Lithium versus valproate initiation and psychiatric service utilisation in bipolar disorder: an overlap-weighted new-user cohort study from the United Arab Emirates

Comparative-effectiveness evidence on lithium and valproate maintenance treatment in bipolar disorder comes largely from Nordic and United Kingdom registers, with little evidence from Middle Eastern health systems. To our knowledge, this is the first study from the United Arab Emirates, the Gulf Cooperation Council, or the wider Arab world to compare lithium and valproate initiation on psychiatric service use with an active-comparator new-user design. We compared 12-month psychiatric service use after lithium versus valproate initiation. We conducted a retrospective active-comparator new-user cohort study using electronic health records of a tertiary psychiatric hospital in the United Arab Emirates from 1 January 2018 to 22 December 2025. Eligible adults had an International Classification of Diseases, Tenth Revision, F31 diagnosis and initiated lithium or valproate as recorded new users. The primary outcome was the first inpatient admission or emergency department contact within 12 months. Adjustment used overlap propensity score weighting across 14 baseline-covariate terms. Emergency department chief complaints were classified to define relapse-indicative outcomes, and dispensing records described adherence. The cohort included 316 new users: 101 lithium and 215 valproate initiators. The overlap-weighted 12-month risk was 54.3% with lithium and 69.4% with valproate (risk difference −15.1 percentage points, 95% CI −31.6 to 1.1; hazard ratio 0.63, 95% CI 0.41–0.94). The difference was largest for inpatient admission (hazard ratio 0.47, 0.27–0.83). Over half of emergency department contacts were medication-related; counting only admissions and contacts with acute symptoms, the hazard ratio was 0.69 (0.42–1.12). In an exploratory analysis, the difference was greatest in the first 30 days. Dispensing-based coverage and persistence were lower among lithium initiators. Estimates favoured lithium across sensitivity analyses, several with intervals including the null. Lithium initiation was associated with less acute psychiatric service use than valproate initiation over 12 months, most clearly for inpatient admission. The composite risk difference was imprecise, medication-related visits contributed to the composite, and residual confounding by indication cannot be excluded. For patients suitable for either drug, the findings are consistent with the established first-line role of lithium where monitoring can be assured; replication in larger multi-site regional datasets is needed.