Stroke-associated pneumonia in Japanese acute care settings: incidence and preliminary validation of risk prediction scores

Abstract Background Stroke-associated pneumonia (SAP) significantly impacts mortality and functional outcomes in acute stroke care. While multiple risk prediction scores have been developed internationally, their performance in Japanese healthcare settings—characterized by high nurse-to-patient ratios and systematic multidisciplinary protocols—remains unexplored. This study aimed to determine SAP incidence in Japanese acute stroke care and validate established risk prediction scores. Methods We conducted a retrospective cohort study of consecutive patients with first-ever acute stroke admitted within 7 days of onset to a 199-bed acute care hospital with a dedicated 24-bed stroke care unit (April 2022–March 2023). SAP was diagnosed using modified CDC criteria requiring radiographic evidence. Three validated risk scores (A 2 DS 2 , Pneumonia score, ISAN) were calculated and compared using receiver operating characteristic analysis. Independent predictors were identified through multivariable logistic regression using directed acyclic graph-based variable selection. Results Among 847 patients analyzed (median age 75 years, 54.2% male), SAP developed in 22 patients (2.6%, 95% CI 1.5–3.7). Most cases (77.3%) occurred within 72 h. All three scores demonstrated good discrimination: A 2 DS 2 achieved the highest AUROC of 0.825 (95% CI 0.78–0.88), followed by Pneumonia score and ISAN (both AUROC 0.798). No statistically significant differences were observed between scores. An A 2 DS 2 cutoff ≥ 6 provided optimal balance (sensitivity 86.4%, specificity 65.8%, NPV 99.5%). Independent predictors included male sex (OR 3.87, 95% CI 1.41–10.61), NIHSS score (OR 1.08 per point, 95% CI 1.03–1.12), dysphagia (OR 5.31, 95% CI 1.16–24.34), and mechanical ventilation (OR 5.33, 95% CI 1.28–22.18). Conclusions SAP incidence in Japanese acute stroke care (2.6%) was substantially lower than international reports, likely reflecting high nursing standards and systematic preventive protocols. Despite low baseline risk, the A 2 DS 2 score demonstrated excellent discrimination and high negative predictive value, enabling reliable risk stratification. The predominance of early SAP onset supports intensive monitoring during the critical first 72 h. These findings support implementation of A 2 DS 2 -based risk stratification protocols in Japanese stroke care settings.