Data The Prozac Approval Process

As noted earlier in the chapter, although several thousand patients were involved in studies of various kinds, I counted only 286 who actually finished the three placebo-controlled protocols (groups of studies) used for approval. Many patients dropped out because of adverse stimulant reactions. Prozac seldom proved any better than placebo and was not as good as the older antidepressants. It was so stimulating that sedatives were often given along with it. 

In perhaps the most important study, called Protocol 27, the results indicated that Prozac by itself had no efficacy. To get a positive result, the FDA had to allow the drug companies to include all the patients who, 

Protocol 27 was conducted by several separate investigators at sites in different cities. The individual study sites could not show that Prozac was any better than placebo, so the FDA allowed the negative results to be pooled and manipulated until a positive result was barely achieved. 

Lilly employees Stark and Hardison (1985) eventually published Protocol 27 in the Journal of Clinical Psychiatry. They did so without mentioning (a) that four of the five individual centers produced negative results before the data were pruned and pooled, (b) that even the pooled data were negative when Prozac patients taking sedatives and tranquilizers were excluded, (c) that the FDA had many criticisms of the study and its practices, or (d) that even the apparent success of the drug was marginal. The publication by Stark and Hardison claimed that Prozac was comparable to Tofranil in efficacy—a myth that gained considerable currency in the profession—when in fact, the older tricyclic outperformed Prozac most of the time. 

---