3096 — Longitudinal Medication Adherence Instability and the Risk of Psychiatric Admission in Veterans with Schizophrenia
Zeber JE, and Stock ES, Central Texas Veterans Healthcare System; Pugh MJ, VERDICT; Tsan JY, Waco PTSD Center of Excellence; Pugh JA, VERDICT; Morissette SB, Waco PTSD Center of Excellence; Copeland LA, CTVHCS;
Poor medication adherence remains a major barrier to effective treatment for chronically ill patients. 40% of Veterans with schizophrenia are poorly compliant with their antipsychotics during any year, sharply increasing the likelihood for psychiatric admission and other adverse events. However, studies examining adherence and negative outcomes rarely analyze longitudinal medication behavior. Using an innovative methodological approach, this study seeks to construct and validate measures of adherence instability over time, determining how fluctuations influence risk of future admissions.
Veterans (n = 5,131) with schizophrenia having surgery in FY06 and taking antipsychotic medication were followed for four years. Medication Possession Ratios (MPR) were calculated annually per patient, with two measures of adherence stability created: 1) an absolute value summation of continuous MPR changes over time; and 2) variation across commonly used discrete cut-points ( < 0.5, 0.5-0.8, 0.8-1.1 and > 1.1). Using the latter measure, patients were classified as “most stable” (never switched MPR categories), “moderately stable” (1-2 switches) or “very unstable” ( > 2 switches). Logistic regression modeled the likelihood of an FY09 psychiatric admission as a function of adherence group, controlling for demographics, priority group, comorbidities, and prior hospitalization.
Mean MPRs ranged from 0.56-0.58 across four years, with 13% of patients having an FY09 admission. Regarding adherence, 42% were classified as most stable by the discrete-change measure, while 20% vacillated significantly over time. In multivariable analysis, a significant association was observed between admission risk and adherence variability, a higher likelihood for moderately stable patients (OR = 2.08); the most unstable group experienced a nearly three-fold increased hospitalization risk (OR = 2.78). Findings were robust for the continuous MPR measure.
A wide range of adherence instability was observed in these postoperative veterans with schizophrenia. As suggested by exploratory work here, regardless of the static cross-sectional compliance level, variability itself may represent an additional important dimension when monitoring adherence and subsequent outcomes. Further work is needed to refine instability measures across chronic conditions, and their contributions to timely interventions.
Recognizing the challenge of encouraging appropriate antipsychotic use, alternative approaches to monitoring longitudinal adherence behavior can take advantage of the VA’s comprehensive datasets to help identify subgroups at higher risk for adverse events.