Medication Supports at Transitions Between Hospital and Other Care Settings: A Rapid Scoping Review

Introduction

The World Health Organization (WHO) has identified transitions in care (TiC) as a key focus area for improving patient safety and minimizing preventable medication-related harm.¹ TiC, in which patients move from one care setting to another, often involves changes to medication regimens.² During transitions, discrepancies and gaps in communication can lead to medication errors,³ which in turn, can cause patient harm and lead to emergency department visits, hospitalization, and even death.⁴ A study assessing patient’s recall of medication changes one week after discharge found less than half were able to recall whether and which medications changed during their hospital stay, and that nearly 10% experienced changes in their perceived informational needs during their care transition.⁵ As inadequate knowledge about medication regimens can lead to incorrect use and non-adherence,⁶ these findings suggest that many patients may be at risk of harm and/or suboptimal therapy following discharge. Furthermore, certain patient groups are at higher risk for adverse events during transitions in case. Patients with language barriers or low literacy levels often have difficulty understanding their discharge medications.^7,8 Older people and those with intellectual or physical disabilities experience high rates of multimorbidity and are often prescribed many medications, which need to be reviewed frequently.^9–11

Medication support interventions are key to successful transitions.¹² We define medication support interventions as those in which patients and caregivers partner with health-care providers to review medication changes with the goal of increasing the patients’ knowledge and confidence about taking their medications. Current knowledge and understanding of the optimal tools for providing medication support during TiC is limited. Further study is required to understand both the design and implementation of successful support tools. As part of this effort, we performed a rapid scoping review to describe medication support tools and processes that have been studied, understand the extent to which elements of user-centered design were included in their development, and to identify opportunities for future research. This rapid scoping review is a component of a multi-method research project to understand the needs and co-develop guidelines and prototypes of potential medication support tool(s) together with patients and providers, with the goals of improving people’s medication experiences and optimizing medication management at TiC.

Objective

This rapid scoping review sought to address the following research question: What patient-centered or caregiver-centered medication support tools or processes have been studied within three months following TiC between hospitals and other care settings?

Methods

We undertook a rapid scoping review with the intent of identifying evidence in a timely fashion to facilitate the integration of our findings into our complementary co-design work to create medication support tools for use during TiC.¹³ Our methods were guided by the information and recommendations outlined in the Joanna Briggs Institute Manual for Evidence Synthesis on scoping reviews.¹⁴ Reporting of our work was guided by the Preferred Reporting Items for Systematic Reviews and Meta-analysis for Scoping Reviews (PRISMA-ScR).¹⁵

Eligibility Criteria

Study eligibility criteria were as follows.

P - Population: We included studies that involved patients experiencing TiC, and their caregivers. Caregivers were defined as any person who gives care to people who need help taking care of themselves. Examples included family members, friends, or members of the clergy.¹⁶ Only patients experiencing transitions from hospitals to other care settings (home/community, primary care, long-term care, rehabilitation facilities, etc.) were included because this represents a population at high risk for experiencing medication errors and adverse events.^17,18

I–Intervention: Studies that involved or assessed a medication support intervention were included. A medication support intervention was defined as any process or tool that aims to increase patients and caregivers’ understanding of their medications, confidence about taking their medication, or aims to increase providers’ understanding of a patient’s preferences and values with respect to their medications. Broader interventions, such as discharge summaries, with medication information were included.

C - Comparator: Studies with or without controls were eligible for inclusion.

O - Outcomes: Studies that assessed patient and caregivers’ knowledge/comprehension, patient attitudes (satisfaction, preferences, values) and confidence (self-efficacy), patient behaviour (self-management, adherence) and health system outcomes (health care utilization (eg, physician visits, pharmacy visits, readmission) and mortality) were included. Only studies that measured outcomes within 3 months of discharge were included because this time frame is frequently cited as the period in which the outcome may be associated with the intervention administered at discharge.¹⁹

S - Study Designs: Eligible study designs included: experimental studies (randomized controlled trials, quasi-experimental studies with control(s)), observational studies (prospective and historical cohort, case-control, case-crossover, case-cohort, case-time controlled, self-controlled case-series design, cross-sectional studies, time-series, pre-post studies), qualitative studies, and mixed methods studies.²⁰ Descriptive quantitative studies were excluded as these studies do not assess the effectiveness of the intervention. We included published or in-press peer reviewed articles published as full reports. Book chapters, abstracts, conference proceedings were excluded.

Report Characteristics: Given the short time to conduct the review, we confined our search to published literature available in English-language only. We limited eligible studies to those published within the previous 15 years (2004 −2019) due to rapid shifts in clinical practice supporting medication management.

Literature Search Strategy

The search strategy was developed and executed by a university-affiliated librarian co-investigator (Appendix 1). A second librarian with expertise in health literature reviewed the search strategies using the PRESS EBC Checklist.²¹ The following databases were searched between July 26 and 30, 2019:

• Ovid MEDLINE (1946-present including Epub ahead of print, in-process, and other unindexed citations) Ovid MEDLINE (1946-present including Epub ahead of print, in-process, and other unindexed citations),
• Ovid EMBASE (1947-present),
• Cochrane Library,
• EBSCO CINAHL Plus with Full Text (1981-present).

Search results were exported by the librarian into EndNote and duplicates were removed. After duplicate removal, citations from only the prior 15 years (ie, 2019 to 2004) were identified.

Study Selection

Citation titles and abstracts from the searches were screened to determine study eligibility within Covidence^®.²² Interrater agreement was assessed, and any conflicts resolved.

Standardized relevance screening forms with instructions were prepared and piloted to ensure uniform application of eligibility criteria. Titles and abstracts of identified references were reviewed by an experienced member of the study team. A second reviewer verified a subset of the records (5%) examined by the first reviewer and was also consulted for citations about which the first reviewer is uncertain (DG). Agreement on this subset of records was above a minimum kappa value of 0.80 taken to represent substantial agreement.²³

Full-text articles were retrieved and reviewed using the process described above for any abstract that required further clarification or was related to the study research question. Disagreements were discussed until consensus was reached or reconciled by a third reviewer.

Team members did not review studies they authored, co-authored or consulted upon.

A review log was created to document all decisions made in relation to selecting studies for inclusion including recommendations of the reviewer(s) and reason(s) for exclusion to facilitate development of a flowchart, recommended by PRISMA.²⁴

Data Charting Process and Data Items

We collected data about study characteristics (eg study design, citation), population characteristics (eg number of patients, age, gender, whether the population included was at high risk for adverse events due to presence of disabilities or language barriers), intervention characteristics (eg physical, verbal, or electronic modes of delivery, components of intervention such as videos, counselling, medication list, follow-up, content of intervention, layout (design), process of the interventions, transition type), study outcomes (eg patient and caregiver knowledge/comprehension, patient and caregiver attitudes and confidence, patient and caregiver behaviour and experience, health system outcomes, other outcomes), and conclusions.

Fields for data abstraction were based on those recommended by the Cochrane Collaboration and the Centre for Reviews and Dissemination.^24,25 We were also guided by a systematic review conducted by members of our study team that explored the impact of patient-centered discharge tools.¹⁹ Further, the extent of user-centered design approaches used to develop each tool was assessed with the UCD-11 (User-Centred Design for Patient-Centred Tools) instrument.²⁶ The UCD-11 instrument was used to generate a descriptive measure of an intervention’s user-centeredness through assessing the extent of end user (eg patients, caregivers, health professionals) involvement in the design, development, and refinement stages of the intervention.²⁶ If the criteria was not explicitly reported in the article, it was assumed this criteria was not a part of tool development.

A preliminary data abstraction form and accompanying guide were created. Cochrane’s process for developing the data extraction form guided this process: a paper form was designed, two reviewers independently extracted data from 10 studies, and then met to determine the necessary revisions. Feedback was used to modify the form and repeat the pilot before creating the form electronically. Concurrently, a guide with instructions for completing the extraction and decision rules for coding data was created and revised iteratively. Once the fields for extraction and their accompanying coding rules were agreed upon, an electronic form was developed to facilitate data entry and management.

Due to the time constraints in which this rapid scoping review was conducted, one experienced reviewer abstracted data . A second reviewer was consulted in cases of uncertainty . Disagreements that could not be resolved through consensus were reconciled through discussion with the two reviewers in the presence of a third person .

Mechanisms for recording, assessing, and correcting data entry errors were devised.

A descriptive numerical summary and a narrative synthesis were completed to describe study designs, intervention (and comparator intervention, if applicable) characteristics, target population characteristics, and study outcomes, as stated in our study objective. Results were classified based on main conceptual categories (eg timing, mode of delivery, health-care providers involved). Summary tables were created for study characteristics, intervention characteristics, and development of the interventions (eg UCD-11).

Risk of Bias in Individual Evidence Sources

Our primary interest in this body of literature was to create an inventory of design, layout, and process considerations. Consistent with established scoping review methods, we did not appraise the risk of bias for included studies.^14,15

Results

Search Results

The search yielded 8231 reports (Figure 1). After removing duplicates, 7671 articles were assessed for eligibility and after title and abstract screening, 113 articles were marked for full-text review. After full-text review, 53 articles were excluded, and 60 articles remained eligible.^27–86 For a large number of articles (n = 45/53), the full text could not be found. Other reasons for excluding studies after full-text review included ineligible design (n = 3) or outcome (n = 3) and duplicates (n = 1) that were not identified during the title and abstract screening stage. None of the included studies were co-authored by a team member.

Figure 1 PRISMA flow diagram.

Study Characteristics

Of the 60 studies included, 50.0% (n = 30) were randomized controlled trials,^{28,29,33–36,39,43,45–47,50,53,54,61–64,68–74,76,80,83,84,86} 23.3% (n = 14) were non-randomized trials,^{30,31,37,40,48,49,51,56–58,60,75,77,78} 11.7% (n = 7) were cohort studies,^{27,42,52,59,66,67,79} 8.3% (n = 5) were before-and-after studies,^{32,41,55,81,82} 5.0% (n = 3) were quasi-randomized trials,^38,44,85 and 1.7% (n = 1) were case-control studies.⁶⁵ The majority of studies (53.3%, n = 32/60) were conducted in the United States of America (USA).^{29,30,32–34,37,39,42,43,45,48,49,51,53,57,59–61,64–67,69–71,75,78,79,81–83,85} Other countries of origin for more than one study included Australia (n = 4),^38,40,55,84 Canada (n = 4),^44,47,76,80 Brazil (n = 2),^36,68 Croatia (n = 2),^62,63 France (n = 2),^56,72 and Switzerland (n = 2).^52,58 The remaining 12 studies originated from other unique countries. Table 1 describes the participants, intervention, comparator and outcome(s) of the included studies. Table 2 provides additional information regarding the interventions of included studies ie, specifies health-care providers delivering the intervention, the timing of the intervention, the various components of multi-pronged interventions, etc.

Table 1 Summary of Studies Included

Table 2 Key Characteristics of Studies Included

Table 3 UCD-1126 Criteria of Included Studies

Many studies selected patients based on age (76.7%, n = 46/60)^{27–30,33–45,47,49–53,55–57,60–64,66,67,71–76,78,80,81,83–86} and presence of a particular medical condition (50.0%, n = 30/60).^{29–31,33,34,36,37,40,43–45,47,48,50,53,54,59,60,63,66–69,71,76,78,82–84,86} Fewer studies selected patients based on the medication-class prescribed (30.0%, n = 18/60)^{29,30,39–41,44,47,48,53,64,68–71,74,78,84,85} or the patients’ number of medications (31.7%, n = 19/60).^{28,30,33,35,43,46,48,49,52,55,58,61,62,65,70,74,77,84,85} Few studies selected patients based on the presence of an intellectual or physical disability (3.3%, n = 2/60),^45,84 low literacy level (n = 1/60),⁴⁸ or language barrier (n = 1/60).⁸⁴ There were also only a few studies that included a subgroup analysis involving patients with low literacy (8.3%, n = 5/60),^{34,38,53,69,71} intellectual or physical disability (5.0%, n = 3/60),^34,45,84 and language barriers (n = 1/60).⁸⁴

UCD-11 Reporting

Only 25% of studies (n = 15/60) explicitly discussed the development of the intervention,^{29,31,33–36,43,47,51,52,64,76,81,82,86} making it difficult to apply the UCD-11 criteria. Of the 60 articles, none met all 11 criteria and only 1 met 10 of the 11 UCD-11 criteria (Table 3).⁴⁷ Twenty percent (n = 12/60) of articles met more than 3 of the 11 UCD-11 criteria.^{29,33–36,43,47,52,64,76,81,86} Of the 15 studies which discussed intervention development, 1 study reported consultation of expert opinion in the development process (n = 1/15),³⁴ and 3 studies reported usage of three or more iterative cycles in which outside opinions were consulted and prototype revisions were made accordingly (n = 3/15).^36,47,81 In contrast to these under-reported criteria, end user involvement in evaluation of the intervention, part of our study eligibility criteria, was consistently reported by all 15 studies (n = 15/15) that explicitly discussed intervention development.^{29,31,33–36,43,47,51,52,64,76,81,82,86}

Content

Physical/Electronic Medication List Interventions

For details regarding interventions including physical or electronic medication lists, see “Table 4”. These included any interventions that contained the patient’s medication lists in either a physical (eg paper) or electronic format. While 63.3% (n = 38/60) interventions included a physical or electronic document containing information regarding the patient’s medications (eg customized medication list, medication information pamphlet on a particular medication),^{28,29,33–43,45–47,50,52,53,55,56,58–68,71,76,77,83,85,86} only 36.7% (n = 22/60) interventions included a list of the patient’s medications.^{28,29,33–37,43,45,53,55,56,58,60,61,65–67,77,83,85,86} Of the 38 interventions with a physical or electronic document pertaining to the patient’s medications, the content related to medication information included varied. Many included an indication/reason for use (52.6%, n = 20/38),^{29,33–37,40,43,47,52,53,56,58,61–64,66,76,77} timing of administration (42.1%, n = 16/38),^{34–37,40,43,52,53,55,60–64,77,86} frequency of administration (42.1%, n = 16/38),^{34,36,37,40,43,47,52,53,56,58,60,61,64,66,77,86} and side effects (34.2%, n = 13/38).^{34,36,39,40,50,52,56,58,61–64,66} Ten percent (n = 6/60) of interventions included an electronic component.^{29,39,44,54,69,81}

Table 4 Characteristics of Written Information in Physical/Electronic Interventions*

Design

Mode of Delivery

Interventions included in the study were classified based on their mode of delivery into either physical, verbal, electronic or a combination of these modalities. The majority of interventions consisted of multiple components (78.3%, n = 47)^{27–38,41–43,45–51,53,55–60,63–72,74,75,78,79,82,83,85,86} with 41.7% (n = 25) of interventions including a follow-up component.^{27,30,33,34,36,37,42,44,45,48,51,53,59,65–67,70,73–75,79,82–85} Sixty-percent (n=37/60) of interventions, used physical and verbal means to deliver various components (eg medication list and patient education).^{28,29,33–43,45,46,49,50,53,55,56,58–60,62–69,71,74,82,83,85,86} Three of these interventions also included an electronically delivered component.^29,39,69 Some interventions had only one mode of delivery: thirty percent (n = 13/60) of interventions were only verbally communicated to the patient,^{30–32,48,51,57,70,72,73,75,78,79,84} while 10% (n = 6/60) were solely physically provided to the patient^{47,52,61,76,77,80} and 5% (n = 3/60) were only electronically delivered.^44,54,81 A small number of interventions did not involve direct delivery to patients. One of the interventions consisted of a letter mailed to the patient⁷⁶ and family physician while another intervention involved a referral to outpatient providers if a need was identified.²⁷

Personalization

Interventions were considered “personalized” if they contained any information specific to the patient, rather than a particular medication or medical condition. Most of the interventions (83.3%, n = 50/60) were personalized to the patient.^{27–30,32–38,42–49,51,53–63,65–70,72–79,82–86} The most common methods of personalization were including the patient’s specific medications list. However, only 13.3% (n = 8/60) involved assessment of the patient’s health priorities (eg if they wanted the medication list in a particular language).^{33,34,36,43,45,47,49,79} Even fewer (11.7%, n = 7/60) reported integrating the patient’s values and preferences into the care plan.^{33,34,36,43,45,49,79} Few interventions (20.0%, n = 12/60) addressed the patient’s ability to access medications financially in the community.^{32–34,36,37,48,51,56,70,73,75,79}

Process

Patient Involvement

A large portion (81.7%, n = 49) of interventions included patient counselling.^{28–46,48–51,53,55–60,62,63,66–75,78,79,82–86} Of these interventions, 53.1% (n = 26/49) directly involved the patient during intervention delivery through strategies such as assessing patient understanding during the delivery or engaging patients in creation of a management plan.^{33,34,36–38,42,44,45,49,53,55,56,58,60,66,68,70,71,73,74,79,82–86} Twenty-two (44.9%) interventions assessed the patient for understanding,^{33,34,37,38,42,45,49,56,60,66,68,70,71,73,74,79,82–86} while 22.4% (n = 11/49) of these interventions specifically used teach-back (eg confirm patients’ understanding by asking them to “teach it back” to the health-care provider),^{34,38,42,45,49,53,71,74,79,82,83} and 12.2% (n = 6/49) assessed the patient’s health priorities (eg values, preferences).^{33,34,36,45,49,79} Other strategies used included increasing responsibility based on patient’s successful compliance, patient’s own management plan using a template, and positive reinforcement.

Timing

The majority of interventions were delivered to the patient before leaving the hospital (51.7%, n = 31).^{28,29,31,32,35,38–41,43,46,47,49,50,52,55–58,61–64,68,71,72,75,77,80,81,86} Thirty percent (n = 18) of interventions had elements delivered before and after the patients left the hospital,^{30,34,36,37,42,45,48,51,66,67,70,74,78,79,82–85} 16.7% (n = 10) of interventions were delivered exclusively after the patients left the hospital,^{27,33,44,53,54,59,60,65,73,76} and for one of the studies, it was not clear when the intervention was delivered to the patient based on the manuscript.⁶⁹

Healthcare Providers Involved and Number of Providers

Many interventions involved delivery from pharmacists (61.7%, n = 37),^{27–32,34–38,41,42,46,48,51–53,55–57,59,65–68,70,72–75,78,79,82,84–86} physicians (25.0%, n = 15),^{27,31,40,47,58–60,62–64,66,67,69,77,82} and nurses (25.0%, n = 15).^{30,32,33,39,43,44,49,50,53,55,59,61,66,67,83} Thirteen interventions (21.7%) involved delivery from multiple health-care providers.^{27,30–32,53,55,59,65–67,79,82,83} Only 26.7% (n = 16/60) studies reported training the staff administering the intervention.^{33,34,36,38,40,43,45,48,49,53,60,64,67,68,71,83} Twenty percent (n = 12) of interventions were shared with other health-care practitioners outside the intervention setting.^{35,47,48,53,57,60,72,73,76,83–85}

Reported Outcomes

“Table 5: Outcome Measure of Studies” presents details regarding outcome measures ie, which of the included studies assessed a particular outcome category, and those that demonstrated a clinically significant impact in the intervention group. Overall, 63% (38/60)^{27–38,42,45–49,51,54,56,57,60,62,65–68,70,71,73–75,78,79,81,83,85} of included studies reported health system outcomes (eg readmission rate, mortality rate) and 52% (31/60)^{28,30,33,36,37,40,43–47,49,50,53–56,58,59,62–64,68,69,71,74,75,80,84,86} explored patient and caregiver behavior and experience (eg behaviours related to adherence, life-style management). Thirty-percent (18/60)^{28,31,33,35,39,41,43,45,50,52,53,58,61,64,70,77,81,84} examined patient and caregiver knowledge/comprehension (eg, understanding of condition and medication), 28% (17/60)^{33,38,42,46,49,52–54,58,60,61,69,73–75,80,81,} reported patient and caregiver attitudes and confidence (eg, satisfaction, preferences, values and self-efficacy; and 27% (16/60)^{30,31,35,40,43,50,51,54,63,65,70,75,77,79,82,85} explored other outcome measures (eg, health-care provider insights, economic impact).

Table 5 Outcome Measures of Studies*

With respect to included studies reporting significant findings, 68% (21 of 31 studies)^{28,30,33,36,40,46,47,50,55,56,59,62,64,68,69,71,74–76,84,86} of studies reported significant impact on patient and caregiver behavior and experience. Sixty-seven percent (12 of 18 studies)^{31,35,39,41,45,50,52,58,61,64,77,84} assessing patient and caregiver knowledge, 55% (21 of 38 studies)^{27–33,35,36,47–49,60,62,66,70,71,75,78,79,85} assessing health system outcomes, 44% (7 of 16 studies)^{30,31,50,51,54,77,85} assessing other outcome measures (eg health-care provider insights, economic impact) and 41% (7 of 17 studies)^{33,38,49,69,74,75,81} assessing patient and caregiver attitudes and confidence demonstrated a clinically significant impact in the intervention group.

Discussion

Our rapid scoping review of medication support-related TiC interventions identified 60 relevant studies. The majority of studies were randomized controlled trials involving verbal counselling and/or physical document delivered to the patient before discharge. Most studies assessed health system outcomes and patient/caregiver behavior and experience, with the majority of studies showing a significant improvement in these outcomes. We identified four key findings, which differ from previous reviews on the topic of medication support and care transitions:^87,88 1) the need for improved reporting of intervention development and characteristics; 2) the need to include high-risk populations when designing studies; 3) the potential for future research to apply user-centred design approaches to intervention development, 4) the opportunity for future studies to examine electronic interventions.

Our first key finding is that included studies lacked reporting on the development and characteristics of interventions. The majority of included studies failed to mention the development of their tool entirely, making it difficult to understand the rationale and elements of the interventions, and to assess whether the interventions addressed patient needs. Further, studies were lacking in their description of interventions for both physical and counselling interventions. For example, Lam 2011 mentioned a “medication list with detailed instructions” but did not specify these details.⁵⁵ Also, Chakravarthy 2018 suggested that “discharge instructions” were provided by nursing staff but did not describe the content or components of the instructions.³⁹

Importantly, these gaps in reporting make it difficult to discern what was tested, thereby diminishing the replicability of particular studies in situations where beneficial impact in the intervention group was demonstrated. We recommend increased use of the TIDieR reporting guideline to ensure that authors provide sufficient details when describing their intervention.⁸⁹ Further, similarly detailed reporting of the elements of usual care comparators will help readers better understand the “baseline” and assess applicability to their own health systems. This review identified that few staff administering interventions were trained prior to the study. It was unclear whether training was not reported or if it was not provided. A possible explanation may be that interventions were delivered by health-care providers performing their regular scope of practice and did not require additional training. Regardless, additional reporting in publications regarding training, or the lack thereof, is encouraged.

Second, our scoping review identified an important gap with respect to patient populations that have been studied. Very few studies included or considered at-risk populations beyond older adults, such as patients with low literacy, language barriers, or disabilities, in their patient selection or sub-group analysis. These groups are at increased risk of adverse events and more likely to experience systemic barriers to care.^8,10,90 Based on the theory of design by exception, considering these “extreme” users when designing an intervention, instead of designing for the average, will result in more equitable interventions that are compatible with the large majority of users and does so more efficiently than other mainstream design approaches.^13,91

Third, we found that very few studies met the UCD-11 criteria, which is a way to describe an intervention’s user centredness ie, explores end user involvement in design, development and refinement stages of an intervention.²⁶ This may be partially but not fully explained by poor reporting of how interventions were developed. The strengths of participatory action and user-centered design approaches when developing complex interventions are increasingly recognized, as evidenced, for example, by their inclusion in recent UK Medical Research Council guidance.⁹² Members of our research team have successfully used these approaches to engage patients, caregivers, and health-care providers in co-developing patient-oriented discharge instructions, which has subsequently undergone wide-scale implementation and evaluation at the provincial level with positive results.^90,93 The application of these approaches when developing medication-focused interventions at care transitions is an important opportunity for researchers to explore.

Finally, we note that most interventions included in this review involved providing written information and verbal instructions or education to patients and caregivers. Electronic interventions were the least frequently used. The paucity of electronic interventions was somewhat unexpected considering the shift toward technology in health care. Considering that the technology is available for various tools, such as medication time reminder, missed medication alerts, applications across multiple devices and much more, future assessment and use of these tools is another opportunity for future studies.

Strengths and Limitations

Limitations of this study are consistent with those known for rapid reviews.⁹⁴ After piloting the process for title and abstract screening to attain agreement on 5% of the studies, subsequent studies were assessed for eligibility by one reviewer. Our review was limited to articles published between January 2004 and July 2019 in English, articles indexed in well-known databases, and those known to context experts. Grey literature sources were not searched and as an optional component of scoping reviews, risk of bias for included studies was not assessed.

Despite these limitations, our study has several strengths building on previous research conducted examining medication interventions during care transitions. For example, Tomlinson et al⁸⁷ conducted a systematic review and meta-analysis of randomized controlled trials evaluating interventions that support medication continuity in older adults at or started within one month of discharge from hospital. Consistent with scoping review methods, we used a broader search strategy (ie, included more constructs), and included study designs beyond randomized controlled trials, more eligible populations, and studies with a range of outcomes. Another systematic review and meta-analysis by Daliri et al⁸⁸ examining the impact of prospective studies of medication-related interventions at care transitions on outcomes, is complementary to our review. They used meta-analyses to estimate the impact of interventions on outcomes while we focused on describing patient and caregiver-centered medication support interventions and how user-centered design elements were considered in the development of these interventions. Our study builds on their review by expanding the breadth of study designs and outcomes eligible for inclusion.

Conclusion

There is global interest in understanding the components and impact of medication-related interventions on patient outcomes during care transitions, as evidenced by recent systematic reviews exploring this topic. In this rapid scoping review, focused on describing patient and caregiver-centered medication support interventions and the role of user-centred design, the majority of studies included were randomized controlled trials involving verbal counselling and/or physical document delivered to the patient before discharge. Most studies selected patients based on age or medical condition, and few studies considered patients at high-risk of adverse events. Few studies reported on the development of the intervention. The majority of studies included multiple components, most commonly involving verbal or physical modalities, rather than electronic interventions. With respect to outcomes, most studies assessed health system outcomes and patient and caregiver behavior and experiences. Less commonly explored were patient and caregiver knowledge, attitudes or confidence. Studies assessing medication support interventions at TiC would benefit from improved reporting on development, characteristics of the intervention and usual care comparators, and training administered. As well, active involvement of people with higher medication-related needs, such as those with low literacy, language barriers, and disabilities, is highly encouraged.