Epidemiological screening for potentially carcinogenic drugs.

This paper describes a unique program for the systematic screening of commonly used prescription drugs for possible carcinogenic effects, by following up a large cohort of patients with computer-stored pharmacy data for incidence of cancer. Among the most interesting findings in recent analyses are an association of several antibiotics with subsequent lung cancer, and negative associations of prescribed vitamin E and diazepam with certain cancers. Analyses of additional data do not clearly indicate that these represent causal relationships to the drugs themselves. Also of interest is our continuing negative evidence regarding reserpine and metronidazole. The planned computerization of all pharmacies in our medical care program now serving over 2.2 million subscribers, should greatly increase our drug surveillance capabilities.