PARTIALLY-SIGHTED - INVESTIGATOR GUESSES IN A DOUBLE-BLIND CONTROLLEDTRIAL

In a double-blind, randomized trial of two antidepressants, lofepramin e and fluoxetine, and a subsequent placebo-controlled trial of lithium augmentation, investigators made guesses as to which drug subjects we re taking. Preliminary guesses made during the course of treatment sho wed no association with the drugs taken, but 'best possible' estimates made after the conclusion of the study, based on biochemical test res ults as well as systematic knowledge of the side-effects of the drugs, were correct in 68% of cases (p < 0.001). On the preliminary guesses, subjects taking lofepramine who suffered side-effects considered to b e specific for lofepramine were more likely to be correctly identified than those without such side-effects. Subjects taking lofepramine who se treatment condition was correctly guessed had a higher recovery rat e; non-compliance with treatment was probably the common factor behind the lower rates of both side-effects and recovery from illness. In th e second phase of the study, 69% of the subjects taking lithium and pl acebo were correctly distinguished (p < 0.05). Trials comparing two ac tive treatments probably have little advantage over placebo-controlled trials as far as preserving blindness is concerned.