Finding the optimal length of stay for veterans in substance use disorder residential treatment using generalized propensity score modeling

Substance use disorder (SUD) residential treatment provides critical services to patients with complex clinical needs. The Department of Veterans Affairs (VA) has over 70 SUD residential programs to meet the needs of veterans with severe SUD. Prior research is mixed on what SUD residential length of stay (LOS) duration maximizes average treatment outcomes, with some studies advocating for around 30 days and others advocating for at least 90 days. Much of this research is limited by methodological issues and fails to consider both outcomes for established patients and timely access for prospective patients. The present study sought to clarify the relative benefits of different LOSs to be considered by SUD residential program decision-makers as default LOSs, upon which to be personalized by individual patient needs and clinical assessment. Using medical record data associated with N = 15,889 veterans discharging from a VA SUD residential center between 10/1/2021 and 9/30/ 2022, we identified an optimal LOS range balancing average treatment outcomes and ensuring new patient access/minimizing established patient diminished returns from treatment. Using a generalized propensityweighted dose-response curve, we identified key LOSs between 35 and 49 days with different tradeoffs between enhancing outcomes and access.