Colon Leakage Score and Preoperative Albumin for Predicting Anastomotic Leakage after Colorectal Cancer Resection in Indonesia: A Predictive Accuracy Study

Abstract

Background: Anastomotic leakage (AL) is the most serious complication of colorectal cancer resection. The Colon Leakage Score (CLS) has no biochemical component and has not been tested in Indonesian patients, many of whom have hypoalbuminemia.

Objective: To measure how accurately CLS >11 combined with preoperative albumin <3.5 g/dL predicts AL, and to quantify what albumin adds to the CLS.

Methods: This retrospective STARD-guided study included 60 adults who underwent colorectal cancer resection with primary anastomosis and no diverting stoma at Dr. Mohammad Hoesin General Hospital, Palembang (January 2022–June 2024). Accuracy measures carry Wilson 95% confidence intervals (CIs). Albumin had not been cross-classified against AL. Every contingency table matching the observed data was therefore enumerated, with Firth regression across 19,977 admissible datasets.

Results: Thirteen patients (21.7%; 95% CI 13.1–33.6) developed AL, and 43 (71.7%) were hypoalbuminemic. Six of seven composite-positive patients leaked, as did 7 of 53 composite-negative patients (odds ratio 35.6; exact 95% CI 3.6–1850.2; p < 0.001). Sensitivity was 46.2% (23.2–70.9), below the 80% benchmark (p = 0.007). Specificity was 97.9% (88.9–99.6), positive predictive value 85.7% (48.7–97.4) and negative predictive value 86.8% (75.2–93.5). Seven of eight CLS-high patients were hypoalbuminemic; albumin reclassified one. CLS >11 was associated with AL in all 15 admissible tables (odds ratio 19.3–53.7). The albumin association (0.23–15.5) could not be identified.

Conclusion: The composite works as a high-specificity rule-in criterion, and the CLS carries its discrimination. It misses more than half of all leaks. Widespread hypoalbuminemia is a separate nutritional target that deserves prospective evaluation.