Association and predictive performance of inflammatory biomarkers for post-stroke pneumonia: a systematic review and meta-analysis

Stroke remains a leading cause of death and long-term disability worldwide. Post-stroke pneumonia (PSP), one of its most serious complications, lacks robust tools for ultra-early prediction. Inflammatory biomarkers quantify the systemic inflammatory and immunosuppressive response triggered by acute stroke, but their predictive value for PSP remains unclear. We evaluated the association and predictive value of inflammatory biomarkers for PSP. This systematic review was prospectively registered with PROSPERO and conducted per the PRISMA guidelines. We searched PubMed, Web of Science, China National Knowledge Infrastructure (CNKI), and Wanfang Data. Study quality was assessed using the Newcastle–Ottawa Scale (NOS), and prediction-study risk of bias was additionally assessed with PROBAST. Outcomes were: (1) associations of inflammatory markers (continuous variables) with PSP risk; (2) adjusted odds ratios (ORs) per 1-unit increase; and (3) predictive performance, measured by the area under the receiver-operating characteristic curve (AUC). Forty-nine studies involving 22,796 patients were included. Within 24 h of stroke onset and before PSP diagnosis, patients with PSP had higher systemic inflammation response index (SIRI), systemic immune-inflammation index (SII), neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), interleukin-6 (IL-6), lipopolysaccharide-binding protein (LBP), D-dimer, and fibrinogen than controls. In adjusted analyses per 1-unit increase, SIRI and NLR remained significantly associated with PSP; MLR was nominally significant before but not after Benjamini–Hochberg (BH) correction; SII and PLR showed no significant association owing to the scale of composite indices, and LMR owing to statistical imprecision. Prognostic nutritional index (PNI) could not be pooled. For prediction, NLR (AUC 0.78) and MLR (AUC 0.77) yielded the highest pooled AUCs, followed by SII and SIRI (both AUC 0.73). After BH correction, 16 of 21 primary analyses remained significant. Several inflammatory biomarkers were significantly associated with PSP and showed moderate predictive value. Given the moderate discrimination, variability in optimal cut-offs, and substantial residual heterogeneity, findings are hypothesis-generating: whether dynamic biomarker monitoring adds value to clinical prediction tools remains to be established in prospective multicenter studies. Systematic Review Registration: https://www.crd.york.ac.uk/prospero/ ; CRD420251004401.