SDL Index Predicts Stroke-Associated Pneumonia in Patients After Endovascular Therapy

Abstract

Objective: This study aimed to develop and validate a novel index to predict SAP for AIS patients who underwent endovascular treatment. Methods: A study was conducted in an advanced comprehensive stroke center from January 2013 to December 2019 aiming to develop and validate a novel index to predict SAP for AIS patients who underwent endovascular treatment. This cohort consisted of a total of 407 consecutively registered AIS patients who underwent endovascular therapy, which was divided into derivation and validation cohorts. Multiple blood parameters as well as demographic features, vascular risk factors, and clinical features were carefully evaluated in the derivation cohort. The independent predictors were obtained using multivariable logistic regression. The scoring system was generated based on the β-coefficients of each independent risk factor. Results: Ultimately, a novel predictive model: the SDL index (stroke history, dysphagia, lymphocyte count < 1.00 × 10³/μL) was developed. The SDL index showed good discrimination both in the derivation cohort (AUROC: 0.739, 95% confidence interval, 0.678-0.801) and the validation cohort (AUROC: 0.783, 95% confidence interval, 0.707-0.859). The SDL index was well-calibrated (Hosmer-Lemeshow test) in the derivation cohort (P = 0.389) and the validation cohort (P = 0.692). We therefore divided our population into low (SDL index = 0), medium (SDL index = 1), and high (SDL index ≥ 2) risk groups for SAP. The SDL index showed good discrimination when compared with two existing SAP prediction models. Conclusions: The SDL index is a novel feasible tool to predict SAP risk in acute ischemic stroke patients post endovascular treatment.

Keywords: acute ischemic stroke; endovascular therapy; lymphopenia; prediction scoring system; stroke-associated pneumonia.

Figure 1

A study flowchart. Of the 786 patients screened, 407 patients were included in the final analysis. 258 patients were allocated to a derivation cohort and 149 patients in the validation cohort.

Figure 2

Differences in 3-month outcome according to patients with SAP or non-SAP. Among 94 patients with SAP, 80 patients got unfavorable outcome (mRS 3–6). Among 119 non-SAP patients, 46 patients got unfavorable outcome. SAP is significantly associated with unfavorable outcome at 3-month (p < 0.001).

Figure 3

Validation of discrimination in the derivation cohort. ROC, receiver operating characteristic curve; SDL indicates Stroke history, Dysphagia, Lymphocyte count < 1.00 × 10³/μL; A²DS² indicates age, atrial fibrillation, dysphagia, sex, stroke severity (National Institutes of Health Stroke Scale); ISAN, prestroke Independence (modified Rankin scale), Sex, Age, National Institutes of Health Stroke Scale.

Figure 4

Validation of discrimination in the validation cohort. ROC, receiver operating characteristic curve; SDL indicates Stroke history, Dysphagia, Lymphocyte count < 1.00 × 10³/μL; A²DS² indicates age, atrial fibrillation, dysphagia, sex, stroke severity (National Institutes of Health Stroke Scale); ISAN, prestroke Independence (modified Rankin scale), Sex, Age, National Institutes of Health Stroke Scale.

Figure 5

The prevalence of SAP in the derivation, validation cohort, and in total. The error bars indicated 95% confidence interval for the prevalence of SAP in each category. SDL indicates Stroke history, Dysphagia, Lymphocyte count < 1.00 × 10³/μL.