• 
• 

Three Part Question

IN [a population of UK emergency department frequent attenders] HOW [effective is case management] AT [reducing the number of ED attendances]?

Clinical Scenario

A 42-year old woman presents to the emergency department (ED) with an intentional paracetamol overdose. She is a known frequent ED attender due to paracetamol or insulin overdoses, but would often refuse the appropriate treatment. She has emotionally unstable personality disorder, with high levels of anxiety and distress when she comes into the department. You wonder if implementation of case management can reduce her number of ED visits due to overdoses.

Search Strategy

Evidence was searched using Medline and Embase via OVID interface.
MEDLINE (R) 1946 – April week 3 2016
1. Emergency service, hospital/ or emergency medical services/ or emergencies/ or emergency medicine/ or evidence-based emergency medicine/
2. [emergency or “emergency department*” or ED or “emergency service*” or healthcare or “health care” or “health service*” or clinic*].ti.
3. 1 or 2
4. [overutilis* or overutiliz* or overuse* or over-use* or super-use* or frequent use*].tw.
5. [frequen* or heavy or repeated* or chronic or over or high].ti.
6. [attend* or use* or visit* or utilis* or utiliz* or present* or flyer*].ti.
7. 5 and 6
8. [overutili* adj10 [emergency or emergency department* or ED or emergency service* or healthcare or health care or health service* or clinic*]].tw.
9. [[[frequent adj3 user*] or frequent ED or frequent emergency or super-user* or over-user*or overuser*] adj10 [emergency or emergency department* or ED or emergency service* or healthcare or health care or health service* or clinic*]].tw.
10. Health services misuse/
11. Utilization review/
12. 4 or 7 or 8 or 9 or 10 or 11
13. [intervention* or method* or strateg* or program* or “case management” or “care management” or “care plan*”].ti. or case management/ or patient care planning/ or patient care management/ or patient care bundles/
14. 3 and 12 and 13
15. Limit 14 to English Language
16. Limit 15 to 1996-present

Search strategies were adapted and modified from Soril LJJ et al. PLoS One 2015;10(4): 1-18.

Search Outcome

331 papers were found in MEDLINE and 265 in EMBASE according to the search strategy. We focused our subsequent search on a general adult frequent ED attender population with frequent attendances defined as those with ≥ 4 ED visits in 12 months. 20 relevant papers were identified, 3 of which were systematic reviews and the remainder were primary studies. 1 additional relevant primary study was found through hand-searching of the systematic reviews. The systematic reviews were not included in the analysis. 3 primary studies directly answered the three-part question, but along with the majority of the primary studies had small patient numbers and used pre-post intervention studies without control groups. As the results were likely to be subjected to regression to the mean, it was decided to exclude them. Here, we present the results of 4 randomised controlled trials and 2 pre-post intervention studies.

Relevant Paper(s)

Author, date and country	Patient group	Study type (level of evidence)	Outcomes	Key results	Study Weaknesses
Bodenmann et al. 2014 Switzerland	250 patients with ≥5 ED visits in 12 months. Randomised to 125 in CM group (usual ED care + CM) and 125 in control group (usual ED care). Patients identified through ED tracking system.	RCT, 12-month follow-up (CM team delivering an assertive clinical CM model that involves a continuum of hospital- and community-based services)	ED attendances	This study reported a standardised regression coefficient of attendances: b. Both control and CM groups showed a significant reduction in the number of ED visits (b=-0.56, p<0.001), but the decrease was significantly more important for the intervention group (b=-0.025, p=0.012).	Selection bias: informed consent was obtained, 84 declined to participate.
Reinius et al. 2013 Sweden	268 patients with ≥3 ED visits in 6 months. Randomisation of 211 to CM group and 57 to control group; only 106 subsequently agreed to intervention. Patients extracted from electronic records according to attendances, followed by exclusion of certain populations, e.g. dementia, end-of-life and severe psychiatric disease.	Zelen’s design RCT, 12-month follow-up (Assignment of a CM nurse who designs care plans and makes intense weekly/biweekly phone contact with patients)	ED attendances	Moderate relative risk reduction in total number of ED visits (RR 0.77; 95% CI 0.69-0.86).	No absolute values quoted in results, only presented in events/person years. The study adopted the Zelen’s design whereby patients in the control group were not informed of the participation in the study. This raised problems with nonparticipation (significant when only 106 out of 211 in the participant group consented to intervention) especially when intention-to-treat principle was applied during analysis. It risks underestimating efficacy of the intervention and decreases overall validity. Selection bias: manual selection of patients who would more likely benefit from the intervention, requirement for informed consent and exclusion of certain populations. No attempt made at blinding, especially during data collection.
			Costs	Total costs per patient were 45% (p=0.004) lower in the intervention group. ED costs per patient were also 49% (p=0.09) lower.
			Outpatients	Moderate relative risk reduction in total number of outpatient visits (RR 0.80; 95% CI 0.75-0.84).
			Inpatient admissions	No significant decrease relative risk in the total number or ED hospital admissions. Total number of inpatient days lower in CM group due to shorter hospital stay.
			Health status self-assessment	Significant increases (p<0.05) in 6 out of 8 parameters including general health perception, emotional role functioning, physical role functioning, bodily pain, social role functioning and vitality.
Shumway et al. 2008 United States	252 patients divided into 2 groups: 5-11 ED visits and ≥12 ED visits in 12 months. Randomisation of 167 to CM group and 84 to control (usual ED care). Patients chosen by ED research assistants and case management supervisor; must have psychosocial problems addressable by CM to be eligible.	RCT, 24-month follow-up (Assignment of a psychiatric social worker who implements CM including crisis intervention, linkage to medical care providers and ongoing assertive community outreach to maintain continuity of care)	ED attendances	Statistically significant (p<0.01) difference in the mean number of ED visits between CM group (0.9 ± 1.5) and control (2.0 ± 3.4). Intervention is similarly effective for all levels of prior ED use (5-11 and ≥12).	No details of how randomisation was achieved. No mention of the change in the mean number of ED visits, both groups have different starting numbers. Selection bias: patients selected at discretion of ED staff and only between 8am – 5pm. Unavailability of case managers also means that some eligible patients would not be included. Potential for reporter bias in reporting psychosocial problems. No attempt at blinding, especially during data collection.
			Costs	Significantly lower ED costs (p<0.01) in CM group (247 ± 455) compared to controls (647 ± 1089). No difference in overall hospital costs. So cost-effective, but not cost-saving.
			Outpatients	No significant difference in number of outpatient visits between CM and control group.
			Inpatient admissions	Lower (p=0.08) mean number of hospital admissions in CM group, but no difference in use of other healthcare services.
			Psychosocial factors	Statistically significant reduction in levels of homelessness, problem alcohol use, lack of health insurance and lack of social security income in CM group (p<0.05).
Spillane et al. 1997 United States	70 patients with >10 ED visits in 12 months. Randomisation of 33 to intervention group and 37 to control group; only 25 and 27 patients respectively were evaluated. Patients extracted from electronic records according to attendances.	RCT, 12-month follow-up (Individual ED care plans on hospital system to guide appropriate medical and social management and CM to coordinate hospital- and community-based care)	ED attendances	No significant difference in the median number of ED visits between intervention and control groups (7 and 6 respectively). Median number of visits decreased by 7 in both groups.	Small sample size. Did not apply intention-to-treat principle. No information (e.g. timing within the study period) about patient deaths or attempt to correct for it during data analysis. Missing records, only 62% of individual ED visits were available for review. No mention of blinding especially of data analysers.
Murphy and Neven 2014 United States	144 patients divided into frequent users: 3-11 ED visits in 12 months, and extreme users: ≥12 ED visits in 12 months. Patients enrolled through ED staff referrals and Medicaid care plans.	Pre-post intervention study, 12-month follow-up (Individual ED care plans on hospital system to guide appropriate medical and social management which is shared with all EDs in the region and coordinating hospital- and community-based care, especially primary care)	ED attendances	Extreme group: median number of ED visits decreased by 79%. Frequent group: median number of ED visits decreased by 71%.	No control group or randomisation, but used statistical analysis to correct for regression to the mean. No mention of patient number in each group. Selection bias: patients were referred at the discretion of ED staff and Medicaid care plans. Preference of enrolment was also given to managed-care Medicaid patients. No mention of proportion of ED referrals to Medicaid patients or whether there were any overlaps. Not a concurrent study, patients were entered into the programme at different times within a 3-year period, introducing confounding. Unclear as to whether data was collected from regional EDs.
			Costs	Extreme group: 76% decrease in hospital treatment costs. Frequent group: 55% decrease in hospital treatment costs. Overall total direct costs were lower post-intervention, which also yielded a net income.
Shah et al. 2011 United States	258 patients with ≥4 ED visits/≥3 inpatient admissions/≥2 admissions and 1 ED visit in 12 months. 98 received CM; 160 age- and ethnicity-matched controls Patients identified through monthly report based on attendances; must have low-income and be uninsured.	Pre-post intervention study, 18-month period (Assignment of a care manager who worked intensely with patients helping them to navigate primary care services, social care and third sector providers)	ED attendances	CM lowers the risk of ED visits by 32% when compared to the control group (Poisson regression reported to adjust for the covariates between groups). Unadjusted comparisons showed no difference between groups (p=0.8).	Retrospective design. No randomisation. Both groups were not matched for health status whereby the Charlson comorbidity index is significantly higher in the CM group. Selection bias: due to the nature of the program (i.e. a Medicaid demonstration project), biased towards the lower socioeconomic group. There was also selection due to informed consent.
			Costs	26% reduction in ED costs and 65% reduction in inpatient admissions cost post-CM. None calculated for control group.
			Inpatient admissions	No significant difference in risk of additional admissions in intervention group compared to controls (Poisson regression). Unadjusted comparisons showed no difference (p=0.7).

Comment(s)

Frequent attenders are a small subpopulation of emergency department (ED) users who present to the department at a disproportionately high frequency. These patients often have complex medical and psychosocial needs, and utilise emergency services either for non-emergent care or upon reaching a crisis. Whilst EDs are able to manage these problems acutely, the department is only able to provide episodic care on each independent visit, which on multiple visits becomes a cost burden for the health service and at the same time, doesn’t address the underlying cause. Case management is recognised as the optimal method of managing these patients. It is a holistic model that provides continuity of care extending from hospital to the community, which ultimately aims to improve their overall social wellbeing and in turn, reduce the number of ED attendances. Most studies in this field are pre-post intervention cohort studies without a control group. Shumway et al., Spillane et al. and Shah et al. all indicated a significant difference between ED attendances at the start and end periods in both control and intervention groups, showing that if there is no intervention, people’s crises diminish and that regression to the mean comes into play. Hence, if not taken into account, it risks wrongly attributing natural reduction as positive outcomes. Overall, the results indicate that case management produces a reduction in the number of ED attendances. However, only very minimal reduction was seen when compared to controls and one RCT showed no difference between the intervention and control groups. One other study similarly also included individual ED care plans, although it is unclear whether these provided any additional benefit. ED attendances also decreased in studies which only managed the psychosocial factors with no medical advice. Many of the studies evaluated other parameters including use of other healthcare services, cost-effectiveness and clinical and social outcomes, which generally yielded positive outcomes. These findings suggest that case management can improve the general well-being of the patient as well as improve the cost-effectiveness of healthcare services. Further support for the intervention can be obtained if it can be shown to also decrease societal costs.

Clinical Bottom Line

Case management is effective at achieving a modest reduction in the number of frequent attender ED visits, although there was variability in the amount of reduction that can be achieved. Furthermore, all of the analysed studies were conducted outside of the UK and there was heterogeneity in patient population. Most studies also provided intensive case management. It remains to be proven whether a similar approach will be effective and cost saving in the NHS.

References

1. Bodenmann P, Velonaki V-S, Baggio S, Iglesias K, Moschetti K, Ruggeri O, et al. Frequent users of the emergency department in a universal health coverage system: A randomized controlled trial of a case-management intervention.
2. Reinius P, Johansson M, Fjellner A, Werr J, Ohlén G, Edgren G. A telephone-based case-management intervention reduces healthcare utilization for frequent emergency department visitors. Eur J Emerg Med. 2013 Oct; 20(5):327–34
3. Shumway M, Boccellari A, O’Brien K, Okin RL. Cost-effectiveness of clinical case management for ED frequent users: results of a randomized trial. Am J Emerg Med. 2008 Feb; 26(2):155–64.
4. Spillane LL, Lumb EW, Cobaugh DJ, Wilcox SR, Clark JS, Schneider SM. Frequent users of the emergency department: can we intervene? Acad Emerg Med. 1997; 4(6):574–80
5. Murphy SM, Neven D. Cost-Effective: Emergency Department Care Coordination with A Regional Hospital Information System. J Emerg Med. 2014; 47(November 2013):1–9
6. Shah R, Chen C, Rourke SO, Lee M, Mohanty SA, Abraham J. Evaluation of Care Management for the Uninsured. Med Care. 2011; 49(2):166–71