From colorectal to general surgeon in the management of left colonic perforation: A cohort study

Background: Management of left colonic perforation in emergency depends largely upon the attending surgeon. The primary endpoint of this observational, retrospective study analyses surgical technique chosen by the colorectal specialized (CS) or general surgeon (GS) and changes over time. Materials and methods: Interventions for left colonic perforation from 2004 to 2015 are grouped for CS or GS. Type of operation (Hartmann (HP), primary anastomosis (RPA) +/- covering ileostomy (IL)), year, Peritonitis Severity Score (PSS), morbidity, mortality, anastomotic dehiscence and stoma closure were recorded. Results: 190 patients were included. CS performed RPA +/- IL in 83 pts (74.1%) and HP in 29 pts (25.9%) while GS performed RPA +/- IL in 26 pts (33.3%) and HP in 52 pts (66.7%), (p < 0.001). CS performed over time more RPA with covering ileostomy to the detriment of HP. No differences were observed between the two surgeon-groups in terms of overall morbidity and mortality. Anastomotic dehiscence was higher among GS (20% vs 4.8%, p = 0.046). Mortality after HP overtrumped RPA (26.8% versus 11.0%, p = 0.009). Regression analysis showed that HP's probability increased 3.7 times by GS, 2.3 times by each PSS point and decreased 32.5% every forthcoming year (p < 0.001). A multinomial logistic model illustrates evolution of surgical management over time, CS leading towards extension of reconstructive techniques, subsequently adopted by GS. Conclusions: CS attempt bowel reconstruction in more patients than GS in left colonic perforation without differences in overall postoperative morbidity or mortality. CS introduced covering IL to further indicate primary anastomosis avoiding HP. GS stepwise adopted this management although results are improved by CS. These findings favor primary anastomosis with/without covering ileostomy in left colonic perforation in selected patients where PSS can be used as a tool to discriminate best candidates.