Washington University Emergency Medicine Journal Club – August 18th, 2022
It’s another long Saturday night in EM1 when you are sent the last patient in the waiting room. You open up his chart to find a 40-year-old male with PMHx of polysubstance use, severe CHF with EF <15%, and COPD. Today, he is here for leg swelling. You note multiple ED visits and admissions in the last two months for possible CHF exacerbations. His vitals reveal that he is afebrile with BP 175/90, HR 90, RR 15, satting 92% on RA. He tells you his medications were stolen again at the shelter last week, so he needs a refill and would also like some juice and a turkey sandwich. Also, he tells you he has missed the last several appointments with his cardiologist because he is too short of breath to walk there. He was discharged with similar vitals and a stably elevated BNP last week, and his edema was described as chronic on that discharge summary. You know the medicine team could raise objections if you were to pursue admission.
We care for complex care patients in the emergency department on a daily basis. Any patient with a decompensated chronic condition with insufficient access to housing, insurance, or primary care can be classified as a complex care patient. In the ED, complex care patients often disproportionately use ED resources and are thus called “frequent utilizers” in the literature, although this term has no clear definition. These patients often have a higher prevalence of multiple comorbid chronic illness—including psychiatric illnesses—lower socioeconomic status, and higher rates of adverse health outcomes. Efforts targeting complex care patients to improve their healthcare access while decreasing avoidable utilization typically involve a multidisciplinary team of physicians and other healthcare staff, such as social workers and case managers, who can connect patients with needed resources.
This month, we will review key articles in the literature to attempt determine what interventions can be taken to effectively identify complex care patients and reduce ED utilization among patients who are heavy users.
Article 1: Finkelstein A, Zhou A, Taubman S, Doyle J. Health Care Hotspotting – A Randomized, Controlled Trial. N Engl J Med. 2020 Jan 9;382(2):152-162. doi: 10.1056/NEJMsa1906848. PMID: 31914242; PMCID: PMC7046127. Answer Key.
Article 2: Powers BW, Modarai F, Palakodeti S, Sharma M, Mehta N, Jain SH, Garg V. Impact of complex care management on spending and utilization for high-need, high-cost Medicaid patients. Am J Manag Care. 2020 Feb 1;26(2):e57-e63. doi: 10.37765/ajmc.2020.42402. Erratum in: Am J Manag Care. 2020 Mar;26(3):132. PMID: 32059101. Answer Key.
Article 3: Johnson TL, Rinehart DJ, Durfee J, Brewer D, Batal H, Blum J, Oronce CI, Melinkovich P, Gabow P. For many patients who use large amounts of health care services, the need is intense yet temporary. Health Aff (Millwood). 2015 Aug;34(8):1312-9. doi: 10.1377/hlthaff.2014.1186. PMID: 26240244. Answer Key.
Article 4: Ku BS, Fields JM, Santana A, Wasserman D, Borman L, Scott KC. The urban homeless: super-users of the emergency department. Popul Health Manag. 2014 Dec;17(6):366-71. doi: 10.1089/pop.2013.0118. PMID: 24865472. Answer Key.
Healthcare expenditure in the US has long been allocated disproportionately; in 2012, 5% of the population accounted for nearly half of all healthcare dollars spent, and 1% of the population accounted for nearly one quarter of healthcare expenditure (Cohen 2012). As a result, there has been recent focus on so-called “superutilizers” of the healthcare system who account for a disproportionately high rate of healthcare costs compared to others. Some research has sought to identify and better understand this population of super utilizers, while others have studied interventions at aimed at decreasing unnecessary healthcare spending in patients with medical, psychiatric, or social factors that put them at high risk of disproportionate spending.
A group out of Denver Health, the largest provider to people in Colorado with Medicaid or no insurance, conducted a cross-sectional, longitudinal study to describe superutilizers within their system, assess persistence of super utilizer status over time, and quantify cost trends under current care models (Johnson 2015). They defined superutilizer status as 3 or more hospitalizations over a 12-month look-back period or both a mental health diagnosis and 2 or more hospitalizations over the same time frame. They were able to identify 1682 subjects who qualified as superutilizers in the first month and 4774 over the entire 24-month study period. Clinical, demographic, and financial population-level characteristics of the adult superutilizers remained stable over the 24 months of the study. Approximately 3% of the superutilizers accounted for 30% of total charges. Individually, superutilizers were not stable over the study period, cycling in and out of superutilizer status on a monthly basis; out of 1682 subjects who met superutilizer criteria in the first month of the study, only 472 (28%) remained superutilizers at the end of the first year and just 240 (14%) remained at the end of the second year. Baseline spending for the 1682 subjects who met superutilizer criteria in the first month was high ($113,522 per capita), but fell almost 60% after two years. The instability of “superuser” status at the individual level has been previously reported, which is why it makes sense to look at those that meet criteria (which unfortunately is not uniform across studies) over an extended period of time of at least 12 months.
A second cross-sectional study conducted at Thomas Jefferson University Hospital in Philadelphia, PA sought to quantify and compare spending between homeless and non-homeless frequent users seen in the ED (Ku 2014). Frequent user status was defined as 5 or more ED visits during the study year. Out of all 61,124 ED visits during 2006, 542 frequent users were responsible for 8.9% of visits. Of these, 74 (13.7%) were identified as homeless and were responsible for 845 (15.5%) of the frequent visits. The total charges and payments for all homeless frequent users were $4,812,615 and $802,600, respectively. The median charge for homeless frequent users was lower than the median charge for non-homeless frequent users ($1478 vs. $2125) and the median payment for homeless frequent users was also lower than the median payment for non-homeless frequent users ($272 vs. $348). There was some risk of inadvertent inclusion of non-homeless patients and a higher risk of exclusion of truly homeless patients from this group (selection bias) based on criteria to define homelessness.
This study may have been improved by reporting comparisons between frequent users who were “homeless,” non-frequent users who were “homeless,” and the general ED populations. Clinical characteristics such as time of presentation, ESI score, and chief complaint should be considered in relationship to each of these groups to clarify whether these demographic characteristics are unique to the “homeless” population. . It would also be helpful to know how many “homeless” ED users were not frequent users to better determine if the 13.7% of frequent users that are homeless is indicative of this as a risk factor.
While such cross-sectional studies may help with our understanding of the clinical, demographic, and social characteristics of superutilizers, they do not necessarily demonstrate feasible ways to decrease expenditure in a way that does not worsen—and ideally improves—healthcare and patient satisfaction for these patients. The Camden Core Model, a care management intervention aimed at superutilizers in Camden, New Jersey, was evaluated in a randomized controlled trial (Finkelstein 2020). Admitted patients with at least one hospital admission to any of four Camden-area hospitals in the 6 months prior to the index admission were eligible if they had at least 2 chronic medical conditions and 2 or more of the following traits: use of 5 or more outpatient medications, difficulty accessing services, lack of social support, a coexisting mental health condition, an active drug habit, and homelessness. These patients were then randomized to usual care or the multidisciplinary Camden Core Model aimed at reducing hospital readmission. There were 800 patients enrolled, with 399 assigned to the treatment group and 401 assigned to the control group. The authors found no significant difference in the primary outcome, 180-day readmission, between the intervention group (62.3%) and the control group (61.7%); adjusted difference -0.81%, 95% CI -5.97% to 7.61%. There was also no difference in the difference in number of readmissions (-.02, 95% CI -0.29 to 0.26), hospital days (-0.59, 95% CI -2.49 to 1.31) or hospital charges ($1,654, 95% CI -$25,523 to $28,831).
While these results are disappointing, a couple of key limitations of this study may explain the lack of benefit seen from the Camden Core Model. Primarily, the study used an overly broad definition of superutilizers, including patients who were unlikely to benefit from an intensive, multidisciplinary intervention. Additionally, the primary goals of the intervention were not met in a significant number of patients (60% for the first goal, 36% for the second goal, and 28% for both goals). Use of a narrower, more appropriate patient population and improved compliance with the program’s goals may have resulted in a significant benefit from the intervention. Besides failure to meet the primary goals of the intervention in a sizable percentage of the interventional cohort, there was no difference in 2 of the 3 “benefits of participation” (receipt of temporary assistance for needy families and receipt of general assistance) and minimal difference in the 3rd (SNAP). This possible “cross-contamination” between the intervention group and the control group may also have led to an inability to reject the null hypothesis.
An additional randomized controlled trial was identified which included from patients whose primary care was through the CareMore Health Centers in Memphis, Tennessee (Powers 2020). Medicaid patients meeting at least one of the inclusion criteria (top 5% of total medical expenditures in the prior 12 months, top 5% of Chronic Illness Intensity Index (CI3) score, or care team member nomination) were randomized to complex care management or usual care. There were 160 patients randomized to usual care. Out of 93 patients randomized to complex care, only 53 consented and were enrolled; after excluding 22 patients lost to follow-up, the remaining 71 patients (including those who did not enroll in the program) were included in the intention-to-treat analysis. The complex care management program was associated with lower total medical expenditures (adjusted difference –$7732 per member, 95% CI, –$14,914 to –$550), fewer IP bed days (–0.32, 95% CI, –0.54 to –0.11), fewer IP admissions (–0.32, 95% CI, –0.54 to –0.11), and fewer specialist visits (–1.35, 95% CI, –1.98 to 0.73 ) in the 12 months following enrollment. It is also possible that additional benefit would have been seen if enrollment in the treatment group had been higher.
Complex care management, often utilizing a multidisciplinary approach to address medical and psychosocial patient needs, certainly has the potential to decrease unnecessary healthcare expenditure while improving patient care and satisfaction. Unfortunately, none of the authors of the studies included in this review were emergency physicians, whose exclusion from work so closely entwined with our daily practice is concerning. As a result, neither of the randomized controlled trials reviewed considered ED visits as an outcome of interest and none of the studies considered ED interventions as an integral component of complex care plans. As many of the frequent users seen in our ED have no primary care physician, any complex care intervention that does not include the ED will fail to reach a large number of superutilizers likely to benefit. Future research should include a focus on the ED and potential interventions for this particular group of patients.