Firstly, 3735 studies were identified by using PubMed, Ovid Medline, Cochrane Library, and Scopus databases. After removing duplicates, 1701 titles and abstracts were screened. Of these, 1275 were excluded from titles and abstracts. Based on the remaining 426 records, 405 articles were eliminated by reading the full text due to a lack of to be analyzed. Finally, 21 articles were included in the final analysis [4, 5, 7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25], as shown in Fig. 1. The detailed literature search strategies in different databases were displayed in Supplementary material 1.

Flow chart presenting the steps of literature search and selection

The detailed characteristics of the studies were shown in Tables 1 and 2. Among the 21 studies included in the analysis, 17 were retrospective observational studies and 4 were randomized controlled studies [4, 5, 7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25]. Overall, 363,470 patients were analyzed in this systematic review and meta-analysis. 228,713 patients received general anesthesia and 134,757 patients received regional anesthesia. The sample sizes of the included studies varied widely, ranging from 129 to 104,088. Regional anesthesia included the use of a neuraxial technique (spinal anesthesia, epidural anesthesia, continuous spinal anesthesia, continuous epidural anesthesia, or combined spinal-epidural anesthesia) with or without the use of a nerve block or multiple nerve blocks. One study did not provide a definition of regional anesthesia [16].

Analysis of the risk of bias for randomized controlled studies and retrospective observational studies are shown in Fig. 2. The authors’ judgments about each risk of bias item for each included study were described. The details were shown in Supplementary material 2 and 3.

Cochrane collaboration risk of bias for A randomized controlled studies and B retrospective observational studies

The in-hospital mortality was examined by 6 retrospective observational studies after hip fracture surgery in adults [8, 10, 11, 15, 16, 21]. 4 studies showed that there was no significant difference in in-hospital mortality in patients receiving either general or regional anesthesia [8, 11, 15, 21]. But Neuman and his colleagues (n = 18,158, general anesthesia = 12,904, regional anesthesia = 5254) revealed that the in-hospital mortality rate in the general anesthesia group was higher than regional anesthesia group [16]. The study of Chu and his colleagues (n = 104,088, general anesthesia = 52,044, regional anesthesia = 52,044) reported a significantly higher incidence of in-hospital mortality in the general anesthesia group [10]. Our meta-analysis, including the above 6 studies, showed a higher in-hospital mortality in the general anesthesia group than in the regional anesthesia group (OR = 1.21; 95% CI 1.13–1.29; P < 0.001, n = 191,511) without heterogeneity (I2 = 0%). The details were shown in Fig. 3.

Forest plots showing pooled effect estimates for in-hospital mortality when comparing general with regional anesthesia. The odds ratio was calculated with a fixed effect method

Thirteen studies investigated the effect of general versus regional anesthesia on 30-day mortality after hip fracture surgery in adult patients including 2 prospective randomized controlled studies and 11 retrospective observational studies [4, 7,8,9, 14, 17,18,19,20, 22,23,24,25]. The study of Radcliff and his colleagues (n = 5683, general anesthesia = 3353, regional anesthesia = 2330) reported a significantly higher risk of 30-day hospital mortality in the general anesthesia group [19]. The other studies showed that there was no significant difference in the 30-day mortality between general and regional anesthesia in the above studies [4, 7,8,9, 14, 17, 18, 20, 22,23,24,25]. Our meta-analysis of these 13 studies showed no statistically significant difference in 30-day mortality (OR = 1.00; 95% CI 0.96–1.05; P = 0.95, n = 163,811) without heterogeneity (I2 = 0%). The subgroup analysis for 2 randomized controlled studies (OR = 0.90; 95% CI 0.41–2.00; P = 0.80, n = 1264) and 9 retrospective observational studies (OR = 1.00; 95% CI 0.95–1.04; P = 0.89, n = 155,797) also indicated no statistical difference in the 30-day mortality between the two groups, as shown in Fig. 4.

Forest plots showing pooled effect estimates for 30-day mortality when comparing general with regional anesthesia. The odds ratio was calculated with a fixed effect method. 4.1.1: The odds ratio of the randomized controlled studies represented a subgroup. 4.1.2: The odds ratio of the retrospective observational studies represented a subgroup

Eight studies investigated the incidence of pneumonia after hip fracture surgery in adults, including 3 randomized controlled studies and 5 retrospective observational studies [4, 7, 8, 12, 13, 16, 18, 21]. Among them, Neuman et al. (n = 18,158, general anesthesia = 12,904, regional anesthesia = 5254), Fields et al. (n = 6628, general anesthesia = 4813, regional anesthesia = 1815) and Shih et al. (n = 335, general anesthesia167, regional anesthesia = 168) found a higher incidence of pneumonia in the general anesthesia group [12, 16, 21]. The other studies revealed no significant difference in the incidence of pneumonia between the two groups [4, 7, 8, 13, 18]. These 8 studies were eligible to be included in the meta-analysis. There was no statistical difference in the incidence of pneumonia between the two groups (OR = 0.93; 95% CI 0.82–1.06; P = 0.28, n = 36,743) without heterogeneity (I2 = 0%). The subgroup analysis for 3 randomized controlled studies (OR = 0.78; 95% CI 0.21–2.91; P = 0.71, n = 1651) and 5 retrospective observational studies (OR = 0.93; 95% CI 0.83–1.06; P = 0.30, n = 35,092) indicated no statistical difference in the incidence of pneumonia between the two groups, as shown in Fig. 5.

Forest plots showing pooled effect estimates for pneumonia when comparing general with regional anesthesia. The odds ratio was calculated with a fixed effect method. 5.1.1: The odds ratio of the randomized controlled studies represented a subgroup. 5.1.2: The odds ratio of the retrospective observational studies represented a subgroup

Three prospective randomized controlled studies and 1 retrospective observational study evaluated the incidence of postoperative delirium between general anesthesia and regional anesthesia in adults. As there was only 1 retrospective study, subgroup analysis was not possible. There was no significant difference in the occurrence of postoperative delirium between general and regional anesthesia [4, 5, 18, 21]. Our meta-analysis of these 4 studies also showed no statistically significant difference between the two groups regarding the incidence of postoperative delirium (OR = 0.94; 95% CI 0.74–1.20; P = 0.61, n = 2861), without heterogeneity (I2 = 39%). The results were shown in Fig. 6.

Forest plots showing pooled effect estimates for delirium when comparing general with regional anesthesia. The odds ratio was calculated with a fixed effect method

We used Stata 12.0 to perform sensitivity analysis using the one-by-one elimination method. For in-hospital mortality, the combined results of those 6 retrospective observational studies [8, 10, 11, 15, 16, 21] were greatly influenced by the study of Chu et al. [