Decompressive craniectomy in adults with moderate to severe traumatic brain injury: A systematic review and meta-analysis of randomised controlled trials Authors: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Paule Lessard Bonaventure, Antoine Ouellet, Olivier Costerousse, Charles Francoeur, Ryan Zarychanski, Shane W English, Michael Verret, François Lauzier, Alexis F Turgeon Published: 2026 ABSTRACT AbstractObjectiveThe aim of this study was to evaluate the impact of decompressive craniectomy (DC) on functional outcomes, mortality, and quality of life in adults with moderate to severe traumatic brain injury (TBI).DesignSystematic review and meta-analysis of randomised controlled trials (RCTs).Data sourcesWe searched MEDLINE, Embase, Web of Science, Cochrane CENTRAL, and ClinicalTrials.gov from inception to August 1, 2026, for RCTs comparing DC with medical or surgical management in adults with moderate to severe TBI.Review methodsTwo reviewers independently extracted data and assessed risk of bias using the Cochrane Risk of Bias 2 tool. Certainty of evidence was evaluated with the Grading of Recommendations, Assessment, Development, and Evaluation framework. Risk ratios (RRs) with 95% confidence intervals (CIs) were pooled using random-effect models. A trial sequential analysis was performed to assess the robustness of the findings.ResultsEight RCTs including 1326 patients met the eligibility criteria. Seven trials, including 1296 patients, reported neurological outcomes. DC did not reduce the risk of unfavourable neurological outcome, defined as Glasgow Outcome Scale scores of 1–3 or Glasgow Outcome Scale–Extended scores of 1–4, at the longest available follow-up (3–12 months) (RR: 1.00, 95% CI: 0.83–1.20; I2 = 68%; low certainty). Overall mortality did not differ significantly between groups (RR: 0.84, 95% CI: 0.59–1.19; I2 = 65%; low certainty). Subgroup analyses suggested a possible mortality benefit when DC was performed for refractory intracranial hypertension or when compared with medical management, however, with no improvement in functional neurological outcome.ConclusionsIn adults with moderate to severe TBI, DC does not reduce unfavourable neurological outcomes or overall mortality. A possible mortality benefit in selected subgroups cannot be excluded, but this may occur at the cost of survival with severe disability. Decisions should be individualised based on each patient’s balance between survival and quality of life considerations.Keywords: Traumatic brain injury, Decompressive craniectomy, Systematic review and meta-analysis, Intracranial hypertension, Neurologic outcome assessment, Mortality Introduction Traumatic brain injury (TBI) is a major global public health issue and remains one of the leading causes of mortality and long-term disability worldwide.1 Elevated intracranial pressure is a major complication of TBI and impairs cerebral perfusion, contributing to secondary brain injury which may ultimately lead to brain herniation and death.2 Decompressive craniectomy (DC) is a surgical procedure in which part of the skull is removed, performed either to prevent intracranial hypertension or as an urgent last-tier intervention when other therapies have failed.3,4 Although DC lowers intracranial pressure,5 its impact on functional recovery remains controversial. The possibility of prolonging life at the expense of severe neurological disability continues to raise significant ethical and clinical concerns.6,7 Recent guidelines recommend the use of DC in selected clinical situations (i.e., for late refractory intracranial hypertension), based on low-certainty evidence suggesting benefits in terms of intracranial pressure control and mortality reduction.8 Most patients in whom DC is considered would not have been eligible for the trials on which these recommendations are based, highlighting a persistent gap between evidence and clinical practice.9 Previous systematic reviews and meta-analyses have inconsistently reported a reduction in mortality, on which they based their conclusion and recommendations, when functional recovery is the key outcome for patients, families, and clinicians.[10], [11], [12], [13], [14], [15] However, none of these reviews included trials published after 2022 and therefore did not account for the most recent large-scale study assessing DC in acute subdural haematoma.16 In addition, prior reviews included observational studies,10,12,13 often focused exclusively on comparisons with medical management,12,14 included a trial enrolling only paediatric populations,12,15 or incorporated studies that were not strictly randomised or clinically relevant.12,15 These important limitations justify a methodologically rigorous synthesis restricted to adult randomised controlled trials (RCTs), encompassing all relevant comparators and the most recent evidence available to date. We thus conducted a systematic review and meta-analysis of RCTs in adult patients with moderate to severe TBI to evaluate the effect of DC on functional neurological outcomes, mortality, and quality of life. Methods Study design Our systematic review and meta-analysis was conducted in accordance with the methodological guidance of the Cochrane Handbook for Systematic Reviews of Interventions17 and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement.18 The study protocol was registered prospectively in the International Prospective Register of Systematic Reviews (PROSPERO) database (registration number: CRD42024565322). Eligibility criteria We included RCTs that compared DC to any alternative management (medical or surgical) in adult patients with moderate to severe TBI, as defined by each trial’s criteria. RCTs were eligible if they included at least 80% of patients aged ≥18 years and reported on at least one of our predefined outcomes of interest. No restriction was applied regarding language, date of publication, or publication status. Outcomes Our primary outcome was an unfavourable functional neurological outcome, measured using the Glasgow Outcome Scale (GOS) or its extended version (GOS-E)19,20 at the latest available follow-up time point. We defined an unfavourable outcome as a GOS score of 1–3 or a GOS-E score of 1–4. Secondary outcomes included all-cause mortality, intensive care unit (ICU) and hospital lengths of stay, and quality of life (using any validated scale). Search strategy A comprehensive and systematic literature search was conducted across four major databases: MEDLINE (via Ovid), Embase, Web of Science, and Cochrane CENTRAL from inception to August 1, 2026. In addition, we searched for ongoing and unpublished clinical trials on ClinicalTrials.gov. The search strategy was developed in collaboration with an experienced medical information specialist and peer-reviewed according to the PRESS (Peer Review of Electronic Search Strategies) guidelines to ensure sensitivity and methodological rigour.21 The full MEDLINE search strategy is provided in Supplementary Table S1. Trial selection and data extraction Pairs of reviewers (FLB, MB, and AO) independently screened all records for eligibility in two phases: first by title and abstract and then by full-text review. Discrepancies were resolved through consensus or, when necessary, adjudicated by a third reviewer (AFT). Data extraction was performed in duplicate by the same reviewers using a standardised and pilot-tested data collection form to ensure accuracy and reproducibility. Extracted data included trial characteristics (i.e., publication year, country, and sample size), patient demographics (age, sex, TBI severity, and inclusion and exclusion criteria), intervention and comparator details (type and timing of DC and cointerventions), outcome measures, and key methodological elements such as randomisation procedures, blinding strategies, and duration of follow-up. Risk of bias assessment Pairs of reviewers (FLB, MB, and AO) independently assessed the risk of bias for each included trial using the Cochrane Risk of Bias 2.0 (RoB 2) tool, which evaluates five domains of potential bias: randomisation process, deviations from intended interventions, missing outcome data, outcome measurement, and selection of the reported result.22 An outcome-specific assessment was not performed as all primary and secondary outcomes were objective or semiobjective measures. The overall risk of bias judgement for each trial was determined by the highest level of risk observed across these domains. Any discrepancies between reviewers were resolved through discussion, with arbitration by a third reviewer (AFT) when necessary. Strength of evidence We evaluated the certainty of evidence for each outcome using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework.23 This approach considers five key domains, risk of bias, inconsistency, imprecision, indirectness, and publication bias, to rate the certainty of evidence as high, moderate, low, or very low. Assessments were conducted in duplicate (by FLB and MB) to ensure consistency and transparency. Statistical analysis We conducted the main analyses using Review Manager Web (RevMan) Web, version 10.4.0; The Cochrane Collaboration 2026, available at https://revman.cochrane.org/). For dichotomous outcomes, such as neurological function and mortality, we calculated risk ratios (RRs) with 95% confidence intervals (CIs). For continuous outcomes, including ICU and hospital length of stay, we reported mean differences (MDs). When results were provided as medians with interquartile ranges or CI, we estimated means and standard deviations using established conversion methods.24,25 We applied a random-effect model using the DerSimonian and Laird method with inverse-variance weighting for all analyses.26 To assess heterogeneity, we used the I2 statistic, with values greater than 50% considered as indicative of substantial heterogeneity.27 To explore potential sources of heterogeneity, we performed prespecified subgroup analyses based on key clinical and methodological factors: presence of intracranial haemorrhage, comparator type (medical vs. surgical), type of DC performed regardless of intracranial pressure (primary DC) vs. performed for refractory intracranial hypertension (secondary DC), and risk of bias (low vs. high or unclear). To determine whether the accumulated evidence was sufficient to draw firm conclusions, we conducted a trial sequential analysis (TSA) for the primary outcome using TSA software version 0.9.5.10 Beta (Copenhagen Trial Unit, Center for Clinical Intervention Research, Copenhagen, Denmark, 2011).28 We calculated the required information size based on event rates, effect size, and heterogeneity observed in the meta-analysis. The required information size was calculated based on the primary outcome data, and cumulative Z-scores were computed using the O’Brien–Fleming alpha-spending function. Analyses were conducted using a two-sided test with a 5% significance level and 80% power. Ethical approval Ethical approval was not applicable. This meta-analysis involved secondary data from published studies, exempting it from institutional review board approval and consent. The data used are in the public domain. Registration This study was registered with the International Prospective Register of Systematic Reviews; the registration number was CRD42024565322. Results Literature research and characteristics of trials Our search strategy identified a total of 7448 citations. After the final screening, eight trials were included in the systematic review, for a total of 1326 patients.16,[29], [30], [31], [32], [33], [34], [35] (Fig. 1). Three trials were single-centre trials conducted in Pakistan,35 India,30 and Egypt,31 while the remaining five were multicentre.16,29,[32], [33], [34] Included patients were predominantly male (65%–89%) with a mean or median age ranging between 23 and 49 years. Four trials enrolled only patients with severe TBI[32], [33], [34], [35] (GCS score ≤8). Four studies compared DC to standard medical management,[32], [33], [34], [35] whereas four used surgical comparators: craniotomy16,29,31 and cisternostomy.30 Primary DC (performed without regard to intracranial pressure) was evaluated in five trials16,[29], [30], [31],33 and secondary DC (performed for refractory intracranial hypertension) in the remaining three.32,34,35 Seven trials reported neurological outcomes, assessed at 3 months,30 6 months,29,[32], [33], [34], [35] or 12 months16 and these time points were used in the meta-analysis as the longest available follow-up for each trial. One trial used the GCS score after surgery as a surrogate outcome.31 While two trials excluded traumatic intracranial haemorrhage,34,35 others included all types of lesions.16,[29], [30], [31], [32], [33] Additional characteristics of the included trials are reported in Table 1. Flow chart of included trials. Table 1Characteristics of the included trials.Table 1StudiesPrimary outcomeInclusion criteriaNaAge (year)Intervention groupComparator groupCooper et al. 2011 (DECRA)MulticentricAustralia, New-Zealand and Saudi ArabiaGOS-E at 6 monthsSpontaneous increase in ICP (>20 mmHg) for more than 15 min within a 1-h period, despite optimised first-tier interventionsAge between 15 and 59 yearsSevere, nonpenetrating TBI (GCS ≤8 or Marshall class III)Intracranial haemorrhage excluded155Median with IQRI: 23.7 (19.4–29.6)C: 24.6 (18.5–34.9)Procedure: Secondary DCN randomised: 73Male gender: 81%Procedure: Standard medical care (sedation, normalisation of arterial carbon dioxide pressure, mannitol, hypertonic saline, neuromuscular blockade, external ventricular drainage. Second-tier options for refractory ICP included mild hypothermia (to 35 °C) or barbiturates)N randomised: 82Male gender: 74%Moein et al. 2012MulticentricIranGOS at 6 monthsSevere TBI (GCS ≤8)Presence of midline shift or basal cistern closure in a brain CT scanIntracranial haemorrhage included20MeanI: 34.6C: 31.8Procedure: Primary DCN randomised: 10Male gender: UnknownProcedure: Standard medical care (sedation, analgesia, muscle relaxation, normothermia (36–36.5 °C) and normoglycemia (70–105 mg/dL)N randomised: 10Male gender: UnknownHutchinson et al. 2016 (RESCUEicp)Multicentric (20 centres)71% of patients from United KingdomGOS-E score at 6 monthsRaised ICP (>25 mmHg for > than 1–12 h) refractory to initial medical treatment measuresAge between 10 and 65 yearsSevere TBI (abnormal CT scan requiring ICP monitoring)Patients may have had an immediate operation for a mass lesion but not a “decompressive” craniectomy.Intracranial haemorrhage included408Mean with SDI: 32.3 ± 13.2C: 34.8 ± 13.7Procedure: Secondary DCN randomised: 206Male gender: 82%Procedure: Standard medical care (stage 1: ventilation, sedation, analgesia, head up ± paralysis, monitoring (central venous pressure, arterial line, ICP). Stage 2 if ICP >25 mmHg: Ventriculostomy, inotropes, mannitol, hypertonic saline, loop diuretics, therapeutic hypothermia and barbiturates)N randomised: 202Male gender: 80%Hashmi et al. 2022UnicentricPakistanGOS at 6 monthsICP >20 mmHg for more than 15 min despite medical managementAge between 15 and 60 yearsDiffuse TBI (GCS ≤8)Injury within 12 h at randomisationIntracranial haemorrhage excluded170Mean with SDI: 34.3 ± 14.1C: 34.5 ± 13Procedure: Secondary DC (unilateral or bilateral)N randomised: 81Male gender: 65%Procedure: Standard medical care (sedation, analgesia, ventilation, head up, mannitol, hypertonic saline)N randomised: 89Male gender: 66%Chandra et al. 2022UnicentricIndiaGOS at 3 monthsAge between 18 and 64 yearsGCS score ≥4Post-traumatic diffuse oedema with mass effect and midline shiftIntracranial haemorrhage included50Mean with SD:I: 42.84 ± 13.90C: 44.48 ± 12.48Procedure: Primary DCN randomised: 25Male gender: UnknownProcedure: CisternostomyN randomised: 25Male gender: UnknownHutchinson et al. 2023 (RESCUE ASDH)Multicentric (40 centres)55% of patients from the United KingdomGOS-E at 12 monthsAdults (aged >16 years) with severe TBIAcute subdural haematoma on CT that needs to be evacuated by either a craniotomy or decompressive craniectomyIntracranial haemorrhage included450Mean with SDI: 48.8 ± 16.6C: 48.3 ± 16.5Procedure: Primary DCN randomised: 222Male gender: 81%Procedure: CraniotomyN randomised: 228Male gender: 78%Azouz et al. 2023UnicentricEgyptGCS after surgeryAge between 20 and 50 yearsGCS between 6 and 12Traumatic acute subdural haematoma fitting surgical criteria (midline shift ≥5 mm, >1 cm in thickness)Intracranial haemorrhage included30Mean with range (all included patients):40 (16–60)Procedure: Primary DCN randomised: 15Male gender (overall study): 70%Procedure: CraniotomyN randomised: 15Male gender (overall study): 70%Feng et al. 2026 (PREDICT-AEDH)MulticentricChinaGOS-E at 6 monthsAge between 18 and 65 yearsTraumatic acute epidural haematoma (volume >30 mL, >1,5 cm in thickness, midline shift >5 mm) who required urgent surgical evacuation <12 h of injuryObliterated ambient cistern confirmed by CTClinical signs of transtentorial herniationIntracranial haemorrhage included120Mean with SD:I: 44.24 ± 11.14C: 42.16 ± 11.25Procedure: Primary DCN randomised: 58Male gender: 86%Procedure: CraniotomyN randomised: 62Male gender: 89%C = comparator group; CT = computed tomography; I = intervention group; DC = decompressive craniectomy; GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale; GOS-E = Glasgow Outcome Scale-Extended; ICP = intracranial pressure; IQR = interquartile range; SD = standard deviation; TBI = traumatic brain injury.aNumber of patients randomised at the beginning of the study. Characteristics of the included trials. C = comparator group; CT = computed tomography; I = intervention group; DC = decompressive craniectomy; GCS = Glasgow Coma Scale; GOS = Glasgow Outcome Scale; GOS-E = Glasgow Outcome Scale-Extended; ICP = intracranial pressure; IQR = interquartile range; SD = standard deviation; TBI = traumatic brain injury. Number of patients randomised at the beginning of the study. Risk of bias Risk of bias assessment showed variability across the included trials. Four trials were judged at low risk of bias,16,29,32,34 mainly because the primary outcome (GOS-E) was assessed by blinded evaluators. In the remaining trials, the absence of blinded outcome assessment, together with concerns regarding the randomisation process and deviations from intended interventions, led to an overall judgement of high risk of bias.30,31,33,35 A detailed evaluation of the risk of bias for the included studies is presented in Fig. S1. Primary outcome Seven trials (n = 1296) reported functional neurological outcomes, of which four used the GOS-E16,29,32,34 and three used the GOS.30,33,35 An unfavourable outcome was observed in 61.5% (397/646) of patients in the DC group and 60.2% (391/650) in the control group, yielding an RR of 1.00 (95% CI: 0.83 to 1.20; I2 = 68%), with low certainty of evidence (Fig. 2, Table 2, Table 3). The lack of effect was consistent across predefined subgroups based on comparator type (medical vs. surgical), presence of intracranial haemorrhage, and type of DC (primary vs. secondary) and RoB (Fig. 3, Fig. S3A–D). The TSA showed that the cumulative Z-curve crossed the futility boundary, suggesting that the current evidence is sufficient and conclusive (required information size = 973; Fig. S2). Meta-analysis for unfavourable neurological outcome. CI = confidence interval; DC = decompressive craniectomy; IV = inverse variance. Table 2GRADE Summary of findings.Table 2OutcomesAnticipated absolute effectsd (95% CI)Relative effect (95% CI)№ of participants (studies)Certainty of the evidence (GRADE)CommentsRisk in control groupsRisk in decompressive craniectomy groupUnfavourable neurological outcome602 per 1000602 per 1000 (499–722)RR: 1.00 (0.83–1.20)1296 (seven RCTs)⨁⨁◯◯Lowa,bThe evidence suggests that decompressive craniectomy results in little to no difference in unfavourable neurological outcome.All-cause mortality337 per 1000283 per 1000 (199–401)RR: 0.84 (0.59–1.19)1326 (eight RCTs)⨁⨁◯◯Lowa,cDecompressive craniectomy may reduce all-cause mortality slightly.ICU length of stayThe mean ICU length of stay was 13.32 daysMD: 1.69 lower (4.16 lower to 0.78 higher)–1266 (six RCTs)⨁◯◯◯Very lowa,b,cThe evidence is very uncertain about the effect of decompressive craniectomy on mean intensive care unit length of stay.Hospital length of stayThe mean hospital length of stay was 30.01 daysMD: 0.88 lower (4.06 lower to 2.30 higher)–461 (four RCTs)⨁◯◯◯Very lowa,b,cThe evidence is very uncertain about the effect of decompressive craniectomy on mean hospital length of stay.CI = confidence interval; GRADE = Grading of Recommendations, Assessment, Development, and Evaluation; ICU = intensive care unit; MD = mean difference; RCT = randomised controlled trial; RR = risk ratio.GRADE Working Group grades of evidence.High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.ExplanationsaDowngraded due to the risk of bias.bDowngraded due to considerable and unexplained heterogeneity across studies (high I2 without clear clinical or methodological justification), reducing confidence in the consistency of the effect.cDowngraded due to imprecision as the wide confidence interval includes both a meaningful benefit and no effect (or harm), limiting the certainty of the estimate.dThe risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). GRADE Summary of findings. CI = confidence interval; GRADE = Grading of Recommendations, Assessment, Development, and Evaluation; ICU = intensive care unit; MD = mean difference; RCT = randomised controlled trial; RR = risk ratio. GRADE Working Group grades of evidence. High certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. Explanations Downgraded due to the risk of bias. Downgraded due to considerable and unexplained heterogeneity across studies (high I2 without clear clinical or methodological justification), reducing confidence in the consistency of the effect. Downgraded due to imprecision as the wide confidence interval includes both a meaningful benefit and no effect (or harm), limiting the certainty of the estimate. The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). Table 3Secondary outcomes.Table 3Secondary outcomeTrialsParticipantsAssociation measureStatistical method, analysis modelAssociation [95% CI]I2GRADE certaintyAll-cause mortality81326Risk ratioIV, random0.84 [0.59, 1.19]65%⨁⨁◯◯, lowICU LOS61266Mean differenceIV, random−1.69 [−4.16, 0.78]85%⨁◯◯◯, very lowHospital LOS4461Mean differenceIV, random−0.88 [−4.06, 2.30]51%⨁◯◯◯, very lowCI = confidence interval; DC = decompressive craniectomy; GRADE = Grading of Recommendations, Assessment, Development, and Evaluation; ICU = intensive care unit; IV = inverse variance; LOS = length of stay. Secondary outcomes. CI = confidence interval; DC = decompressive craniectomy; GRADE = Grading of Recommendations, Assessment, Development, and Evaluation; ICU = intensive care unit; IV = inverse variance; LOS = length of stay. Subgroup analysis for unfavourable neurological outcome. Primary DC was performed without regard to intracranial pressure; secondary DC was performed for refractory intracranial hypertension. CI = confidence interval; DC = decompressive craniectomy; IV = inverse variance. Secondary outcomes Mortality All eight trials (n = 1326) reported mortality outcomes assessed at 3,30 6,29,[32], [33], [34], [35] and 12 months,16 with one trial reporting mortality at an unknown time point.31 Overall, DC was not associated with a statistically significant reduction in mortality compared to standard care (26.5% vs. 33.7%) (RR: 0.84, 95% CI: 0.59 to 1.19; I2 = 65%) with low certainty of evidence (Fig. S4, Table 2, Table 3). Subgroup analyses suggested a mortality benefit when DC was compared to medical management (RR: 0.59, 95% CI: 0.43 to 0.81; I2 = 25%; four trials[32], [33], [34], [35]) but not when compared to surgical interventions (RR: 1.12, 95% CI: 0.87 to 1.43; I2 = 0%; four trials16,[29], [30], [31]) (Fig. S5A). Similarly, secondary DC was associated with reduced mortality (RR: 0.61, 95% CI: 0.42 to 0.87; I2 = 47%; three trials32,34,35), whereas primary DC showed no significant effect (RR: 1.10, 95% CI: 0.86 to 1.41; I2 = 0%; five trials16,[29], [30], [31],33). The effect of DC on mortality was not modified in other subgroups (Fig. S5B and D). ICU length of stay Six trials16,29,30,32,34,35 reported ICU length of stay (n = 1266). Overall, DC was not associated with a significant reduction in ICU stay compared to standard care (MD: −1.68 days; 95% CI: −4.65 to 1.29; I2 = 88%), with very low certainty of evidence (Fig. S6, Table 2, Table 3). Subgroup analyses suggested potential effect modification with a reduction in ICU stay when DC was compared to medical management (MD: −3.26 days; 95% CI: −6.36 to −0.15; three trials32,34,35), and in trials including patients with traumatic intracerebral haemorrhage (MD: −4.94 days; 95% CI: −6.15 to −3.74; two trials34,35) (Fig. S7A–C). Hospital length of stay Four trials29,30,34,35 (n = 461) reported hospital length of stay. DC was not associated with a significant reduction in hospital length of stay compared to control (MD: −0.88 days; 95% CI: −4.06 to 2.30; I2 = 51%) with very low certainty of evidence (Fig. S8, Table 2, Table 3). We did not perform any subgroup analysis for this outcome. Quality of life Two trials reported quality of life outcomes.16,29 In one trial,16 quality of life was assessed using the Euroqol 5-Dimension 5-Level (EQ-5D-5L), a standardised instrument yielding a utility index score ranging from −0.594 (a health state considered worse than death) to 1 (perfect health); deceased patients were assigned a score of 0.36 The MD in EQ-5D-5L scores between the DC group and the control group was not significant: 0.058 (95% CI: −0.024 to 0.141). In the other trial,29 at 6 months, health-related quality of life (EQ-5D-5L sum score) was similar between the DC and standard craniotomy groups (median: 6 [interquartile range: 5–9] vs 5 [5–9]; p = 0.13). Cognitive function (Mini-Mental State Examination (MMSE), range: 0–30, with <24 indicating cognitive impairment) also did not differ: 7 out of 52 (13%) vs 7 out of 56 (13%) among 6-month survivors (OR: 0.92 [95% CI: 0.25–3.34]; p = 1.0). Discussion In our systematic review and meta-analysis of RCTs, DC was not associated with a decreased unfavourable neurological outcome in adults with moderate to severe TBI. The TSA indicated that further studies are unlikely to change the conclusion of no effect for this outcome. Overall, no mortality benefit was observed. These results differed when DC was performed in patients with refractory intracranial hypertension or was compared with medical treatment in which a potential mortality benefit with DC was observed. The overall certainty of evidence was low to very low for all outcomes. Our results differ from the most recent systematic review and meta-analysis on DC in TBI, which reported a reduced mortality in the DC group without showing a beneficial effect on neurological function.15 One way to explain this discrepancy is that our review includes five additional RCTs published since 2022 (832 additional patients). In contrast, the previous review considered five RCTs (665 patients in total) published up to 2021, including one conducted in a paediatric population37 and one trial comparing two different DC techniques (unilateral temporoparietal vs. frontoparietotemporal craniectomy). These two trials were excluded in our review.38 Our results also differ from previous meta-analyses reporting a reduction in mortality with DC11,12,14 and, for two of them, a higher rate of unfavourable neurological outcomes.11,12 The 2019 Cochrane review14 included only three RCTs (n = 590) comparing DC exclusively with medical management, whereas our analysis encompassed all relevant comparators and performed predefined subgroup analyses comparing medical versus surgical controls. Similarly, other meta-analyses12 combined adult and paediatric populations, included a nonrandomised trial,39 and restricted comparisons to medical treatment only. In contrast, our review focused exclusively on RCTs in adults, is the only one to incorporate RCTs published since 2022, and provides a broader and methodologically stronger evaluation of both functional and mortality outcomes. Interestingly, our results are not aligned with those of the most recent Brain Trauma Foundation Guidelines for the Management of Severe Traumatic Brain Injury recommending secondary DC only in cases of late refractory intracranial hypertension.8 This recommendation is not supported by the published data or results of this meta-analysis. One main explanation is that it was largely based on the RESCUE-ICP trial,32 and five additional RCTs have since been published.16,[29], [30], [31],35 Nevertheless, our findings align with previous systematic reviews suggesting that DC may reduce mortality compared to medical management in certain patient subgroups but at the cost of a higher rate of severe disability among survivors.[9], [10], [11], [12], [40] This underscores a central dilemma in neurocritical care: balancing saving a life with respect for patient autonomy, which encompasses individual values and perceptions of quality of life. In TBI, this ethical tension is particularly important as survival may often come at the cost of severe long-term disability. Decisions to perform DC are frequently made in emergency contexts with high emotional stress, which might influence the consent process.41,42 These situations are especially difficult because most patients are young, reinforcing the impulse to pursue high-intensity care interventions. The high probability of survival with profound disability must be clearly exposed so an informed decision can be made. It is essential to involve families in a decision-making process that respects the patient’s values and preferences. Beyond clinical outcomes, the long-term psychosocial and economic burden on family caregivers, including emotional distress, financial strain, and caregiving responsibilities, deserves greater consideration.43,44 Our systematic review has several strengths. The protocol was prospectively registered and developed according to rigorous methodological standards, enhancing transparency. We used a comprehensive and peer-reviewed search strategy across multiple databases, without language or date restrictions, thereby minimising the risk of publication bias. Unlike most previous reviews, we included all relevant comparators, both medical and surgical, providing a more complete and clinically applicable synthesis of the available evidence. The selection of a patient-centred primary outcome specific to the TBI population, along with the inclusion of quality-of-life data, which are rarely reported in prior analyses, adds further relevance. We conducted predefined subgroup analyses to explore potential heterogeneity and effect modifiers. We also performed a TSA to assess the sufficiency of the accumulated evidence. An important limitation of current evidence concerns patients undergoing DC during an indicated craniotomy, most often for large epidural or subdural haematomas. This scenario accounts for a substantial proportion of DCs in practice, yet available RCTs were not designed to address it and do not clearly distinguish last-tier DC for refractory intracranial hypertension from DC as an extension of haematoma evacuation. The recent RESCUE-ASDH trial16 compared primary DC versus craniotomy for acute subdural haematoma and found no evidence of improved functional outcomes with DC, despite a reduction in the risk of further surgery. This gap in the literature underscores the need for trials specifically evaluating outcomes in patients requiring emergent craniotomy for mass lesions, where the decision to leave the bone flap off remains frequent in daily neurotrauma practice. Another limitation of our study is the substantial heterogeneity observed across several outcomes, which could not be fully accounted for by subgroup analyses. Consequently, the certainty of evidence for both primary and secondary outcomes was rated as low or very low. Although a TSA showed that the cumulative Z-curve crossed the futility boundary for neurological outcomes, this result should be interpreted with caution. The low certainty of evidence, driven by inconsistency across studies and risk of bias, suggests that residual uncertainty persists, and further high-quality research may still be needed in specific patient populations. Conclusion In adults with moderate to severe TBI, DC does not reduce the risk of unfavourable neurological outcomes or mortality. A possible mortality benefit cannot be excluded in patients with refractory intracranial hypertension or when compared to medical management, though this may come at the cost of survival with severe disability and dependency. DC should not be a standard of care procedure; rather, its use should be individualised, guided by informed consent that reflects the patient’s preferences on survival versus quality of life. CRediT authorship contribution statement Conceptualisation: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Alexis F. Turgeon. Data curation: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Antoine Ouellet. Data extraction: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Antoine Ouellet. Statistical analyses: Marwan Bouras, Frédérique Lamothe-Boucher. Methodology: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Alexis F. Turgeon. Supervision: Alexis F. Turgeon. Funding acquisition: Alexis F. Turgeon. Data interpretation: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Paule Lessard Bonaventure, Antoine Ouellet, Olivier Costerousse, Charles Francoeur, Ryan Zarychanski, Shane W. English, Michael Verret, François Lauzier, Alexis F. Turgeon. Writing – original draft: Marwan Bouras, Frédérique Lamothe-Boucher, Alexis F. Turgeon. Writing – review & editing: Marwan Bouras, Frédérique Lamothe-Boucher, Jihad Abou Jamous, Paule Lessard Bonaventure, Antoine Ouellet, Olivier Costerousse, Charles Francoeur, Ryan Zarychanski, Shane W. English, Michael Verret, François Lauzier, Alexis F. Turgeon. Data sharing statement All data relevant to the study are included in the article or uploaded as supplementary information. Funding This work was funded by the Canadian Institutes of Health Research (Foundation Scheme Grant (#148443)). The funding source had no role in this work. Conflict of interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements MB received a fellowship training grant from the Fondation des Gueules Cassées, the Société Française d’Anesthésie Réanimation (France) and from the Canada Research Chair in Critical Care Neurology and Trauma. FLB was supported by a Graduate Scholarship Award from the Canadian Institutes of Health Research (CIHR). RZ is supported by a Canada Research Chair in Clinical Trials. AFT is the chairholder of the Canada Research Chair in Critical Care Neurology and Trauma. Footnotes Supplementary data to this article can be found online at https://doi.org/10.1016/j.ccrj.2026.100220. Contributor Information Marwan Bouras, Email: marwan.bouras@chu-brest.fr. Alexis F. Turgeon, Email: alexis.turgeon@crchudequebec.ulaval.ca. Appendix A. Supplementary data The following are the Supplementary data to this article: Multimedia component 1 References Associated Data Supplementary Materials Multimedia component 1