Determination of optimized multidisciplinary care team for maximal antiretroviral therapy adherence

Abstract

Objective: We seek to determine the optimized multidisciplinary care team (MDCT) composition for antiretroviral therapy (ART) adherence.

Methods: We analyzed all new regimen starts (n = 10,801; 7071 ART naive, 3730 ART experienced) among HIV-positive patients in Kaiser Permanente California from 1996 to 2006. We measured 12-month adherence to ART (pharmacy refill methodology) and medical center-specific patient exposure to HIV/infectious disease specialist (reference group), non-HIV primary care provider, clinical pharmacist, nurse case manager, non-nurse care coordinator, dietician, social worker/benefits coordinator, health educator, and mental health worker. We used recursive partitioning to ascertain potential MDCT compositions associated with maximal mean ART adherence. We then employed mixed linear regression with clustering by provider and medical center (adjusting for ART experience, age, gender, race/ethnicity, HIV risk, hepatitis C virus coinfection, ART regimen class, and calendar year) to test which potential MDCT combination identified had statistically significant association with ART adherence.

Results: We found maximal increase in adherence with pharmacist plus coordinator plus primary care provider combination (8.1% ART adherence difference compared with reference; 95% confidence interval: 2.7% to 13.5%). Other MDCT teams with significantly (P < 0.05) improved adherence compared with specialist only were nurse plus social worker with primary care provider (7.5%; 5.4% to 9.7%), specialist plus mental health worker (6.5%; 2.6% to 10.4%), pharmacist plus social worker plus primary care provider (5.7%; 4.1% to 7.4%), and pharmacist plus primary care provider (3.3%; 0.8% to 5.8%). Among these MDCTs, there were no significant differences in mean adherence, odds of maximal viral control, or CD4+ changes at 12 months (except pharmacist plus primary care provider).

Conclusions: Various MDCTs were associated with improved adherence, including ones that did not include the HIV specialist but included primary care plus other health professionals. These findings have application to the HIV care team design.

Figure 1. Recursive Partitioning Results

Regression tree of potential MDCT combinations that maximal ART adherence through twelve months post-regimen initiation. The tree was “pruned” to obtain a set of mutually exclusive groups of patients characterized by different MDCT components with minimized unexplained variance. For parsimony, we chose the smallest tree within one standard error of the best-performing tree.