• 
• 

Three Part Question

IN [a population of adult UK emergency department frequent attenders] WHAT [effective interventions are there] AT [reducing the number of ED attendances]?

Clinical Scenario

A 22-year-old woman presents to the Emergency Department (ED) with suicidal ideation and ongoing Medically Unexplained Symptoms (MUS). This is her second attendance today, and her fourth this week. This patient is a known high-intensity user/Frequent Attender (FA) due to multiple previous episodes of Deliberate Self-Harm and intentional overdose. She has a background of emotionally unstable personality disorder, depression & anxiety, and a learning disability- all of which contribute to her distress while in the department, meaning she frequently leaves before being seen by a clinician. You wonder if there is an effective intervention that you could implement, to reduce her number of ED visits.

Search Strategy

Evidence was searched using Medline, EMBASE, CINAHL and EMCARE, 1994- April 2021, via OVID interface.
Same strategy used for all databases.
1. exp *"EMERGENCY MEDICAL SERVICES"/ OR exp
2. *"EMERGENCY SERVICE, HOSPITAL"/
3. (frequen* OADJ2 attend*).ti,ab
4. (frequen* OADJ2 flyer*).ti,ab
5. ((high ADJ1 intensity) ADJ1 user*).ti,ab
6. (2 OR 3 OR 4)
7. (1 AND 5) [DT 2016-2021] [Languages English]
8. exp "UNITED KINGDOM"/ OR uk
9. (1 AND 5 AND 7)

Search Outcome

The search term yielded 445 articles in total. Grey literature was also reviewed, and the reference lists of all relevant papers were screened. Once the titles were screened, 102 abstracts were reviewed, 98 remained after removal of duplicates, and 6 relevant papers met the eligibility criteria and were included in the review- three Systematic Reviews (which reviewed 5x RCT’s, 30x Comparative Cohort studies, 9x Case Series, over the 3 studies) and three cohort studies. Studies were excluded if: the intervention was not based in the ED; did not focus on adult (18yrs and older) frequent presenters; or did not include any data on, or examine the impact of, an intervention to reduce the ED utilisation of frequent ED users. Bibliographies of the systematic reviews were reviewed before being included, to ensure that this review did not ‘double count’ individual papers.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Huang et al. 2020 UK	38 pts initially identified (as having presented to the ED >4 times in last 12 months + met exclusion criteria) as the cohort. The participants were then sent a letter identifying them as frequent attenders and inviting them to a tailored multi-disciplinary team meeting. Only the impact of the letter was assessed (not the MDT meeting itself).	Cohort study with 6 months follow-up. Paired sample for means t-test was used to compare attendances pre- and post-intervention.	ED Attendance	There were a wide range of presenting complaints to ED, most of which were psychiatric related – overdose, deliberate self-harm, alcohol/drug related. The intervention letter improved the attendance rate the most in this cohort, whereas it had a negligible effect on attendances with physical complaints. The average fall in attendance was 33% in the 6 months post patients receiving the intervention letter. This reached significance level with p=0.039.	• Small sample size (38 pts). • Exclusion criteria eliminated large subsection of typical FA profile (Exclusion criteria: pt is under 18yro, has known long term medical conditions and known safeguarding concerns). • Only intervention implemented was the sending out of a letter- MDT as intervention not assessed. • This study used pre- and post-intervention with the same cohort and therefore may not adequately account for secular trends or regression to the mean.
Heelas L 2020 UK	Systematic review of 11 studies that looked at reducing FA’s presenting with persistent pain to the ED. This review covered multiple types of intervention: MDT Review; Referrals to primary care/other services; Behavioural care +/- addiction care package; Case management including advice on medical, behavioural and social issues; Medication restriction, care plans in notes; Chronic pain care plan or pain management programme) from a range of levels of evidence.	Systematic review (11 studies included- 1x RTC, 1x cohort/ control, rest case series. A Population, Intervention, Comparator, Outcome and Study design methodology was utilised).	ED Attendance	There is low to moderate evidence to suggest that MDT case management and treatment with specialists in pain management and mental health is effective in reducing frequent Emergency Department attendance for persistent pain. (Grade A and B evidence for reducing frequent attendance was for a multidisciplinary team to offer a care package, including referral to specialists in both chronic pain and mental health. Grade C evidence suggests that multidisciplinary, prescription restriction or a pain management programme were effective in reducing ED FA.)	• While ‘persistent or chronic pain’ is a common reason for FA’ to attend the ED, given the heterogeneous nature of this group- FA’s- this evidence may only be applicable to a small subsection of the group (ie. not generalisable). • Studies including homelessness or substance abuse were excluded, large subsections of FA group. • Most studies included were observational and did not qualify as an experimental design.
Gerdtz et al. 2019 Australia	All FA’s (>5 visits/year) attending a major metropolitan hospital in Australia between November 2010 to September 2014, looking at ED attendance data 12 months before and after implementation of a ‘Management plan’ in these participants. This study also surveyed the perspectives of staff who use ‘management plans’ between July to November 2014.	A descriptive observational design including before and after measures of attendance, and survey of staff perceptions.	ED Attendance	Fifty-seven patients made 1482 ED attendances. Of these 830 occurred in the 12 months before the management plan was implemented and 652 during the 12 months after. The number of attendances per patient decreased from a median of 11 to 4.	• Observational study design with no control, poor level and grade of evidence. • Small sample size. • Selection bias within the management plan staff survey. • No mention of statistical testing, intervals or power; only ‘association’ noted in reduced attendances; correlation found but causation lacking in the study.
			Staff Perceptions	Staff considered management plans to be beneficial to care planning practices and individual patient outcomes.
Flowers & Shade 2019 USA	A retrospective cohort study design was used to analyse a multidisciplinary care coordination program (ie. MDCM) on 58 patients between Jan 2015 - Aug 2018 at a single hospital ED in Northern California. Patients were identified from a high-utilisation report when they had 10 or more ED visits in a 6-month period, were 18 years of age or older, and members of the integrated delivery system's health plan.	Retrospective Cohort study design. Data was collected at initiation and 6 months postintervention. The pre-/post-analysis consisted of descriptive statistics, Wilcoxon signed ranks test and binary logistic regression.	ED Attendance	There was a statistically significant pre-/post- difference of 7.7 ED visits (95% confidence interval [CI] = 4.44-10.97, p < .001).	• Relatively small sample size and only conducted at one centre limiting generalisability. • Dissimilar healthcare system set up (USA vs UK), this sample was taken from participants from a particular health plan, this may have biased the data collected. • This study only focused on ‘very high-intensity ED users’ (>/= 10 in a 6-month period), this intervention may therefore not be applicable or as effective for ‘lower level’ FA’s.
			Inpatient Admissions	The program did not result in statistically significant reduced hospital admissions (95% CI =-1.24 to 1.45, p = .875).
Soril et al. 2015 Canada	Systematic review of 17 articles to establish the effectiveness of interventions to reduce frequent ED use among a general adult high ED-use population. Study design, patient population, intervention, the frequency of ED visits, and costs of frequent ED use and/or interventions were extracted and narratively synthesized. Studies were included if they: had a control group (controlled trials or comparative cohort studies), were set in an ED or acute care facility, and examined the impact of an intervention to reduce frequent ED use in a general adult population.	Systematic Review of the literature from 1950- Jan 2015. Seventeen RCT’s (4) & Comparative Cohort studies (13) included. Three intervention categories were identified: case management (n = 12), individualized care plans (n = 3), and information sharing (n = 2).	ED Attendance	Case Management (12): Ten studies examining case management reported reductions in mean (-0.66 to -37) or median (-0.1 to -20) number of ED visits after 12-months; one study reported an increase in mean ED visits (+2.79); and one reported no change. Individualised Care Plans (3): Only 1 study evaluating individualized care plans examined ED utilization and found no change in median ED visits post-intervention Information Sharing (2): 1 study reported no change in mean ED visits, whereas the other reported a decrease in mean ED visits (-16.9).	Case management had the most rigorous evidence base, yielded moderate cost savings, but with variable reductions in ED use- the two RCT’s in this group yielded only minor changes in mean and median number of ED visits between the intervention and control groups, with remaining 10 comparative cohort studies (with smaller sample sizes) reporting greater reductions. Given the non-randomized nature of these studies and the small, selective study populations, there may be other uncontrolled confounding or modifying factors influencing the observed reduction in ED utilization. The limited evidence base for the remaining two frequent user intervention categories was somewhat inconclusive. Studies conducted within a higher quality RCT setting reported no change in median or mean ED visits following individualized care plans or information sharing; versus the Cohort studies which showed large decrease in mean visits. • Only examined general adult FA’s, studies looking at specific subgroups of FA’s were excluded. • Most studies conducted in the USA, limiting generalisability. • Risk of bias was consistently high across all four studies in the areas of allocation concealment, and blinding.
			Cost Utility	Case Management (12): Of these, 6 studies also reported reduced hospital costs between £477-£5302/patient. Individualised Care Plans (3): Costs following individualized care plans were also only evaluated in 1 study, which reported savings in hospital costs of £526/patient. Information Sharing (2): one of the studies did not report on cost-saving, the other reported ED cost savings of £11,006/patient.
Korczak et al. 2019 Australia	Systematic review of 16 articles conducted to assess interventions for frequent presenters to the ED and the associated costs of the intervention programs outlined. Papers were included from all countries and languages, in adult populations over 18 years of age and if the study included more than 10 subjects. All interventions were variations of a case management approach.	Systematic Review of all the relevant literature until Dec 2018 (no start date). PRISMA guidelines were used. 16 studies found, majority Cohort pre- and post- intervention study design, only 5 studies randomly assigned participants, and only 7 had a control group.	ED Attendance	All of the studies- apart from one which showed mixed results- yielded findings of decreased ED use/attendance, however this varied significantly between studies. As this was the secondary measure of the review, caution should be taken when interpreting these results. Due to heterogeneity of reporting, no statistical testing was performed within this review, on this data- narrative review only.	• Vast majority of the studies were done at single centre sites (15/16) and were mainly conducted within the USA (13/16) limiting generalisability for other countries with a different health care system. • Little homogeneity within the studies, (eg. Defining FA’s within the studies ranged from between 2 and 10 ED attendances/year) made systematic review difficult and may have reduced validity of results. Unable to extrapolate from this data. • No mention of statistical testing, intervals or power with the systematic review itself, narrative review of the literature only. • No data on actual ‘cost-effectiveness’ of these particular interventions, only cost analysis. • The majority of the studies were pre and post with the same cohort and therefore may not adequately account for secular trends or regression to the mean.
			Cost Utility	The majority of the studies yielded findings of decreased ED costs but this varied in the way it was reported (ie. cost saved per patient per month vs. by hospital per year) and amounts saved. Due to heterogeneity of results, no statistical testing was performed to evaluate an average cost saving or utility- narrative review only. It was unclear whether the costs reported in the studies incorporated the costs of the program.

Comment(s)

The proportion of patients attending ED who fit the definition of attending frequently has risen substantially. An ED visit is not always beneficial and over time becomes expensive. The most commonly accepted definition for a “Frequent Attender” is a patient who comes to the ED 5 or more times per year. Frequent Attenders make up a significant percentage of all attendances, and tend to have a higher triage category, greater rates of admission, and a greater burden of chronic disease, when compared to matched groups. The population of patients who are frequent attenders is relatively heterogenous, however those with chronic mental health problems combined with social problems and alcohol/drugs tend to make up the very high frequency patients. They can be vulnerable and may struggle to access other services. Frequent Attenders were found to have double the mortality of non-frequent attenders. The number of patients who attend frequently stays fairly static but patients come and go from this group. This makes measurement of attendances and any intervention problematic as patients tend to come to our attention whilst in crisis and then attendances drop off as the crisis resolves. Based on the literature evaluated in the present review, three main types of interventions have been evaluated: case management/MDCM; individualised ED care plans; and information sharing/referral to primary care or other services. The impact of the three types of frequent ED user interventions was variable, but modest at best. Case management had the most rigorous evidence base, yielded moderate cost savings, but with variable reductions in frequent ED use. While some of the case series included in this review have shown marked reduction in attendances after case management, the majority did not have a control group which would allow the effect to be properly quantified. Of 4 RCT’s of Case Management for frequent attenders, two showed a modest relative risk reduction but small absolute risk reduction, one showed a reduction that did not meet statistical significance and one that combined care plans which showed no difference. This is probably reflective of the complexity of issues that patients face. It may also show that a short period of case management is not sufficient to change patients’ lives and suggests patients need long periods of ongoing support. In terms of cost utility of the interventions reviewed, considering the significant costs and resources required for implementation, the present evidence suggests that none of the examined interventions are likely to yield substantial, overall cost savings for the healthcare system. The most clinically beneficial and cost-effective intervention to deter frequent ED use remains unclear given the overall variability in reported outcome and cost data. Findings from this review further indicate that prior to implementing any given intervention, thorough identification of prevalent risk factors of frequent ED use, among ED populations, must first be conducted to determine inefficiencies or gaps in the delivery of health services and the resultant appropriateness of interventions. Such personalising and tailoring of interventions and models of care, rather than standardisation of care, may prove to be most effective at reducing high ED utilisation.

Clinical Bottom Line

There is no sound, high-level evidence to suggest that any of the standardised interventions reviewed here, provided an effective- or cost-effective- reduction in ED attendance or service use by Frequent Attenders. Given the heterogeneous nature of this patient population, perhaps a more focused or personalised intervention, may prove to be most effective at reducing high ED utilisation.

References

1. Huang F, Kiberu Y, Das S. Reducing frequent attenders to the emergency department at West Suffolk Hospital. Postgraduate Medical Journal 2020;96(1132):119.3-120
2. Heelas L. A rapid review of evidence for management of patients that frequently attend emergency departments with chronic pain. Physiotherapy 2020;107:e148
3. Gerdtz M, Kapp S, Michael E, Prematunga R, Virtue E, Knott J. An evaluation of the use of management care plans for people who frequently attend the emergency department. Australasian Emergency Care. 2019;22(4):229-235.
4. Flowers A, Shade K. Evaluation of a Multidisciplinary Care Coordination Program for Frequent Users of the Emergency Department. Professional Case Management 2019;24(5):230-239.
5. Soril L, Leggett L, Lorenzetti D, Noseworthy T, Clement F. Reducing Frequent Visits to the Emergency Department: A Systematic Review of Interventions. PLOS ONE 2015;10(4):e0123660.
6. Korczak V, Shanthosh J, Jan S, Dinh M, Lung T. Costs and effects of interventions targeting frequent presenters to the emergency department: a systematic and narrative review. BMC Emergency Medicine 2019;19(1):83.