Approaches to Reducing Bias from Unobservable Prescription Drug Exposure in Medicare Part D Data
Using prescription drug claims data, such as Medicare Part D, may present methodological challenges for health services researchers. Researchers cannot observe medications taken by the patient while in institutional care settings because Part D event data covers prescriptions only on an outpatient basis. Past literature has described the importance of accounting for hospitalized days in such situations, but few strategies are tested and recommended for dealing with this issue. In “Accounting for Unobservable Exposure Time Bias When Using Medicare Prescription Drug Data,” published in Volume 3, Issue 4, of the Medicare & Medicaid Research Review, the authors test and recommend ways to identify and account for bias in results due to unobservable prescription drug exposure time when analyzing Medicare Part D prescription drug event data.
Cook, E. A., Schneider, K. M., Chrischilles, E., & Brooks, J. M. (2013). Accounting for Unobservable Exposure Time Bias When Using Medicare Prescription Drug Data. Medicare & Medicaid Research Review, 3(4), E1–E18.
