CLINICIAN'S CAPSULE
What is known about the topic?
Older patients with complex care needs and limited personal and social resources are heavy users of acute care services.
What did this study ask?
What interventions have demonstrated effectiveness in decreasing emergency department (ED) use and hospital admissions in older patients?
What did this study find?
Community-based interventions that include comprehensive geriatric assessments and multidisciplinary teams with a geriatrician were more likely to reduce acute care.
Why does this study matter to clinicians?
Specific interventional strategies, including home visits and community-based comprehensive geriatric assessments, show promise at reducing ED and hospital use.
INTRODUCTION
High utilization of acute care services, particularly emergency department (ED) visits, remains an important topic for healthcare providers and health policymakers within Canada and abroad.1,Reference Roberge, Pineault, Larouche and Poirier2 While EDs remain integral to providing rapid access to care for those with acute medical needs, prior work has shown that a small proportion of patients account for the majority of ED visits.Reference LaCalle and Rabin3,Reference Pines, Asplin and Kaji4 The healthcare needs of this patient population have been studied extensively in hopes of improving ED efficiency and decreasing ED wait times and costs, while also improving care for this high-risk group of patients.Reference Kumar and Klein5
Existing literature has shown that older patients with complex medical needs and limited personal and social resources are heavy users of ED services and are often admitted when they present to the ED; this is particularly true for those who are frail and in assisted-living facilities.Reference McCusker, Tousignant and Borges Da Silva6–Reference Gruneir, Bell and Bronskill8 Further, older patients often experience adverse outcomes such as medical complications, functional and cognitive decline, loss of socialization, and care fragmentation when admitted to the hospital.Reference Boyd, Landefeld and Counsell9–Reference Gillick, Serrell and Gillick11 Identifying strategies to reduce the use of ED services and hospital care, while ensuring that older patients receive the most appropriate care in the most appropriate setting, is crucial to ensuring optimal use of limited healthcare resources.Reference Gruneir, Bell and Bronskill8
There have been many efforts to develop ED alternatives for older patients living either in the community or within supportive environments, along with strategies to avoid hospital admission for those who present to the ED. While prior systematic reviews have addressed hospital avoidance strategies,Reference Flores-Mateo, Violan-Fors, Carrillo-Santisteve, Peiro and Argimon12,Reference McCusker and Verdon13 they are either a decade old, have not focused specifically on older patients, or include both interventional and observational designs. Given the strain being placed on ED and hospital services by an aging population, up-to-date information on the most effective strategies to appropriately avert ED presentation and hospital admission among these patients is needed. For this reason, we conducted a systematic review to identify interventions that have demonstrated effectiveness in decreasing ED use and hospital admissions, while also ensuring optimal outcomes for older patients.
METHODS
We performed this systematic review using a predetermined study protocol (PROSPERO Registration ID: CRD42017064894) in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.Reference Moher, Liberati, Tetzlaff, Altman and Group14
Data sources and search strategy
Literature searches were conducted in July 2019, in electronic databases, including Ovid MEDLINE, EMBASE, CINAHL, and the Cochrane Central Register of Controlled Trials, with no language or date restrictions (Appendix 1). A research librarian developed the search and recruited another librarian to peer review the MEDLINE search strategy using the PRESS (Peer Review of Electronic Search Strategies) template.Reference McGowan, Sampson and Salzwedel15 We also searched the reference lists of prior systematic reviews and included articles and consulted an expert in the field (JHL) regarding any missing or ongoing trials that had not been identified within our database search.
Study selection
Two pairs of reviewers (AN/NS and MS/MB) independently reviewed all identified abstracts for eligibility. All original research reporting on interventions to avoid ED use and/or hospital admissions among older patients were selected for a full-text review. We excluded commentaries, review articles, and observational study designs. Disagreements were resolved by consensus.
The same pairs of reviewers then performed a full-text review of articles that met our broad inclusion criteria and those with uncertain eligibility. Articles were retained in the systematic review if they met the following criteria for 1) population: older patients (aged 65 years and older) with acute illness defined as a sudden deterioration in health, warranting medical intervention. We included both community and facility-living patients; 2) intervention: provider or system-level strategies whose primary purpose was to avert presentation to the ED and/or admission to the hospital for those who presented to the ED. We excluded pharmaceutical interventions (e.g., interventional studies evaluating different doses of a medication among older patients) or interventions that did not include an outpatient/community-based component; 3) comparator: usual care (as defined within individual studies); 4) primary outcomes: ED visits and hospital admissions. Secondary outcomes included hospital readmission, mortality, morbidity, re-presentation to ED within 7 or 30 days, cost, and patient-reported outcomes (e.g., experience, quality of life, functional status, change in living arrangement); and 5) study designs: limited to interventional designs – randomized and non-randomized controlled trials (including pre-post and randomized cross-over designs) and interrupted time series.
Data extraction and quality assessment
Two reviewers (CP, AN) extracted study data using a pre-specified standardized data extraction form (Appendix 2). Authors of the selected studies were contacted if methods or data required clarification. Study quality was assessed using selected questions from the Cochrane Risk of Bias AssessmentReference Higgins, Altman and Gotzsche16 for randomized interventions and from the ROBINS-I for non-randomized interventions.Reference Sterne, Hernan and Reeves17
Data synthesis
Given the heterogeneity in study populations, interventions, and outcome measures, a meta-analysis of included studies was not performed. Rather, we used semi-quantitative tables to show trends across intervention types. Specifically, we reported the proportion of studies that reported an increase, decrease, no change/non-significant difference across the nine intervention types (described previously) for the four most commonly reported outcomes (i.e., primary: ED visits and hospitalizations; secondary: hospital re-admission, and mortality). Detailed findings for all other outcomes from the individual studies were reported within the appendices.
RESULTS
Our initial search yielded 7,943 unique citations. Of the 444 full-text articles reviewed, we identified 53 unique studies for inclusion, which examined interventions to prevent ED use and/or hospital admission in older patients(E1–53) (Figure 1). (See Appendix 3 for a complete list of included studies.) Table 1 provides details of the included studies. Publication dates ranged from 1994 to 2018 with nearly half (43%) conducted in the United States. Of the selected studies, 26 (49%) were randomized controlled trials, 8 (15%) were cluster-randomized controlled trials, and 19 (36%) were controlled before-after trials. The most commonly used intervention types were home visits (49%), education (47%), and case management/care coordination (47%). Interventions and follow-up periods were between 1 month and 3 years, with an average of 10 months. The most common outcomes measured in the studies were ED visits, hospitalization, hospital readmission, and mortality, but exact definitions of these outcomes varied substantially between studies. Almost two-thirds (62%) of the studies conducted at least part of the visit in the patient's home and about one-fifth (21%) occurred at least partly in a primary care environment. A detailed description of each study is provided in Appendix 4.
Figure 1. PRISMA flow diagram.
Table 1. Characteristics of included studies (n = 53)
Study quality
Quality of the selected studies varied substantially. On average, the randomized controlled trials and cluster-randomized controlled trials met three of the five selected quality assessment criteria, as did the controlled before-after trials (Figures 2 and 3). The majority of included studies protected against selection and detection bias, using random sequence generation, concealment of allocation, and blinded assessments of primary outcomes. About one-third of studies obtained outcome measures from an automated system or administrative database. Studies generally scored low regarding follow-up of professionals and patients (suggesting weak protection against exclusion bias, or incomplete outcome data) (Appendix 7, 8).
Figure 2. Study quality assessment randomized controlled trials/cluster-randomized controlled trials.
Figure 3. Study quality assessment controlled before-after trials.
Effect of interventions on the most commonly reported primary and secondary outcomes
ED visits were the most commonly reported outcome (70% of studies) and appeared to decrease across multiple intervention types. However, the overall effects of different intervention types on hospitalization and readmission rates were unclear. Similar findings were observed for all-cause mortality (Table 2). Interventional factors more likely to show decreases in ED use were home visits and comprehensive geriatric assessments (Figure 4). These studies reported statistically significant decreases in mean ED use (for interventions versus controls) ranging from -0.12 to -1.32 visits/patient and -6 to -178 visits/1000 patient-years. Interventions that included home visits were more likely to show decreases in hospitalizations with reductions in hospital admissions ranging from -6% to -14%. (Appendix 5). Interventions with primary care integration also showed reductions in hospital admissions, though the number of studies exploring this interventional strategy was low. Telemedicine had inconclusive effects on overall ED use but was more likely to show decreases in hospitalization and hospital readmission rates. However, the number of studies that employed this intervention type and reported on all outcomes was also low. Education, case management, and patient/caregiver counselling had inconclusive effects on ED use, hospitalization, and hospital readmissions.
Figure 4. Graphical representation of the effectiveness of hospital avoidance interventions on ED utilization by intervention type. Interventions that included home visits and comprehensive geriatric assessments were proportionally more likely to report a decrease in ED use. Statistically significant decreases in mean ED use/patient (for interventions versus controls) ranged from -0.12 to -1.32. Studies reporting rates reported decreases ranging from -6 to -178 visits/1000-patient-years. Reporting varied substantially across the included studies with authors using absolute values, rates (with varying denominators), ratios (OR, HR, RR), means, and percent differences to quantify the effect of interventions on ED use.
Table 2. Summary of the effectiveness of ED avoidance strategies by intervention type
Numerical values represent the total number of reported outcomes across intervention types. Multiple interventions within individual studies were reported separately. Individual studies with co-interventions (e.g., CGA and case management) were also reported in both intervention rows.
Interdisciplinary assessment strategy
Of the five studies that used an interdisciplinary assessment strategy(E8, E14–16, E39) and showed a decrease in ED use, four(E8, E14, E15, E39) of the teams included a geriatrician. Of the six(E4, E19, E32, E44, E45, E47) studies that used an interdisciplinary assessment that did not demonstrate significant effects on ED use, only two(E4, E19) had a team that included a geriatrician. Single-discipline interventions were largely undertaken by nurses. A small number of studies focused on the pharmacist's role in the ED and the community. Single-discipline intervention studies tended to be of short duration and had a varied period of follow-up, from 3–12 months. Some single-discipline interventions showed slight reductions in ED use and hospitalization; however, findings were inconsistent across these studies.
Intervention setting
Intervention setting appeared to influence study findings. Specifically, home-based, outpatient, and/or primary-care-based strategies were more likely to show reductions in ED visits and hospitalizations. Of the 25 studies that reported positive findings in at least one of the primary outcomes, 15 used home visits as part of their intervention. Of the studies conducted in-home, outpatient, and/or primary-care-based settings, those that included comprehensive geriatric assessments, home visits by various healthcare team members within dual or interdisciplinary teams, or regular contact with a nurse, general practitioner, or geriatrician had better outcomes. These interventions also tended to be of longer duration (6 months to 3 years). Hospital-based models, where patients were assessed prior to discharge, showed mixed results in acute care service use. These interventions tended to be shorter in duration (1–4 weeks) and period of assessment (3–6 months follow-up).
Composition of intervention teams
With respect to the composition of the intervention team members, their effect on ED use, hospitalizations, and readmission was mixed. Of the 19 studies that used multidisciplinary teams, 8 either involved social workers or explained how patients’ social needs were met by the intervention(E4,E6,E8,E15,E32,E47–49) (Appendix 8). However, the inclusion of social workers did not appear to have a conclusive impact on ED use, with only three studies showing a higher likelihood of success in decreasing ED visits.
Additional secondary outcomes
While all studies sought to measure ED and hospital use, they varied widely with respect to secondary outcomes examined. These included time to the next ED visit, hospital length of stay, functional assessment scores, self-reported quality of life, and cost-effectiveness. The large majority of effects on secondary outcomes were not significant (p > 0.05, or as otherwise defined by study authors). Detailed study-specific results are provided in Appendix 6.
DISCUSSION
We conducted a comprehensive systematic review of the literature to investigate interventions aimed at reducing ED use and hospitalization among older adults. Among the 53 studies included in our review, we identified key trends in intervention types, setting, and team composition that appear to reduce acute care use within this population. Specifically, home visits and comprehensive geriatric assessments appeared to reduce ED use. Interdisciplinary assessments and teams that included a geriatrician were also more likely to show reduced ED use compared with those that did not include geriatricians. Finally, the use of community-based strategies that included regular contact with a nurse, general practitioner, or geriatrician led to better outcomes for patients. The effect of these interventions on secondary outcomes was mixed.
Many of our findings are consistent with prior research in this field. A prior review by McCusker et al.Reference McCusker and Verdon13 also examined intervention setting and found that community-based interventions were more effective than those conducted in-hospital. The authors suggested that this finding could be due to longer interventions and a greater ability to affect continuity of care. Our findings support this explanation of longer intervention length influencing the success of community-based interventions. McCusker et al.Reference McCusker and Verdon13 also proposed that patients who present to and are admitted to the hospital are more likely to have prior ED visits, have problems accessing primary care, and could have higher medical complexity. These confounding factors may explain why hospital-based interventions do not appear to be as effective as community-based interventions.
We identified some novel interventional characteristics that require further exploration. Specifically, the use of comprehensive geriatric assessments as an interventional component appeared to have promising effects on lowering ED use. These assessments are done by an interdisciplinary team working together to develop a coordinated plan for treatment and follow-up of older patients.Reference Eamer, Taheri and Chen18, Reference Ellis, Whitehead, Robinson, O'Neill and Langhorne19 Because patients can present with a wide variety of overlapping issues, the use of assessors and care providers from different fields collaborating to identify, plan for, and treat patients is paramount and may explain the effect of this interventional strategy in older populations.Reference Ellis, Whitehead, Robinson, O'Neill and Langhorne19 Further, among the 19 studies that used a multidisciplinary team as part of their intervention, 10 had a geriatrician or geriatric fellow on the team, and those were more likely to show a decrease in either ED use, hospitalization, or hospital readmission. The inclusion of a geriatrician has been identified as a key aspect of effective comprehensive geriatric assessments, due to geriatricians’ expertise of aging-related physiological changes and the challenges associated with frailty, dementia and polypharmacy, as well as the role they play in facilitating appropriate access to other services or specialists.Reference Kocman, Regen and Phelps20, Reference Sucher, Mangram and Dzandu21 However, it should be recognized that geriatricians do not work in isolation, and therefore the influence of team composition on outcomes within these studies is hypothesis generating. Future work is required to understand how team composition and team dynamics influence success within these complex interventions.
Strengths and limitations
This systematic review has a number of strengths. Our comprehensive search across multiple databases provides up-to-date evidence on the effectiveness of interventions to reduce ED/hospitalization among older patients. Furthermore, we examined various outcomes, identified the composition of multidisciplinary teams, and stratified results by intervention type to identify key trends. This expands upon previous reviews that focused on intervention setting or the skills and training needed for multidisciplinary teams, rather than the composition of the team.Reference McCusker and Verdon13, Reference Ellis and Sevdalis22
However, these findings should be interpreted in light of the study limitations. Firstly, there was a lack of standard reporting of outcomes across studies, which precluded a quantitative synthesis of results. Within the 53 included studies, there was considerable variability in study methodology, sample size, secondary outcomes, measurement of outcomes, and study quality. Secondly, the majority of studies employed co-interventions, which makes it difficult to isolate the independent effects of intervention components on healthcare use and patient outcomes. We recognize that interventions are often adapted to local context and need, which inherently makes this a difficult area of work to make meaningful comparisons and generalizable recommendations. Whether the efficacy of these complex interventions differs among community- versus facility-living individuals and/or patient populations with specific chronic conditions versus undifferentiated populations remains to be determined. While standardized approaches to the measurement of common outcomes could help with future interpretation, the fact that interventions are shaped by context, available resources, and healthcare structure suggests that conducting pooled analyses in this research space may be elusive unless restricted to specific patient subgroups and interventional factors.
CONCLUSIONS
This comprehensive review identified 53 studies aimed at reducing acute care utilization among older patients. While models of care that include comprehensive geriatric assessments and home visits may reduce acute care utilization, standardization of outcome measures is needed to further delineate which parts of these complex interventions are contributing to efficacy. The potential effects of multidisciplinary team composition on patient outcomes also warrant further investigation.
Supplemental material
The supplemental material for this article can be found at https://doi.org/10.1017/cem.2020.4.
Acknowledgements
We would like to thank the librarian, Heather Ganshorn, at the University of Calgary, for her assistance in our database search.
Financial support
This project was funded by the Alberta Health Services – Emergency Strategic Clinical Network 2017 Systematic Review Grant Competition, which was used solely for stipends for two student investigators (NS, MS).
Competing interests
None declared.
