2048. Optimizing a Medication Adherence Intervention for Schizophrenia
TJ Hudson, Center for Mental Healthcare and Outcomes Research and University of Arkansas for Medical Sciences, RR Owen, Center for Mental Healthcare and Outcomes Research and University of Arkansas for Medical Sciences, CR Thrush, Center for Mental Healthcare and Outcomes Research and University of Arkansas for Medical Sciences, X Han, Centers for Mental Health Research and University of Arkansas for Medical sciences

Objectives: Non-adherence to medication continues to be a major cause of hospital readmission and increased cost for veterans with schizophrenia.  This study uses qualitative and quantitative methods to identify strategies to optimize a medication adherence intervention.

Methods: These data were drawn from a multi-site study that compared a usual guideline implementation strategy with an enhanced strategy in which nurses worked individually with subjects to identify and overcome medication adherence barriers. The Positive and Negative Symptom Scales (PANSS), the Schizophrenia Outcomes Module, the SF-36, and the Barnes akathesia scale (BAS) were completed at baseline and at 6-month follow-up. The study nurse completed a barrier assessment at baseline and every 30 days. Using baseline data, demographic and clinical characteristics and medication adherence data,  subjects with low barriers (<2) were compared to those with high barriers (>2). Risk for non-adherence was modeled with logistic regression analysis. Qualitative interviews were completed with study nurses (n=3).

Results: Data were available from 153 veterans who were a mean age of 45; 94% were male, 75% were African American, 22% were Caucasian, and 20% were married. Using baseline measures, veterans with >2 barriers had significantly higher BAS (3.8 vs 2.7 p<.05) and PANSS total scores (89.8 vs 82.3 p<.05), were more likely to use drugs or alcohol (53.6% vs 34.8% p<.05) and were less adherent based on patient report (56.6% vs 79.7% p<.05) and on medical records (28.6% vs 49.3%) compared to patients with low barriers. Logistic regression showed that lower education level (OR=5.05, p<0.05), high barriers for subjects with PANSS scores ¡Ü 81 (OR is greater than 2.5, p<0.05), substance abuse, and higher total PANSS scores for people with < 2 barriers (OR=1.052, p<0.05) were associated with increased risk of non-adherence.  Qualitative results suggested that building trust and tools such as pill organizers, behavioral tailoring and patient education are important components of an adherence intervention.

Conclusions: These results suggest that greater symptom severity and current substance abuse are associated with having two or more barriers to adherence. Future strategies to improve medication adherence should incorporate assessment of adherence barriers.

Impact: The adherence intervention will be modified and tested in a subsequent project. The intervention components and results of the current study will be disseminated nationally.