BACKGROUND: Many deaths in the intensive care unit (ICU) occur after a decision to withdraw or withhold life-sustaining therapies (WLSTs). We aimed to explore the differences in the incidence and timing of WLST between patients with and without acute brain injuries (ABIs). METHODS: We did a secondary analysis of two prospective, international studies that recruited patients who were invasively or non-invasively ventilated between 2004 and 2016 from 40 countries. ABI was defined as brain trauma, ischaemic stroke, intracranial haemorrhage, seizures, or meningitis-encephalitis. The comparator group included non-ABI conditions. Time to WLST was evaluated by use of cumulative incidence curves. Differences in WLST were analysed by use of multilevel logistic regression. FINDINGS: Between March 11, 2004, and Dec 17, 2016, we recruited 21 970 patients (16 791 in the WLST analysis), of whom 13 526 (61·6%) were male and 8444 (38·4%) were female and 2896 (13·2%) had ABI. WLST occurred in 2056 (12·2%) of 16 791 patients) and was more common in patients with ABI versus without (372 [17·0%] of 2191 vs 1684 [11·5%] of 14 600
risk difference 5·5%
95% CI 3·8-7·1
odds ratio [OR] 2·42
1·89-3·12). WLST decisions occurred earlier in patients with ABI versus patients without ABI (median, 4 days [IQR 2-9] versus 6 days [2-13] after ICU admission
absolute difference, 2 days
95% CI 1-3). Findings were similar across different ABI subgroups, world regions, and cohort years. Variability among ICUs in WLST decisions for patients with ABI and patients without ABI was high (respectively, median OR, 3·04
95% CI 2·54-3·67, and median OR 2·59
2·38-2·78). INTERPRETATION: Our findings suggest that WLST decisions are significantly more common in patients with ABI versus patients without ABI and occur earlier in this group. The rationale for early WLST following ABI warrants further exploration, accounting for additional neurological factors that were not available in the present analysis. FUNDING: Canadian Institutes of Health Research.