Twelve randomised controlled trials (RCTs), with a total of 3,809 participants, were included. All trials had a Jadad quality score of 3 points. The funnel plot suggested there may have been some publication bias.
Overall survival: The pooled hazard ratio showed significantly higher survival rates for chemotherapy plus surgery compared to surgery only (hazard ratio 0.78, 95% confidence interval (CI): 0.71, 0.85; 12 RCTs). Of the individual trials, most showed trends for a survival benefit, but only four trials showed a significant result in favour of chemotherapy. No significant heterogeneity was detected in the main meta-analysis (p=0.15).
Subgroup analyses: Significant hazard ratios in favour of chemotherapy were found in all subgroup analyses.
Overall survival was between 13 and 56% for trials from Western countries and 70 to 84% for the Japanese trials (based on ten RCTs with five-year survival rates, one RCT with three-year survival rates, and one RCT with 10-year survival rates).
No significant difference in hazard ratios appeared to exist for subgroups according to percentage of T1 tumours, percent lymph node positive rate, and type of surgery.
Three trials that used oral fluorouracil agents, showed statistically significant results for survival (hazard ratio 0.63, 95% CI: 0.52 to 0.78). The remaining eight intravenous regimens showed no significant result for survival (hazard ratio was 0.80, 95% CI: 0.71 to 0.89).