McCarthy JF (SMITREC / Ann Arbor HSR&D COE), Post EP
(SMITREC / Ann Arbor HSR&D COE), Austin K
(SMITREC / Ann Arbor HSR&D COE), Kim HM
(Ann Arbor HSR&D COE), Yano EM
(Sepulveda HSR&D COE), Zivin K
(SMITREC / Ann Arbor HSR&D COE), Hoggatt KJ
(University of Michigan), Kilbourne AM
(SMITREC / Ann Arbor HSR&D COE), Blow FC
(SMITREC / Ann Arbor HSR&D COE), Valenstein M
(SMITREC / Ann Arbor HSR&D COE)
Veterans Affairs (VA) patients with depression have high suicide risks. In the past two years, VA has implemented national suicide prevention initiatives and RFP-funded initiatives to enhance mental health treatment in primary care. Concurrent implementation of these initiatives complicates assessment of their impact on suicide risks. We examined risks in the period FY99-04 (prior to these initiatives) among patients with depression at facilities that were early adopters (EAs) of primary care-mental health (PCMH) integration models. We compared risks at these versus other facilities and, for EA facilities, we compared risks pre- versus post-implementation.
Data from the National Registry for Depression identified all patients treated for depression between 4/1/99-9/30/04. Patients had > = 2 encounters with depression diagnoses, or one encounter plus an antidepressant medication fill. National Death Index searches identified suicide deaths. VA Clinical Practice Organizational Survey responses identified facilities that adopted PCMH integration models between FY99-04. These included the Translating Initiatives in Depression into Effective Strategies (TIDES) and Behavioral Health Laboratory (BHL) care management models and co-located collaborative care models. Using Cox proportional hazard models, we assessed suicide risks at EA sites versus other sites and, within model categories of EA facilities, in the pre- versus post-implementation phases. Covariates included age, sex, race/ethnicity, service connection, PTSD, substance abuse/dependence, inpatient psychiatric stay in the year prior to study entry, and Charlson comorbidities. Covariance sandwich estimators adjusted variance estimates for clustering within facilities.
Among all patients (N = 807,694), 1,683 suicide deaths were observed. Controlling for model covariates, patients at EA sites had lower suicide risks (hazard ratio = 0.60; 95%CI [0.45,0.82]) than those at other facilities. Among patients at TIDES facilities (N = 82,920), risks were lower post-implementation than pre-implementation (hazard ratio = 0.52; 95%CI [0.37,0.72]). Similarly, post-implementation risks were lower at BHL facilities (N = 11,537; hazard ratio = 0.38; 95%CI [0.15,0.96]) and co-located collaborative care facilities (N = 132,168; hazard ratio = 0.54; 95%CI [0.38,0.78]).
Suicide risks among patients with depression were lower at facilities with early adoption of PCMH integration models. Within EA facilities, risks were lower following program implementation.
PCMH integration program initiatives to enhance access and quality of mental health services in primary care may reduce suicide risks among patients with depression.