Store and prescription characteristics associated with primary medication nonadherence

Primary medication nonadherence (PMN) is any instance whereby patients fail to initiate a pharmacotherapy regimen after receiving a prescription for new therapy. The Pharmacy Quality Alliance (PQA) has proposed a standardized definition for PMN and a quality measure to assess the rates of PMN in community pharmacies. To (a) measure PMN using the proposed PQA measure with data available from a pharmacy dispensing system and (b) identify the prescription-level (patient, prescriber, and medication) and store-level (store and neighborhood) characteristics associated with PMN. This study was approved by a southern university institutional review board, and a data use agreement was in place. A large national pharmacy grocery chain provided de-identified, transactional data for 2010 through January 2012, for 100 pharmacies (de-identified unique patient and store codes were available). The proposed PQA-PMN measure was used, and PMN rates were calculated. Investigators examined adult individuals with a new electronic prescription for any of the included medications during the measurement period and determined whether the medication or an appropriate alternative was claimed within 30 days. Multilevel logistic regression with a random intercept was used to evaluate prescription-level and store-level predictors of PMN. Prescription-level variables included prescriber type, PQA-defined drug class, patient gender and age, whether the prescription was accompanied by another prescription on the same day, payment source, and out-of-pocket costs. A daily average prescription volume variable was calculated for each pharmacy as a store-level variable. Additional store-level variables were derived from the 2007-2011 American Community Survey, available from the U.S. Census Bureau (median household income, educational level, percentage of minorities, and percentage aged 65 years and over in the census tracts where the pharmacies are located). Of the e-prescriptions during the 1-year measurement period, 29,238 were for new therapies as defined by the PMN measure, and 3,570 (12.2%) of those new prescriptions were not claimed within a 30-day period. There was significant variability among the pharmacies (intraclass correlation coefficient=0.140). In the adjusted multilevel model, the estimated odds of an unclaimed prescription were significantly different among drug classes comprising the PQA-PMN measure and were higher as out-of-pocket costs increased, when the prescription was