Concordance of Medicaid and pharmacy record data in inner-city children with asthma

Background: Management of asthma involves adherence to medication regimens. Assessing adherence is difficult for health care providers and researchers. Self-reported medication use is subjective, so objective methods of data collection for medication use are frequently used in asthma research. The aim of this project is to examine the concordance between asthma medication pharmacy data culled from Medicaid claims data ("Medicaid pharmacy data") and patient pharmacy record data obtained from individual pharmacies ("pharmacy record data"). Methods: Medicaid pharmacy data and pharmacy record data were obtained from inner-city children enrolled in a prospective study of children with persistent asthma. A subject level comparison of pharmacy records and Medicaid pharmacy data pharmacy records was done to determine concordance between the 2 data collection methods. Results: Of 513 children recruited for inclusion, 221 were consented and randomized. Medicaid claims data were collected on 72.8% (n = 161) of the 221 enrolled subjects. Pharmacy record data were available on 96.8% (n = 214) of the 221 subjects. Data presented represent the 159 subjects who had both Medicaid claims data and pharmacy records data available throughout the study period. There was complete agreement between Medicaid pharmacy data and pharmacy record for 26% (n = 42) of subjects. A total of 1858 asthma medication claims were captured by the Medicaid pharmacy data. Medicaid pharmacy data missed 149 claims that were capture by the pharmacy record data. Medicaid pharmacy data failed to capture a single claim on 4.4% (n = 7) of subjects. The pharmacy record data captured a total of 1627 asthma medication claims and missed 371 claims that were captured by the Medicaid pharmacy data. Pharmacy record data failed to capture a single claim in 1.9% (n = 3) of subjects. Conclusions: There was overlap between the pharmacy data captured by the Medicaid pharmacy dataset and pharmacy record dataset, but the overall concordance between the two data collection methods was low. Pharmacy records collected directly from the pharmacy included data on more subjects and pharmacy data culled from Medicaid claims captured more total number of claims. In spite of the differences in the methods used to collect data, pharmacy fill records are a rich source of data with both clinical and research applications. Clinicians and researchers must weigh the benefits and limitations of each method used to collect pharmacy data.