Prognostic Nutritional Index and Postoperative Complications After Colorectal Surgery: A Retrospective Cohort Study

Background: The prognostic nutritional index (PNI) is an inexpensive marker derived from serum albumin and peripheral lymphocyte count. Although low PNI has been associated with adverse surgical outcomes, its incremental value beyond basic clinical variables and the stability of commonly used thresholds remain uncertain in heterogeneous colorectal surgical populations. We evaluated the association between preoperative PNI and postoperative complications and examined whether adding PNI improved a basic clinical model. Methods: This retrospective single-center cohort included 205 consecutive adults undergoing colorectal surgery between January 2024 and March 2026. PNI was calculated as albumin (g/L) + 5 × lymphocyte count (109/L). PNI was modeled primarily as a continuous predictor; PNI < 45 was examined secondarily. ROC analysis, multivariable logistic regression, nested-model comparison, calibration assessment, bootstrap internal validation, complete-case sensitivity analysis, and exploratory subgroup analyses were performed. Results: Fifty-six patients (27.3%) had PNI < 45. Postoperative complications occurred in 67 patients (32.7%) overall and were more frequent with PNI < 45 (51.8% vs. 25.5%; RR 2.03, 95% CI 1.40–2.95; OR 3.14, 95% CI 1.65–5.95; Fisher p < 0.001). PNI alone showed modest discrimination (AUC 0.659); the Youden cutoff was 45.45 (sensitivity 50.7%, specificity 78.3%). In the complete-case multivariable model (n = 191), each 5-point decrease in PNI was associated with higher odds of complications (OR 1.52, 95% CI 1.22–1.90; p < 0.001). Adding continuous PNI to age, BMI, and operative approach increased AUC from 0.663 to 0.711 and improved model fit (likelihood-ratio χ2 = 15.89, p < 0.001). Bootstrap resampling showed substantial cutoff variability (95% percentile interval 32.05–52.00). Conclusions: Lower preoperative PNI was associated with postoperative complications and added discriminatory information to a basic clinical model. However, its stand-alone discrimination was modest, the data-derived cutoff was unstable on bootstrap resampling, and residual confounding and clinical heterogeneity limit causal or treatment-directed interpretation. PNI should therefore be considered a risk marker rather than a stand-alone decision rule. Prospective interventional studies in well-defined patient groups are needed to determine whether PNI-guided nutritional optimization improves outcomes.