The prevalence of wholly attributable alcohol conditions in the United Kingdom hospital system: a systematic review, meta-analysis and meta-regression.

Background and Aims The prevalence of alcohol-related conditions is often reported as higher in hospital in-patients compared with the general population. However, formal prevalence estimates are commonly derived from small studies which report highly varied results. This systematic review and meta-analysis, within the UK hospital system, aimed to estimate the pooled prevalence of the 26 ICD-10 conditions that are wholly attributable to alcohol in in-patient settings. Methods We searched Medline, Embase, PsychINFO and CENTRAL from database inception until 1 May 2018. We included studies of any design that reported the prevalence of one of 26 wholly attributable alcohol conditions defined by the ICD-10. Studies were required to be conducted in one or more of the constituent nations of the United Kingdom and in an in-patient setting (general wards, intensive care units, accident and emergency departments or mental health in-patient units). Estimates were pooled using random-effects meta-analysis, and meta-regression tested study and patient factors contributing to variation. Quality was assessed using the Grading of Recommendations Assessment Development and Evaluation (GRADE) framework. Results A total of 124 studies were included, reporting on a total of 1 657 614 patients. The majority of studies reported on harmful use of alcohol and alcohol dependence, for which the pooled prevalence was 19.76% [95% confidence interval (CI) = 15.61-24.26%] and 10.25% (95% CI = 7.06-13.96%), respectively. Mean patient age and type of in-patient setting were identified as the main sources of variation in prevalence estimates, but not date of data collection. Both estimates were deemed very low quality according to GRADE. Conclusions An estimated one in five patients in the UK hospital system use alcohol harmfully, and one in 10 are alcohol-dependent.