Effects of using different rural measurements on estimates of hospitalizations for depression and substance use

Abstract

Background/purpose: To examine how the choice of rural measurements affects estimates of hospitalization rates for depression and substance use disorders (SUD). Methods: We conducted cross-sectional analyses using the 2018 State Inpatient Database (SID) for 5 states, including Arizona, Kentucky, Maryland, Washington, and Florida, to determine how (1) estimates of hospitalization rates for depression and SUDs; and (2) patient characteristics among those hospitalized differ. Five measurements of rurality including rural-urban commuting areas (RUCA) codes, core-based statistical areas (CBSA), urban-rural category four (URCategory4) and two definitions of rural urban continuum codes (RUCC) were used. For each measurement, we calculated frequencies and percentages for age, race, sex, and insurance type. We conducted Spearman's rank correlations to compare associations and internal agreement. We created an UpSet chart to visualize the overlap in different measurements. Results: There were 152,771 hospitalizations for depression and 43,760 hospitalizations for SUDs. The percentage of hospitalizations for depression or SUD differed significantly (3.2-8.1% for depression and 5.0-11.6% for SUDs ) based on rurality measure. Race and insurance characteristics of those identified as rural varied by rural measurement for depression and SUD hospitalizations. Spearman's correlations were higher for hospitalizations for SUD than for depression, ranging from r = 0.61 (RUCC and RUCA) to r = 0.99 (CBSA and URCategory4). Conclusions: Different rurality measurements result in differing estimates of hospitalizations for SUD or depression. Stakeholders should be aware that the choice of rural measurements can impact policy decisions and resource allocation for programs intended to improve care in rural areas.