BCG prophylaxis against cancer: methodological evaluation of a continuing controversy

A possible protective effect of tuberculosis against cancer, together with the known immune-stimulating properties of BCG, have motivated 14 investigations of BCG prophylaxis against cancer. The results have been conflicting: 3 studies found high levels of protection, 8 found no effect, and 3 suggested that BCG may actually raise the risk of cancer. To assess whether the inconsistent results could have a methodological basis, we analyzed each study according to several basic criteria necessary for scientific validity in cause-effect research. The analyses showed two major problems: (1) None of the studies used safeguards against detection bias by ensuring that the diagnosis of cancer in vaccinees and non-vaccinees was made with equal follow-up and surveillance procedures, and in an accurate and unbiased fashion; (2) Studies alleging that BCG is carcinogenic were weakened by the inappropriate use of cancer mortality data, by the analysis of multiple BCG-cancer associations without appropriate statistical adjustment, and by post hoc creation of new cancer risk categories after inspection of the data. Since these methodologic inadequacies preclude any firm conclusions about the role of BCG in preventing cancer, a resolution of the controversy will require new research, designed and conducted with greater attention to methodologic principles necessary to ensure scientific validity.