Principal findings
We have developed ten theories to explain how, why, when, where and for whom PERMs are designed, implemented or used in practice at care transitions that impacts on MedRec based on data from 19 systematically sourced articles. These theories describe the contexts and mechanisms that impact on outcomes.
By mapping the broad contexts and mechanisms that influenced the design, implementation or use of PERMs to improve MedRec at care transitions identified in this review to Pawson et al’s four I’s grouping of contexts outlined earlier in this article, we can hypothesise why the introduction of PERMs might succeed in one context and not in another:
Individuals (characteristics and capacities of the various stakeholders):
Consideration should be given to the individual user’s levels of confidence, trust, engagement and acceptance of PERMs. Users may have differing needs in relation to the amount, flexibility and format of training throughout the development and implementation life-cycle and therefore need differing levels of on-demand support in relation to both the MedRec process itself and the function and purpose of PERMs to support that process. Identification and comprehension of the user’s existing cognitive and decision-making processes and level of application of these to PERMs are important. For example, displaying pre-admission and in-hospital medication lists side by side on the screen reduces the cognitive burden and facilitates the task to compare these medication lists. An awareness of the changing roles of individuals undertaking MedRec is also important. For example, the expanding roles of the pharmacist and pharmacist technician.
Interpersonal relationships (the stakeholder relationships that carry the intervention):
The level of inclusive engagement of all stakeholders, including the individual users, in the design and implementation of PERMs; the level of awareness and respect for each other’s roles in the processes; and the level of clarity around who is responsible for the various stages of the MedRec processes, may all influence the successful introduction of PERMs.
Institutional Settings (the rules, norms and customs local to the intervention):
The quality of the MedRec processes in place, the quality and complementarity of PERMs to support it and the complexity of the contexts in which PERMs will be used will influence success.
Infrastructure (the wider social, economic and cultural setting of the intervention):
The level of interoperability of systems supporting PERMs; the level of existing and future legislation and governance supporting MedRec and use of PERMs; the quality of the sources of data for MedRec; and the extent of resourcing including budgetary and human resources, required to allow PERMs to achieve their full potential, will influence success.
Many of the theories could be applicable to any electronic system, including in a healthcare setting, where multiple users are using the system for different purposes (Theories 1, 2, 4, 5, 6). However, some are more particular to PERMs (Theory 4, 9 and 10), or others again specifically for PERMs for MedRec at care transitions (Theory 3, 7 and 8).
Focusing purely on the final group, Theory 3 states that for PERMs to become embedded into normal work practices they should complement an existing MedRec process. Some examples from the studies of the elements that could be complemented were: organisation and display of information following the natural order of events in a patient encounter; supporting workflow and accreditation requirements; medication information ordered by clinical importance; access to the history of a person’s previous medication use; inclusion of a function to prompt the user to record medication changes and reason for starting or stopping medications; inclusion of a field to capture details identified about a person’s medication adherence, allergies, or posologies; date of initiation of prescription and first prescriber; access to the prescribers’ or other HCPs’ contact details and the ability to produce a discharge summary and/or patient friendly print out.
Elements of PERMs, identified in the studies, that were expected to improve the frequency and integrity of MedRec, cognitive burden and decision making were: use of generic drug names thereby reducing confusion and duplication, organisation of information from different sources with the ability to see several lists at the same time facilitating rapid comparisons, information displays tailored to the needs of different users, ease of identifying medication changes, alerts i.e. drug-drug interactions or drug-disease interactions, reasonable balance of alerts, checks or warnings, action buttons (i.e. for stopping, modifying or continuing drugs), edit, sort (i.e. by therapeutic intent), drag and drop, or on-demand and additional information options, 24hr access to information, smooth processes to get patient consent to access their information. Barriers included number of clicks needed, too much text, unclear icons and lack of understanding or training in relation to how PERMs worked.
In relation to Theory 7, the quality of the data sources and level of interoperability between them, allowing safe and accurate sharing of information, was determined to be crucial to the successful introduction of PERMs.
Theory 8 identified that in order for the full potential of MedRec using PERMs to be realised, proper investment in the technology and sources of information must be provided, ensuring interoperability and access to accurate information for MedRec. For this to occur, increased patient safety and decreased hospital readmission rates and preventable events must be sufficiently valued.
Comparisons with other literature
A number of systematic reviews have examined the use of electronic MedRec at care transitions (8, 45-48). Both Wang et al and Mekonnen et al reported equivocal findings regarding the impact of electronic MedRec tools on the prevalence of medication discrepancies or the proportion of patients experiencing (unintentional) discrepancies (46, 47). Mekonnen reported that effective MedRec likely requires a multi-faceted approach involving people, process and technology, a finding which endorses the importance of our rapid realist review to identify CMOCs and generation of theories to test in the field. This multi-faceted approach should be addressed in future research. In their systematic review, Marien et al compiled a list of recommendations for the successful development and implementation of electronic MedRec tools (8). Although theirs was not a realist review, the recommendations reported mirror many of the contexts identified within our review, for example, development and implementation contexts, design features and functionalities. Our rapid realist review builds on this work by configuring contexts with mechanisms to influence a broad range of outcomes. Wang, Marien and Mekonnen all refer to the absence of evidence about usability, user satisfaction and user adherence (8, 46, 47). Several of the theories generated in our review address these issues and future work to validate our theoretical framework in the field should therefore contribute to this evidence gap.
The scoping review by Monkman et al examined both the contextual and human factors perspectives of using PERMs for MedRec and had several findings similar to those found in Bassi’s,(45) Marien’s (8)and our review; successful implementation of electronic MedRec systems requires well designed systems, attention to implementation features and standardisation of the MedRec process (48). Specifically identified were interoperability, design and layout of the system, clear MedRec processes and workflow, identification of who is responsible for MedRec and involvement of the user in the design especially for complex systems (8, 45, 48).
Comparing this review to the general health information technology (HIT) literature, Ammenwerth et al’s framework focused on what they described as the “fit” of three key elements; individuals, task and technology (FITT framework) (49). They identified what in realist terminology are the contexts and mechanisms of these three interacting elements with the aim of better understanding the reasons for information technology (IT) introduction failures. They identified that the user must be motivated, flexible and open to new ways of working; knowledgeable about the task and use of IT and trained to use the technology. The organisation must have a team culture and support the introduction of the HIT. The task must be organised; the complexity of the task to be completed must be considered. The technology must be functional, interoperable and usable. These are all consistent with the findings of this review and support theories 1. Engage Stakeholders, 2. Inclusive Design, 5. Tailored Training, 7. Interoperability and 10. Patients as users of PERMs.
Yen et al considered how HIT implementation is evaluated. They suggest including evaluation of technology acceptance, communication and collaboration, work productivity, training and competency, leadership, existing policy, the organisational culture, the level of social support provided and the idiosyncrasies across contexts (50). Marien et al identified many reports lacked adequate information and recommended that study reports should carefully describe each component and that validated measures of usability should be reported (8). The findings of our review, which included nine studies published since the Marien et al review, are that such measures have still not been adopted. The importance of evaluating sociotechnical factors and usability outlined in these reviews are consistent with the finding of our review as outlined in theories 1. Engage Stakeholders, 2. Inclusive Design, 4. Build Trust, and 9. Positive impact of Legislation or Governance.
Shachak et al described what the provided support for PERMs should look like, describing it under three headings; functional support: e.g. assistance in learning how to use the various features of the system; data support: e.g. activities intended to ensure the completeness, accuracy and consistency of data input and finally; training and education support, which they combined and stressed was an essential part of end-user support (51). Such recommendations were re-iterated by Marien et al, specific to implementation of electronic MedRec tools (8). This is consistent with the findings of this review specifically theories 5. Tailored Training and 6. Support and on-demand training.
The HIT literature features many of the contexts identified in this review as impacting on the design, implementation and use of PERMs and supports the validity of evaluating outcomes under the four I groupings referred to earlier as outlined by Pawson et al; the individuals, interpersonal relationships, institutional settings and infrastructure (19, 49-51).
Acknowledging the associated human and financial resources required to deal with the increased effort, volume and quality of data gathered and generated when using PERMs for MedRec was identified in this review as being important. This finding was unique in the literature. The increased data potentially available provides opportunities for risk identification, management, and analysis; if the value of these opportunities is recognised by leaders/ management from the outset then they will understand the need for additional resources. (Theory 8. Resource investment). However, consistent with Sevick et al’s (52) finding regarding electronic discharge communications, we identified the lack of evidence and the importance of evaluating the cost effectiveness of using PERMs for MedRec. Until such evidence is available, convincing individual organisations or countries of the additional benefits will be difficult. Careful consideration of the outcomes to be assessed in any cost analysis is also needed (52).
Strengths and limitations
A RRR, by its nature does not include all relevant literature on the topic of interest, however, we strived to include the richest data in this review, and the reviewers feel that saturation for each of the theories was achieved. If time allowed, a specific search through the remaining 32 articles included in the full text reading for other potential contexts and mechanisms identified by the panels during the review of the final theories, but lacking evidence in the 19 included articles, could have provided enough evidence to support the development of additional theories. Examples of such contexts include organisational culture or level of reliance on PERMs. We recommend this as a future study.
The quality assessment of realist data, using relevance, rigour and richness, could be considered a subjective assessment of the quality of the data within the articles included. However, the research team have extensive knowledge and experience in the field under review allowing expert assessment of these elements. We followed the RAMESES guidelines for the quality assessment process by providing a transparent procedure for the rating systems used to appraise the evidence used within the review; the real quality has been determined within the act of synthesis (53). The iterative nature of realist reviews resulted in the richness rating, the method used to focus the review, for some articles being revised downwards at the data extraction stage.
This RRR has allowed additional valuable data to be extracted from existing primary research, with minimal resources, that may impact positively on the future design, implementation and use of PERMs at care transitions for MedRec and of research in this area. The main outcomes formally considered in the included articles focused on either the impact on patient safety or the user’s experiences during implementation or use of a s. The use of realist methods has generated data to support identification of additional outcomes from the articles which were not formally identified by the authors in their aims and objectives. The additional outcomes included changes to interprofessional relationships, changes to awareness of others’ roles, changes to the users’ cognitive burden, changes to the consent process when sharing patient information, changes to the level of interaction with patients, changes to the patients level of awareness regarding the need for their consent, their opinions in relation to the benefits of interoperability/sharing their information and their level of engagement with PERMs. Consideration of this evidence has added richness to the knowledge identified in the systematic reviews referred to earlier, validating our chosen methodology. Future research should include appraisal of these socio-technical outcomes.
Practical implications
The product of realist reviews are theories, developed from evidence extracted from the literature and the input of experts in the field. These theories should subsequently be incorporated into an intervention or mapped to existing interventions and evaluated to further test their validity and refined or rejected based on that evaluation.