Predictors of underlying carcinoma in patients with suspected acute diverticulitis.

Abstract

A subset of patients who undergo colon resection for suspected diverticulitis will unexpectedly be found to have cancer. We hypothesized that a subset of variables could be used to help predict a diagnosis of cancer preoperatively. National Surgical Quality Improvement Program data (2012-2018) were used to identify all patients with a preoperative diagnosis of diverticulitis who had unexpected cancer using postoperative staging data. Key characteristics of the study groups were compared with χ² tests and multivariate logistic regression modeling. A total of 17,368 patients were identified with an operative indication of acute diverticulitis. Of these, 164 (0.94%) had an unexpected postoperative diagnosis of cancer. Most cancer patients had locally advanced tumors (T1-2: 15%; T3: 39%; T4: 45%), and 37.1% had positive lymph nodes. Rates of margin positivity and inadequate lymph node harvest were 1.2% and 15.9%, respectively. In bivariate analyses, cancer patients had increased age (P < .01), decreased albumin (P < .001), and increased rates of preoperative anemia (P = .01), sepsis (P < .01), and weight loss (>10% in 6 months) (P < .001). The only variables significantly associated with cancer in multivariate regression analysis were sepsis (odds ratio 2.14, 95% confidence interval [1.3-3.6]; P < .01), weight loss (odds ratio 2.31, 95% confidence interval [1.1-4.4]; P = .01), and preoperative albumin level (odds ratio 0.64, 95% confidence [0.45-0.92]; P < .01). An unexpected postoperative diagnosis of cancer occurs in a small percentage of patients with suspected diverticulitis. Surgeons should have a high index of suspicion for cancer in patients with sepsis or malnutrition.