Abstract
A study was undertaken to discover potential markets for U.S. pharmaceutical firms by predicting potential market opportunities of new drug entities. The study was composed of two phases. This article, as the first of a two-part series, describes regression model-building for demand prediction of a drug entity. The second part will discuss the exploratory use of the regression model to predict potential demand of a new drug entity by means of quasisimulation. Regression models were constructed on the basis of actual use of existing generic drugs in U.S. physicians' offices, using data from the 1980 National Ambulatory Medical Survey Micro-Data Tapes. Model building involved the drug market share, the age of a drug, clinical superiority and use in combination. The best model explained 59% of the variance in the drug market share. Emphasis was placed on discussions of the methodological issues of the study.