Major Article

The pneumonia score: A simple grading scale for prediction of pneumonia after acute stroke

The current medical practice is more responsive rather than proactive, despite the widely recognized value of early disease detection, including improving the quality of care and reducing medical costs. One of the cornerstones of early disease detection is clinically actionable predictions, where predictions are expected to be accurate, stable, real-time and interpretable. As an example, we used stroke-associated pneumonia (SAP), setting up a transformer-encoder-based model that analyzes highly heterogeneous electronic health records in real-time. The model was proven accurate and stable on an independent test set. In addition, it issued at least one warning for 98.6 % of SAP patients, and on average, its alerts were ahead of physician diagnoses by 2.71 days. We applied Integrated Gradient to glean the model's reasoning process. Supplementing the risk scores, the model highlighted critical historical events on patients' trajectories, which were shown to have high clinical relevance.

The study aims to systematically evaluate the risk factors of pulmonary infection in elderly patients with acute stroke.

PubMed, CENTRAL, CINAHL, Web of Science, Embase, and four Chinese databases (CNKI, SinoMed, VIP, and Wanfang databases) for studies involving risk factors of pulmonary infection in elderly patients with acute stroke were searched. Then, two researchers independently read the article titles and abstracts to screen the literature, extracted relevant research data, and evaluated the methodological quality. Finally, a meta-analysis was performed.

In total, fifteen studies were included, with medium and high-grade quality. Meta-analysis results showed: age [OR = 1.70, 95% CI (1.27, 2.29), P = 0.0004], type of stroke [OR = 1.30, 95% CI (1.21, 1.40), P < 0.00001], conscious disturbance [OR = 2.27, 95% CI (1.44, 3.58), P = 0.0004], dysphagia [OR = 3.24, 95% CI (2.06, 5.12), P < 0.00001], diabetes mellitus [OR = 2.35, 95% CI (1.23, 4.48), P = 0.010], hypertension [OR = 2.05, 95% CI (1.83, 2.31), P < 0.0001], chronic obstructive pulmonary disease (COPD) [OR = 2.69, 95% CI (1.90, 3.81), P < 0.00001], hyperlipidemia [OR = 1.29, 95% CI (1.19, 1.39), P < 0.00001], invasive procedure [OR = 3.37, 95% CI (2.30, 4.94), P < 0.00001], hospital stays [OR = 1.41, 95% CI (1.22, 1.62), P < 0.00001], bedridden time [OR = 1.51, 95% CI (1.36, 1.68), P < 0.00001], and National Institute of Health Stroke Scale (NIHSS) score [OR = 1.67, 95% CI (1.02, 2.75), P = 0.04] were independent risk factors. Glasgow Coma Scale (GCS) score was not a risk factor. However, the relationship between atrial fibrillation, smoking history, and pulmonary infection in elderly patients with acute stroke needs further proof.

Age, type of stroke, conscious disturbance, dysphagia, diabetes mellitus, hypertension, COPD, hyperlipidemia, invasive procedure, hospital stays, bedridden time, and NIHSS score were risk factors for pulmonary infection in elderly patients with acute stroke.

Few studies have assessed the sex-specific and age-specific risk of aspiration pneumonia (AP) in patients with stroke and evaluated whether mental disorders may increase this risk. In this population-based cohort study, we investigated the sex-specific and age-specific risk of AP in association with stroke and the joint effects of stroke and mental disorders on the risk of AP.

We included 23,288 patients with incident stroke admitted between 2005 and 2017 and 68,675 matched nonstroke controls. Information on mental disorders was obtained from medical claims data within the 3 years before the stroke incidence. Cox proportional hazards models considering death as a competing risk event were constructed to estimate the hazard ratio of AP incidence by the end of 2018 associated with stroke and selected mental disorders.

After ≤14 years of follow-up, AP incidence was higher in the patients with stroke than in the controls (11.30/1000 vs. 1.51/1000 person-years), representing a covariate-adjusted subdistribution hazard ratio (sHR) of 3.64, with no significant sex difference. The sHR significantly decreased with increasing age in both sexes. Stratified analyses indicated schizophrenia but not depression or bipolar affective disorder increased the risk of AP in the patients with stroke.

Compared with their corresponding counterparts, the patients with schizophrenia only, stroke only, and both stroke and schizophrenia had a significantly higher sHR of 4.01, 5.16, and 8.01, respectively. The risk of AP was higher in younger stroke patients than those older than 60 years. Moreover, schizophrenia was found to increase the risk of AP in patients with stroke.

Swallowing disorders (dysphagia) is common in stroke patients. However, the epidemiology of post-stroke dysphagia (PSD) is poorly described. We herein synthesize the data of eligible studies on occurrence rate of dysphagia in Asian populations with stroke.

We searched the electronic databases (PubMed, Embase and Web of Science) to collect the studies on the prevalence of PSD. We used the Newcastle–Ottawa Scale (NOS) to estimate the quality of studies. The pooled dysphagia occurrence rate was obtained in Asian stroke patients.

40 studies (including 43 observations) from 2318 initial references were selected in the synthetic analysis. The pooled occurrence rate of dysphagia in post-stroke patients was 36.3% (95% CI, 33.3%–39.3%). Meta-regression analysis showed that the “country” and “developing level” may influence the pooled occurrence rate of PSD.

Dysphagia is common in Asian post-stroke patients. Our meta-analysis may raise concern about evaluating and managing dysphagia in stroke patients.

Pneumonia is a common complication after stroke, causing an increased length of hospital stay and death. Therefore, the timely and accurate prediction of post-stroke pneumonia would be highly valuable in clinical practice. Previous pneumonia risk score models were often built on simple statistical methods such as logistic regression. This study aims to investigate post-stroke pneumonia prediction models using more advanced machine learning algorithms, specifically deep learning approaches.

Using a hospital’s electronic health record(EHR) data from 2007-2017, 13,930 eligible patients with acute ischaemic stroke (AIS) were identified to build and evaluate the models (85% of the patients were used for training, and 15% were used for testing). In total, 1012 patients (7.23%) contracted pneumonia during hospitalization. A number of machine learning methods were developed and compared to predict pneumonia in the stroke population in China. In addition to the classic methods (i.e., logistic regression (LR), support vector machines (SVMs), extreme gradient boosting (XGBoost)), methods based on multiple layer perceptron (MLP) neural networks and recurrent neural network (RNNs) (i.e., attention-augmented gated recurrent unit (GRU)) are also implemented to make use of the temporal sequence information in electronic health record (EHR) systems. Prediction models for pneumonia were built for two time windows, i.e., within 7 days and within 14 days after stroke onset. In particular, pneumonia occurring within the 7-day window is considered highly associated with stroke (stroke-associated pneumonia, SAP).

The attention-augmented GRU model achieved the best performance based on an area under the receiver operating characteristic curve (AUC) of 0.928 for pneumonia prediction within 7 days and an AUC of 0.905 for pneumonia prediction within 14 days. This method outperformed the other machine learning-based methods and previously published pneumonia risk score models. Considering that pneumonia prediction after stroke requires a high sensitivity to facilitate its prevention at a relatively low cost (i.e., increasing the nursing level), we also compared the prediction performance using other evaluation criteria by setting the sensitivity to 0.90. The attention-augmented GRU achieved the optimal performance, with a specificity of 0.85, a positive predictive value (PPV) of 0.32 and a negative predictive value (NPV) of 0.99 for pneumonia within 7 days and a specificity of 0.82, a PPV of 0.29 and an NPV of 0.99 for pneumonia within 14 days.

The deep learning-based predictive model is feasible for stroke patient management and achieves the optimal performance compared to many classic machine learning methods.

Stroke is the second leading cause of death globally that predisposed to sepsis. Therefore, this study was aimed to assess the risk factors and epidemiologic features of sepsis in ischemic stroke patients admitted to ICUs.

Throughout this prospective study, we investigated all severe ischemic stroke patients admitted to ICUs of Namazi and Ali-Ashghar Hospitals in Shiraz. After ICU admission and diagnosing stroke by a neurologist according to NIHSS (National Institute of Health Stroke Scale) criteria, sepsis work-up was performed in all patients suspected to have sepsis. Then the incidence of sepsis and its risk factors in ICU admitted stroke patients were determined.

A total of 149 patients were screened in this study. The mean age of the participants was 65.37 ± 15.40 years old and 57.4% of them were male. Hypertension was the most common coexistent disease (74.6%) in stroke patients. Seventy-six patients (62.3%) were diagnosed with sepsis and pneumonia was the most common infection leading to sepsis in stroke patients. Our data showed significant differences between two groups in terms of APACHE-IV score (P < 0.001), NIHSS and APS (P < 0.001) before ICU admission (P < 0.001) and NIHSS at admission (P < 0.001); however, age (P = 0.07) and sex (P = 0.17) were not significantly different between the groups. Logistic regression analysis displayed that severe stroke (NIHSS = 21–42, OR = 49.09) and severe loss of consciousness (GCS < 8, OR = 27.95) at admission were the most essential predictive factors for sepsis after ischemic stroke.

This study showed that ICU patients with severe ischemic stroke were more susceptible to sepsis during the hospital course.