Tudies from Tel Aviv [31,42,43], third the studies from Glostrup [20,44] and at least the studies of Boetto and Deras et al. [22,27]. Furthermore, the EXEL-2880 custom synthesis results from our meta-analysis are dominated by two larger retrospective studies with 611 [34], respectively 477 patients [43] and one prospective study with 511 patients [55]. This was partially taken into account in our meta-analysis with the use of the random effects model, which applies less weight to large studies than fixed AM152 web effect models. The meta-analyses revealed no statistically significant differences of AC failures, intraoperative seizures, new neurological dysfunctions, and the composite outcome (AC failure, intraoperative seizure, mortality) depending on the use of SAS or MAC technique. We found a substantial heterogeneity between the included studies and only the heterogeneity for conversion to GA showed a possible significant connection to the anaesthesia technique in the logistic meta-regression analysis. This analysis suggested significantly more unplanned conversions into GA with the use of SAS than MAC anaesthesia technique. However, this result was mainly depending on one high risk of bias retrospective SAS study with 6 events in 102 patients [57]. Removing this study abolishes the significant difference between the techniques. Of note, two of the patients in this study required conversion into GA due to an air embolism, which was most likely related to the halfsitting patient position and not the used anaesthesia technique [57]. Although air embolism was not analysed in detail in our SR, this was the only study, which reported a failure of AC due to this life-threatening adverse event. Furthermore, the sensitivity analysis, which included only prospective studies, confirmed the weakness of the result obtained by the main metaregression analysis. A significant difference between the used anaesthesia techniques in regard to conversion to GA could not be revealed by the sensitivity analysis anymore. The decision to perform a sensitivity analysis by including only prospective studies and not the largest ones, was justified by the increased risk for confounding in larger studies due to a prolonged study duration. The most studies with more than 100 AC procedures, where performed during 5? [22,31,42,43,45,46,52] or 10?8 years [34,35,37,55,57]. The probability of a continuously same anaesthesia or AC surgery conduction in these observational studies during the large timespans is very low. Our sensitivity analysis did also not reveal any statistical significant difference for the other four outcomes, which were included in the meta-analyses. Of note, the new neurological dysfunction outcome was only presented by one prospective study [38] in the SAS group. Therefore, we could not estimate the proportions for this outcome in the meta-analysis (S1 Fig). However, the main analysis included also only six studies in the SAS group [23,37,38,51,53,57] and the result was dominated by this prospective observational study of Li et al. in a Chinese population [28]. Although 53.8 of the 91 patients exhibited new neurological dysfunctions, these dysfunctions remained permanent only in 1 patient, which suggests that the aim of safe resection was achieved in the longer-term. Furthermore, the generalizability of their results is unclear, due to possible differences in the distribution of the Chinese language areas compared to Non-Chinese people. Therefore we suggest interpreting our result.Tudies from Tel Aviv [31,42,43], third the studies from Glostrup [20,44] and at least the studies of Boetto and Deras et al. [22,27]. Furthermore, the results from our meta-analysis are dominated by two larger retrospective studies with 611 [34], respectively 477 patients [43] and one prospective study with 511 patients [55]. This was partially taken into account in our meta-analysis with the use of the random effects model, which applies less weight to large studies than fixed effect models. The meta-analyses revealed no statistically significant differences of AC failures, intraoperative seizures, new neurological dysfunctions, and the composite outcome (AC failure, intraoperative seizure, mortality) depending on the use of SAS or MAC technique. We found a substantial heterogeneity between the included studies and only the heterogeneity for conversion to GA showed a possible significant connection to the anaesthesia technique in the logistic meta-regression analysis. This analysis suggested significantly more unplanned conversions into GA with the use of SAS than MAC anaesthesia technique. However, this result was mainly depending on one high risk of bias retrospective SAS study with 6 events in 102 patients [57]. Removing this study abolishes the significant difference between the techniques. Of note, two of the patients in this study required conversion into GA due to an air embolism, which was most likely related to the halfsitting patient position and not the used anaesthesia technique [57]. Although air embolism was not analysed in detail in our SR, this was the only study, which reported a failure of AC due to this life-threatening adverse event. Furthermore, the sensitivity analysis, which included only prospective studies, confirmed the weakness of the result obtained by the main metaregression analysis. A significant difference between the used anaesthesia techniques in regard to conversion to GA could not be revealed by the sensitivity analysis anymore. The decision to perform a sensitivity analysis by including only prospective studies and not the largest ones, was justified by the increased risk for confounding in larger studies due to a prolonged study duration. The most studies with more than 100 AC procedures, where performed during 5? [22,31,42,43,45,46,52] or 10?8 years [34,35,37,55,57]. The probability of a continuously same anaesthesia or AC surgery conduction in these observational studies during the large timespans is very low. Our sensitivity analysis did also not reveal any statistical significant difference for the other four outcomes, which were included in the meta-analyses. Of note, the new neurological dysfunction outcome was only presented by one prospective study [38] in the SAS group. Therefore, we could not estimate the proportions for this outcome in the meta-analysis (S1 Fig). However, the main analysis included also only six studies in the SAS group [23,37,38,51,53,57] and the result was dominated by this prospective observational study of Li et al. in a Chinese population [28]. Although 53.8 of the 91 patients exhibited new neurological dysfunctions, these dysfunctions remained permanent only in 1 patient, which suggests that the aim of safe resection was achieved in the longer-term. Furthermore, the generalizability of their results is unclear, due to possible differences in the distribution of the Chinese language areas compared to Non-Chinese people. Therefore we suggest interpreting our result.