Seventy percent of patients with panic disorder (PD) have sleep impairment. Cognitive behavior therapy (CBT) effectively treats PD, but the concomitant effect on sleep is understudied. Physical exercise (PE) improves sleep, but this has not been investigated in patients with PD.

To compare the effects of CBT or PE on sleep in PD-patients, and to determine potential mechanisms of action.

Thirty-six PD-patients were randomized to either group CBT for PD or regular PE. Sleep was assessed pre/post with the Pittsburgh sleep quality index. Effects were investigated with repeated measures ANOVA and t-tests. Expected mediators were added to the general linear model to assess mediation.

The effect of time was significant, F(1.33)=10.11, P=0.003, but not the interaction (Time×group), F(1.33) =.48, P=0.49. Symptoms were significantly reduced from pre- to post-treatment: PE, t(16)=3.03, P =.008, and CBT, t(17)=2.18, P=0.044. CBT-patients changed significantly (P<0.05) on Sleep quality, t(17)=2.47 and Sleep disturbance, t(17)=2.38. PE-patients changed significantly on sleep duration, t(16)=2.58 and sleep disturbance, t(16)=2.58. A significant interaction with change in fear of bodily symptoms, F(1.16)=5.53, P=.032, and with change in depression-level, F(1.16)=12.13, P=0.003 was only found for CBT. A significant interaction with change in physical fitness, F(1.15)=5.01, P=.041, was only found for PE.

Both interventions improve sleep in PD-patients, but differently. The findings also suggest that these changes are related to different mechanisms for PE and CBT.