Impact of a Student-Run Clinic on Emergency Department Utilization

Student-run clinics (SRCs) provide primary care access to patients who would otherwise pursue expensive avenues of care, such as visits to emergency departments (EDs).1,2 Decreasing inappropriate ED utilization is a major opportunity to generate health care value. However, the impact of SRCs on ED utilization is unknown.3-8 To address this gap, we studied the value created by the Crimson Care Collaborative (CCC), a Harvard Medical School-affiliated network of seven SRCs providing after-hours primary care in the Boston area. Established in 2011, the CCC clinic at Massachusetts General Hospital-Internal Medicine Associates operates from 5:30 pm to 7:30 pm on Tuesdays for 49 weeks each year. For 39 weeks, 7 to 10 patients are seen each night; for 10 weeks during the summer, the clinic operates at half capacity. We hypothesized that increased outpatient access through CCC would be associated with decreased ED utilization.

A retrospective longitudinal analysis was conducted to quantify the change in ED utilization before and after patients enrolled at CCC.

The study population included all 796 adult patients who enrolled at CCC between October 1, 2010 and September 30, 2015. Annual patient cohorts consisting of newly enrolled CCC patients were created for each fiscal year.

Patient data was obtained from the hospital’s clinical data registry.9 CCC visits were obtained from the clinic’s RedCap database. Socioeconomic status was approximated using US census data, which lists the mean per capita income associated with a patient’s home zip code. The Deyo adaptation of the Charlson Comorbidity Index (CCI) was used as a proxy for health status.10

The primary outcome—change in per patient ED utilization—was calculated using annualized ED utilization in the 18 months before and after CCC enrollment. Independent variables included gender, age, language, race, distance to clinic, marital status, insurance type, socioeconomic status, smoking status, body-mass index, CCI, and annual CCC utilization.

Descriptive analyses characterized the cohorts. Two-sided Wilcoxon signed-rank tests estimated if changes in ED utilization were significant. Regression analysis assessed the influence of the named demographic and clinical covariates on changes in ED utilization. SAS v 9.4 software was used for statistical analysis.11

Cost savings were calculated using the median ED charge of $1,233.12 The 2015 inflation adjusted charge of $1,357 was multiplied by the average number of avoided visits per year to compute annual cost savings. Avoided visits were calculated by multiplying the average annual cohort size by the observed average annual per patient reduction in ED utilization. The Partners Healthcare Institutional Review Board approved this study.

Patients were predominantly under age 65 years, white, English-speaking, commercially-insured, lived within 10 miles of CCC, and had annual per capita incomes between $15,000 and $40,000 (Table 1).

In the 2012 cohort, average per patient ED utilization increased by 80% (P<0.03) after CCC enrollment. For the three subsequent cohorts, average per patient ED utilization decreased by 23%, 50%, and 48% (P<0.001) respectively. During these 3 years, ED utilization decreased by an average 0.39 visits per year per patient (Figure 1).

Regression analysis revealed that for each additional CCC visit, ED utilization decreased by 0.41 visits per patient (P<0.0001). Although age and smoking were correlated with ED utilization (r=-0.09, 0.11; P=0.008, 0.003, respectively), they were not significant in the final model. No other characteristics were related to change in ED utilization.

Assuming a 5-year annual cohort size of 159 patients and an annual reduction of 0.39 visits per patient, 62.01 ED visits were avoided each year, amounting to $84,148 in annual savings—68% of the clinic’s direct operating costs.

In the 2013, 2014, and 2015 cohorts, the observed decrease in ED utilization following CCC enrollment suggests that patients will seek after-hours outpatient care, if available. The increase in ED utilization in the 2012 cohort is likely due to low patient engagement, fragmented staffing, and leadership turnover during the clinic’s pilot period. From 2013 onwards, clinic staffing and leadership stabilized and a sustained decrease in ED utilization was observed.

Our results are notable for three reasons: (1) quantifying the clinic’s impact on ED utilization can guide QI initiatives, (2) quantification of value provides justification for the SRC model, and (3) it is proof-of-concept that SRCs can train future health care professionals in value-based care.

QI initiatives can educate patients around appropriate ED use and incentivize longitudinal use of CCC. A key next step is understanding the drivers for ED and CCC utilization. This could drive disease-specific initiatives to shift care to appropriate settings.

Although the financial implications of reduced utilization do not directly impact CCC’s operating budget, shifting care to a lower-cost setting benefits the host hospital and can justify budget increases for CCC QI initiatives.

Our study suggests SRCs create value to payers and providers by providing alternative, more appropriate sources of care. We assume reductions in ED visits reflect shifts of nonurgent care from the ED to the outpatient clinic, freeing up ED capacity for higher-acuity cases. CCC likely creates value for patients by providing social and behavioral health services in the clinic, helping patients establish primary care relationships, and offering after-hours care.13

Medical school is the ideal time to educate young physicians about payment models, and the value of care they deliver. The nearly 100 medical schools hosting SRCs provide an ideal opportunity to apply theoretical knowledge around value-based care.14

Although we only examined ED utilization within the Partners Healthcare system, given Partners’ broad network, we believe our results account for the majority of patients’ ED usage. Because we did not classify ED visits as urgent or nonurgent, we are unable to draw conclusions about the acuity of avoided ED visits. Lastly, given that our study looks at a snapshot in time, it is possible that as time progresses and the SRC helps high-ED utilizers access primary care, cost savings may diminish.

This study is the first to examine the impact of SRCs on ED utilization. Our approach is a proof of concept that SRCs can be positioned as a platform for teaching students about value in health care delivery.