Evaluating and identifying changes in the rate of unplanned discharge among Department of Veterans Affairs (VA) facilities providing substance use disorder residential care

Introduction

Quality improvement (QI) methods play a critical role in ensuring that patients receive high-quality and timely care. Healthcare systems should use valid and reliable measures to inform QI efforts. Mental health settings including substance use disorder (SUD) residential programs have been slow to develop and incorporate quality measurement into routine practice. Unplanned discharge is of particular concern because this event is associated with harm including suicide. Healthcare systems require criteria that they can use to operationalize unplanned discharge as a quality measure in SUD residential programs.

Methods

The study included all discharges from the Department of Veterans Affairs (VA) residential SUD programs between 2018 and 2022. The study calculated crude and adjusted rates of irregular discharge. The study used the first two years of observation (2018–2019) in a logistic regression model to determine the parameter estimates for three important covariates, age, risk for homelessness, and principal diagnosis. The study tested permutations of bin size (N) and days (D) per bin to identify a single set of parameters to enable small and large facilities to have sufficient power to detect out-of-control processes (i.e., significant worsening or improvement in rates). Aligned with standard nomenclature, the study calculated the control limits based on three standard deviations (SD). Values that fell above or below three SD were statistically significant.

Results

The cohort included 56 facilities (26,361 discharges). Irregular discharge was associated with younger age (18–40 years) and a principal diagnosis of a drug use disorder. Testing parameter values of 100 discharges (N) over 120 days (D) would yield enough power to detect modest relative changes to the irregular discharge rate for small and large facilities while testing frequently enough to make the evaluations temporally relevant. Because secular trends such as staff changes over time will impact results, the quality control method should allow for real-time feedback to those most proximal to the event.

Conclusions

The study created a set of parameters and a methodology that residential SUD programs can use to operationalize unplanned discharge locally. These data could assist programs in conducting QI work to address unplanned discharge and related harms.