Predictors of chlordiazepoxide response in anxiety

A step‐search multiple‐regression procedure was employed in an effort to predict treatment response in a group of 111 patients treated with chlordiazepoxide and 201 patients treated with placebo, using as potential predictors several forms of demographic information, assessments of the severity of initial psychopathology, and measures of patient and physician attitude. Patients were drawn from 2 hospital clinic settings and the private practices of several general practitioners and psychiatrists. They participated in a double‐blind drug trial of 4 weeks' duration. All were psychoneurotic and most complained of moderately severe anxiety. Outcome criteria included both a global improvement measure, combining physician ratings of over‐all response to treatment after 2 and 4 weeks, and the 4 week change score of the 10 item physician questionnaire. Treatment differences were highly significant for both improvement measures. Also, for both improvement measures, greater drug‐placebo differences were found in patients who: (1) were more severely ill; (2) were of higher socioeconomic class; and (3) had an illness of 6 months' or more duration. Irrespective of treatment agent, greater improvement was seen in those patients receiving a more favorable prognosis. Several additional variables were predictive for one but not the other measure of improvement, and their implicatiOns, more tentative than those for the more consistent predictors, have been explored.