Primary anastomosis vs nonrestorative resection for perforated diverticulitis with peritonitis: a prematurely terminated randomized controlled trial

Abstract

Aim This randomized controlled trial (RCT) was performed to test the hypothesis that adverse event rates following primary anastomosis (PRA) are not inferior to those following nonrestorative colon resection for perforated diverticulitis with peritonitis. Method Patients admitted for perforated diverticulitis with peritonitis were randomly assigned to PRA (left colon resection with PRA and loop ileostomy) or nonrestorative colon resection (left colon resection with end colostomy). The endpoint was adverse events defined as mortality and morbidity following PRA or nonrestorative colon resection and stoma reversal. The estimated sample size was 300 patients in each study arm (alpha 0.10; 90% power). Results During a 9‐year period, 90 patients were randomly assigned to undergo PRA or nonrestorative colon resection in 14 centres in eight countries. Thirty‐four PRA patients were comparable to 56 nonrestorative colon resection patients for age (P = 0.481), gender (P = 0.190), APACHE III (P = 0.281), Hinchey stage III vs IV (P = 0.394) and Mannheim Peritonitis Index (P = 0.145). There were no differences in operating time (P = 0.231), surgeries performed at night (P = 0.083), open vs laparoscopic approach (P = 0.419) and litres of peritoneal irrigation (P = 0.096). There was no significant difference in mortality (2.9 vs 10.7%; P = 0.247) and morbidity (35.3 vs 46.4%; P = 0.38) following PRA or nonrestorative colon resection. After a similar lag time (P = 0.43), 64.7% of PRA patients and 60% of nonrestorative colon resection patients underwent stoma reversal (P = 0.659). Adverse event rates following stoma reversal differed significantly after PRA and reversal of nonrestorative resection (4.5 vs 23.5%; P = 0.0589). Conclusion No conclusions may be drawn on preference of one treatment over another from this RCT because it was prematurely terminated following accrual of 15% of its sample size.