Predictors of Anastomotic Leak in Elderly Patients After Colectomy: Nomogram-Based Assessment From the American College of Surgeons National Surgical Quality Program Procedure-Targeted Cohort

Abstract

BACKGROUND: Elderly patients undergoing colorectal surgery have increasingly become under scrutiny by accounting for the largest fraction of geriatric postoperative deaths and a significant proportion of all postoperative complications, including anastomotic leak. OBJECTIVE: This study aimed to determine predictors of anastomotic leak in elderly patients undergoing colectomy by creating a novel nomogram for simplistic prediction of anastomotic leak risk in a given patient. DESIGN: This study was a retrospective review. SETTINGS: The database review of the American College of Surgeons National Surgical Quality Improvement Program was conducted at a single institution. PATIENTS: Patients aged ≥65 years who underwent elective segmental colectomy with an anastomosis at different levels (abdominal or low pelvic) in 2012-2013 were identified from the multi-institutional procedure-targeted database. MAIN OUTCOME MEASURES: We constructed a stepwise multiple logistic regression model for anastomotic leak as an outcome; predictors were selected in a stepwise fashion using the Akaike information criterion. The validity of the nomogram was externally tested on elderly patients (≥65 years of age) from the 2014 American College of Surgeons National Surgical Quality Improvement Program colectomy-targeted database. RESULTS: A total of 10,392 patients were analyzed, and anastomotic leak occurred in 332 (3.2%). Of the patients who developed anastomotic leak, 192 (57.8%) were men (p < 0.001). Based on unadjusted analysis, factors associated with an increased risk of anastomotic leak were ASA score III and IV (p < 0.001), chronic obstructive pulmonary disease (p = 0.004), diabetes mellitus (p = 0.003), smoking history (p = 0.014), weight loss (p = 0.013), previously infected wound (p = 0.005), omitting mechanical bowel preparation (p = 0.005) and/or preoperative oral antibiotic use (p < 0.001), and wounds classified as contaminated or dirty/infected (p = 0.008). Patients who developed anastomotic leak had a longer length of hospital stay (17 vs 7 d; p < 0.001) and operative time (191 vs 162 min; p < 0.001). A multivariate model and nomogram were created. LIMITATIONS: This study was limited by its retrospective nature and short-term follow-up (30 d). CONCLUSIONS: An accurate prediction of anastomotic leak affecting morbidity and mortality after colorectal surgery using the proposed nomogram may facilitate decision making in elderly patients for healthcare providers.