Network meta-analysis of protocol-driven care and laparoscopic surgery for colorectal cancer.

Abstract

Laparoscopic approaches and standardized recovery protocols have reduced morbidity following colorectal cancer surgery. As the optimal regimen remains inconclusive, a network meta-analysis was undertaken of treatments for the development of postoperative complications and mortality. MEDLINE, Embase, trial registries and related reviews were searched for randomized trials comparing laparoscopic and open surgery within protocol-driven or conventional perioperative care for colorectal cancer resection, with complications as a defined endpoint. Relative odds ratios (ORs) for postoperative complications and mortality were estimated for aggregated data. Forty trials reporting on 11 516 randomized patients were included with the network. Open surgery within conventional perioperative care was the index for comparison. The OR relating to complications was 0·77 (95 per cent c.i. 0·65 to 0·91) for laparoscopic surgery within conventional care, 0·69 (0·48 to 0·99) for open surgery within protocol-driven care, and 0·43 (0·28 to 0·67) for laparoscopic surgery within protocol-driven care. Sensitivity analyses excluding trials of low rectal cancer and those with a high risk of bias did not affect the treatment estimates. Meta-analyses demonstrated that mortality risk was unaffected by perioperative strategy. Laparoscopic surgery combined with protocol-driven care reduces colorectal cancer surgery complications, but not mortality. The reduction in complications with protocol-driven care is greater for open surgery than for laparoscopic approaches. Registration number: CRD42015017850 (https://www.crd.york.ac.uk/PROSPERO).