsistencies can also be observed within the same study, adding to the complexity of the findings [29] . However, as Table 1 suggests, in general, HIT has a positive impact on mortality and patient readmissions in hospitals.
3.4.3. Efficiency
Evidence on the effects of HIT on efficiency also shows mixed results. HIT is found to reduce costs [2] [30] [36] [38] and the number of radiology exams [37] [38] . However, studies also suggest that HIT increases hospital costs and nurse staffing levels [6] [29] . Contrary to expectation, studies showed mixed results to reduce length of stay [15] [29] [30] [34] [35] [36] . HIT increases resource use [4] and hospitals had lower productivity gains compared to facilities that have not yet implemented HIT [39] . Also the use of HIT leads to a higher number of patients that face diagnosis related groups, indicating that HIT use could lead to higher patient costs through up coding [40] . For more information see Table 2.
3.4.4. Medical Professional Satisfaction
Studies suggests positive outcomes of HIT on medical professional satisfaction, support of decision making when prescribing mediations, and ease of requesting laboratory tests [43] [45] [46] . However, medical professionals also experience HIT as cumbersome to use and adding to their workload [26] [33] [45] [47] . For a complete overview of the studies and these effects of HIT, see Table 3.
![]()
Table 2. HIT functionalities and effects on efficiency in hospitals. Explanation of symbols and colours: ↑ higher, ↓ lower, colour green positive, colour red negative. The IQI 91 is a hospital-wide quality indicator that measures multiple quality indicators.
![]()
Table 3. HIT functionalities and effects on medical professional satisfaction. Explanation of symbols and colours: ↑ higher, ↓ lower, colour green positive, colour red negative.
![]()
Table 4. HIT functionalities and effects on patient satisfaction. Explanation of symbols and colours: ↑higher, ↓ lower, colour green positive, colour red negative..
3.4.5. Patient Satisfaction
As for patient satisfaction, studies show positive effects of HIT use on patient satisfaction and patient loyalty [1] [16] [42] [43] [48] . However, some HIT functions, such as documentation and health information exchange improve patient outcomes, whereas clinical decision support functions negatively affect these outcomes [32] . Meyerhoefer, Sherer, Deily et al. (2018) [46] specifically found that HIT systems negatively impacted patient satisfaction during implementation. For a complete overview of the studies and these effects of HIT, see Table 4.
3.4.6. Other
We found seven performance indicators [17] [21] [27] [33] [40] [50] that do not fit within the four previously mentioned dimensions. For example, number of lawsuits, [17] malpractice insurance premium [21] , and reuse of data [33] . We bundled these performance indicators into the category “other”. For more information see Appendix B.
3.4.7. Influencing Hospital Characteristics
The literature review reveals several hospital characteristics that may affect the relationship between HIT and hospital-level outcomes. First, Agarwal, Gao, DesRoches et al.’s (2010) [9] research suggests that future studies should differentiate between the various types of hospitals, such as ownership status, location, teaching status, system affiliation and hospital size. Of the 46 quantitative studies included in our research, ten studies do not examine the impact of HIT on hospital performance but focus on studying HIT usage and factors for satisfaction. In six of these 46 studies the hospital population consisted of only one or a few hospitals, therefore these studies show no statistically relevant results. Six other studies did not distinguish between hospital characteristics. The remaining 24 did distinguish between different hospital characteristics, although sometimes only as a control variable. Only 14 studies explicitly indicate whether they discover variances, and these results show a fragmented picture [1] [2] [3] [5] [7] [21] [28] [30] [34] [35] [36] [40] [44] [49] . For example, HIT more positively affects process quality in small rural hospitals [7] , HIT more positively affects costs and readmissions in large hospitals that treat less complex cases [36] and HIT leads to a higher amount of readmissions and mortality in for profit hospitals than in not for profit hospitals [30] .
Furthermore, the impact of HIT on a single performance indicator may conceal trade-offs between indicators. For example, dissatisfaction of medical professionals with HIT and difficulties incorporating HIT into patient care may negatively impact patient satisfaction [46] .
Also, HIT consists of many subsystems, which may lead to varying influence on performance metrics. We found four reasons for these variations: hospitals implemented subsystems in a different sequence [5] , hospitals implemented subsystems with a different strategy (bottom up versus top down or big bag versus phased) [20] [39] , hospitals implemented subsystems to support different type of illness (chronic or acute) [5] and hospitals implemented different combinations of subsystems [5] [32] [35] [41] .
Finally, the duration of HIT usage also affects performance indicators. This duration is called a “lag”; the time between implementing a system and the moment of measuring its influence on hospital performance. Many researchers discuss that including a lag is important, although they have not always done so themselves [2] [7] [25] [27] [34] . In studies that do include a lag, it varies in time: up to a year after implementation [4] [29] [30] [37] [41] [42] [48] , one to one and a half years after implementation [51] , two years after implementation [1] [5] [29] , three years after implementation [29] [44] , and two to six years after implementation [41] .
4. Discussion
Our literature review identifies four HIT functionalities and five dimensions of hospital performance indicators, highlights their respective impacts as described in literature, and offers a conceptual research framework to better understand how these technologies are used. Figure 3 summarises all the suggestions.
Our review reveals several issues in the HIT literature. First, our research shows that comparing outcomes from previous studies is challenging because of differences in HIT definitions. Therefore, forthcoming studies should establish a unified definition of HIT to facilitate further advancement in the field. We believe that the identified types of HIT in this study are able to properly incorporate new technological developments in this domain. Additionally, an exploration is warranted into how diverse combinations of HIT applications [5] [32] [35] [41] , their support of chronic versus acute medical conditions [5] , their implementation sequencing [5] , and implementation strategies [20] [39] impact hospital performance.
Second, our research underscores that simply implementing HIT is not enough, HIT must be properly used to influence performance [7] [22] [39] . Yet, only a few studies to date have examined the combination of HIT functionalities, usage and performance indicators. And when they did, they did not measure use of HIT the way it was intended, which calls for more research into its use. As hospitals may concurrently implement other procedural enhancements alongside HIT functionalities, forthcoming research can integrate process indicators to measure immediate outcomes of HIT use [6] [17] [22] [39] .
Third, our research shows that previous studies show a partial understanding of hospital performance, by reducing outcome to one or two performance indicator dimensions, such as quality of care and efficiency. And even within dimensions, most studies focus on only one or two performance indicators. The question arises whether a single indicator is representative of an entire dimension. Consequently, more research is needed that examines more performance indicators simultaneously and future research can also examine trade-off effects or interactions between hospital level outcomes [2] [3] [6] [7] [29] [43] [51] . Future research must also differentiate between hospital characteristics, such as ownership type (for-profit or not-for-profit), teaching status, healthcare system affiliations and the duration of use (lag).
Our research is not without limitations. First, we conducted a literature search using a broad search strategy. Although this strategy allowed us to include a wide range of studies, it also required us to select studies from 81 unique journals, excluding other studies. Second, we cannot make generic statements about the influence of HIT on hospital performance because HIT definitions are not standardised and different outcome measures are used. Our study thus provides a good overview of the current state of research, but also shows that much remains to be researched.
5. Conclusion
The value of HIT has been extensively studied, and our literature review provides an overview of what is known about how HIT influences hospital performance. Unfortunately, the results of previous studies contradict each other: some are positive, some neutral and some negative. Our findings suggest that different definitions circulate in the existing literature, and therefore the scope of studies differs, which makes it hard to compare results. Additionally, results of previous studies may be distorted, as studies examine HIT with a limited number of performance indicators, distinguish different kind of hospital characteristics, and rarely measure the combination of HIT functionalities, usage and performance indicators. Given the amount of time and money spent by hospitals on implementing HIT, we propose that an intensified exploration into the value of HIT is imperative, encompassing actual use analysis and the establishment of uniform HIT definitions. The proposed framework could help hospitals and researchers to make decisions regarding HIT functionalities and the effects of HIT use in hospitals.
Acknowledgements
The authors are most grateful to the editors and the reviewers for guidance provided on their work.
Contributors
AW: concept and design, data acquisition and analysis, drafting of the manuscript, revision of the manuscript, and approval of the final version. CH: concept and design, revision of the manuscript and approval of the final version. WJB: concept and design and approval of the final version.
Appendix A
Research String
We used the following search string: (health information technology) or HIT or (electronic health records) or (electronic health record) or EHR or (electronic medical record) or (electronic medical records) or EMR or (health it) or (healthcare IT) or (health care IT) AND (quality of patient care) or (quality of care) or (quality) or (patient safety) or efficient or efficiency or performance or (value based healthcare) or VBHC or satisfaction or productive or productivity or cost or costs or (patient flow) or usage AND hospital or hospitals.
Appendix B
![]()
Table 1. Literature review overview.
Source: * Abbreviations: HIT Health Information Technology, HIS Hospital Information System, CPOE Computerized Physician Order Entry, CDS Clinical Decision Support, EHR Electronic Health Record, EMR Electronic Medical Record, PMIT Patient Management IT, TSIT Transactional Support IT, CIT Communications IT, AIT Administrative IT, HIE Health Information Exchange; ** HIT scope Clinical means for example documenting, viewing and ordering, decision support means one or more decision support systems for medical professionals, administrative means administrative systems for example Enterprise Resource Planning and data analytics, patient engagement means systems like tele monitoring and a portal for patient self-collected data, n/a means in the publication a definition of the HIT is lacking.; *** Usage or adoption measured as mentioned in the publication. “–” means that usage and adoption are not measured, “=’’ means that usage and adoption is in these articles is the same as HIT implementation and are not separately measured (authors therefore use usage/adoption and implementation as interchangeable definitions).CMS MU means Centers for Medicare and Medicaid Services Meaningful Use; *** Explanation of symbols and colors: ↑ higher, ↓ lower, n neutral, colourgreen positive, colour red negative.
NOTES
1The American Hospital Association IT survey database is often used in quantitative research, allowing researchers to include thousands of hospitals in their research. The AHA IT survey focuses on the clinical domain and particularly investigates 31 HIT functionalities and other functionalities (such as telehealth and remote patient monitoring).
2The Healthcare Information and Management Systems Society (HIMSS) is a global non-profit organization and collects data in the US on the functionalities and use of HIT. The different surveys that are used based on the HIMSS multiple databases include HIT functionalities such as EHR, ERP, clinical decision support, radiology and nurse staffing (Gardner 2015, Oh 2018), depending on the database that is used.