Health Care Provider Adoption of eHealth: Systematic Literature Review

1Poor health care outcomes lead to increased levels of morbidity and mortality, and obstruct countries’ prosperity and business profitability (eg,). eHealth is an application of information and communication technologies (ICT) across health-related functions. The benefits of eHealth, such as improved operational efficiency, higher quality of care, and positive return on investments have been well documented in the literature.

2eHealth is an emerging field at the intersection of medical informatics, public health, and business, and refers to health services and information delivered or enhanced through the Internet and other related technologies. Different eHealth applications have been used across countries, corresponding to their health needs and priorities. The World Health Organization (WHO) eHealth for Health Care Delivery (eHCD) program, for example, targeted primary health care in a number of countries in the Asia-Pacific region. Some of these countries have instigated telemedicine as a means of bringing specialist health care to rural communities, whereas some others have endeavoured to improve the safety and continuity of patient care through the use of electronic health records (EHR).

3While there has been high interest in eHealth, the adoption and acceptance rates have not been high enough for health care systems to experience the maximal benefits eHealth has to offer. Past experience of eHealth adoption in the United States, for example, informed us that the low adoption rate could be attributed to both macro-level factors (eg, supportive policies) from the perspective of the public, health care organization, and system, and micro-level barriers from the perspective of health care providers (eg, physicians’ perception about technological complexity,).

4A broad spectrum of research methodologies have been used to study eHealth adoption and acceptance factors based on information provided in published studies. The methodologies include quantitative surveys, observations, qualitative focus groups, ethnographic studies, and personal intuition and experience. According to the results of these studies, different eHealth adoption factors may have led to difficulty for decision makers to explicitly understand, measure, and decrease inhibiting factors or enhance facilitating forces. Hence, there is a need to synthesize those insights and provide decision makers with a holistic view of eHealth adoption.

5Health care providers are the key driving force in pushing eHealth initiatives. eHealth implementation represents a disruptive change in the health care workplace. The change does not occur simply from the introduction of ICT infrastructure but may also require remodelling of the job design of interconnected health professionals to effectively and efficiently incorporate technology. Without the presence of motivational forces (eg, health care providers’ dissatisfaction with the status quo), it is unlikely that the innovation process would be initiated. If health care providers resist change or do not possess attributes necessary for change (eg, adaptability and growth-orientation), the change process is less likely to proceed. The objective of this paper was to identify and synthesize the factors influential to health care providers’ acceptance of various eHealth applications.

6In light of the guidelines originally proposed by and already applied in several systematic reviews (eg,), we conducted a systematic literature review on eHealth adoption. For the specific objective of this study, the guidelines have been modified and 4 steps were taken: (1) identification of resources, (2) selection of relevant papers, (3) data extraction, and (4) data analysis and validation.

7A literature search was conducted between October and November 2011 using 8 online databases: Medline, Cinahl, Web of Science, PubMed, PsychInfo, ERIC, ProQuest Science Journals, and EMBASE. These databases were thought to be the most likely to publish eHealth adoption related work. All search fields available from each search service were used. In each database, the search was repeated 3 times using the following phrases (operators came before keywords): [“e-Health” AND “Adoption” OR “User Acceptance”] or [“eHealth” AND “Adoption” OR “User Acceptance”] or [“EMR” AND “Adoption” OR “User Acceptance”] or [“EHR" AND “Adoption” OR “User Acceptance”].

8The terms “electronic medical records” (EMR) and EHR were separately used to search papers. This is because the EMR/EHR consists of patient health related information and forms the core of eHealth systems. The inclusion of those papers increased the validity of the findings. Table 1 lists the number of papers found in each database using the search phrases. In summary, a total of 3315 papers were found, of which 420 papers were duplicated. The selection process excluded the repeated papers from the archive and produced a list of 2895 papers.

9The full texts of the selected papers were reviewed for relevance. Papers with the following criteria were filtered out:

10articles not written in English

11articles that did not directly use the terms “adoption” and “eHealth” or related terms in the title, abstract, or entire text, with casual referencing of eHealth adoption related issues.

12articles without empirical evidence

13articles which discussed adoption or user acceptance of eHealth but not from the health care provider’s perspective

14This examination process had two iterations. Finally, 93 relevant papers were selected.

15The key information was extracted from the 93 papers. The extracted data included: (1) characteristics of the study (eg, year of publication and health care settings where the studies were conducted), (2) the study results and output—eHealth adoption factors. Relevant text was extracted or retyped verbatim and was added to a database.

16Figure 1 illustrates the analysis process of the data collected in Step 3. Based on the terminologies or terms utilized in the papers, 49 eHealth adoption/acceptance factors were initially extracted. All citations used to identify the results were noted. The next activity was to study the definitions used in the papers. Factors with close relevance were combined, generating a list of 40 factors. For example, “time required to select, purchase, and install the eHealth system”, “time involved in learning to use the eHealth system and additionally required to become familiar with the system operation”, and “the degree to which use of the innovation is perceived as being time consuming” were all grouped to “time cost”.

17Based on the perceived commonality of the themes, the 40 factors were analyzed and organized according to the Unified Theory of Acceptance and Use of Technology (UTAUT) by Venkatesh et al. The UTAUT set out to integrate the fragmented theory and research on individual acceptance of information technology into a unified theoretical model, which highlights the importance of contextual analysis in developing strategies for technology implementation within organizations. This model accounts for 70% of the variance in usage intention—a substantial improvement over any of the original 8 models and their extensions. Within the UTAUT, 3 core constructs that impact on behavioral intention, and consequently use behavior, are performance expectancy, effort expectancy, and social influence, whereas the other core construct facilitating conditions has a direct impact upon use behavior. Four moderators (ie, gender, age, voluntariness of use, and experience) have also been incorporated in the UTAUT. Apart from the 4 core constructs and 4 moderators, another cluster of eHealth adoption factors, which could not be mapped against the UTAUT, was identified. Accordingly, the factors were initially grouped into 9 clusters (Figure 1).

18To search for convergence among multiple sources of information and methods of data collection and analysis, a validity procedure was applied. First, the eHealth adoption factors were reanalyzed within and across the clusters to ensure consistency and independence. The factors were regrouped into 7 clusters:

19health care provider characteristics (eg, IT experience and knowledge, gender, age, and years in practice)

20medical practice characteristics (eg, practice size and teaching status)

21voluntariness of use

22performance expectancy (eg, perceived usefulness and needs)

23effort expectancy (eg, perceived ease of use)

24social influence (eg, subjective norm)

25facilitating or inhibiting conditions (eg, legal concerns)

26The clusters were then given labels and reviewed once more for consistency. Reassessment and relabelling were performed for some papers. This step was repeated until a consensus was reached on the labels for clusters. In the final analysis, papers were reassigned to appropriate clusters. The resulting clusters represented another level of abstraction.

27This section presents the results of statistical analyzes on the characteristic data extracted from the 93 papers, including: (1) the growth of publications by years, (2) distribution by geographical areas, (3) types of research methodologies employed, (4) eHealth applications studied, (5) health care settings selected, and (6) study participants.

28Figure 2 shows the growth in the publications. The growth represented by the curve was not linear, with a dramatic rise in the number of papers published after 2005.

29The majority of the studies (72/93, 77%) were conducted in North America, followed by Europe (9/93, 10%), and Asia (7/93, 8%).

30Quantitative methodology was predominately used by 57/93 studies. The number was nearly twice as large as that of qualitative studies.

31The 93 papers addressed a wide range of eHealth applications. 57 targeted the EHR/EMR, which was defined as computerized medical information systems that collect, store, and display patient information. Telemedicine/Telehealth was the second most popular application studied (addressed by 7/93 studies). Telemedicine frequently referred to the use of a wide array of technologies to deliver a range of medical services to persons at some distance from a health care provider. The remnant studies examined the acceptance of other eHealth applications such as Intensive Care Information System (ICIS), e-discharge which helps inpatient physicians to track pending tests at hospital discharge, Anesthesia Information Management System (AIMS), and electronic logistics information system.

32The majority of the studies were conducted in hospitals and office-based clinics (primary care). In some studies, multiple health care settings of different types were chosen to examine the eHealth acceptance issue. For example, Jha et al used survey data from stratified random sample of all medical practices in Massachusetts in 2005 to determine rates of EHR adoption and perceived barriers to adoption.

33The majority of the studies (ie, 68/93) focused on physicians. Nurses and other health workers were recruited in 25 research projects on eHealth adoption and acceptance.

34Through the data analysis and validation process, 40 factors were identified to be influential to the health care providers’ acceptance of eHealth and grouped into 7 clusters (Figure 3 and Table 2). A brief description of each cluster is provided below.

35A health care provider’s characteristics included his/her information technology (IT) experience and knowledge, years in medical practice, professional role, age, gender, and race. Characteristics in relation to a health care provider’s medical practice included the practice size, teaching status, location, single or multi-specialty, practice level, types of third party payers, and patient age range. Voluntariness of use was defined as “the degree to which use of the innovation is perceived as being voluntary or of free will”. Performance expectancy was defined as the degree to which a health care provider believes that using the eHealth system will help him or her to attain gains in job performance. It included the perceived usefulness and needs, relative advantage, job-fit, and reimbursement and financial incentive. Effort expectancy was defined as the degree of ease associated with the use of the eHealth system. It included perceived ease of use, ease of use, and complexity. Social influence was defined as the degree to which a health care provider perceives that important others believe he or she should use the new eHealth system. It included the subjective norm, competition, supportive organizational culture for change, and friendship network. Facilitating or inhibiting conditions were defined as the degree to which a health care provider believes that an organizational and technical infrastructure exists to support use of the eHealth system. It included the computer self-efficacy, computer anxiety, legal concerns, financial constraints, availability of ICT infrastructure, time cost, eHealth interoperability, IT support, eHealth and business process alignment, end user involvement, management commitment and support to change, uncertainty about IT vendor, professional autonomy, interference with the health care provider and patient relationship, and patient privacy concerns.

36Of the 93 papers, 57 examined the adoption/acceptance issue of EHR/EMR. EHR/EMR is a repository of health information in relation to a subject of care (ie, patient) in a computer processable form. Li et al explained that electronic patient records form the core of any other eHealth applications and thus the success of these is very much dependent on the EHR/EMR adoption. Although EHR/EMR can be utilized by all groups of health care providers (eg, physicians, nurses, and pharmacists), physicians were study participants among an overwhelmingly large number of publications.

37After 2002-2004, there was a sharp increase in the number of publications. A majority of these studies were conducted in the United States. According to Burt et al, EHR adoption in the United States was significantly low until 2005, with less than 18% of physicians used EHR at their office. After 2005, there was a great increase in EHR adoption levels across the United States, making more health care settings available for eHealth acceptance research.

38Most of the 93 studies used a quantitative research methodology to measure eHealth adoption/acceptance variables and test hypotheses. A small percentage applied models or theories on individual acceptance of information technology (eg, Technology Acceptance Model, TAM). The results supported the models in predicting the adoption behavior in the health care context. The most applied model was the TAM, which proposed a method of evaluating user acceptance through his/her beliefs, attitudes, intentions, and actual technology adoption behavior. Within these studies, the factors influential to health care providers’ acceptance of eHealth included their perceived usefulness and needs, perceived ease of use, and all of the facilitating or inhibiting conditions.

39Few studies (eg,) have successfully tested the applicability of the UTAUT model by Venkatesh et al [21]. Using the definition of the UTAUT constructs, we analyzed and organized the eHealth acceptance factors that we found. The mapping work demonstrated that the UTAUT model is a useful framework for applying and organizing literature, which is of great benefit to readers interested in learning more on the topic. Nevertheless, it was found that half of the health care provider characteristics (years in practice, role, and race) as well as medical practice characteristics identified from this literature review have not yet been covered in the UTAUT. Further, some studies also showed significant correlations among the identified factors. Perceived usefulness had the strongest impact on health care providers’ behavior intention, whereas their perceived usefulness was influenced by the perceived ease of use, eHealth and business process alignment, end user involvement, management commitment and support to change, health care provider-patient relationship, and IT experience and knowledge. The variance of the perceived ease of use was associated with the computer self-efficacy, end user involvement, management commitment and support to change, as well as health care provider-patient relationship. These correlations have not been incorporated in the UTAUT. Our efforts to map eHealth acceptance research results against the UTAUT model suggested that health care settings could potentially extend theories on information technology adoption due to their complex contextual dynamics.

40In some of the papers, significant correlations were not necessarily found between acceptance factors on the list (particularly those of individual characteristics and medical practice characteristics) and health care providers’ usage intention or actual use of eHealth. Chavis’s study, for example, did not demonstrate a significant positive correlation between individual characteristics (ie, job role and age) and technology adoption. This result can be explained with the UTAUT model: the age acts as a moderator rather than a factor directly impacting upon the behavioral intention or use behavior. Russell et al found that health care providers in large practices were not more likely to use an EMR. Others argued against that, suggesting that larger practices tended to “have access to the potentially greater resources” (financial and human resources) required for the eHealth system delivery and adoption, and have extensive internal IT assistance and training.

41Apart from the contradicting findings among these studies, some acceptance factors can also be context sensitive. Given that most of the 93 studies were conducted in the United States, the types of third-party payers (which is by definition the proportion of patients who are privately insured, Medicaid, Medicare, or uninsured), for example, reflects the health insurance scheme specifically in the United States context. In the future, further studies particularly in health care settings of other countries, are required in order to improve the understanding of eHealth adoption phenomenon in a global context, as well as to extend the theory and research on individual acceptance of information technology.

42Here are a few major limitations of this literature review. Although efforts were made to include all research papers on health care providers’ acceptance of various eHealth applications, some may not have been identified due to selected search phrases. In order to at least include those papers, which can help us increase the validity of the findings, the supplementary search keywords “EHR” and “EMR” were both used as previously discussed.

43The review was limited also due to the selection of the databases. Although they are the outlets that were deemed most likely to publish eHealth acceptance-related work, some papers may have been missed. We tried to compensate for this potential loss by ensuring that all selected databases were searched to their full extent.

44Mapping the identified eHealth adoption factors against the UTAUT model can be subjective. We attempted to maximize the accuracy and appropriateness of our mapping work by applying the validity procedure.

45The study results could help decision makers at the health care setting systematically understand facilitating forces and inhibiting factors influential to the health care providers’ acceptance of eHealth, and thus proactively introduce interventions for the adoption success. For example, health care providers may lack the adequate computer skills to use eHealth systems or had previous negative technology experiences. IT support before, during, and after initial eHealth implementation can provide a smooth transition to their reengineered job routine and overcome their technology phobia, hence facilitating eHealth acceptance and use (eg,). IT support includes, but is not limited to training, provision of guideline documents, and troubleshooting [50,123,124].

46Training can take various forms such as group training or one-on-one training, which is ideal in all circumstances. One-on-one training needs to set expectations, teach health care providers about the eHealth system features, customize the technology for each particular specialty, and help them to integrate the system (eg, e-Prescribing) into their medical practice workflow.

47Guideline documents as a knowledge source promote authentic translation of domain knowledge and reduce the overall complexity of the implementation task. Each care provider should be provided with a manual containing step-by-step instructions for the system’s use.

48Real time troubleshooting (especially through internal resources) facilitates the effective use of the eHealth system and becomes essential to the system success in terms of actual usage. Health care providers need to know how to access it when required. A feedback mechanism (eg, online help) allows health care providers to document a problem that they are having with the system and then to receive prompt feedback. Compared with external support services from the IT vendor, internal IT staff is more familiar with the work environment and related needs, and may respond more quickly to an urgent request.

49Another example is eHealth/business process alignment. Workflow is associated with routine processes, characterized by a fixed definition of tasks and an order of execution. The eHealth system needs to be designed in close collaboration with health care providers so that it truly assists their medical practice. The collaboration between IT vendors and clinical sites is to understand the site's workflow and determine the most suitable IT solution. After the workflow is analyzed thoroughly with health care providers’ involvement, their participatory process is also essential to fine-tune the system’s capabilities. Extensive software testing of the vendor's claims for the baseline functionality and system adaptability to local needs is critical before the implementation, as health care providers' frustration from software problems can promptly escalate and result in resistance to continue using the system.

50By synthesizing the evidence from the literature, our study may also assist policy makers at the health sector in refining or developing relevant policies to push eHealth innovation. eHealth adoption and ongoing maintenance requires a large capital investment [131-133]. While the government in some cases funds the start-up cost of an eHealth project (eg, the EMRX system in Singapore), health care providers may still need to undertake the operation and enhancement cost of their system. In small or independent medical practices, there is lack or absence of internal capacity for system maintenance; eHealth vendors alternatively provide all these services but often charge high fees. Due to financial constraints, system maintenance represents a vulnerable spot for the entire effort of eHealth and many practices underperform. To address this challenge, the development of programs such as zero-interest or revolving loans that make capital available to health care provider groups at low interest rates is essential, particularly in small or independent practices.

51Another important issue is interoperability. Bates commented that the interoperability between eHealth applications and seamless and reliable clinical information exchange is a key to making EHR use a cornerstone of practice. Even if physicians started to use an EHR system, they might still be unable to seamlessly share some other patient information (such as laboratory and radiology results stored in Laboratory Information Systems, LIS, and Picture Archiving and Communication Systems, PACS) for clinical decisions. According to a recent analysis, $77.8 billion USD could be saved annually by interoperable clinical information exchange among key stakeholders in the health care delivery system. The government should take stronger position to create a database of eHealth vendors whose products meet certain standards and enable clinical information exchange and to certify these products. The certification effort would also minimize health care providers’ uncertainty over the selection of a viable and sustainable product from hundreds of IT vendors in the market.

52Legal and regulatory changes can be required to address eHealth adoption related issues. For example, the Medicines Regulations (1984) and the Misuse of Drugs Regulations (1977) in New Zealand, which governs respectively the form of medication prescriptions and controlled substances, stated that indelible text and practitioners’ handwritten signature was required for a legitimate prescription. To facilitate the adoption of electronic prescribing and dispensing of medicines, the Health Department of Commonwealth has amended the National Health (Pharmaceutical Benefits) Regulations. These amendments came into effect from March 1, 2007 and the electronic prescribing and dispensing process has been additional and separate to the already existing paper-based process. The states and territories have continuously been taking steps to remove any legal barriers to the adoption of the electronic process in each jurisdiction.

53In this 4-step literature review, 40 factors were identified to be influential to health care providers’ acceptance of eHealth and organized in accordance with the UTAUT model. The findings may help decision makers at health care settings and policy makers at the health sector to better understand eHealth adoption issues and take action to facilitate the eHealth innovation process. Our work also suggests further studies to extend theories on information technology adoption.