Association of delayed swallow screening with stroke-associated pneumonia and in-hospital outcomes in acute stroke patients in the neurocritical care unit.

BACKGROUND: Stroke-associated pneumonia (SAP) is a frequent and devastating complication after acute stroke. Although early dysphagia screening is strongly recommended by guidelines, its implementation is often delayed in the Neurological Intensive Care Unit (Neuro-ICU). The precise continuous dose-response relationship between screening delay time and SAP risk has not been adequately quantified. METHODS: We retrospectively enrolled 300 consecutive acute stroke patients admitted to the Neuro-ICU, excluding those requiring intubation within 24 h to avoid reverse causation. Hierarchical multivariable logistic regression, adjusting for consciousness level (Glasgow Coma Scale, GCS) and systemic inflammation (categorical Neutrophil-to-Lymphocyte Ratio, NLR) while excluding clinical mediators, evaluated the association between delayed screening (>24 h) and SAP. Restricted cubic spline (RCS) analysis modeled the non-linear dose-response relationship with 4 h as the reference point. Subgroup analyses used Firth's penalized regression to address separation bias. RESULTS: Overall SAP incidence was 48.3%. In the fully adjusted model, delayed screening remained an independent risk factor for SAP (adjusted OR = 3.69, 95% CI: 2.12-6.51, p < 0.001). The RCS curve showed a significant non-linear trajectory (p for non-linearity <0.05), with SAP odds remaining stable during the early buffer window but increasing steeply after the 24-h threshold. This effect was consistent across ischemic stroke (adjusted OR = 3.10, 95% CI: 1.59-6.10) and hemorrhagic stroke (adjusted OR = 4.69, 95% CI: 1.80-13.00) subgroups. CONCLUSION: Delayed dysphagia screening beyond 24 h was associated with a non-linear increase in the odds of SAP. The 24-h mark may represent a clinically relevant risk threshold, although residual confounding cannot be excluded.