Several CRHC faculty have expertise in conducting health insurance claims analyses and apply innovative methods like latent class analysis to study the dual use of VA and non-VA care. Dr. Radomski, for example, is developing, validating, and applying a patient-centered metric of low-value prescribing (LVP) using administrative data in order to measure and reduce low-value care. Dr. Gellad is applying novel machine learning approaches to administrative claims data to predict who is at risk of problematic prescription opioid use and overdose.
Dr. Gellad’s research focuses on physician prescribing practices and policy issues affecting access and adherence to medications for patients. Two of his current projects use novel machine learning approaches to determine risk of adverse events among patients who are prescribed opioids. Data for these studies include Medicare, Pennsylvania Medicaid, and Allegheny County service use claims.
Dr. Radomski conducts research on health care delivery, financing, and utilization. He has adapted a medication-based risk adjustment method in order to overcome limitations of claims-based risk adjustment using ICD-9 codes. His current work involves studying the prevalence and determinants of low-value test and procedure use among Veterans with a focus on the dual use of VA and Medicare services.