Trajectory of serial neutrophil-to-lymphocyte ratio predicts neurological outcome after out-of-hospital cardiac arrest

Aim To evaluate whether serial neutrophil-to-lymphocyte ratio (NLR) trajectory patterns and individual time-point NLR values predict neurological outcome after out-of-hospital cardiac arrest (OHCA). Methods This retrospective study analyzed 414 comatose OHCA survivors treated with targeted temperature management at a tertiary center (2009–2024). NLR was calculated at 0, 24, 48, and 72 h after return of spontaneous circulation. The primary outcome was 6-month Cerebral Performance Category dichotomized as good (1–2) or poor (3–5). Patients with complete serial data ( n = 277) were classified into trajectory groups. A linear mixed-effects model tested group-by-time interaction, and DeLong tests compared discriminative performance across time points. Results Among 414 patients, 131 (31.6%) achieved good outcomes. NLR did not differ between groups at 0 or 24 h but diverged at 72 h (median 6.0 vs 13.9; P < 0.001; area under the curve 0.785), driven by failure of absolute lymphocyte count recovery in poor-outcome patients (median 0.86 vs 1.28 × 10³/µL; P < 0.001). Trajectory analysis identified early resolution (60.6% good outcome), sustained elevation (31.9%), and late rise (10.5%; adjusted odds ratio 22.6; P < 0.001). Log-transformed NLR at 72 h was independently associated with outcome (odds ratio 5.13; 95% confidence interval 2.88–9.14), and its addition to the clinical model significantly improved discrimination (area under the curve from 0.824 to 0.891; DeLong P = 0.001). In a combined model with neuron-specific enolase at 48 h, NLR at 72 h showed a borderline independent association ( P = 0.050). Conclusions Failure of NLR to resolve by 72 h was associated with poor neurological outcome in this cohort. As NLR is derived from routine blood counts at no additional cost, it may serve as an accessible complement to specialized prognostic biomarkers, particularly in settings where neuron-specific enolase is unavailable. These findings require prospective multicenter validation.