Online Tool Predicts Mortality Risk Among Patients With Anastomotic Leak

A tool for predicting post-esophagectomy anastomotic leak (AL) mortality was found to perform well, according to study findings published in Diseases of the Esophagus.

Researchers in the Netherlands sourced data from an international retrospective cohort study that collected data from 71 centers in 20 countries. Patients (N=1509) who underwent esophagectomy with gastric tube reconstruction for esophageal or gastroesophageal cancers between 2011 and 2019 and had AL were evaluated for AL-associated mortality at 90 days.

A total of 27 patient characteristics were evaluated by multivariate logistic regression as potential predictors for mortality. An internal bootstrapping strategy was used to validate the prediction tool. Anastomotic leak was defined as a full thickness gastrointestinal defect of the esophagus, anastomosis, staple line, or conduit.

Among the study participants, the median age was 66 (IQR, 59-71) years, 72% had adenocarcinoma, 56% underwent transthoracic intrathoracic anastomosis esophagectomy, 56% had an American Society of Anesthesiologist (ASA) score of 2, and 44% had an Eastern Cooperative Oncology Group (ECOG) Performance Status (PS) of 0.

The overall mortality rate at 90 days was 11.7%.

Out of all considered variables, 16 patient- and leak-related characteristics were included in the model as predictors. The strongest predictors of AL-associated mortality included renal failure at AL diagnosis (odds ratio [OR], 2.16), an ECOG PS of at least 2 (OR, 2.05), an ASA of at least 3 (OR, 1.95), ASA of 2 (OR, 1.80), and AL diagnosis in the intensive, medium, high, or post anesthesia care units (OR, 1.69).

After internally validating the model, mortality was predicted with a c-index of 0.79 (95% CI, 0.75-0.83). It was noted that the model slightly underestimated mortality risk.

The researchers have incorporated this predictive model into an online application where clinicians can obtain the likelihood of patient mortality using simple predictors.

This study may have been limited by not using an independent cohort for validation.

The study authors conclude, “The proposed clinical prediction model combines patient-related and leak-related factors to predict 90-day postoperative mortality in patients with AL. Internal validation showed good performance, but external validation is warranted. This easily accessible tool is useful for clinicians during counselling of patients and their families and may aid identification of high-risk patients at diagnosis of AL.”

Disclosure: Multiple authors declared affiliations with industry. Please refer to the original article for a full list of disclosures.