OBJECTIVE  To analyze the risk factors for hospital-acquired pneumonia (HAP) in patients with acute stroke, and develop and validate a risk prediction model for early identification of high-risk patients.

METHODS  A total of 1 481 patients with acute stroke admitted to the Stroke Center of the First Affiliated Hospital of Shihezi University from Jan. 2021 to Dec. 2023 were enrolled. Patients were randomly divided into training and validation sets at a 7∶3 ratio. Logistic regression was used to identify HAP risk factors in patients with acute stroke and construct a prediction model. The model was evaluated with receiver operating characteristic (ROC) curves and visualized through nomogram.

RESULTS  Nasogastric tube feeding (OR=6.953, 95%CI: 4.340−11.140), acid suppressant use (OR=2.197, 95%CI: 1.410−3.425), stroke type (OR=37.991, 95%CI: 4.031−358.066), surgical intervention (OR=4.351, 95%CI: 2.448−7.730), GCS score (OR=1.057, 95%CI: 1.006−1.110) and blood glucose (OR=1.057, 95%CI: 1.009−1.108) were identified as risk factors for HAP in patients with acute stroke, while increased albumin-globulin ratio (OR=0.173, 95%CI: 0.097−0.311) served as a protective factor. The prediction model and nomogram were constructed based on these factors, and internal validation was performed. The prediction model demonstrated good performance: the area under the ROC curve (AUC) was 0.868 (95%CI: 0.843−0.894), with a sensitivity of 85.70% and a specificity of 74.00%. The model exhibited excellent calibration, as indicated by the Hosmer-Lemeshow test (P=0.674).

CONCLUSION  This study developed a prediction model for HAP risk in patients with acute stroke, providing clinicians with an effective and practical tool for early high-risk identification, which holds significant value for HAP prevention and reduction.