Ruling out anastomotic leakage after digestive resection: diagnostic accuracy of postoperative day 3 and day 5 C-reactive protein in a single-centre cohort including several anastomotic sites

Aims: To evaluate postoperative C-reactive protein (CRP) for ruling out anastomotic leakage (AL) after digestive resection in a cohort including several anastomotic sites.Patients & methods: Prospective single-centre diagnostic accuracy study (STARD 2015) of 82 consecutive patients with an unprotected digestive anastomosis (January 2024–August 2025). CRP was measured on postoperative days (POD) 1, 3 and 5. No analysis was formally prespecified. The primary analysis excluded leaks diagnosed on or before POD 5; a strict analysis also excluded four leaks with an undated diagnosis.Results: AL occurred in 12 patients (14.6%). CRP discriminated on POD 3 (area under the curve [AUC] 0.83; 95% CI: 0.71–0.95) and POD 5 (AUC 0.88; 0.78–0.98) but not on POD 1 in the 71 patients with an available value (AUC 0.59). In the primary pre-diagnosis analysis (79 patients, 9 leaks), an exploratory, data-derived POD 5 cutoff of 127 mg/L had a sensitivity of 77.8% (40.0–97.2) and a negative predictive value of 96.4% (87.7–99.6); in the strict analysis (75 patients, 5 leaks), sensitivity was 80.0% and NPV 98.2%. In the full cohort, sensitivity was 83.3% (apparent) and 76.5% (optimism-corrected). Both false negatives followed gastric resection.Conclusion: A low POD 5 CRP makes AL unlikely, particularly after non-gastric resection, but these data-derived thresholds are exploratory, do not support any change in discharge practice and require external validation.