We use a number of different data sources to collect four types of information: drug prices, regimen market shares, typical drug dosage amounts for each regimen, and regimen attributes. IMS Health collects information on the sales in dollars and the quantity of drugs purchased by 10 different types of customers (e.g., hospitals, physician offices, retail pharmacies) from wholesalers in each quarter from 1993 through the third quarter of 2005. Prices and quantities are reported separately by National Drug Classification (NDC) code, which are unique for each firm-product-strength/dosage-package size. We calculate the average price paid per milligram of active ingredient of a drug by averaging across the different NDC codes for that drug. IMS Health reports the invoice price a customer actually pays to a wholesaler, not the average wholesale price (AWP) that is set by a manufacturer and often differs substantially from the true transaction price.
The price we calculate does not include any discounts or rebates a customer may receive from a manufacturer after purchasing the product from the whole- saler. Based on interviews with a few oncologists, we do not believe that man- ufacturers offered substantial rebates during this period. Although we have in- formation on 10 different types of customers, we focus on the prices paid by the two largest customers - hospitals and physician offices. Because most oncology drugs are infused in a physician’s office or hospital clinic, nursing homes and retail pharmacies purchase relatively little.
Most colon cancer patients are treated with regimens that combine two or more drugs. The IMS Health data contain information on market share by drug, but not market share for the combinations of drugs (regimens) actually used on patients. We rely, therefore, on two different sources for regimen-specific market shares. IntrinsiQ is a company that provides information systems to oncologists to help them determine the proper chemotherapy dosing for their cancer patients. As a result, IntrinsiQ collects monthly data from its oncology clients on the types of