Chronobiological models postulate that abnormalities in circadian rest/activity rhythms (CRAR) are core phenomena of bipolar disorders (BDs). We undertook a meta‐analysis of published studies to determine whether self‐ or observer ratings of CRAR differentiate BD cases from comparators (typically healthy controls [HCs]).

We undertook systematic searches of four databases to identify studies for inclusion in random effects meta‐analyses and meta‐regression analyses. Effect sizes (ES) for pooled analyses of self‐ and observer ratings were expressed as standardized mean differences with 95% confidence intervals (CIs).

The 30�studies meeting eligibility criteria included 2840 cases and 3573 controls. Compared with HC, BD cases showed greater eveningness (ES: 0.33; 95% CI: 0.12–0.54), lower flexibility of rhythms (ES: 0.36; 95% CI: 0.06–0.67), lower amplitude of rhythms (ES: 0.55; 95% CI: 0.39–0.70) and more disturbances across a range of CRAR (ES of 0.78–1.12 for general and social activities, sleep and eating patterns). Between study heterogeneity was high (I^2�>�70%) and evidence indicated a potential publication bias for studies using the Biological Rhythms Interview of Assessment in Neuropsychiatry. Meta‐regression analyses suggested significantly larger ES were observed in studies using observer ratings or including BD cases with higher levels of depressive symptoms.

This meta‐analysis demonstrates that BD is associated with higher levels of self‐ or observer‐rated CRAR disturbances compared with controls. However, further studies should examine the respective performance of individual instruments when used alone or in combination, to clarify their applicability and utility in clinical practice.