Non-Randomized Controls in Cancer Clinical Trials

Abstract Quantitative, comparative clinical trials can sometimes be better accomplished with technics other than randomization for selection of a control group. These include selection of literature controls, matched controls and controls from a previous study. Such controls have permitted detection of therapeutic activity for many anti-cancer agents. Circumstances of greatest utility are as follows: when the primary purpose of study is to estimate effectiveness of treatment; when large differences in response rate are expected on the basis of preliminary studies; or when the new therapy would only be interesting if it was substantially more effective than the standard. When a new therapy is to be evaluated, clinical investigators must consider both ethical responsibility to administer best available therapy and scientific responsibility to conduct a proper evaluation. Selected control patients can permit valid evaluation of a new therapy without the requirement of prescribing control therapy known to be ineffective. (N Engl J Med 290:198–203, 1974)