3097 — Measuring the Quality of Substance Use Disorder Treatment: Assessing the VA Continuity of Care Performance Measure

Humphreys K (Center for Heath Care Evaluations, VA Palo Alto HCS), Bowe T (Center for Heath Care Evaluations, VA Palo Alto HCS), Kivlahan D (CESATE, VA Puget Sound Health Care System), Finney J (Center for Heath Care Evaluations, VA Palo Alto HCS)

Objectives:
To examine the patient- and facility-level associations between meeting the Veterans Affairs (VA) Continuity of Care (CoC) substance use disorder (SUD) performance measure and improvements in alcohol and drug problems.

Methods:
With administrative and survey data from a nationally representative sample of 5,546 VA SUD inpatients and outpatients from 73 VA facilities, we compared patients who did and did not meet the COC measure on improvements (baseline to 7 months post-treatment) in alcohol and drug functioning assessed by the Addiction Severity Index (ASI) using propensity score weighted regression models. Separately, facility performance on the measure was used to predict average patient improvement in outcomes in multi-level regression models.

Results:
Patient-Level Results: Meeting the CoC measure was not associated with improvement in most alcohol measures (ASI composite, days of intoxication, or days of problems). However, CoC was associated with better odds of recent alcohol abstinence at follow-up for patients not abstinent at baseline (but not for the 30% of patients abstinent at baseline). Satisfying the CoC criterion was not associated with improvement in the ASI drug composite or days of drug problems. However, CoC was associated with better odds of drug abstinence at follow-up for patients not abstinent at baseline (but not for the 48% of patients abstinent at baseline). For patients not abstinent from both alcohol and drugs at baseline (81%), meeting the criterion was associated with increased odds of abstinence at follow-up, but not for 19% of patients who were abstinent at baseline. Facility-Level Results: Facility-level CoC was negatively associated with average improvements in ASI Alcohol and Drug Composites and not associated with average follow-up abstinence rates.

Implications:
On the patient-level, meeting the CoC criterion was related to follow-up abstinence but not to improvement on other substance use outcomes. Contrary to expectations, better performance on the VA CoC measure was negatively associated with facility-level outcomes.

Impacts:
These results emphasize the inappropriateness of assuming performance measures derived from patient-level evidence will discriminate facility-level performance in the expected direction. Translating patient-level research into quality measures requires post-construction validation, such as that provided here. Alternative performance measures should be validated such as number of visits in the first month of outpatient care.