1008 — Improving Long-Term Quality of Care for Serious Mental Illness through Collaborative Care Models

Bauer MS (VA Boston Healthcare System), Biswas K (Perry Point Cooperative Studies Coordinating Center), Kilbourne AM (Ann Arbor VAMC)

Objectives:
Quality of care for chronic illness is suboptimal both in the VHA and beyond, a situation which is particularly problematic for chronic mental illnesses. Collaborative chronic care models (CCMs) improve outcome and care quality for chronic medical illnesses and for depression treated in primary care. We previously reported at this meeting 2 long-term RCTs demonstrating that CCMs for bipolar disorder, a chronic, severe mental illness, improve clinical, functional, and quality of life outcomes. However, nothing is known about whether CCMs can improve care quality for such chronic mental illnesses, and whether such gains endure over the long term.

Methods:
In Cooperative Study #430, 306 participants with bipolar disorder across 11 sites were randomized to the bipolar CCM vs. continued usual care and followed for 3 years. The CCM consisted of 3 components: patient self-management skill enhancement, provider support through simplified practice guidelines, and augmented access/continuity of care via nurse care manager. Guideline concordance with anti-manic medicine management (lithium, valproate, carbamazepine, second-generation antipsychotics) was determined for each participant in six 6-month follow-up epochs. Concordance was defined (yes/no) for each subject in each epoch by whether they (a) received any of the above medications and (b) had guideline-concordant serum levels (lithium, valproate, carbamazepine) or doses (antipsychotics) in that epoch. Measures were analyzed using GEE and an autoregressive correlation structure was used to model the correlations between concordances in successive time points. Baseline concordance, status of bipolar type, and age were covariates.

Results:
A significant treatment main effect favored CCM (OR 1.74 with 95% CI: 1.05 – 2.88; p = 0.047), without time or time-treatment interaction effects; there was no difference between bipolar type I vs. II. Among those treated with any anti-manic medication in a given epoch, the main effect was similar (OR 1.83 with 95% CI: 1.10 – 3.05, p = 0.033).

Implications:
The bipolar CCM improved guideline-concordant care compared to usual care, effects which began promptly, extended across bipolar types, and endured over 3 years.

Impacts:
CCMs support guideline concordance by addressing patient, system, and provider factors. Conceptualizing guideline concordance as multi-factorial, rather than purely provider-driven, may lead to more effective interventions, and merit further exploration.