3013 — From Clinic toward Community: Care for Complex, Chronic Cases
Bauer MS (Boston VA Healthcare System), Kilbourne AM
(SMITREC / Ann Arbor VAMC), Biswas K
(Perry Point CSPCC / VA Maryland Healthcare System)
It is typical to find a “voltage drop” in impact as promising interventions move from efficacy trials to dissemination in more complex populations and settings. Recently, two multi-year effectiveness trials, studying over 700 participants from population-based or high-risk samples, have demonstrated the impact of collaborative chronic care models in improving outcome in bipolar disorder under “typical” clinical practice conditions. Despite these encouraging findings, the question remains whether the intervention had its effect primarily on the less ill and less complex participants or, critical for dissemination potential, also had impact on the more severely ill and more complex participants.
Moderators are those participant characteristics present at the time of treatment initiation that impact intervention effect. Using the Kraemer rubric (Arch Gen Psychiatry 2002; 59:877-883), we investigated moderators in outcome data from VA Cooperative Study #430, an 11-site, 3-year randomized controlled trial of collaborative chronic care for bipolar disorder vs. usual care. We focused on four characteristics that predict outcome in bipolar disorder: current substance use (SUD) or anxiety (AnxD) disorders, psychosis, and cardiovascular risk (CVD).
Among the 306 participants (mean age=46.6) complexity was common: SUD 34%, AnxD 38%, psychosis 34%, and CVD.33% No moderator effect of SUD or AnxD was demonstrated, indicating that such complex comorbid participants benefited as greatly as those without these comorbidities. Notably, psychosis was associated with an augmented effect on weeks in affective episode (beta=-1.07, p=0.04) and weeks manic episode (beta=-0.85, p=0.04). CVD reduced treatment effect on the secondary outcome of physical quality of life (beta=-6.11; p=0.04).
Effects of the collaborative chronic care model on bipolar outcome are not limited to the less ill, less complex participants, but rather extends to the type of complex patient typically found in clinical populations; in fact, effects were augmented in those who entered the trial psychotic. Future revisions of the model should incorporate more focused attention on physical risk and CVD.
Further scrutinizing clinical trials for moderator effects provides important information to: (a) delineate those populations to which the intervention might successfully be disseminated and (b) identify intervention characteristics for further improvement to enhance and broaden clinical impact.