Inpatient Reliance Ratio Versus Emergency Department Reliance Ratio in a Sickle Cell Disease Medical Home

Background: The Emergency Department (ED) Reliance Ratio (ERR, defined as ED visits/ED + ambulatory + inpatient visits) measures the percentage of all health care visits that occur in the ED in relation to primary care, including avoidable ED visits, and may therefore discriminate among high-ED-user populations. Adults with sickle cell disease (SCD) are known to have high ERR (by convention, >0.33), likely driven by a complex mix of factors. Like in other ambulatory-care-sensitive conditions, intense SCD ambulatory case management may drive down the ERR. But we found no validated measure of hospital utilization percentage analogous to the ERR. We therefore created the inpatient reliance ratio (IRR, defined as inpatient visits/ED + ambulatory + inpatient visits), as the percentage of all health care visits that occur in the hospital in relation to primary care and ED care, including avoidable hospital visits. We tested the performance of the ERR versus the IRR in an adult SCD sample.

Method: In 2018, as part of a multi-component intervention in an SCD Adult Medical Home, we assigned Patient Navigators (PNs), as case managers to each of the program's top 50 highest utilizers, but also began periodically managing the remaining medical home patients ad hoc. Services provided included behavioral health services, dedicated inpatient care, an infusion unit, and inpatient and ED individualized care plans. The study sample was Adult SCD patients at Virginia Commonwealth University from 2011-2021 (annual N's from 299 to 641). Using Analysis of Variance, we studied the effect of the intervention over time on total annual charges, the annual ERR, and the annual IRR. We also performed Spearman's correlations associating annual total charges with either annual ERR or annual IRR. Last, we performed linear regression with annual total charges as the outcome variable (linearly distributed), to test the relative predictive contribution of ERR vs. IRR.

Results: (Figure 1) After increasing steadily from 2011-2017 pre-program onset, both the annual ERR and the annual IRR decreased roughly successively from 2017-21 (ANOVA p-value=0.0089 for ERR, ANOVA p-value=0.0002 for IRR). Annual total charges correlated strongly with ERR (Spearman R=0.44, P<0.0001) but more strongly with annual IRR (Spearman R=0.64, P-value<0.0001). Similarly, linear regression indicated that IRR contributed more strongly to annual total charges than did ERR (beta coefficients = 97906 vs 23175).

Conclusion: We demonstrate that the IRR, similar to the ERR, appears to be a measure of avoidable care: It varies as a function of total costs, but it is sensitive to intense ambulatory case management. It more closely correlates with total costs and more closely predicts total costs than the ERR. The benchmark for what makes a high or low IRR has not been defined as it has for ERR. But future studies could further show its relationship to utilization management and get population norms to set an appropriate management goal for the IRR.

Smith:Imara: Research Funding; Pfizer: Consultancy, Research Funding; Novo Nordisk: Other: DSMB; Novartis: Consultancy, Honoraria; Global Blood Therapeutics: Consultancy, Honoraria, Research Funding, Speakers Bureau; Emmaus: Consultancy; Forma Therapeutics: Consultancy, Research Funding; Agios: Research Funding.