S22 The Diagnostic Accuracy of Neutrophil-to-Lymphocyte Ratio (NLR) for Assessing Severity of Acute Cholecystitis in the Preoperative Period: A Bayesian Latent Class Meta-Analysis

Introduction: Severe acute cholecystitis carries high risks of morbidity and mortality, yet no universally agreed upon standard method to assess its severity at admission, especially in low-resource emergency settings. The neutrophil-to-lymphocyte ratio (NLR), a simple and inexpensive blood marker, reflects the intensity of inflammation and may indicate complications like gangrene or organ dysfunction. Its low cost and quick turnaround make NLR a practical tool for early triage. This meta-analysis aimed to evaluate the diagnostic accuracy of NLR in predicting severe cholecystitis. Methods: Five databases were searched to identify relevant studies. Given the lack of a uniform reference standard, a Bayesian latent class meta-analysis model was employed which estimates sensitivity and specificity without assuming a perfect gold standard and accounts for uncertainty in the true disease status. Optimal threshold for NLR was determined using a hierarchical summary receiver operating characteristic model. Results: A total of 27 studies met the inclusion criteria. Three reference standards were considered i.e., Tokyo guidelines, histopathology, and presence of organ complications. Bayesian latent class meta-analysis produced posterior median estimates for sensitivity of 74.7 (95% CrI: 67.1–83.6) and specificity of 76.8 (95% CrI: 71.2–82.6), indicating moderate to high diagnostic performance across diverse thresholds and reference standards. The subgroup analysis based on reference standard revealed similar results in all subgroups. The cut-point parameter (0.392; 95% CrI: -0.291–4.267) and accuracy parameter (2.55; 95% CrI: 1.869–8.837) suggested substantial between-study variability in the NLR thresholds used. The optimal NLR threshold determined via hierarchical summary receiver operating characteristic model was 9.250, yielding a sensitivity of 72.2% (95% CI: 56.3–84.0) and specificity of 70.3% (95% CI: 59.4–79.3). Conclusion: NLR showed moderate diagnostic performance in predicting severe acute cholecystitis; however, significant between-study heterogeneity reduces the certainty of the evidence. While further high-quality studies are needed to confirm its clinical utility, given its moderate effectiveness, NLR may still be valuable as part of a broader diagnostic algorithm, alongside markers like CRP and ultrasonography, for effective preoperative triage.