Five RCTs (n=717) were included in the review. Sample sizes ranged from 74 to 237 patients. The mean Jadad score was 3; limitations arose from justification of the sample size, absence of allocation concealment and double-blinding, and the subjectivity of reporting return of gastrointestinal function.
The pooled analysis of three RCTs showed that time to flatus was shorter in the group without decompression (WMD 0.16, 95% CI -0.02 to 0.34 days, p=0.081). Pooled analysis of the same three RCTs showed that time to starting oral diet was shorter in the group without decompression (WMD 0.43, 95% CI 0.23 to 0.62 days, p<0.001). There was no statistically significant heterogeneity in either analysis.
There were no significant differences between patients who did or did not receive decompression for the other outcomes assessed: anastomotic leakage (p=0.610; five RCTs); pulmonary complications (p=0.161; five RCTs); length of hospital stay (p=0.153; three RCTs); morbidity rates (p=0.442; four RCTs); and mortality rates (p=0.650; five RCTs). There was no significant heterogeneity in any of the analyses. Results for anastomotic leakage were similar when stratified for subtotal and total gastrectomy. There were no reported adverse events that related to tube insertion, pneumothorax and oesophageal perforation in any of the included studies. Sensitivity testing for outcomes deemed to be the most important (anastomotic leakage, pulmonary complications and morbidity) did not substantially change the original results.
The funnel plot did not show evidence of publication bias.