The Effect of Unproved Cancer Therapy in Advanced Cancer

To the Editor: The difficulty of randomizing conventional cancer treatments has led Cassileth et al. (April 25 issue) 1 to adopt a convoluted study design that cannot, even if the major defects in its statistical analysis are rectified, provide reliable evidence as to whether the particular unconventional method studied affects survival in advanced cancer. Their comparison of conventional with unconventional treatment (the Livingston-Wheeler treatment) in 78 pairs of patients with advanced cancer was neither randomized nor large, so it may be subject to both bias and substantial random error. Hence, although it may provide useful evidence against absurdly big expectations of . . .