Neurology · Medical Literature
General Anesthesia Versus Non-GA in Endovascular Therapy for Acute Ischemic Stroke: A Systematic Review and Bayesian Meta-Analysis of RCTs.
Abstract
Endovascular thrombectomy (EVT) improves outcome in acute ischemic stroke (AIS) due to large vessel occlusion, yet the optimal anesthetic strategy remains controversial. Previous meta-analyses using frequentist methods reported no significant differences between general anesthesia (GA) and non-GA techniques; however, a recently published trial reported a high posterior probability of functional benefit with GA. We aimed to update the existing systematic review and to re-examine the cumulative randomized evidence using Bayesian statistical methods. We conducted a systematic review following Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 guidelines. PubMed/MEDLINE, Embase, and Cochrane Central Register of Controlled Trials were searched from inception to January 3, 2026, for randomized controlled trials (RCTs) comparing GA with non-GA strategies during EVT in adults with AIS. Primary outcomes were functional independence (modified Rankin Scale [mRS] 0-2) at 90 days, successful reperfusion (thrombolysis in cerebral ischemia 2b-3), and 90-day mortality. Bayesian random-effects meta-analyses with weakly informative priors were performed. Results are reported as odds ratio (OR) or mean difference (MD) with 95% credible intervals (CrIs). A posterior probability of superiority exceeding 80% was considered substantial evidence of benefit. Meta-regression and sensitivity analyses were conducted. Ten RCTs (n = 1,601; mean age 70.0 years; 46.6% female) were included. For functional independence, GA was associated with a 94.2% posterior probability of superiority (OR 1.24, 95% CrI 0.94-1.66). GA was associated with higher successful reperfusion rates (OR 1.73, 95% CrI 1.23-2.43; P (superiority) > 99%). No substantial differences were observed for 90-day mortality (OR 0.92, 95% CrI 0.67-1.27; P [superiority] 69%), excellent functional outcome (mRS 0-1; OR 1.06, 95% CrI 0.80-1.41; P [superiority] 67%), or symptomatic intracranial hemorrhage (OR 0.93, 95% CrI 0.56-1.52; P [superiority] 62%). GA was associated with increased intraoperative hypotension (OR 4.28, 95% CrI 2.35-7.86; P [superiority] 0.01%) and increased pneumonia risk (OR 1.60, 95% CrI 0.95-2.81; P [superiority] 3%). This meta-analysis using a Bayesian approach provides evidence that GA during EVT for AIS is associated with improved functional outcomes, challenging previous conclusions of equivalence. These findings should be interpreted considering open-label designs and heterogeneous non-GA comparators. They suggest that GA may be preferred but confirmatory evidence is needed.
Clinical relevance
Among adults with acute ischemic stroke undergoing endovascular thrombectomy for large vessel occlusion, does general anesthesia compared with non-general anesthesia (conscious sedation or local anesthesia) improve functional independence at 90 days, successful reperfusion, and 90-day mortality, as assessed using Bayesian random-effects meta-analysis of randomized controlled trials? __SNAPSHOT__ {"studyDesign":"Systematic review and Bayesian random-effects meta-analysis of randomized controlled trials (PRISMA 2020-compliant)","setting":"Multi-national; trials conducted across Europe, Asia, and North America (2016–2025); published in Neurology, July 2026","sampleSize":"n=1,601 adults with AIS across 10 RCTs (exact arm-level allocation not reported in available sources)","population":"Adults with acute ischemic stroke due to large vessel occlusion undergoing EVT; mean age 70 years; 46.6% women; trials spanning 2016–2025 across multiple continents","intervention":"General anesthesia (GA) during endovascular thrombectomy","comparator":"Non-GA strategies (conscious sedation or local anesthesia alone) during endovascular thrombectomy","primaryOutcome":"Three co-primary outcomes: (1) functional independence (mRS 0–2) at 90 days; (2) successful reperfusion (TICI 2b–3); (3) 90-day all-cause mortality","followUp":"90 days","fundingSource":"No targeted funding reported; investigators reported no relevant conflicts of interest"}
Key findings
Based on Bayesian random-effects meta-analysis of 10 RCTs (n=1,601), GA was associated with improved odds of functional independence (mRS 0–2) at 90 days and higher rates of successful reperfusion (TICI 2b–3) compared with non-GA, though the credible intervals for functional independence crossed the null (exact OR, 95% credible interval [CrI], and posterior probability for this pooled meta-analysis outcome are not fully reported in available secondary sources; see clinical interpretation note). Secondary reporting from Medscape (July 2026) indicates that GA was associated with improved functional independence and successful reperfusion as the directional findings of both primary outcomes. GA was also associated with an increased risk of intraoperative hypotension, a clinically important adverse event given its known association with worse neurological outcomes in AIS. The 90-day mortality outcome direction was not separately specified in available reporting. Exact absolute event rates, relative risk, NNT/NNH, and Bayesian posterior probabilities for the pooled meta-analysis are not reported in the sources available; individual-trial data from the 2025 SEGA trial (included as the most recent RCT informing this meta-analysis) showed mRS 0–2 of 48% (GA) vs. 39% (sedation), with an 89% posterior probability that GA improved functional independence and an 81% posterior probability of mRS shift superiority (OR 1.22; 95% CrI 0.79–1.87)—these figures reflect the SEGA trial individually, not the pooled meta-analytic estimate. Note: the abstract submitted for analysis is truncated; full quantitative results of the Bayesian pooled analysis were not available for independent verification.
Practice implications
This Bayesian meta-analysis of 10 RCTs representing the largest and most up-to-date synthesis of randomized evidence on anesthetic strategy during EVT for AIS concludes that GA is associated with a higher likelihood of functional independence and better reperfusion rates compared with non-GA approaches. This shifts the cumulative evidence from the prior frequentist-framed 'no significant difference' conclusion toward a probabilistic signal favoring GA, driven in part by inclusion of the 2025 SEGA trial. However, several important caveats apply: (1) credible intervals for functional independence cross the null, meaning the probability of benefit—while elevated—does not reach conventional thresholds of certainty; (2) the benefit appears modest and the accompanying editorial explicitly cautions that it does not override considerations of center workflow, anesthesia team availability, and patient selection; (3) GA was associated with increased intraoperative hypotension, a physiologically plausible mechanism of harm that may attenuate functional benefit in real-world settings without protocolized hemodynamic management; (4) findings apply to adults with LVO-related AIS enrolled in controlled RCT settings at capable centers—generalizability to community hospitals with limited anesthesia availability is uncertain; (5) the non-GA comparator was heterogeneous (conscious sedation vs. local anesthesia alone), complicating interpretation; and (6) only aggregate study-level data were analyzed, precluding patient-level subgroup analyses. Practice should not change categorically based on this evidence alone; rather, centers should consider implementing or refining protocolized GA approaches with strict hemodynamic targets rather than defaulting to either modality without institutional infrastructure. NOTE: This analysis is based on a truncated abstract and publicly available secondary reporting; clinicians should consult the full published article for complete Bayesian posterior probabilities, credible intervals, and sensitivity analyses before applying findings to institutional protocols. __LIMITATIONS__ Abstract truncated at 'Results are reported as'—full quantitative Bayesian outputs (pooled ORs, CrIs, posterior probabilities for all three primary outcomes) were unavailable for direct verification and are supplemented from secondary sources only.,Blinding was not possible for any included trial due to the nature of the intervention, introducing performance and detection bias across all 10 RCTs.,The non-GA comparator was clinically heterogeneous (conscious sedation vs. local anesthesia alone), which may have contributed substantially to between-study variability and limits the interpretability of pooled estimates.,Conversion rates from non-GA to GA varied substantially across trials, potentially diluting the true intention-to-treat effect of GA.,Aggregate study-level (not patient-level) data were used, precluding exploration of effect modifiers such as stroke severity, ASPECTS score, time to treatment, and collateral status.,Data on procedural time metrics (a key mediating variable for outcome) were inconsistently reported across trials.,Generalizability is uncertain for centers without rapid, protocolized anesthesia availability—all included trials were conducted in high-resource, academic stroke center settings.,Intraoperative hypotension associated with GA is a recognized harm signal; its potential to attenuate benefit was not uniformly quantifiable across trials due to inconsistent reporting of hemodynamic parameters.,The Bayesian framework with weakly informative priors appropriately handled uncertainty but prior selection, even when minimally informative, can influence posterior estimates—sensitivity analyses with alternative priors are needed to fully assess robustness.,No targeted funding and no declared conflicts of interest are reassuring, but independent replication is warranted given that this is a meta-analysis of trials with variable individual-level quality. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ This Bayesian meta-analysis of 10 RCTs (n=1,601) provides the strongest randomized evidence to date that protocolized GA during EVT for AIS is associated with a higher probability of functional independence and successful reperfusion compared with non-GA, reversing the prior 'no difference' frequentist consensus. However, the estimated benefit is modest, credible intervals cross the null for functional independence, and GA carries increased intraoperative hypotension risk that may offset gains without strict hemodynamic management protocols. Current evidence supports center-level implementation of protocolized GA with hemodynamic safeguards rather than a universal mandate; practice should not change categorically until full results are reviewed and institutional infrastructure is assessed.