Methodological considerations in estimating adherence and persistence for a long-acting injectable medication.

BACKGROUND Measures of medication adherence and persistence are important for researchers and policymakers to assess quality of care. Lack of adherence has been associated with adverse outcomes and higher costs of care. Long-acting medication formulations, including injectable forms, have been proposed as interventions to increase adherence and in turn improve health outcomes and costs. Standard measures of adherence/persistence were developed for orally administered medications. Methods for assessing adherence/persistence of long-acting injectable dose forms are understudied. OBJECTIVE To compare the consistency between standard measures of adherence/persistence versus proposed variations that consider the data quality and injectable administration method for a long-acting injectable second-generation antipsychotic (SGA) using an orally administered SGA as the reference. METHODS Standard adherence/persistence measures were designed for oral tablet formulations, in particular accounting for accumulation of pills caused by early refills. To address this limitation and the accuracy of the days supply field for long-acting injectable SGAs in pharmacy claims, 2 alternatives are proposed. The first approach calculates days supply using the labeled dosing schedule for the given injectable. The second approach builds on the first and sets days supply to the minimum of the time between injections and the time frame according to the labeled dosing schedule. Administrative health care claims data from the Missouri Medicaid system were analyzed to compare adherence/persistence measures between formulations. Common adherence/persistence measures, including medication possession ratio (MPR) and proportion of days covered (PDC), were evaluated in this study. The analysis cohorts comprised 195 adult patients with schizophrenia who initiated a long-acting injectable SGA (LA-SGA) and 369 patients initiating an oral SGA (O-SGA) from August 1, 2009, through April 30, 2010. Chi-squared tests, the Kruskal-Wallis test, and Kaplan-Meier curves were used to compare adherence/persistence measures between cohorts. RESULTS Days supply was most frequently recorded as 30 days for O-SGA and 28 days for LA-SGA. Time between claim fills was most commonly 28 days for both cohorts. Using the LA-SGA pharmacy claims data, MPR was 0.91 and did not vary significantly from MPR of O-SGA (0.90; test statistic = 0.29, P = 0.590). When applying the labeled dosing schedule to compute days supply, the LA-SGA MPR rose to 0.97 and varied significantly from MPR of O-SGA (test statistic = 9.60, P = 0.002). Additionally controlling for the inability for excess medication accumulation, MPR for LA-SGA dropped to 0.86, which varied significantly from MPR of O-SGA (test statistic = 4.01, P = 0.045). PDC varied from 0.55 to 0.61 for LA-SGA but was consistently significantly different from the 0.37 PDC value of O-SGA (P  less than  0.05 for each comparison). CONCLUSIONS Standard medication adherence/persistence measures yielded different conclusions when comparing a LA-SGA and an O-SGA, depending on the measure and underlying assumption for days supply. Adherence/persistence measures that address pharmacological differences in terms of formulation and duration of therapeutic drug levels between medications may be necessary and are particularly important as more injectable antipsychotic medications are approved in the United States. Therefore, payers and investigators should consider sensitivity analysis using different adherence/persistence definitions when making product comparisons to ensure confidence in conclusions.