The value of self-report assessment in studies of anxiety disorders.

This review compares the sensitivities of a physician-rated scale, the Hamilton Anxiety Scale (HAS), and a patient-rated scale, the Hopkins Symptom Checklist (HSCL), in detecting the anti-anxiety effects of benzodiazepines in a large sample of placebo-controlled trials. Scales and subscales were compared within the same study, a methodologic feature unique to this review. The total score, psychic factor, and somatic factor of the HAS were equally sensitive to the effects of benzodiazepines. The total score, anxiety factor, and somatization factor of the HSCL also were equally sensitive. The HAS total score, however, was consistently more sensitive than any of the HSCL scores. There was no evidence that physicians used side effects to make their ratings more sensitive. Sedative side effects, however, adversely affected the sensitivity of patient ratings. The data suggest that patient ratings reflect a cost-benefit computation taking account of both antianxiety and sedative effects.