Sex-specific non-linear association between admission neutrophil-to-lymphocyte ratio and coagulopathy in traumatic brain injury: a large retrospective cohort study

Systemic inflammation is associated with coagulation abnormalities after traumatic brain injury (TBI), but the sex-specific and potentially non-linear relationship between admission neutrophil-to-lymphocyte ratio (NLR) and international normalized ratio (INR) remains unclear. This single-center retrospective cohort study included a cross-sectional analysis of contemporaneously measured admission NLR and INR and secondary exploratory analyses of subsequent in-hospital outcomes among 2394 TBI patients admitted between March 2018 and September 2024. The association between admission NLR and INR was examined using multivariable linear regression. Restricted cubic spline analysis using the Model 3 covariate set was the primary method for assessing non-linearity, whereas segmented regression using the same adjustment set was used only as a secondary exploratory analysis. The extended Model 4, which additionally included brain surgery, ventilator therapy, and concurrent injury as available severity-related proxies and treatment-related indicators, was evaluated as a sensitivity analysis. Analyses were stratified by sex. Missingness diagnostics, multiple-imputation sensitivity analysis, bootstrap estimation of the candidate inflection point, and exploratory prediction analyses including discrimination, calibration, Brier score, decision curve analysis, net reclassification improvement, and integrated discrimination improvement were performed. Among 2,394 patients, 1707 were male and 687 were female. In the primary Model 3 adjustment, admission NLR was positively associated with INR in males (β = 0.0016, 95% CI 0.0008–0.0024, P < 0.001) and females (β = 0.0017, 95% CI 0.0005–0.0029, P = 0.005). Restricted cubic spline analysis showed evidence of non-linearity in males ( P for non-linearity < 0.001), but not in females (P for non-linearity = 0.375). A formal sex-by-spline interaction test supported a sex-specific difference in curve shape (P for interaction = 0.016). As a secondary exploratory analysis, segmented regression in male patients identified an internally derived candidate inflection point at NLR 7.15; however, the bootstrap 95% CI remained wide (6.16–17.17), indicating substantial instability in the estimated location of the inflection point. Male patients with NLR > 7.15 had higher rates of composite poor outcome and survivor severe disability, whereas mortality alone did not differ significantly. In an exploratory comparison, adding NLR and INR to a limited available-covariate model was associated with higher apparent discrimination (AUC 0.645 vs. 0.759; optimism-corrected AUC 0.755) and a lower Brier score (0.1513 vs. 0.1725). However, the limited available-covariate model lacked established neurological and imaging predictors of TBI outcome. Admission NLR was associated with INR in patients with TBI, with evidence of a sex-specific non-linear pattern in males. The candidate NLR inflection point of 7.15 was highly uncertain and should be regarded as an exploratory finding rather than a reproducible biological threshold or clinically applicable cutoff. The prognostic utility of NLR and INR remains exploratory because established TBI severity measures were unavailable, and external validation is required.