Background and Purpose- Delirium is associated with increased mortality, length of stay, and poor functional outcome following critical illness. The epidemiology of delirium in stroke is poorly described. We sought to collate evidence around occurrence (incidence or prevalence) of delirium in acute stroke. Methods- We searched multiple cross-disciplinary electronic databases using a prespecified search strategy, complemented by hand searching. Eligible studies described delirium in acute (first 6 weeks) stroke. We compared delirium occurrence using random-effects models to describe summary estimates. We assessed risk of bias using the Newcastle-Ottawa tool, incorporating this in sensitivity analyses. We performed subgroup analyses for delirium diagnostic method (confusion assessment method scoring, clinical diagnosis, other), duration and timing of delirium assessment (>1 or <1 week), and performed meta-regression based on the year of publication. Results- Of 8822 titles, we included 32 papers (6718 participants) in the quantitative analysis. Summary estimate for occurrence of delirium was 25% (95% CI, 20%-30%; moderate quality evidence). Limiting to studies at low risk of bias (22 studies, 4422 participants), the occurrence rate was 23% (95% CI, 17%-28%). Subgroup summary estimates suggest that delirium occurrence may vary with assessment
Method: confusion assessment method, 21% (95% CI, 16%-27%); clinical diagnosis, 27% (95% CI, 19%-38%); other, 32% (95% CI, 22%-43%) but not with duration and timing of assessment. Meta-regression suggested decline in occurrence of delirium comparing historical to more recent studies (slope, 0.03 [SE, 0.004]; P<0.0001). Conclusions- Delirium is common, affecting 1 in 4 acute stroke patients. Reported rates of delirium may be dependent on assessment method. Our estimate of delirium occurrence could be used for audit, to plan intervention studies, and inform clinical practice. Clinical Trial Registration- URL: http://www.crd.york.ac.uk/PROSPERO/. Unique identifier: CRD42015029251.