Dr. Glass described various research studies involving clinical trials, drug safety, and placebo data that produced very interesting findings. One of his studies looked at the prescribing pattern of physicians who had functioned as principle investigators for compounds (new drugs) versus those who had not participated as an investigator in Phase 3 clinical trials. His study found that investigators “sustain a significantly higher product market share for the tested drug” than non-investigators at 3, 6, and 18 months after the drug’s release. To more critically analyze these results, Glass and his associates looked at the prescribing patterns for four “problem” drugs (three of which were eventually pulled from the market). The same pattern (investigators with a higher market share) emerged, even though investigators tend to be more educated and experienced physicians.  The reasoning is thought to be that investigators tend to see the positive effects of the drug rather than the negative and small sample sizes (as many clinical trials have) may not generate enough adverse drug events to be considered significant. Therefore, the investigators continue to prescribe the drug. 

After a drug reaches the market and adverse events occur, it is difficult to determine whether the drug caused the problem or if the occurrence is within a normal population incidence rate. This is because we do not have accurate data to estimate the natural occurrence of certain events in the population. Glass’ idea is to pool together the placebo group data from clinical trials and use this to extrapolate population estimates because claims data, though large in number, are spotty records and subject to upcoding. The placebo group data are a smaller set, but the information is more accurate and complete. The problem with placebo data is that each pharmaceutical company collects data in a different manner and the cost to sift through and standardize this information is estimated to be several million dollars. However, if standardization is done, the dataset could be a valuable resource for healthcare organizations.