Interactive model of therapeutic response in panic disorder: moclobemide, a case in point

It was proposed that pre-post regression slopes be used to index treatment response when the effect of baseline scores differed among treatments (interaction between treatment and baseline score). Reanalyses of two studies using imipramine and fluoxetine in panic disorder showed doserelated decreases in pre-post slopes for the frequency of unexpected panic attacks, but not for the frequency of situational panic attacks or measures of agoraphobia. This report presents similar analyses of data from a study using moclobemide. Patients (N = 452) with panic disorder were randomized to placebo or a fixed dose of moclobemide (75, 150, 300, 600, or 900 mg/day). They were treated double-blindly and evaluated at baseline and 1, 2, 3, 4, 6, and 8 weeks later. The authors analyzed the frequency of unexpected and situational panic attacks compiled from a daily diary, and fear and avoidance ratings based on the patient's main phobia using baseline (pre) and end-point (post) values for all randomized patients. Adjoining dose groups were combined. Both unexpected and situational panic attacks showed systematic doserelated suppression of pre-post treatment slopes. Neither pre-post slopes nor adjusted posttreatment means for fear and avoidance differed reliably between treatment arms. This study replicates the authors' earlier findings, except for situational panic attacks, which probably were not reliably identified. Antidepressants selectively suppress panic attacks, especially unexpected attacks, but not agoraphobia. The findings are consistent with the hypothesis that panic disorder with agoraphobia has clinically separable biologic and cognitive components that respond differentially to treatment. Antidepressants benefit primarily patients with many unexpected panic attacks. Investigators should evaluate pre-post treatment slopes before comparing adjusted posttreatment means (analysis of covariance).