Literature Identification
Ovid Medline, Ovid EMBASE, CENTRAL and Web of Science were systematically searched through December 12, 2018. The search resulted in 2,885 articles. After initial evaluation, 597 studies were removed for being duplicates, 2,204 for being irrelevant (as determined by reading the abstracts), and 67 studies were excluded for reasons determined by reading the full text (Additional File 2). 16 studies[18-33] were included in the final meta-analysis. Fig. 1 shows the work flow for the selection of studies.
Study and ERAS Characteristics
Studies were included in the meta-analysis when they adhered to consensus guidelines for ERAS protocols [6, 34]. The basic characteristics of the included studies are shown in Table 1. Table 2 shows the details of the key elements of ERAS protocols for all of the studies, including the type of disease and the surgical site. It also summarizes ERAS protocol items, and it details the primary endpoints and follow-up times. Two studies reported comparing laparoscopic to open surgery[19, 23]; the other reported on patients aged 45 to 74 years and 75 to 89 years[21]. Finally, a total of 19 RCTs from 16 studies, included 1,830 patients, of whom 907 were in the ERAS arm and 923 were in the control arm, were found to be studies that compared ERAS to conventional care. Gastric cancer surgery was reported in 14 RCTs from 11 studies, and esophagectomy was reported in 5 studies.
Quality assessment
The results of the quality assessment are shown in Additional File 3. It is notable that none of the 19 RCTs can blind the surgeon or the patient during the surgery. In addition, all 19 RCTs were quite similar in their risk of bias. Nine of the RCTs did not report random sequence generation, only 2 RCTs had blinded outcome assessments.
Primary Outcomes
Lung infection
Fifteen RCTs including 1,496 patientsreported postoperative LI. Pooling the resultssuggested that ERAS protocols significantly decreased the incidence of postoperative LI compared to conventional care (Fig. 2, RR = 0.50, 95%CI: 0.33 to 0.75). The test of heterogeneity (I2 = 0%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that the incidence of LI after surgery was significantly decreased by using ERAS protocols (Fig. 2, RR = 0.57, 95%CI: 0.34 to 0.95, I2 = 0%). Among RCTs performed in the area of esophagectomies, the incidence of LI after surgery was significantly reduced by using ERAS protocols (Fig. 2, RR = 0.41, 95%CI: 0.21 to 0.79, I2 = 0%). However, based on subgroup analyses of the surgical procedure and scope of gastrectomy, Table 3 showed that there were no statistical differences in all subgroup analyses of LI.
Urinary tract infection
Ten RCTs included 824 patients diagnosed with postoperative UTI. Pooling the results suggested that ERAS protocols did not increase the incidence of urinary tract infection compared to conventional care (Fig. 3, RR = 0.59, 95% CI: 0.31 to 1.11). The test of heterogeneity (I2 = 0%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that the incidence of postoperative UTI was not increased by ERAS protocols (Fig. 3, RR = 0.60, 95%CI: 0.31 to 1.16, I2 = 0%). There were too few RCTs about esophagectomies to calculatethe incidence of postoperative UTI in this area. However, the results of subgroup analyses of UTI based on the surgical procedure and scope of gastrectomy were no statistical differences in Table 3.
Surgical site infection
Fifteen RCTs included 1,555 patients who reported postoperative SSI. Pooling the resultssuggested that ERAS protocols did not increase the incidence of postoperative SSI compared to conventional care (Fig. 4, RR = 0.80, 95%CI: 0.47 to 1.37). The test of heterogeneity (I2 = 0%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that the incidence of SSI after surgery was not increased by ERAS protocols (Fig. 4, RR = 0.86, 95%CI: 0.46 to 1.61, I2 = 0%). Among RCTs performed in the area of esophagectomiesy surgery, the incidence of SSI was not increased (Fig. 4, RR = 0.67, 95%CI: 0.24 to 1.86, I2 = 0%). However, Table 3 demonstrated that there were no statistical differences in all subgroup analyses of SSI based on the surgical procedure and scope of gastrectomy.
Postoperative anastomotic leakage
Fourteen RCTs including 1,414 patients reported postoperative anastomotic leakage. Pooling the results suggested that ERAS protocols did not increase the incidence of postoperative anastomotic leakage compared to conventional care (Fig. 5, RR = 0.80, 95%CI: 0.44 to 1.45). The test of heterogeneity (I2 = 0%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that the incidence of anastomotic leakage after surgery was not increased by ERAS protocols (Fig. 5, RR = 1.27 95%CI: 0.60 to 2.66, I2 = 0%). Among RCTs performed in the area of esophagectomies, postoperative anastomotic leakage (Fig. 5, RR = 0.31, 95%CI: 0.09 to 1.01, I2 = 2%) was not increased by ERAS protocols. However, there were no statistical differences in all subgroup analyses of postoperative anastomotic leakage based on the surgical procedure and scope of gastrectomy in Table 3.
Postoperative ileus
Thirteen RCTs (1,313 patients) reported postoperative ileus. Pooling the results suggested that ERAS protocols did not increase the incidence of postoperative ileus compared to conventional care (Fig. 6, RR = 1.43, 95%CI: 0.78 to 2.65). The test of heterogeneity (I2 = 0%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that the incidence of ileus after surgery was not increased by ERAS protocols (Fig. 6, RR = 1.56, 95%CI: 0.82 to 2.97, I2 = 0%). In the area of esophagectomies, postoperative anastomotic leakage (Fig. 6, RR = 1.56, 95%CI: 0.82 to 2.97, I2 = 0%) was not increased by ERAS protocols. However, the results that there were no statistical differences were found in all subgroup analyses of postoperative anastomotic leakage based on the surgical procedure and scope of gastrectomy (Table 3).
Secondary Outcomes
Postoperative length of stay
Eighteen RCTs (1,716 patients) reported PLOS. Pooling the results suggested that ERAS protocols significantly decreased the postoperative length of stay compared to conventional care (Fig. 7, MD = -2.53, 95%CI: -3.42 to -1.65). The test of heterogeneity (I2 = 97%) indicated that there was a high degree of heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that PLOS was significantly reduced by ERAS protocols (Fig. 7, MD = -1.77, 95%CI = -2.29 to -1.24, I2 = 85.8%). Among RCTs performed in the area of esophagectomies, PLOS was significantly reduced by ERAS protocols (Fig. 7, MD = -5.12, 95%CI: -5.40 to -4.83, I2 = 34%). Based on the surgical procedure and scope of gastrectomy, all subgroup analyses in PLOS, including radical gastrectomy (MD = -1.79, 95%CI: -2.59 to -0.99), distal gastrectomy (MD = -1.64, 95%CI: -2.60 to -0.33), laparoscopic surgery (MD = -1.95, 95%CI: -2.99 to -0.91), and open surgery (MD = -1.83, 95%CI: -3.01 to -0.66), showed statistical differences by ERAS protocols in Table 3.
The duration of intestinal function recovery
Thirteen RCTs (1,072 patients) reported the time until the first postoperative flatus. Pooling the results suggested that ERAS protocols significantly decreased the time until the first postoperative flatus compared to conventional care (Fig. 8, MD = -0.65, 95% CI: -0.85 to -0.45). The test of heterogeneity (I2 = 82%) indicated that there was significant heterogeneity among these trials.
Five RCTs (539 patients) reported the time until the first postoperative defecation. Pooling the results suggested that ERAS protocols significantly decreased the time until the first postoperative defecation compared to conventional care (Fig. 9, MD = -1.10, 95% CI: -1.74 to -0.47). The test of heterogeneity (I2 = 87%) indicated that there was significant heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated that ERAS was associated with a significant reduction in the time until the first postoperative flatus (Fig. 8, MD = -0.68, 95%CI: -0.85 to -0.45, I2 = 74%) and until the first postoperative defecation (Fig. 9, MD = -1.11, 95%CI: -1.85 to -0.36, I2 = 90%). However, for the esophagectomies, ERAS protocols significantly decreased the time until the first postoperative defecation (Fig. 9, MD = -1.09, 95%CI: -1.88 to -0.30, I2 = Not applicable), but it wasn’t decreased the time the first postoperative flatus (Fig. 8, MD = -0.62, 95%CI: -1.32 to 0.09, I2 = 84%). Based on the surgical procedure and scope of gastrectomy, the subgroup analyses of first postoperative flatus, including radical gastrectomy (MD = -0.75, 95%CI: -1.09 to -0.41), distal gastrectomy (MD = -0.45, 95%CI: -0.62 to -0.28), and open surgery (MD = -0.599, 95%CI: -1.08 to -0.27), showed statistical differences by ERAS protocols, but laparoscopic surgery (MD = -0.81, 95%CI: -2.04 to 0.43) wasn’t statistical differences in Table 3. Subgroup analyses of first postoperative defecation, including radical gastrectomy (MD = -1.63, 95%CI: -2.79 to -0.47) and open surgery (MD = -1.05, 95%CI: -1.45 to -0.65), showed statistical differences by ERAS protocols, but laparoscopic surgery (MD = -1.36, 95%CI: -3.05 to 0.34) wasn’t statistical differences in Table 3.
Readmission rates
Eleven RCTs (1,211 patients) reported postoperative readmission rates. Pooling the results suggested that ERAS protocols didnot increase postoperative readmission rates compared to conventional care (Fig. 10, RR = 1.29, 95%CI: 0.79 to 2.12). The test of heterogeneity (I2 = 4%) indicated that there was little heterogeneity among these trials.
Among RCTs performed in the area of gastric surgery, analysis indicated readmission was significantly increased by ERAS protocols (Fig. 10, RR = 1.99, 95%CI: 1.06 to 3.73, I2 = 0%). Among RCTs performed in the area of esophagectomies, readmission rates were not increased by ERAS protocols (Fig. 10, RR = 0.63, 95%CI: 0.14 to 2.80, I2 = 44%). However, the results that there were no statistical differences were found in all subgroup analyses of readmission rates based on the surgical procedure and scope of gastrectomy (Table 3).
Publication bias
There was no evidence of asymmetry in these Doi plots (LFK index < |2|) of the risk of postoperative LI, UTI, SSI and postoperative anastomotic leakage. However, the obvious publication bias was observed in postoperative ileus.