Work

The admission neutrophil-to-lymphocyte ratio improves early prediction of poor functional outcome beyond the modified Rankin Scale in viral encephalitis: a cohort study

Year: 2026

Type: article

Abstract: Background The admission modified Rankin Scale (mRS) strongly predicts outcome in viral encephalitis (VE) but does not capture the systemic inflammatory response of severe disease. We tested whether the neutrophil-to-lymphocyte ratio (NLR) improves early prediction of poor functional outcome beyond the admission mRS. Methods We studied 200 patients admitted with VE at a single center; the primary outcome was a poor functional outcome at discharge (mRS 3-6). Among routine admission blood tests we identified the markers most strongly associated with poor outcome and selected the leading one, the NLR, by LASSO logistic regression. Incremental value beyond the admission mRS was assessed primarily by a likelihood-ratio test (LRT) for the nested models and supported by the change in discrimination (ΔAUC), reclassification, and decision curve analysis. We predefined a stratified analysis of whether the NLR predicted outcome within bands of admission mRS. The final model was internally validated with 1000 bootstrap resamples, without repeating the initial marker selection. Results Forty patients (20.0%) had a poor outcome. After correction, 12 of 16 candidate variables were associated with poor outcome, dominated by inflammatory indices. The admission mRS alone discriminated poor outcome with an apparent AUC of 0.884; adding NLR raised it to 0.912. The gain was significant by LRT ( χ 2 = 8.5, p = 0.004; NLR odds ratio 1.14 per unit, 95% CI 1.03–1.25), with a ΔAUC of 0.029 (95% CI 0.012–0.046). In the mRS 3–5 stratum, where 39/40 poor outcomes occurred, NLR remained incrementally predictive ( n = 133; odds ratio 1.15, 95% CI 1.04–1.27, p = 0.007). The two-variable model was well calibrated (bootstrap-corrected slope 0.97) and showed greater net benefit than the mRS alone. Conclusion In VE, adding the admission NLR to the admission mRS modestly but significantly improves early prediction of poor outcome at discharge, with the increment persisting within admission-severity strata (mRS 3–5). A simple two-variable nomogram (a development-stage tool) is presented for early in-hospital risk flagging. Further analysis suggested that this gain was primarily driven by increased confidence in identifying patients with favorable outcomes, rather than by earlier detection of those who would eventually have poor prognosis.

Source: Frontiers in Neurology

DOI: 10.3389/fneur.2026.1948837