Background: Caffeine has been purported to have anxiogenic and panicogenic properties, specifically salient in patients with panic disorder (PD). However, compilations of the magnitude of the effect of caffeine on anxiety and panic attacks are lacking and potential dose-response relationships have not been examined.

Objectives: In the present systematic review and meta-analysis, we aimed to examine the acute effects of placebo-controlled caffeine challenge on occurrence of panic attacks and subjective anxiety in patients with PD and healthy controls (HC), including dose-response relationships.

Methods: Systematic searches were performed in six databases. We included blinded placebo-controlled studies of acute caffeine challenge on panic attacks and/or subjective anxiety in adult patients with PD.

Results: Of the 1893 identified articles, ten met our inclusion criteria. The 9 studies investigating panic attacks included 237 patients, of which 51.1% had a panic attack following caffeine, but none after placebo. Six of these studies compared 128 patients with 115 healthy controls (HC), finding that patients (53.9%) were more vulnerable than HC (1.7%) for panic attacks following caffeine (log RR: 3.47; 95% CI 2.06-4.87). Six studies investigated subjective anxiety in 121 patients and 111 HC following caffeine, with an overall effect indicating increased sensitivity to the anxiogenic effects of caffeine in the patient group (Hedges' g = 1.02 [95% CI: 0.09-1.96]). The restricted range of caffeine employed [400-750 mg] and few studies (3) not using 480 mg prevented any meaningful analysis of a dose-response relationship.

Limitations: Of the ten studies included, only 2 reported anxiety data for the placebo condition, precluding a proper meta-analysis comparing anxiogenic effects of caffeine and placebo. The restricted dose range used prevented assessment of dose-response relationships.

Conclusions: The results confirm that caffeine at doses roughly equivalent to 5 cups of coffee induces panic attacks in a large proportion of PD patients and highly discriminates this population from healthy adults. Caffeine also increases anxiety in PD patients as well as among healthy adults at these doses although the exact relationship between caffeine-induced anxiety and panic attacks remains uncertain. The results suggest that caffeine targets important mechanisms related to the pathophysiology of PD.

Implications: Future studies should employ a wider range of caffeine doses and investigate contributions of biological and psychological mechanisms underlying the anxiogenic and panicogenic effects of caffeine. In the clinic, patients with PD should be informed about the panicogenic and anxiogenic effects of caffeine, with the caveat that little is known regarding smaller doses than 480 mg. Registration. PROSPERO (www.crd.york.ac.uk/prospero) registration number CRD42019120220.