Remember me
Background:
Acute ischemic stroke (AIS) remains one of the major contributors to mortality and disability worldwide. Stroke-associated infection (SAI) is one of the most frequent complications following AIS and has a substantial impact on clinical outcomes, being closely linked to unfavorable prognosis. This study aimed to provide a comprehensive description of SAI, identify independent risk factors, and develop a predictive nomogram for its early identification.
Methods:
This study included 836 AIS patients of the Dalian Single-center Study on Intravenous Thrombolysis for Ischaemic Stroke (DATIS) cohort who received recombinant tissue-plasminogen activator-induced thrombolysis at Central Hospital of Dalian University of Technology between January 2018 and November 2021. Patients were divided into a training cohort (n = 586, 70%) and a validation cohort (n = 250, 30%). Composition and economic features of SAI was explored. Independent risk factors were identified using univariate, multivariate, and multimodal logistic regression analyses. A predictive nomogram was then developed based on these independent risk factors. Model performance was assessed with receiver operating characteristic curves, and calibration curves.
Results:
Among the 836 enrolled patients, 168 (20.1%) developed SAI. Composition of 168 patients with SAI were: 99 pulmonary infections (58.93%), 44 upper respiratory tract infections (26.19%), 15 urinary tract infection (8.93%), 2 gastrointestinal tract infections (1.19%), 1 periodontal infection (0.60%), 1 conjunctival infection (0.60%), and 1 erysipela (0.60%). In addition, 5 patients (2.98%) had multi-site infections (4 pulmonary plus urinary tract infection, 1 pulmonary plus gastrointestinal tract infection). Compared with non-infected patients, the SAI group experienced a significantly longer median hospitalization duration [9 days, IQR (7, 10) vs. 8 days, IQR (7, 9), p < 0.001] and incurred higher median inpatient medical costs [28114.04 RMB, IQR (23230.12, 33379.85) vs. 22292.84 RMB, IQR (19203.53, 25999.63), p < 0.001]. Five variables—higher modified Rankin Scale at admission, male sex, prolonged prothrombin time, elevated blood urea nitrogen and lower thyroid-stimulating hormone—were independent risk factors for SAI. The nomogram constructed based on above predictors achieved an area under the curve of 0.80 in the training cohort and 0.72 in the validation cohort. Calibration curves supported the model’s performance.
Conclusion:
This prospective cohort study comprehensively described composition and economic features, identified risk factors and developed predictive nomogram for SAI in AIS patients receiving intravenous rt-PA. Early identification of high-risk patients may facilitate targeted interventions, potentially reducing infection-related complications and improving clinical outcomes.
1 IntroductionAcute ischemic stroke (AIS) stands as one of the leading causes of disability and death worldwide. Intravenous recombinant tissue-type plasminogen activator (rt-PA) has been approved as the first-line therapy for AIS in the United States, Europe, and China. However, stroke-associated infection (SAI), a common complication of AIS, has been shown to significantly worsen patient prognosis (Rocco et al., 2013). Although several studies focussed on the immune responses predisposing to SAI - immunodepression was found to play a decisive role (Tuz et al., 2024; Roth et al., 2021), the clinical factors predisposing to SAI are still not well defined. Hospital-associated infections are associated with prolonged hospitalization, exacerbation of pre-existing medical conditions, and impaired functional recovery (Fluck et al., 2024), highlighting the critical need for early prediction and prevention of this complication.
The incidence of post-stroke infections has been reported to reach up to 30% (95% CI 24–36%), with pulmonary infections (10, 95% CI 9–10%) and urinary tract infections (UTI) being the most common types (Westendorp et al., 2011). Other types of infections, such as upper respiratory tract, gastrointestinal, periodontal, conjunctival, and skin (erysipelas) infections, have also been reported in AIS patients, although less frequently (Fluck et al., 2024; Grau et al., 2004). Predictive studies have also focused mostly on stroke-associated pneumonia (SAP) and UTI. Clinical score, e.g., Ischemic Stroke-Associated Pneumonia, and machine learning were developed specifically for SAP prediction (Suda et al., 2018a; Smith et al., 2015; Xie et al., 2025). Post-stroke dysphagia, which was considered main contributor to SAP, was confirmed to have risk factors of older age, higher NIHSS, and right-hemispheric stroke (Krekeler et al., 2024). Prediction for UTI has been focused on severe AIS and associated with Foley catheter retention (Jitpratoom and Boonyasiri, 2023). Furthermore, epidemiologic investigations such as that by Krekeler et al. (2024) have identified risk factors for post-stroke dysphagia (a known contributor to respiratory infection).
Currently, comprehensive studies elucidating composition and economic characteristics of SAI - including but not limited to SAP and UTI - and developing clinically practical prediction model are scarce. This prospective cohort aims to elucidate SAI with more clinical details, identify independent risk factors for SAI, and to develop a nomogram prediction model, which enables individualized risk assessment at the bedside. Consequently, it facilitates to support early identification of high-risk individuals, to optimize post-thrombolysis monitoring, and ultimately to improve patient outcomes by enabling timely interventions.
2 Materials and methods2.1 Study design and participantsThis study includes AIS patients who received intravenous thrombolysis with rt-PA and were admitted to the Department of Neurology of Central Hospital of Dalian University of Technology between January 2018 and November 2021. This study is part of the prospective Dalian Single-center Study on Intravenous Thrombolysis for Ischaemic Stroke (DATIS) cohort study that is continuously recruited at the Central Hospital. The Central Hospital of Dalian University of Technology is a major hospital in the city of Dalian, an 8 million inhabitant city in the North-East of China. It therefore has broad access to AIS patients. The study protocol has been registered at ChiCTR2400089803.
The inclusion criteria were as follows: Admission between January 2018 and November 2021; age ≥ 18 years; confirmed diagnosis of AIS by a neurologist in accordance with the Chinese Guidelines for the Diagnosis and Treatment of Acute Ischemic Stroke issued in 2023 (Liu et al., 2023); according to the indications for intravenous thrombolysis, the patients and his family signed the informed consent for thrombolysis and received intravenous thrombolysis with rt-PA.
The exclusion criteria were as follows: Presence of infection within 3 days before stroke onset; taking antibiotics, steroids, immunosuppressants and other drugs before admission; bridging therapy with intravenous rt-PA combined with subsequent endovascular therapy; comorbid tumors and immune system disorders; pregnant women, lactating and preparing for pregnancy; incomplete clinical data.
Clinical judgment of SAI: The disease was diagnosed as acute ischemic stroke by a clinical physician, and infections involving any organ system developed that occurred during hospitalization after disease onset. The diagnostic criteria for SAI were based on internationally accepted standards for defining healthcare-associated infections, specifically the CDC’s National Healthcare Safety Network criteria (Horan et al., 2008; Garner et al., 1988).
2.2 Data collectionA comprehensive panel of 42 clinical and biochemical variables was collected at admission, covering demographic characteristics, neurological symptoms and impairment, vascular risk factors, and routine laboratory indices (hematologic, coagulation, metabolic, renal, hepatic, and thyroid parameters). The complete list and descriptive statistics of these variables are presented in Table 1 (Baseline Characteristics).
VariablesTraining cohort (n = 586)Baseline characteristics: training vs. validation cohorts.
This figure presents the baseline characteristics of patients in the training cohort (n = 586) and validation cohort (n = 250). Data include demographic variables such as age, sex, and medical history, as well as clinical variables including NIHSS, mRS, hospitalization duration, and biochemical variables and indices. The data are presented as medians (P25, P75) and percentages, with p-values for group comparisons. Significant differences between the two cohorts are highlighted. SAI, stroke-associated infection; NI, no infection.
2.3 Statistical analysisQuantitative data following a normal distribution were compared using the t-test and expressed as mean ± standard deviation. Non-normally distributed data were compared using non-parametric rank-sum tests and expressed as medians and interquartile ranges. Categorical variables are expressed as counts and percentages and compared using the chi-square test. Univariate logistic regression was performed to screen potential risk factors for SAI, and variables with p < 0.05 were entered into a multivariate logistic regression model to identify independent predictors (reported as ORs with 95% CIs). To further assess robustness, a series of progressively adjusted multimodal logistic regression models were constructed. Model 1 included basic demographic and lifestyle factors: age, sex, smoking, alcohol use, hypertension, diabetes, and dyslipidemia. Model 2 added vascular comorbidities, including coronary heart disease and previous stroke, to the variables in Model 1. Model 3 further incorporated neurological severity and laboratory markers: National Institutes of Health Stroke Scale (NIHSS) at admission, mRS at admission, homocysteine (HCY), white blood cell count (WBC), neutrophil count (NEUT), neutrophil-to-lymphocyte ratio (NLR), PT, BUN, creatinine clearance (Ccr), aspartate aminotransferase (AST), free triiodothyronine (FT3), TSH. A nomogram was constructed from the final model to predict individual SAI risk, incorporating an optimal cut-off point (determined by maximizing the Youden index on the ROC curve) for risk stratification. Model performance was evaluated by discrimination AUC, and calibration (calibration plot and Hosmer-Lemeshow test). SAI was further categorized into pulmonary, urinary tract, multisite, and other infections for descriptive analysis. Analyses were conducted using SPSS 26.0 and R software, with two-sided p < 0.05 considered significant.
3 Results3.1 Patient selection and cohort divisionA patient categorization flowchart was created, depicting the enrollment of 836 patients receiving rt-PA-induced thrombolysis in the study (Figure 1). These patients were divided into a training cohort (n = 586, 70%) and a validation cohort (n = 250, 30%) at a ratio of 7:3. The training cohort was utilized for developing the prediction model, while the validation cohort was employed for model validation.
Flow diagram of patient enrollment for SAI in rt-PA-treated AIS patients. SAI, Stroke-associated infection; NI, No infection; rt-PA, Recombinant tissue-type plasminogen activator.
3.2 Clinical and economic features of SAI3.2.1 Composition and distribution of SAIIn this study, a total of 836 patients were enrolled, with 668 patients (79.1%) presenting no infection (NI) and 168 patients (20.10%) with confirmed SAI during hospitalization, as illustrated in Figure 2A. Composition of 168 patients with SAI were: 99 patients with pulmonary infection (58.93%), 44 patients with upper respiratory tract infection (26.19%), 15 patients with UTI (8.93%), 2 patients with gastrointestinal tract infection (1.19%), 1 patient with periodontal infection (0.60%), 1 patient with conjunctival infection (0.60%), 1 patient with erysipela (0.60%) and 5 patients with multi-site infection (2.98%), including 4 patients with pulmonary infection plus UTI and 1 patient with pulmonary infection plus gastrointestinal tract infection (Figure 2B).
Composition and temporal dynamics of SAI in rt-PA-treated AIS patients. This figure presents the composition and temporal dynamics of SAI in rt-PA-treated AIS patients. (A) Shows the proportion of SAI and NI patients in the overall cohort of 836 patients. (B) Illustrates the distribution of infections types among 168 SAI patients, and multi-site infections (5, 2.98%). The 5 patients with multi-site infections included 4 cases of pulmonary infections combined with urinary tract infections, and 1 case of pulmonary infections combined with gastrointestinal tract infections. (C) Depicts the daily number of new SAI cases from admission (Day 1, defined as hospital admission) throughout the hospitalization period. (D) Illustrates the temporal dynamics of specific infection types (pulmonary, upper respiratory tract, urinary tract, and gastrointestinal tract), plotting the daily count of new cases against time post-admission (Day). SAI, Stroke-associated infection; NI, no infection; rt-PA, recombinant tissue-type plasminogen activator.
As shown in Figure 2C, among 168 SAI patients, the total number of infections presented a distinct temporal pattern. The number of infections was the highest on day 1 (0–24 h after admission), with 48 cases, then decreased steadily to 13 cases on day 4, followed by a slight rebound to 22 cases on day 5. After day 5, the number of infections continued to decline, maintaining a low level (<5 cases) from day 8 onwards. The dynamic changes in the number of SAI patients with different infection types over the disease course are shown in Figure 2D. Pulmonary infection was the most common type in the early stage (day 1), with 33 cases, followed by a gradual decrease. Upper respiratory tract infection showed a fluctuating trend, peaking at day 1 (11 cases) and day 5 (10 cases). Urinary tract infection and gastrointestinal tract infection occurred less frequently, with the highest number of cases being 5 and 1, respectively, and remained at low levels throughout the observation period.
3.2.2 Length of hospital stay and economic assessmentFor length of hospital stay and economic outcomes, the SAI group had both a significantly longer median hospitalization duration [9 days, IQR (7, 10)] and higher median inpatient costs [28114.04 RMB, IQR (23230.12, 33379.85)] compared with the NI group [8 days, IQR (7, 9); 22292.84 RMB, IQR (19203.53, 25999.63)], with p < 0.001 for both comparisons (Table 2).
VariablesSAI (n = 168)Univariate logistic regression of health economics in SAI patients.
OR, odds ratio; CI, confidence interval. This figure presents the results of a one-way logistic regression analysis examining length of hospital stay (LOS) and hospitalization costs between SAI and NI patients. The analysis includes median values (M) and interquartile ranges (P25, P75) for both groups. The odds ratios (OR) with 95% confidence intervals (CI) and p-values are reported for each variable. Both LOS and hospitalization costs show significant differences between the two groups, with p-values of 0.000, indicating statistical significance. SAI, stroke-associated infection; NI, no infection. *p < 0.01.
3.3 Risk factors for SAI3.3.1 Comparison of baseline characteristics between cohortsThe baseline characteristics of patients in the training and validation cohorts are summarized in Table 1. The two cohorts were well balanced across demographic, clinical, and laboratory variables. These findings indicate that the training and validation cohorts were well matched.
3.3.2 Univariate analysis of potential predictorsUnivariate logistic regression identified following variables significantly associated with SAI (p < 0.05), including NIHSS at admission, mRS at admission, age, sex, smoking, alcoholism, and certain laboratory variables, including HCY, WBC, NEUT, NLR, PT, BUN, Ccr, AST, FT3, and TSH (Table 3). These variables, all of which had variance inflation factor values < 10, were entered into multivariate logistic regression to determine independent predictors of SAI.
VariablesSAI (n = 168)
Comments (0)