Determinants of Pharmaceutical Review, Success and Duration

This paper estimates three duration models, one for each of the three clinical phases of drug development using publicly available data. Specifically, we estimate three discrete time mixed proportional hazard models, with some parameter constraints across models. We present the estimated relationship between particular drug characteristics and the probability of succeeding (and failing) each of the clinical review phases, as well as the expected duration of review. The characteristics we control for include, among others, primary indication, route of administration, and original material as well as fixed effects for vaccines and reformulations. We are also able to identify spill-over effects from sponsorship or licensing by large pharmaceutical companies, and the effect of "competition" in the form of same-disease drugs already in the review process. Additionally, we distinguish drugs being developed in the U.S., versus those who are likely developed abroad. Calendar time fixed-effects (both annual and quarterly) are also estimated which permit us to identify whether average durations are changing independently of the changing profile of the drugs under review, in other words, whether they are changing as a result of policy for example.On balance we find that primary disease indications are usually unrelated to a drug's eventual success or failure in review, but that different original materials and different routes of administration are significant. Competition in each of the phases is a determinant factor in both success and failure. Spillover effects seem to have a small impact, and so does being sponsored by a large pharmaceutical company, and being licensed by one. A major finding of our paper is that drugs developed only in the U.S. versus in the U.S. and in other countries simultaneously, move through the review process faster, but also have a lower probability of succeeding. This model can be used to simulate the path of drugs through the pipeline, and can be used in a variety of situations, including antitrust review, intellectual property, insurance, market valuation and social policy analysis.