Seven RCTs with 614 patients were included.
The randomisation method was only reported in 2 studies, none of the included studies used blinding, and 2 studies violated the intention-to-treat analysis.
The funnel plot (which was not shown) did not show any evidence of publication bias.
1-year survival: chemotherapy resulted in significantly lower mortality than no chemotherapy; the RR was 0.69 (95% CI: 0.60, 0.81, p<0.00001; Q=3.53, P>0.10). The NNT was 5.4 (95% CI: 3.9, 9.2), with a risk difference of -0.1835 (95% CI: -0.2586, -0.1084, p<0.00001).
2-year survival: there was no significant difference between the intervention and control in terms of the RR of 2-year mortality (RR 0.93, 95% CI: 0.87, 1.00, p=0.053). The risk difference was -0.0617 (95% CI: -0.11, -0.0035).
Sensitivity analysis for 1-year mortality results: all of the subgroups appeared to benefit from chemotherapy, except the CEA-only patients (1 study), for which there was insufficient power to exclude a benefit.
Quality of life: this was evaluated in 4 studies by a variety of measures. The trials showed either maintenance of, or improvement in, the scores of the chemotherapy arm compared to the initial observational arm.
Toxicity: this was not quantitatively documented by many of the studies. Where documented, the toxicity was generally mild to moderate.