Nonevaluable patients in clinical cancer research

Analysis of controlled and uncontrolled published papers in clinical cancer therapy has revealed differences in patient data handling, which may reflect both the inherent biases of patient selection for uncontrolled studies and the primary purpose of such studies. A distinction between drug-data oriented research and patient oriented research was advanced to explain these results. If comparable results are to be developed, we urge controlled trials at all phases of treatment development.