Determining how studies are related

A data extraction form was developed to extract key information and concepts from the included studies and to ease comparison between them. Data were firstly extracted to describe the type of study including publication date, e-health domain, healthcare setting, inclusion and exclusion criteria and methods used. Secondly, the main themes from each review relating to factors that influence implementation of e-health were extracted from both results and discussion sections of the included papers. Data from discussions were included as they often contained further interpretations from the authors, which offered important insights and enhanced the richness of the findings. A summary table created from this matrix with key study details, and summaries of main findings are presented in Additional file 2.

Translating the studies into one another

Following inductive analysis of the results and discussion sections of the included papers, it became clear that data were, for the most part, consistent with the constructs of the CFIR. Therefore, these detailed data were extracted from the studies into codes guided by the constructs of the CFIR (see Table 2). A category for data that did not fit into one of these constructs or for data that warranted further discussion between reviewers was created. This ensured that data were not being forced into the constructs where there was not a good fit and allowed for the CFIR to be evaluated as to how well the constructs could account for the data from this review. Cells within the matrix remained empty in cases where there were no relevant data in the paper concerned. As a way of remaining faithful to the meanings and concepts of each study, the terminology used in the original paper was preserved within the matrix. A category was also created for the main explanations or theories arising from the authors interpretations that were relevant to the research question. Data were re-categorised from one construct to another and discussions between JR, EM and FS were held until the reviewers were confident that all data were coded into appropriate constructs or categories.

Synthesising translations

By reading the concepts and interpretations off the matrix, it was possible to establish a sense of the relationships between the studies. It became clear that the relationships between studies seemed to be reciprocal (where concepts of one study encompass another), with many themes occurring across studies which were largely in agreement about the factors that influenced implementation, and from which a line of argument (overarching narrative) could be developed. Following a process of meta-ethnography, meant themes from each of the studies were not just described and compared, but rather interrogated in relation to themes from other studies.

Expressing the synthesis

There were no data that could not be coded to one of the CFIR constructs, meaning the categories of the CFIR were endorsed and no additional original theoretical insights developed.

Search results

A total of 2812 unique citations were identified (see Fig. 1). Of these, 2694 could be excluded on the basis of the title or abstract, leaving 118 citations requiring the full paper before a decision could be made. Of the 118 full papers assessed, 44 [15–58] met the criteria for inclusion and were included in this review (Additional file 2 gives details of all included reviews).

Adaptability

An important factor in vendor and technology selection, reported by many studies from all e-health domains, was the ability of the technology to be adapted to fit the local context [34, 39]. Technologies that can have technical adjustments made to them to suit the constant modifications of the environment may have greater acceptance and adoption [18, 35, 41, 50, 58]. End user input in the design and development of e-health technologies should be considered as a way of overcoming barriers of adaptability [28].

Related to adaptability is the interoperability of systems reported by many studies [16, 18, 21, 22, 24, 25, 28, 38, 48–50, 52, 53]. To promote their acceptance and use, systems must be able to adequately interface with other IT systems and exchange information [18, 27, 49]. For example, a major barrier to the adoption of Electronic Health Records (EHR) was the inability of new systems to exchange information with systems already in place, due to a lack of consistent data standards [18].

Complexity

Complexity factors such as slow system performance [55], software and hardware that were difficult to use [27], the need for extensive software modifications [52], the work involved in transferring records between two systems [18], the inability to provide real-time access [38], data handling, reliability, slow speed, unplanned downtime [41] and connectivity issues [49] influence implementation of systems in healthcare settings. Often issues of complexity were linked to health professionals being unable to master the technologies that were implemented [18, 28]. Vendors of e-health systems should aim to make systems as user-friendly as possible, involving end users in the design and development [53], providing guides to their use [34] and providing technical assistance [43].

Cost

The cost of e-health system and costs associated with their implementation were reported as important implementation factors by the majority of studies across all e-health domains [15–19, 22–25, 27, 28, 30–33, 36, 38, 39, 41, 43, 47, 49, 51–54, 57], with some studies citing cost-related factors as the main barrier to implementation [49, 54]. Cost factors are related to start-up costs, ongoing costs, costs related to a loss of revenue and potential savings to put against these costs.

High set-up costs including purchasing and installation costs were cited as barriers to the initial adoption of e-health systems [49]. Financial incentives to adopt e-health systems from insurers and government agencies facilitated adoption decisions in some case [24, 37, 49].

Concerns about ongoing costs were also reported as barriers to adoption [17, 49]. Evidence of cost-saving and returns on investment were shown to be important in ongoing use of technologies [36, 39]. Establishing cost-effectiveness through formal evaluations, financing of services on a bigger scale, and redesigning business models and incentives were suggested as strategies to help overcome cost-related barriers [20, 24, 51].

External policy and incentives

An absence or inadequacy of legislation and policies and liability concerns may hamper the implementation of e-health systems at the organisational and health professional level [17, 18, 20, 51]. The need for recognised standards for the provision of e-health systems was described by many studies [16, 18, 20, 28, 33, 36, 48, 49, 51]. The creation of standards may serve to reduce health professionals’ concerns over patient data safety and professional liability [17, 33] and facilitate the exchange of electronic health information between systems [16] and organisations while maintaining data integrity [49].

Incentives by government organisations and other external stakeholders may facilitate adoption by healthcare organisations [17, 18, 39, 48, 49]. Financial incentives include the provision of initial funds to cover upfront costs [49], financial sponsorship [39], reimbursements for adoption [19, 30, 31, 36], and pay-for-performance initiatives [24, 36, 37, 49].

Implementation climate

Implementation climate includes the compatibility or general fit between the e-health intervention and the organisation [16, 23, 31, 33, 37, 46, 49, 59]. The fit between e-health systems and workflows in particular was discussed by the majority of studies [15, 16, 18, 21, 23, 25, 27, 28, 36, 40, 41, 43, 47–50, 52, 55–58]. A frequent reason for unsuccessful implementation is that the information systems do not fit well with work practices or daily clinical work [25]. Health professionals’ perceptions that e-health systems disrupt workflows, and the delivery of care, are a barrier to both the implementation and use of these systems [23, 49, 52, 56, 57]. When there is a good fit, or perceived fit, between e-health systems and workflows, and when systems positively influence workplace efficiency, this facilitates use [15, 40, 41]. Incorporating workflow analysis into system design [48, 55], the integration of systems into the usual process of care [15], user-friendly systems [40] and minimising workflow interruptions during implementation [17] may minimise disruptions to workflow.

Alterations to workflows created by the introduction of e-health systems may also disrupt established professional roles, responsibilities [16, 20, 25, 36, 55] and working styles [18]. Physician resistance to e-health implementation is reported by several studies to be related to fear of [18], dissatisfaction with [18] and uncertainty over [43, 47] new roles and responsibilities, created by the introduction of e-health systems [18]. The quality of project management during the implementation period [18], careful study of the downstream effects of implementation on workflow [55], additional training [31, 55], the adaptability of technologies to fit with roles, tasks and workflows [37] and dedicated technical support staff [31] are suggested as strategies to reduce barriers related to disruptions to workflow, roles and responsibilities that e-health implementation may bring.

Leadership engagement at all stages of the development and implementation processes can help improve the effective implementation of e-health systems [43, 54, 56, 58] and a lack of involvement can be a barrier to implementation [18, 53]. Management support is also important for implementation success [25, 37, 50, 54].

Authors described available resources including the availability of suitable infrastructure as important for implementation success. Infrastructure features included electricity supply [51], available bandwidth [31, 51], access to reliable internet connectivity [45, 51], access to computers [34], electrical power [45] and access to phone lines and mobile phones [45]. The availability, or lack thereof, of time to learn new e-health systems, implement them [25, 37, 53] and train staff to use them [49] was reported by several studies as a resource factor important for implementation. Providing a period of transition in which end users can become familiar with and learn how to use new systems has been advocated [39, 50].

Access to knowledge and information was also important for the implementation of systems across all e-health domains. Education was reported to increase staff acceptance of e-health systems [16, 25, 41, 45] including education around anticipated benefits and when those benefits could be expected [54]. A lack of knowledge and a limited understanding of benefits afforded by the systems acted as a barrier to implementation [49, 52, 53]. Nearly all studies made reference to training and support in relation to implementation and acceptance of e-health systems. Generally, access to appropriate, high-quality, well-funded, and easily available training was reported as a facilitator to implementation, whereas it was reported as a barrier when it was non-existent or existent but inadequate [15, 16, 18, 20, 23, 25, 28, 31, 37, 43, 45, 50, 51, 53–58]. Access to ongoing support to use systems was important for system use [21, 25, 28, 31, 37, 39, 43, 54] and a barrier to implementation when it was lacking [21].

Knowledge and beliefs

Attitudes and beliefs were reported to act as both facilitators and barriers to implementation and acceptance of e-health systems across all e-health domains. Positive attitudes of practitioners toward e-health systems and their implementation increased acceptance and implementation [20, 21, 41, 44, 46, 49], whereas negative attitudes and staff resistance acted as barriers [49, 52]. Positive staff attitudes were described as: beliefs that the new systems would benefit patients [56], interest in the technologies, perceived usefulness and motivation in working with the systems [21]. Negative perceptions included beliefs that electronic systems would disrupt the delivery of care [49]; doubts that these systems can improve patient care, clinical outcomes or improve the quality of medical practices [43]; and distrust in the systems [37] as well as a more general staff resistance to change [28, 32, 39, 41, 49, 51–53]. Strategies to challenge negative attitudes included fostering a culture of communication and cooperation, involving the eventual users of systems in the development and implementation [20, 49], leadership [56], friendly and context-aware user interfaces which promote perceived ease of use and usefulness [21], better education [47], and clearly and prospectively communicating intended benefits and realistic expectations for the system [54]. The attitudes of colleagues [25, 41] and patients [25, 27, 41] were also reported to influence staff attitudes with regard to e-health acceptance as were staff demographic factors [23, 46, 56].

Specifically, fears over a loss of autonomy [16, 18, 27, 34, 37, 39, 43], concerns about liability [16, 18, 22, 37, 43, 51], concerns over patient privacy and security being compromised [16, 18, 20, 23, 24, 27, 28, 30, 31, 33, 39, 41, 43, 45, 47, 49, 51–53, 56, 57], and perceived threats to patient and health professional relationships [18, 25, 27, 28, 31, 34, 37–39, 41, 43, 54, 57] through the introduction of e-health systems were repeatedly reported as barriers to use.

Other personal attributes

Healthcare professionals’ computer skills, abilities and experience were cited by several studies as influencing implementation and acceptance of e-health systems [16, 18, 25, 28, 34, 36, 37, 41, 43, 52, 57]. Training [25, 37, 43] and financial incentives [37] were cited as strategies to overcome skill-related barriers. Demographic factors such as age, education, sex, nationality, and clinical experience may influence healthcare professionals’ attitudes towards e-health systems [23, 28, 46, 56, 57]; however, most studies conclude that no clear relationships between these characteristics and attitudes could be established [46, 56, 57].

Planning

Planning for implementation was important for success, whereas the lack of a strategic plan was reported as a barrier to e-health implementation [24, 27, 31, 44, 50]. The work of planning includes the delineation of roles and responsibilities [43], securing time to invest system selection and procurement [18], evaluating other concomitant policy and process changes [40], needs assessment and analysis, development of a business plan [44], early identification and engagement of champions [31], involving end users [16, 27, 41, 51], establishing a guiding philosophy [33], testing organisational readiness [27, 32], development of incentive and innovation structures [27], communication of the strategy to all staff [50], and development of protocols for using the system and for provision of training [31]. Incremental implementation strategies where features are made available to users according to a plan were cited as preferable to ‘big bang’ approaches to implementation within complex organisations [39, 49]. One review emphasised the need for ongoing effort after the initial ‘go-live’ phase, referring to the ‘under-recognised maintenance phase of implementation’ [29].

Engaging

The designation of champions [18, 20, 25, 29, 31, 32, 36, 37, 39, 51, 54, 58] may be important for implementation success. Engagement of key stakeholders in the development and selection of e-health systems and in the planning and execution of implementation processes were important for implementation [16, 20, 28, 29, 34, 41, 44, 49–51, 53–56, 58] through fostering a sense of ownership [20, 44, 51], confidence [28], acceptance [34], enjoyment and self-pride [20] towards the e-health system and increasing buy-in [54].

Reflecting and evaluating

Evaluation was seen as important to ensure system benefits [29, 48], to increase health professional acceptance through demonstration of benefits [31, 34, 41, 43] and to secure ongoing funding [53], whereas a lack of evaluation and evidence may act as a barrier to implementation [53]. Four reviews included data on the effects of the e-health systems implemented [24, 36, 49, 52]; these reported mixed effects.

Recommendations

A summary of recommendations for implementation of an e-health system is presented in Table 3.