Novel Clinical Prediction Model: Integrating A2DS2score with 24-hour ASPECTS and Red Cell Distribution Width for EnhancedPrediction of Stroke-Associated Pneumonia following Intravenous Thrombolysis

The authors declare no competing interests.

Background:

Stroke-associated pneumonia (SAP) is a common leading cause of death during the acute phase. The ²2D²2 score has been widely used to predict the risk of SAP. However, 24-hour non-contrast computed tomography-Alberta Stroke Program Early CT Score (NCCT-ASPECTS) and red cell distribution width (RDW) were not included in this scale. The purpose of the present study was to investigate the prognostic added value of combining 24-hour NCCT-ASPECTS and RDW with the ²2D²2 score.

Methods:

A retrospective study of thrombolyzed acute ischemic stroke (AIS) patients from January 2015 to July 2022. Data on ²2D²2 scores, 24-hour NCCT-ASPECTS, and RDW were collected. Three logistic regression models were created: Model A used only the traditional ²2D²2 score; Model B (²2D²2-c) calculated probabilities using a logistic equation; and Model C (combined ²2D²2-MFP) used multivariable fractional polynomial logistic regression and incorporated the ²2D² score, 24-hour NCCT-ASPECTS, and RDW. Ischemic brain lesions in the middle cerebral artery area were assessed using 24-hour NCCT-ASPECTS after completing 24-hour intravenous thrombolysis.

Results:

Among a cohort of 345 thrombolyzed AIS patients, 70 individuals (20.3%) experienced SAP. The area under the receiver operating characteristic (AuROC) of 24-hour NCCT-ASPECTS and RDW were 0.841 and 0.621, respectively. The combined ²2D²2-MFP calculation was significantly superior to the traditional ²2D² score and ²2D²2-c calculation (AuROC 0.917 vs. 0.880, P P=0.026, and 0.917 vs. 0.888, P P=0.024).

Conclusion: This study found that the 24-hour NCCT-ASPECTS and RDW enhanced the predictive value of the A²DS² score for SAP after IV-tPA. The combined A²DS²-MFP model performed excellently in predictive performance, offering robust early SAP detection and potentially improving patient survival. Implementing this novel model in resource-constrained clinical settings could aid clinicians in effective monitoring, enabling risk stratification to guide clinical management.

Protocol.pdf