Article Text

Download PDFPDF

A systematic review examining the impact of redirecting low-acuity patients seeking emergency department care: is the juice worth the squeeze?
Free
  1. Scott William Kirkland1,
  2. Amir Soleimani2,
  3. Brian H Rowe1,3,
  4. Amanda S Newton2
  1. 1 Department of Emergency Medicine, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Alberta, Canada
  2. 2 Department of Pediatrics, Faculty of Medicine and Dentistry, University of Alberta, Edmonton, Alberta, Canada
  3. 3 School of Public Health, University of Alberta, Edmonton, Alberta, Canada
  1. Correspondence to Dr Brian H Rowe, Department of Emergency Medicine, University of Alberta, Edmonton, AB T6G 2R3, Canada; brian.rowe{at}ualberta.ca

Abstract

Objectives Diverting patients away from the emergency department (ED) has been proposed as a solution for mitigating overcrowding. This systematic review examined the impact of interventions designed to either bypass the ED or direct patients to other alternative care after ED presentation.

Methods Seven electronic databases and the grey literature were searched. Eligible studies included randomised/controlled trials or cohort studies that assessed the effectiveness of pre-hospital or ED-based diversion interventions. Two reviewers independently screened the studies for relevance, inclusion and risk of bias. Pooled statistics were calculated as relative risks (RR) with 95% confidence intervals (CI) using a random effects model.

Results Fifteen studies were included evaluating pre-hospital (n=11) or ED-based (n=4) diversion interventions. The quality of the studies ranged from moderate to low. Patients deemed suitable for diversion among the pre-hospital studies (n=3) ranged from 19.2% to 90.4% and from 19% to 36% in ED-based studies (n=4). Of the eligible patients, the proportion of patients diverted via ED-based diversion tended to be higher (median 85%; IQR 76–93%) compared with pre-hospital diversion (median 40%; IQR 24–57%). Overall, pre-hospital diversion did not decrease the proportion of patients transferred to the ED compared with standard care (RR 0.92; 95% CI 0.80 to 1.06). There was no significant decrease in subsequent ED utilisation among patients diverted via pre-hospital diversion compared with non-diverted patients (RR 1.09; 95% CI 0.99 to 1.21). Of the three pre-hospital studies completing a cost analysis, none found a significant difference in total healthcare costs between diverted and non-diverted patients.

Conclusion There was no conclusive evidence regarding the impact of diversion strategies on ED utilisation and subsequent healthcare utilisation. The overall quality of the research limited the ability of this review to draw definitive conclusions and more research is required prior to widespread implementation.

  • emergency department
  • emergency department utilisation
  • crowding

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is already known on this subject

  • Diverting low-acuity patients away from the emergency department has been proposed as a potential solution for mitigating overcrowding.

  • The effects of diversion strategies on patients’ healthcare utilisation and health outcomes are unknown.

What this study adds

  • An extensive search of the literature identified 15 studies comparing either pre-hospital or emergency department-based diversion strategies to standard emergency care response.

  • This review was unable to demonstrate conclusive evidence regarding the effectiveness of diversion strategies on emergency department use and subsequent healthcare utilisation.

  • At this time there is insufficient evidence to recommend the implementation of diversion protocols as effective or safe strategies to address emergency department overcrowding.

Introduction

It is estimated that between 13.7% and 27.1% of all emergency department (ED) visits in the USA could be safely managed at alternative locations and result in an estimated cost savings of over $4 billion.1 In the UK an estimated 17% of ED visits are considered ‘inappropriate’2 while, in Canada, it is estimated that one in five ED visits could be appropriately managed at a family physician’s office.3 Diversion of low-acuity patients from the ED to alternative sources of care may include pre-hospital diversion, in which patients are diverted prior to presenting to the ED (ie, dispatch-directed diversion, treat and release patients at the scene),4 or ED-based diversion in which patients presenting to the ED are diverted to an alternative source of care.5 Diverting low-acuity patients away from the ED to alternative sources of care is controversial due to its potential safety concerns.6 The safety and effectiveness of ED diversion is unclear in part due to the fact that many studies use a single-group study design and do not compare diversion with standard care.5

The objective of this systematic review was to examine the effectiveness and safety of pre-hospital and ED-based diversion strategies on ED utilisation, non-ED healthcare utilisation and patient outcomes compared with standard emergency care responses.

Methods

Protocol

A protocol was developed a priori and registered on the PROSPERO register (CRD42016033613). Reporting adheres to the PRISMA guidelines.

Search strategies

The search strategies used language (English) and year (1990 to January 2016) restrictions and were applied in seven electronic bibliographic databases: Medline, Embase, Cochrane Library, PsycINFO, CINAHL, Social Services Abstracts and ProQuest Dissertations and Theses Global (see supplementary file 1 for the Medline search strategy). To identify unpublished studies and studies in progress, Google Scholar, Health Services Research Projects in Progress, Health Services/Sciences Research Resources, National Centre for Biotechnology Information Bookshelf and ClinicalTrials.gov were searched. In addition, a SCOPUS forward search of all included studies was completed (2017). Reference lists of included studies and conference proceedings from the Society for Academic Emergency Medicine (2008–2017) and the Canadian Association of Emergency Medicine (2008–2017) were also reviewed.

Supplemental material

Study selection

Two reviewers (AS, SWK) independently screened the search results to identify potentially relevant studies; at least one reviewer independently reviewed the full text of potentially relevant studies to confirm their inclusion. Reviewer discrepancies on study inclusion were discussed and resolved via third party abjudication (ASN, BHR). Randomised/controlled clinical trials and controlled cohort studies (retrospective or prospective) were eligible for inclusion. Case–control or before/after studies were not eligible for inclusion, as recommended by the Cochrane Effective Practice and Organisation of Care Group (EPOC) in this review7 due to difficulties attributing causation in these studies. To be included in the review, studies of these designs had to (1) include patients seeking ED care; (2) assess the effect of any diversion strategy designed to either bypass the ED to an alternative source of care (pre-hospital diversion) or direct patients to a non-ED setting after presentation to an ED (ED-based diversion); and (3) compare the diversion with standard emergency care. What was considered to be standard emergency care could vary among studies and intervention types, which could include, but was not limited to, patient transport to the ED, standard emergency medical service (EMS) assessment, or attending and being assessed by an ED physician. Strategies to divert ambulances away from EDs while the patient was being transported (ambulance diversion) were excluded, as this review attempted to examine the effectiveness of diversion strategies other than those that temporarily closed the ED to incoming ambulance transport. No studies were excluded on the basis of patient age or presenting medical conditions.

The primary outcome of interest was the number of visits to the ED, including initial and subsequent visits. For the purposes of this review, subsequent ED visits referred to additional unplanned return ED visits following the index visit. Secondary outcomes of interest were: patient diversion to, or utilisation of, non-ED-based settings for care (eg, assistance at the scene of the emergency, referral to primary care); hospitalisation; serious adverse events (eg, death, ICU admission); and patient quality of life (eg, health status). Additional outcomes of interest—namely, the percentage of patients considered eligible for diversion, the percentage of patients diverted, compliance with the diversion strategy, patient refusal of diversion and cost-effectiveness—were extracted a posteriori.

Quality assessment

Randomised/controlled clinical trials were assessed using the Cochrane Collaboration Risk of Bias Assessment tool8 while the quality of cohort studies was assessed using the Newcastle–Ottawa Scale.9 Two reviewers (AS, SWK) independently assessed each study’s methodological quality. Disagreements that could not be resolved by discussion were adjudicated via a third party (ASN).

Fidelity of the diversion strategy and its implementation were assessed by one reviewer (SWK) and checked by another (AS) based on the five criteria from the Treatment Fidelity Assessment Grid including: fidelity to theory/literature, provider training, treatment implementation, treatment receipt and treatment enactment.10

Data extraction

Data from the included studies were extracted using standardised forms by one reviewer (AS) and verified for accuracy and completeness by a second (SWK). Discrepancies were resolved by discussion between the reviewers and any disagreements were resolved via third party adjudication (ASN, BHR). In the case of unclear or unreported information in the original studies, primary authors were contacted. For studies assessing diversion strategies across various settings, only the data from arms receiving a pre-ED or ED-based intervention were extracted.

Data analysis

Unadjusted pooled statistical analyses were performed using Review Manager (RevMan Version 5.3; Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). For each reported outcome in which pooling was possible (in which two or more studies reported a similar outcome measure), studies were subgrouped based on whether the study employed a pre-hospital or ED-based diversion strategy. Only studies with the same study design were included in the pooled subgroup analysis. Given that the majority of the included studies were randomised controlled trials (RCTs) or controlled clinical trials, all of the pooled meta-analysis consisted of randomised or controlled clinical trials, with outcomes for the observational cohort studies being reported separately. For dichotomous variables, individual and pooled statistics were calculated as unadjusted relative risks (RR) with 95% confidence intervals (CI) using a random effects model. Heterogeneity was tested using the I2 statistic with I2 values of 25%, 50% and 75% representing low, moderate and high degrees of heterogeneity, respectively.8 In some cases, studies calculated and reported adjusted odds ratios (aOR), adjusted risk difference (aRD) and adjusted relative risks (aRR), which were recorded. Published group differences in the proportion of events using frequencies with corresponding P values or mean differences (MDs) with 95% CIs were reported. Pre-planned subgroup analyses for age and illness/disorder/presenting complaint were not possible given the heterogeneity among included studies. A post-hoc decision was made after the review protocol was registered to group study outcomes according to whether the studies employed pre-hospital or ED-based diversion strategies so that outcome reporting was clinically meaningful. No sensitivity analysis was planned.

Results

Study selection and characteristics of included studies

Figure 1 describes the flow of studies through the selection process. The search strategies identified 7891 studies for screening. The full text of 192 studies identified as potentially relevant were reviewed, 15 of which were included in the review. The 15 studies consisted of 10 clinical trials11–20 and five observational cohort studies (prospective21–23 and retrospective24 25). Three studies (Mason et al (2007),11 Mason et al (2008)15 and Dixon et al (2009)14) assessing a ‘treat and release’ protocol reported different outcomes for the same dataset of patients; for the purposes of this review, Mason et al (2007)11 was reported as the main report while Mason et al (2008)15 and Dixon et al (2009)14 were considered supplemental reports. The methodological quality and fidelity assessment was not completed for studies considered supplemental reports.14 15 Mason et al (2012) assessed the effectiveness of emergency care practitioners across five various healthcare services and, as such, only the data from the ambulance service group (pre-hospital diversion) were extracted for the purposes of this review.19 The characteristics of the included studies are shown in table 1. All of the studies reported enrolling patients with low-acuity or ‘non-serious’ medical conditions; however, the criterion for judging patient acuity was not well reported and what was considered a low-acuity medical condition varied among studies. Snooks et al (2014) and Snooks et al (2017) specifically reported enrolling elderly patients calling for an ambulance due to a fall.12 13 The control groups for all of the studies consisted of low-acuity patients, except the study by Washington et al 21 in which the control group consisted of patients with abdominal pain, musculoskeletal or respiratory infection complaints who were deemed unsafe for diversion. In five of the pre-hospital studies it was noted that some patients in the control group were not transported to the ED due to patients refusing transport to the ED, paramedics treating patients on-scene, or paramedics transporting patients to other non-ED healthcare settings.12 13 16 18 22

Figure 1
Figure 1

Literature search flow diagram. 

Table 1

Overview of study and intervention characteristics and implementation outcomes associated with diversion implementation

Methodological quality of the included studies

Overall, the clinical trials were assessed as having either an unclear17 or high risk of bias11–13 16 18–20 (see online supplementary file 2). High risk of bias in the studies was based on inadequate allocation concealment,18 non-blinding of staff and participants,12 13 19 20 selective outcome reporting11 16 18 and lack of compliance with the study protocol among the participating staff.16 The cohort studies were assessed as being of moderate21–25 quality with the primary methodological limitation being a lack of details on comparability between the cohorts (see online supplementary file 3).21–23 25

Diversion characteristics

Eleven studies evaluated pre-hospital diversion, of which seven studies were clinical trials11–16 18 and four studies were controlled observational cohort studies.19 22 24 25 In the study by Dale et al,18 all emergency calls for non-serious conditions were passed on to a nurse/paramedic following ambulance dispatch who, with the aid of computerised decision support, assessed, triaged and provided advice to the patients, including asking patients triaged and not requiring an ambulance whether they still preferred ambulance transport. In the study by Krumperman et al,24 low-severity calls were diverted from emergency call centres to nurse call centres, in which nurses used evidence-based protocols to provide patients' instructions and referrals to primary care providers or urgent care.24 Nine studies assessed the impact of an EMS-based strategy: five studies assessed a ‘treat and release’ strategy in which paramedics assessed and treated low-acuity patients at the scene11 14 15 19 22 while four studies assessed strategies in which ambulance crews either diverted low-acuity patients to a minor injury unit (MIU),16 a community-based falls service,12 13 or transported intoxicated patients to a detoxification centre.25

Four studies evaluated ED-based diversion, of which one study was a randomised controlled trial17 and three studies were controlled observational cohorts.20 21 23 Ellbrant et al 23 examined the role of an experienced paediatric nurse to identify patients suitable for referral to primary care or discharge home, while Washington et al 21 and Washington et al 17 examined deferral by a triage nurse to a next-day primary care appointment. Doran et al diverted patients via a research specialist and a triage nurse identifying eligible ED patients to be diverted to a same-day appointment at a primary care clinic.20

Diversion fidelity

Steps to ensure fidelity of the diversion strategy and its implementation were lacking in the studies (see online supplementary files 4 and 5). Six studies12 17 18 21 22 25 reported the use of literature and/or experts to develop the strategy; however, the strategies were not necessarily underpinned by theoretical propositions. Eight studies described how personnel in diversion strategy delivery were trained.11 12 15 17 18 21 22 25 Five studies used protocols/guidelines to standardise intervention delivery and receipt12 16 17 21 22; this approach was lacking or unclear in other studies. The majority of studies reported steps taken to ensure treatment enactment although approaches varied.12 16–18 21 22 24

Effects of diversion

ED attendance

There was considerable variation among the studies assessing pre-hospital diversion with regard to the proportion of patients deemed safe and appropriate for diversion, as well as the proportion of patients who were actually diverted away from the ED (table 1). Dale et al assessed a prehospital diversion intervention consisting of telephone-based assessment and advice from nurses/paramedics and found that 52% (n=330/635) of emergency callers with non-serious conditions did not require ambulance transport, of which 63% of callers triaged as not needing an ambulance were transported and attended the ED while only 18% (n=58/325) of these patients accepted the ambulance diversion.18 Snooks et al reported that, among those patients allocated to be diverted, only 10% (n=41/409) were actually diverted from the ED (table 1).16 The study by Ross et al, which diverted intoxicated patients to a detoxification facility, found that 19% (n=138/718) of intoxicated patients were deemed safe for diversion, of which 92% (127/138) were successfully diverted away from the ED with the remaining 11 patients being denied entry into the facility.25 Mason et al reported that 43% (n=257/593) of patients seeking ambulance transport were discharged following consultation with an emergency care physician, while another 14% (n=85/593) were referred by the emergency care physician to a primary care provider.19 A pooled meta-analysis of four pre-hospital diversion studies12 13 16 18 did not identify a difference in the number of patients conveyed to the ED between diverted and non-diverted patients (RR 0.92; 95% CI 0.80 to 1.06); however, significant heterogeneity was identified (I2=86%) (see figure 2). Overall, the median proportion of eligible patients diverted via pre-hospital based strategies was 40% (IQR 24–57%).

Figure 2

The impact of pre-hospital diversion on initial ED use.

There was similar variation with regard to the proportion of patients deemed appropriate for diversion among studies assessing ED-based diversion strategies (table 1). In their study in 2000, Washington et al reported that 19% (n=226/1187) of ED patients were safe for deferral to a next-day primary care appointment, of which 68.1% were diverted.21 In their 2002 study, Washington et al found that 36% (n=421/1176) of patients were considered safe for next-day appointments and that almost half (48%; n=143/229) of ED patients refused the offer of diversion.17 Ellbrant et al reported that nurses identified 33% (n=344/1057) of paediatric patients who did not require ED physician assessment.23 Of those children who did not require assessment by a physician and were seen by a paediatric nurse, 51% were discharged home, 33% were referred for primary care and 16% were referred for other ED care.23 Finally, Doran et al reported that 85% (n=563/662) of patients accepted diversion when offered compared with 15% (n=99/662) of patients who refused and received care in the ED.20 Overall, the median proportion of eligible patients diverted via ED-based diversion strategies was 85% (IQR 76–93%).

Subsequent ED attendance

Overall, there is insufficient evidence to conclude whether or not ED diversion alters patients' subsequent ED utilisation (table 2). A meta-analysis of three pre-hospital studies12 13 15 failed to identify a significant difference in subsequent ED visits among diverted and non-diverted patients within 1 month following the index call (RR 1.09; 95% CI 0.99 to 1.21) (see online supplementary file 6). Heterogeneity was low (I2=0%).

Table 2

Impact of diversion on subsequent ED visits

The effect of ED-based diversion on subsequent visits to the ED was only reported in two studies. Ellbrant et al reported that 7.6% of patients either referred to primary care or discharged home by a paediatric nurse returned to the ED with similar symptoms within 72 hours.23 Doran et al reported no significant differences in subsequent ED visits within 2 weeks of the index visit (RR 0.55; 95% CI 0.29 to 1.06) between diverted and non-diverted patients.20

Use of other healthcare services

Two studies compared subsequent unplanned non-ED healthcare utilisation between diverted and standard ED care patients (table 3, online supplementary file 7). Mason et al reported that elderly patients undergoing a ‘treat and release’ protocol were more likely to subsequently visit other healthcare services (RR 1.21; 95% CI 1.06 to 1.38).11 Snooks et al reported that diverted patients were less likely to make a further emergency service call at 1 month after the index call compared with standard care patients (aOR 0.82; 95% CI 0.71 to 0.94); however, no significant difference was found in the proportion of patients making emergency calls at 6 months after the index call (aOR 0.90; 95% CI 0.80 to 1.01).12

Table 3

Impact of diversion on non-ED healthcare utilisation

Three ED-based diversion strategies reported on subsequent healthcare utilisation (table 3).17 20 21 The study by Washington et al in 2002, which assessed next-day referrals to a primary care clinic, reported no differences in the number of patients making subsequent visits to the ED or primary care clinic (RR 1.09; 95% CI 0.23 to 5.26).17 Washington et al in 200021 and 200217 reported that 90.3% (n=139/154)21 and 95.9% (n=71/74)17 of patients deferred to a next-day appointment at a primary care clinic attended the appointment, respectively. Doran et al reported that patients deferred to a same-day appointment at an on-site primary care clinic were more likely to attend follow-up primary care appointments than patients receiving standard ED care (aRD 11.3; 95% CI 6.0 to 16.5).20

Hospitalisation

Three clinical trials assessing pre-hospital diversion compared the frequency of subsequent hospitalisations within 1 month of the post-index call between diverted and standard care patients and found no differences (RR 0.95; 95% CI 0.82 to 1.09)11–13 (table 4,  online supplementary file 8). Heterogeneity was moderate (I2=68%). Similarly, one study reported no differences in subsequent hospitalisations between diverted and non-diverted patients 1 month after the post-index call (RR 1.01; 95% CI 0.95 to 1.07).12

Table 4

Hospitalisation following diversion and standard emergency care

The effectiveness of ED-based diversion strategies to reduce subsequent hospitalisations was not frequently reported (table 4). In their 2000 study, Washington et al reported a significant decrease in subsequent hospitalisations among patients diverted to a follow-up primary care appointment compared with patients attending the ED (RR 0.15; 95% CI 0.05 to 0.48).

Patient outcomes

Four pre-hospital diversion studies reported mortality,11–13 25 with no differences in mortality within 1 month (RR 0.98; 95% CI 0.82 to 1.18; I2=0%; n=4 studies) or 6 months (RR 1.03; 95% CI 0.91 to 1.16; n=1 study) after the index visit (see online supplementary files 9 and 10). Two studies assessing ED-based diversion strategies reported on mortality. In 2000 Washington et al reported no difference in mortality within 1 month between diverted and non-diverted patients (RR 0.39; 95% CI 0.02 to 6.94)21 while in 2002 they reported no deaths among the participating patients.17 No differences in the risk of worsening health status11 or quality of life12–14 17 were reported among the included studies (see online supplementary file 10).

Cost-effectiveness

Three studies assessing pre-hospital diversion strategies reported on the cost-effectiveness of diversion strategies for all patients allocated to the intervention and control groups. In 2014 Snooks et al reported no differences in the total cost of subsequent healthcare utilisation (MD £247; 95% CI −£247 to £741) among diverted and standard care patients.13 In a follow-up study in 2017, Snooks et al reported similar total costs of healthcare resource use at 1 month (adjusted mean change  £190.24; 95% CI −£13.83 to £394.31) or 6 months (adjusted mean change £24.20; 95% CI −£468.01 to £516.40) between elderly patients diverted to a fall service compared with patients transported to the ED.12 Dixon et al reported a decreased cost associated with reduced ED attendance (MD −£32; 95% CI −£38 to –£26) among diverted patients; however, no significant differences were found with regard to the total costs associated with same day or subsequent healthcare utilisation (MD −£140; 95% CI −£694 to £415), which included same-day and subsequent ED presentations and inpatient costs.14 No studies assessing ED-based diversion strategies examined cost-effectiveness.

Discussion

An extensive search of seven electronic databases and the grey literature was conducted to identify all of the available comparative studies assessing the effectiveness of ED diversion strategies. Overall, the evidence to support or refute the effectiveness and safety of this practice in the EMS setting is lacking. The percentage of patients reported to be suitable for diversion was low,17 18 20 21 23 25 with some studies reporting fewer than half of the suitable patients diverted away from the ED,12 13 16 18 19 22 and many patients refusing diversion.17 18 20 Given the potential costs (eg, additional staffing, training, updating, surveillance) associated with implementing pre-hospital and ED-based diversion strategies, evidence supporting its value in the EMS system is limited.14

Overall, there also appears to be limited evidence that ED diversion alters subsequent healthcare utilisation. Among the studies that could be pooled, no differences in subsequent ED utilisation were found. No change in subsequent hospitalisations was found among four studies assessing pre-hospital diversion strategies, while the study by Washington in 2000,21 which assessed ED-based diversion, reported a significant decrease in subsequent hospitalisations. The effect of diversion strategies on subsequent non-ED-based health services was inconsistent. With respect to the safety of pre-hospital and ED-based diversion, ED diversion appears to be no less safe or harmful for low-acuity patients than if they were treated in the ED.11–14 17 21 25

While the current state of evidence is mixed, the results of this review provide several key directions for future diversion-based research. First, there is a need for additional high-quality RCTs assessing the effectiveness of ED diversion strategies. Of the 15 included studies, eight were clinical trials,11–18 none of which were assessed as having a low risk of bias. Second, this review identified inconsistency in outcome reporting across studies, limiting this review to broad interpretations of outcome trends and in some cases precluded studies from inclusion into a meta-analysis. Studies frequently did not report important clinical and patient outcomes such as subsequent ED utilisation and mortality, which are important to understand the effectiveness and safety of diversion strategies. Moving forward, it is important for future research to report these outcomes to improve the validity of author claims and develop a consensus among the research and clinical communities on the effectiveness and safety of diversion strategies. Third, despite all of the studies employing a control group, several studies did not report outcomes for standard care patients. Future studies must examine the differences between standard ED care and diversion strategies to better understand the safety and effectiveness of diversion strategies. Fourth, the effectiveness of the ED diversion strategies across many of the studies is unknown due to the limited reporting of the number of patients suitable for diversion, as well as compliance with or refusal to accept diversion. In addition, despite the clear resource implications for implementing ED diversion strategies, including training and hiring additional staff, costs of implementing the diversion strategies were infrequently reported.12–14 If a minority of patients are deemed suitable for diversion and an even smaller proportion of patients accept diversion recommendations, the costs of implementing diversion strategies may not be justified. Fifth, there were significant differences in type and approach to diversion across the included studies. Future studies should build on the results of this review to develop and compare similar diversion strategies to standard approaches to mature the evidence base. Should promising diversion strategies emerge from this evidence base, there may be an opportunity for comparative effectiveness studies involving differing diversion approaches. Finally, the criteria to determine patient suitability for ED diversion varied among studies; some studies did not report the criteria for how a patient was judged as presenting with a low-acuity/non-serious medical condition. It is possible that only a select set of medical conditions are suitable for ED diversion, but this could not be assessed in this review due to inadequate reporting. Future studies that are designed to conduct subgroup analyses between patient ages and condition types would provide valuable data to the field.

Study limitations

The methodological quality of the included studies necessitates caution when interpreting the results of this review. As discussed previously, the quality of the studies included in this review ranged from moderate to low. The results of this review are undermined by the limited quality and comprehensiveness of outcome reporting among the studies regarding healthcare utilisation and patient safety. Second, based on the assessment of intervention fidelity, it is possible that the implementation of the ED diversion strategies was suboptimal. It is important that future research ensures that the diversion strategies are implemented as originally intended in order to determine the true effectiveness of diversion strategies. Third, this review did not include assessments of all available diversion strategies. For example, a Canadian e-health study assessed the effectiveness of a computer-based algorithm to assist triage nurses to identify ED patients suitable for diversion to a medical clinic; however, this study did not include a comparison group26 and, as such, could not be included in this review. A final limitation of this review relates to the risk of publication and selection bias. While steps were taken to limit these issues (eg, a comprehensive search of the published and unpublished literature; independent reviewers and adjudication), it is possible that studies could have been missed due to the fact that some limitations based on English language and date of publication were imposed. Finally, despite the extensive search, the use of these interventions is probably underestimated by this review. We suspect many studies assessing diversion interventions may not have been found because formal reporting of their evaluations has not been completed and, when completed organisationally, publication is not a priority.

Conclusions

This review was unable to identify clear evidence regarding the impact of pre-hospital and ED-based diversion on subsequent healthcare utilisation. There was considerable variation in the proportion of patients deemed suitable for diversion across the studies. From the available evidence, there does not appear to be clear increased personal health risks to patients who receive diversion versus standard emergency care responses. At this time, there is insufficient evidence to recommend the implementation of ED diversion strategies to address ED overcrowding. Additional comparative effectiveness studies are urgently needed to examine the ‘reach’ of ED diversion strategies, including patient compliance and refusal with diversion, as well as a better understanding of the costs, changes in healthcare utilisation and patient outcomes prior to widespread implementation of any diversion strategies.

Acknowledgments

The authors thank Robin Featherstone (Alberta SPOR Support Unit) for conducting the study search strategy. We would also like to thank Drs Dixon, Knowles, Mason, Ross and Washington for responding to our requests for additional information. The authors would like to thank Lindsay Gaudet and Maureen Kruhlak for their assistance with the review.

References

Footnotes

  • Contributors SWK drafted the protocol, screened abstracts and articles for inclusion, extracted data, assessed quality, analysed data and wrote the manuscript. AS drafted the protocol, screened abstracts and articles for inclusion, extracted data, assessed quality, analysed data and wrote the manuscript. BHR conceptualised the study, drafted the protocol, provided third party adjudication for inclusion screening and wrote the manuscript. ASN conceptualised the study, drafted the protocol, provided third party adjudication for inclusion screening, data extraction, and quality assessment and wrote the manuscript.

  • Funding Funding for this review was provided by the Emergency Strategic Clinical Network (Alberta). In-kind support of the research librarian services was provided by the Alberta SPOR Support Unit, Health Systems Research, Implementation Research & Knowledge translation platform. ASN holds a New Investigator Award from the Canadian Institutes of Health Research (CIHR). BHR research is supported by a Tier I Canada Research Chair in Evidence-based Emergency Medicine from CIHR through the Government of Canada (Ottawa, ON).

  • Competing interests None declared.

  • Patient consent Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The data reported in this systematic review were retrieved from the published data reported in the included studies.