Multi-Level Predictors of Discharges Against Medical Advice: Identifying Contributors to Variation Using an All-Payer Database

Abstract

There is increasing evidence of the role of non–patient-level factors on discharge against medical advice (DAMA), but limited quantitative information regarding the extent of their impact. This study quantifies the contribution of discharge-level and hospital-level factors to the variation in DAMA. We grouped variables from the 2014 National Inpatient Sample data and ran incremental mixed-effects logit models with grouping at the level of the discharge, the hospital, and the census region. We obtained the intraclass correlation coefficients (ICCs), and evaluated the incremental change in ICC. The final sample included 2,687,430 discharges. 12.8% of the identified variation in the probability of DAMA was associated with the hospital, and 1.2% of the variation was associated with the census division in which the hospital was located. The final, fully-adjusted model had 7.3% of variation in DAMA associated with the hospital-level, with the greatest percentage reductions because of the addition of patient demographics. Even after adjusting for measured patient-level characteristics, there was a contribution of non–patient-level factors to DAMA outcomes. The findings identify a role for a multi-level approach to addressing DAMA.