A reexamination of false-positive rates for carcinogenesis studies.

False-positive rates in the National Cancer Institute/National Toxicology Program (NCI/NTP) two-year carcinogenesis bioassay are examined. Previous studies that have reported high (20-50%) false-positive rates for these bioassays are reviewed. The decision rules used in many of these earlier investigations are far different than the procedures actually employed by the NCI/NTP in the interpretation of bioassay data. Data from 25 recent NTP feeding studies are examined, and the statistical significance of observed tumor increases are compared with the final interpretations regarding the carcinogenic effects of the chemicals under study. Based on this examination, a more realistic decision procedure is formulated. It is shown that under the assumption of a statistical decision rule which appears to mimic more closely the scientific judgment process, the actual overall false-positive rate in NCI/NTP bioassays appears to be no greater than 7% to 8%.