sanofi-aventis Managed Care Digest Series®
Local Trends Summary™ 2009
Data Collection
Through agreements with a variety of data providers, the SDI data warehouse receives between 1.5 billion and 2 billion prescription transactions and 425 million medical claims (both provider and hospital) annually—a total of 8 billion prescription claims representing nearly 200 million unique patients. Our prescription data includes samples from nearly 59,000 pharmacies (a near census) in the U.S. and includes cash, Medicaid and commercial transactions. SDI’s warehouse contains data as reported on NCPDP (prescription), CMS 1500 (physician office claims) and UB-92 (hospital claims) forms. All activity is linked and integrated by SDI’s unique identifier. CMS 1500 and UB-92 claims contain ICD-9 (diagnosis) and CPT-4 (procedure) activity that can be linked across prescription activity, and on a cross section of patients. Other valuable medical metrics include date of service, discharge dates, and location of care (office, ER, etc.).
Prescription Projections
SDI receives prescription information from virtually every retail pharmacy in the United States. SDI receives all prescriptions from about one-third of the stores and an average of approximately one-third of prescriptions for the remaining stores. Each store has a prescription sequence numbering system that is used by SDI to count the missing prescriptions in stores where we receive a sample of their data. This allows us to track the size of every store in the U.S. Projections are stratified by payment type, class of trade and geography. In the first phase of projection SDI uses 814 projection territories aligned to balance coverage proportional to prescribing activity. These territories are stratified by payment type and the sample is expanded to the universe, by strata, in each territory. In the second phase of projection SDI stratifies the country by census division and class of trade. There are four store types that define the classes of trade: Pharmacy Chains, Mass Merchandisers with Pharmacies, Food Stores with Pharmacies, and Independent Pharmacies. Each of the nine census divisions is split into four store types. The sample in each of these thirty-six strata is expanded to the universe. In the third phase of the projection, SDI reconciles the results of the first two phases to produce a single projection factor for each claim. This factor is used to project prescriptions and patients filling a prescription. The projection factors calculated above are used to project prescriptions and patients for each prescriber in all sales territories. Regional and national projections are created by rolling up from the prescriber to the geographic area of interest.
Medical Data Projections
Monthly medical projection factors are calculated by specialty at the national level. Physician specialties are classified into 17 related specialty groups. Each month, the number of office-based physicians in the U.S. and the number of physicians in SDI’s medical database are estimated by specialty. Projection factor is the inverse of the specialty coverage rate (size of the physician universe/size of the sample). For specialties with large projection factor variance over time (specialties of low coverage rate or small size), specialty group projection factors are used to stabilize the projection. Specialty projection factors are applied to claims according to date and physician specialty.
|
