Methodology for Building Physician Target List Using Claims and EHR Data

OBJECTIVES: Target physicians for a drug/ therapy area are usually identified based on either prescription audits, sales data, or claims data. However, building a physician target list for rare disease markets is difficult because the disease may not have a specific ICD-10 diagnosis code; even if specific ICD-10 code is available, not all patients may have documented evidence of the diagnosis; patient may often be misdiagnosed before being correctly identified with the disease. The objective of study is to explore if using both claims and electronic health records (EHR) can help with developing a physician target list for a rare disease market.

METHODS: Both claims and EHR data were used. First, we used EHR data to identify the striking clinical characteristics, treatment patterns, and medical profiles of patients with a rare disease (EHR predictors). Next, we identified the disease patients for whom claims data was also available. This led to creation of a new dataset of disease patients for whom we had both claims and EHR. Next, we looked for the patients’ claims data that was analogous to EHR predictors (claims predictors). The claims predictors were used to build a predictive model for identifying all ‘potential’ patients of the disease in claims database. Each patient received a score of 0-1 depending on their probability of having the disease. An optimal probability cut-off was chosen to identify potential disease patients. Once the list of diagnosed and potential disease patients was created, it was used to identify physicians who provided services to these patients – this comprised our physician target list.

RESULTS: Physician target list generated using both claims and EHR data will be helpful to pharma clients in contacting physicians that treat rare diseases.

CONCLUSIONS: Using both claims and EHR data is helpful in building physician target list for rare diseases.