An Empirical Basis for Standardizing Adherence Measures Derived From Administrative Claims Data Among Diabetic Patients

Abstract

To compare the predictive validity of 8 different adherence measures by studying the variability explained between each measure and 2 outcome measures: hospitalization episodes and total nonpharmacy cost among Medicaid eligible persons diagnosed with diabetes. This study was a retrospective analysis of the Arkansas Medicaid administrative claims data from January 2000 to December 2006. Diabetic (ICD-9-CM = 250.0 x - 250.9 x, where x = 0 or 2) patients were identified in the recruitment period July 2000 through April 2004. Patients had to be >or=18 years old and have at least 2 prescription fills in the index period for an oral antidiabetic drug. : Adherence rates to oral antidiabetic therapy were contrasted using the following 8 measures; including the medication possession ratio (MPR), proportion of days covered (PDC), refill compliance rate (RCR), compliance ratio (CR), medication possession ratio, modified (MPRm), continuous measure of medication gaps (CMG), and continuous multiple interval measure of oversupply (CMOS and continuous, single interval measure of medication acquisition (CSA). Multivariate and univariate linear and logistic regression models were used to prospectively predict nonpharmacy costs and hospitalizations in the follow-up year. A total of 4943 diabetic patients were studied. In predicting any cause hospitalization, univariate models with PDC and CMG had the highest predictive validity (C-statistic: 0.544). Multivariate models with MPR, PDC, CMG or continuous multiple interval measure of oversupply (CMOS) as adherence measures had the highest C-statistics of 0.701 in predicting diabetes specific hospitalizations. None of the adherence measures were significantly associated with nonpharmacy cost. MPR and PDC had the highest predictive validity for hospitalization episodes. These 2 measures should be considered first when selecting among adherence measures when using administrative prescription claims data.