“…Using the Andersen/Davidson model of service utilization, which theorizes that both individual and contextual factors impact service use, we selected covariates at both the client and agency level that might facilitate or hinder accessing services after discharge. Our choice of covariates also was based on prior research on continuity . For residential treatment analyses, we controlled for client's age, gender, race/ethnicity, education, marital status, employment status, past month use of substances (alcohol, marijuana, cocaine, opioids, methamphetamines, and other drugs), age at first use of any substance, and homelessness.…”