Diazepam: how much better than placebo?

In a six week, double-blind, predictive, clinical study 224 neurotic and anxious outpatients were randomly assigned to treatment with diazepam (Valium) of placebo and weekly brief psychotherapy administered by experienced psychiatrists. Although diazepam was significantly more effective than placebo for the first week of treatment, it accounted for only 1-4% of the outcome variance, while nonspecific variables contributed 11-22%. After the first revisit and at endpoint, only nonspecific variables accounted for outcome variance ranging from 20 to 26% over the four outcome measures. According to physicians' global and scaled ratings of improvement, diazepam had a treatment advantage (p less than 0.01-0.05) over placebo at the end of the first week of treatment for both the total and completed samples, but not thereafter. According to patients' global ratings of improvement, diazepam was moderately superior (p less than 0.05) to placebo for the total sample only at the end of the first week of treatment and modestly better (p less than 0.10) than placebo for the completed sample on a highly reliable measure of reported anxiety at the same time point. However, patient ratings did not indicate an advantage for diazepam at any of the six points of observation on global ratings of improvement for the completed sample and on reported anxiety for the total sample.