2007 HSR&D National Meeting Abstract
3012 — Evaluating Implementation of Best Practices for Depression Care: Impacts on Process and Outcomes
Chaney EF (COE-Seattle) , Rubenstein LR
(COE-Sepulveda), Yano E
(COE-Sepulveda), Liu C
(COE-Seattle), Felker B
Over twenty randomized trials show that depression treatment by primary care providers in collaboration with care managers and mental health specialists is effective and cost-effective. However, few practices have adopted effective care models. We hypothesized that the demands of implementing collaborative care models required the assistance of a research clinical partnership, and we tested this by conducting a site-randomized controlled trial of the effectiveness of a VA-adapted model in seven VA community based outpatient clinics (practices) in three VISNs.
We assigned ten matched practices to implementation or usual care. In implementation practices, we used quality improvement methods to engage leadership in adapting evidence-based depression care models. Clinicians began referring patients to care managers six months prior to experimental evaluation. For evaluation, we screened 23,000 primary care patients scheduled for upcoming primary care visits for depression using the PHQ-9. We excluded patients already referred to care management. We enrolled 761 veterans with probable major depression in a computer-assisted telephone interview, and referred eligible, consenting patients at implementation clinics to care managers. We assessed patient outcomes at seven months (72% completion) based on depression symptom severity (PHQ-9) and administrative data on antidepressant use. We measured referring clinician experience with care management based on care manager logs.
The implementation was successful, fully employing care management resources. Antidepressant use significantly increased in implementation compared to usual care practices. At seven months, depression symptom severity improved overall.
The implemented intervention changed the process of care (antidepressant use), but not depression outcomes at seven months. Prior implementation of voluntary clinician referral to depression care managers as part of quality improvement may have delayed care manager uptake of study patients such that expected outcome impacts exceeded our measurement window, and may have resulted in adverse selection such that patients of clinicians who were unlikely to refer to care managers on their own were more likely to be admitted to our sample.
The adaptation process necessary to implement research into routine practice maximized external validity but proved difficult to evaluate using an RCT design. Our results can be used to improve research strategies of quality improvement interventions.