Nonlinear threshold effect of neutrophil-to-lymphocyte ratio for predicting stroke-associated pneumonia in acute ischemic stroke: a retrospective cohort study in afebrile patients at admission

Stroke-associated pneumonia (SAP) is a common and serious complication after acute ischemic stroke (AIS), leading to increased morbidity, mortality, and healthcare burden. The neutrophil-to-lymphocyte ratio (NLR) has emerged as a promising biomarker of systemic inflammation, but its predictive value and potential nonlinear threshold effect for SAP remain unclear. We conducted a retrospective cohort study involving 1,473 consecutive afebrile patients with acute ischemic stroke (AIS) who were admitted to a tertiary stroke center between January 2019 and December 2023. The neutrophil-to-lymphocyte ratio (NLR) was calculated from complete blood counts obtained at admission and log₂-transformed for statistical analysis. Stroke-associated pneumonia (SAP) was diagnosed according to modified criteria from the Centers for Disease Control and Prevention (CDC) and established consensus guidelines. The association between log₂-transformed NLR and SAP was assessed using multivariable logistic regression, adjusting for 22 potential confounders. Threshold effects were explored through piecewise regression analysis. Evaluate predictive performance using receiver operating characteristic (ROC) curve and DeLong test. Among 1,473 patients, 339 (23.0%) developed SAP. Patients with SAP had significantly higher NLR levels (median 4.20 vs. 2.39, P < 0.001). After multivariable adjustment, each doubling of NLR was associated with a 51% increased risk of SAP (OR 1.51, 95% CI 1.28–1.79, P < 0.001). Segmented regression identified an inflection point at NLR = 1.79 (log₂-NLR = 0.84). Below this threshold, NLR showed no significant association with SAP (OR 0.42, P = 0.054), while above the threshold, the association was significant (OR 1.44, 95% CI 1.21–1.71, P < 0.001). The combination model of NLR + CRP + WBC achieved the highest discriminatory ability (AUC = 0.875). Subgroup analyses showed consistent associations across different patient characteristics (all P for interaction > 0.05). Elevated NLR is independently associated with SAP in AIS patients, with a threshold effect at NLR = 1.79. NLR may serve as a simple and readily available biomarker for SAP risk stratification. Patients with NLR ≥ 1.79 may benefit from enhanced monitoring and preventive interventions.