ACDD 4 score: A simple tool for assessing risk of pneumonia after stroke

A Novel Machine Learning Model for Predicting Stroke-Associated Pneumonia After Spontaneous Intracerebral Hemorrhage

BACKGROUND

Pneumonia is one of the most common complications after spontaneous intracerebral hemorrhage (sICH), i.e., stroke-associated pneumonia (SAP). Timely identification of targeted patients is beneficial to reduce poor prognosis. So far, there is no consensus on SAP prediction, and application of existing predictors is limited. The aim of this study was to develop a machine learning model to predict SAP after sICH.

METHODS

We retrospectively reviewed 748 patients diagnosed with sICH and collected data from 4 dimensions-demographic features, clinical features, medical history, and laboratory tests. Five machine learning algorithms-logistic regression, gradient boosting decision tree, random forest, extreme gradient boosting, and category boosting-were used to build and validate the predictive model. We also applied recursive feature elimination with cross-validation to obtain the best feature combination for each model. Predictive performance was evaluated by area under the receiver operating characteristic curve.

RESULTS

SAP was diagnosed in 237 patients. The model developed by category boosting yielded the most satisfactory outcomes overall with area under the receiver operating characteristic curves in the training set and test set of 0.8307 and 0.8178, respectively.

CONCLUSIONS

The incidence of SAP after sICH in our center was 31.68%. Machine learning could potentially provide assistance in the prediction of SAP after sICH.

Machine learning-based survival analysis approaches for predicting the risk of pneumonia post-stroke discharge

BACKGROUND

Post-stroke pneumonia (PSP) is common among stroke patients. PSP occurring after hospital discharge continues to increase the risk of poor functional outcomes and death among stroke survivors. Currently, there is no prediction model specifically designed to predict the occurrence of PSP beyond the acute stage of stroke. This study aimed to explore the use of machine learning (ML) methods in predicting the risk of PSP after hospital discharge.

METHODS

This study analyzed data from 5,754 hospitalized stroke patients. The dataset was randomly divided into a training set and a holdout test set, with a ratio of 80:20. Several clinical and laboratory variables were utilized as predictors and different ML algorithms were employed to model time-to-event data. The ML model's predictive performance was compared to existing risk-scoring systems. A model-agnostic method based on Shapley additive explanations was utilized to interpret the ML model.

RESULTS

The study found that 5.7% of the study patients experienced pneumonia within one year after discharge. Based on repeated 5-fold cross-validation on the training set, the random survival forest (RSF) model had the highest C-index among the various ML algorithms and traditional Cox regression analysis. The final RSF model achieved a C-index of 0.787 (95% confidence interval: 0.737-0.840) on the holdout test set, outperforming five existing risk-scoring systems. The top three important predictors were the Glasgow Coma Scale score, age, and length of hospital stay.

CONCLUSIONS

The RSF model demonstrated superior discriminative ability compared to other ML algorithms and traditional Cox regression analysis, suggesting a non-linear relationship between predictors and outcomes. The developed ML model can be integrated into the hospital information system to provide personalized risk assessments.

A predictive model using risk factor categories for hospital-acquired pneumonia in patients with aneurysmal subarachnoid hemorrhage

Objectives

To identify risk factors for hospital-acquired pneumonia (HAP) in patients with aneurysmal subarachnoid hemorrhage (aSAH) and establish a predictive model to aid evaluation.

Methods

The cohorts of 253 aSAH patients were divided into the HAP group (n = 64) and the non-HAP group (n = 189). Univariate and multivariate logistic regression were performed to identify risk factors. A logistic model (Model-Logit) was established based on the independent risk factors. We used risk factor categories to develop a model (Model-Cat). Receiver operating characteristic curves were generated to determine the cutoff values. Areas under the curves (AUCs) were calculated to assess the accuracy of models and single factors. The Delong test was performed to compare the AUCs.

Results

The multivariate logistic analysis showed that the age [p = 0.012, odds ratio (OR) = 1.059, confidence interval (CI) = 1.013-1.107], blood glucose (BG; >7.22 mmol/L; p = 0.011, OR = 2.781, CI = 1.263-6.119), red blood distribution width standard deviation (RDW-SD; p = 0.024, OR = 1.118, CI = 1.015-1.231), and Glasgow coma scale (GCS; p < 0.001, OR = 0.710, CI = 0.633-0.798) were independent risk factors. The Model-Logit was as follows: Logit(P) = -5.467 + 0.057 * Age + 1.023 * BG (>7.22 mmol/L, yes = 1, no = 0) + 0.111 * RDW-SD-0.342 * GCS. The AUCs values of the Model-Logit, GCS, age, BG (>7.22 mmol/L), and RDW-SD were 0.865, 0.819, 0.634, 0.698, and 0.625, respectively. For clinical use, the Model-Cat was established. In the Model-Cat, the AUCs for GCS, age, BG, and RDW-SD were 0.850, 0.760, 0.700, 0.641, and 0.564, respectively. The AUCs of the Model-Logit were insignificantly higher than the Model-Cat (Delong test, p = 0.157). The total points from -3 to 4 and 5 to 14 were classified as low- and high-risk levels, respectively.

Conclusions

Age, BG (> 7.22 mmol/L), GCS, and RDW-SD were independent risk factors for HAP in aSAH patients. The Model-Cat was convenient for practical evaluation. The aSAH patients with total points from 5 to 14 had a high risk for HAP, suggesting the need for more attention during treatment.

A Novel Index in the Prediction of Pneumonia Following Acute Ischemic Stroke

BACKGROUND

The aim of our study was to search for predictive factors and to develop a model (index) for the risk of pneumonia following acute ischemic stroke.

MATERIAL AND METHODS

This study is an analysis of prospectively collected data from the neurology department of a district general hospital in Poland, comprising 1001 patients suffering from an acute ischemic stroke. Based on the medical data, the formula for the prediction of pneumonia was calculated.

RESULTS

Multivariate assessment for pneumonia occurrence was performed using the new PNEUMOINDEX. The study showed a significant increase in pneumonia risk with an increasing PNEUMOINDEX (OR non-adjusted = 2.738, p < 0.001). After accounting for age and comorbidities as confounders, the effect of the Index on pneumonia changed marginally (OR = 2.636, p < 0.001).

CONCLUSIONS

This study presents factors that show a significant association with the occurrence of pneumonia in patients with acute ischemic stroke. The calculated PNEUMOINDEX consists of data obtained at admission, namely NYHA III and IV heart failure, COPD, generalized atherosclerosis, NIHHS score on admission, and CRP/Hgb ratio, and shows high prediction accuracy in predicting hospital-acquired pneumonia in ischemic stroke patients.

Application of machine learning and natural language processing for predicting stroke-associated pneumonia

Background

Identifying patients at high risk of stroke-associated pneumonia (SAP) may permit targeting potential interventions to reduce its incidence. We aimed to explore the functionality of machine learning (ML) and natural language processing techniques on structured data and unstructured clinical text to predict SAP by comparing it to conventional risk scores.

Methods

Linked data between a hospital stroke registry and a deidentified research-based database including electronic health records and administrative claims data was used. Natural language processing was applied to extract textual features from clinical notes. The random forest algorithm was used to build ML models. The predictive performance of ML models was compared with the A²DS², ISAN, PNA, and ACDD⁴ scores using the area under the receiver operating characteristic curve (AUC).

Results

Among 5,913 acute stroke patients hospitalized between Oct 2010 and Sep 2021, 450 (7.6%) developed SAP within the first 7 days after stroke onset. The ML model based on both textual features and structured variables had the highest AUC [0.840, 95% confidence interval (CI) 0.806-0.875], significantly higher than those of the ML model based on structured variables alone (0.828, 95% CI 0.793-0.863, P = 0.040), ACDD⁴ (0.807, 95% CI 0.766-0.849, P = 0.041), A²DS² (0.803, 95% CI 0.762-0.845, P = 0.013), ISAN (0.795, 95% CI 0.752-0.837, P = 0.009), and PNA (0.778, 95% CI 0.735-0.822, P < 0.001). All models demonstrated adequate calibration except for the A²DS² score.

Conclusions

The ML model based on both textural features and structured variables performed better than conventional risk scores in predicting SAP. The workflow used to generate ML prediction models can be disseminated for local adaptation by individual healthcare organizations.

Braden scale for predicting pneumonia after spontaneous intracerebral hemorrhage

OBJECTIVE

Stroke-associated pneumonia is an infection that commonly occurs in patients with spontaneous intracerebral hemorrhage and causes serious burdens. In this study, we evaluated the validity of the Braden scale for predicting stroke-associated pneumonia after spontaneous intracerebral hemorrhage.

METHODS

Patients with spontaneous intracerebral hemorrhage were retrospectively included and divided into pneumonia and no pneumonia groups. The admission clinical characteristics and Braden scale scores at 24 h after admission were collected and compared between the two groups. Receiver operating characteristic curve analysis was performed to assess the predictive validity of the Braden scale. Multivariable analysis was conducted to identify the independent risk factors associated with pneumonia after intracerebral hemorrhage.

RESULTS

A total of 629 intracerebral hemorrhage patients were included, 150 (23.8%) of whom developed stroke-associated pneumonia. Significant differences were found in age and fasting blood glucose levels between the two groups. The mean score on the Braden scale in the pneumonia group was 14.1±2.4, which was significantly lower than that in the no pneumonia group (16.5±2.6), p<0.001. The area under the curve for the Braden scale for the prediction of pneumonia after intracerebral hemorrhage was 0.760 (95%CI 0.717-0.804). When the cutoff point was 15 points, the sensitivity was 74.3%, the specificity was 64.7%, the accuracy was 72.0%, and the Youden's index was 39.0%. Multivariable analysis showed that a lower Braden scale score (OR 0.696; 95%CI 0.631-0.768; p<0.001) was an independent risk factor associated with stroke-associated pneumonia after intracerebral hemorrhage.

CONCLUSION

The Braden scale, with a cutoff point of 15 points, is moderately valid for predicting stroke-associated pneumonia after spontaneous intracerebral hemorrhage.

ICH-LR2S2: a new risk score for predicting stroke-associated pneumonia from spontaneous intracerebral hemorrhage

Purpose

We develop a new risk score to predict patients with stroke-associated pneumonia (SAP) who have an acute intracranial hemorrhage (ICH).

Method

We applied logistic regression to develop a new risk score called ICH-LR2S2. It was derived from examining a dataset of 70,540 ICH patients between 2015 and 2018 from the Chinese Stroke Center Alliance (CSCA). During the training of ICH-LR2S2, patients were randomly divided into two groups – 80% for the training set and 20% for model validation. A prospective test set was developed using 12,523 patients recruited in 2019. To further verify its effectiveness, we tested ICH-LR2S2 on an external dataset of 24,860 patients from the China National Stroke Registration Management System II (CNSR II). The performance of ICH-LR2S2 was measured by the area under the receiver operating characteristic curve (AUROC).

Results

The incidence of SAP in the dataset was 25.52%. A 24-point ICH-LR2S2 was developed from independent predictors, including age, modified Rankin Scale, fasting blood glucose, National Institutes of Health Stroke Scale admission score, Glasgow Coma Scale score, C-reactive protein, dysphagia, Chronic Obstructive Pulmonary Disease, and current smoking. The results showed that ICH-LR2S2 achieved an AUC = 0.749 [95% CI 0.739–0.759], which outperforms the best baseline ICH-APS (AUC = 0.704) [95% CI 0.694–0.714]. Compared with the previous ICH risk scores, ICH-LR2S2 incorporates fasting blood glucose and C-reactive protein, improving its discriminative ability. Machine learning methods such as XGboost (AUC = 0.772) [95% CI 0.762–0.782] can further improve our prediction performance. It also performed well when further validated by the external independent cohort of patients (n = 24,860), ICH-LR2S2 AUC = 0.784 [95% CI 0.774–0.794].

Conclusion

ICH-LR2S2 accurately distinguishes SAP patients based on easily available clinical features. It can help identify high-risk patients in the early stages of diseases.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-022-03389-5.

Novel Predictors of Stroke-Associated Pneumonia: A Single Center Analysis

Stroke-associated pneumonia (SAP) is a common cause of disability or death. Although the researches on SAP have been relatively mature, the method that can predict SAP with great accuracy has not yet been determined. It is necessary to discover new predictors to construct a more accurate predictive model for SAP. We continuously collected 2,366 patients with acute ischemic stroke, and then divided them into the SAP group and non-SAP group. Data were recorded at admission. Through univariate analyses and multivariate regression analyses of the data, the new predictive factors and the predictive model of SAP were determined. The receiver operating characteristic (ROC) curve and the corresponding area under the curve (AUC) were used to measure their predictive accuracy. Of the 2,366 patients, 459 were diagnosed with SAP. International normalized ratio (INR) (odds ratio = 37.981; 95% confidence interval, 7.487–192.665; P &lt; 0.001), age and dysphagia were independent risk factors of SAP. However, walking ability within 48 h of admission (WA) (odds ratio = 0.395; 95% confidence interval, 0.287–0.543; P &lt; 0.001) was a protective factor of SAP. Different predictors and the predictive model all could predict SAP (P &lt; 0.001). The predictive power of the model (AUC: 0.851) which included age, homocysteine, INR, history of chronic obstructive pulmonary disease (COPD), dysphagia, and WA was greater than that of age (AUC: 0.738) and INR (AUC: 0.685). Finally, we found that a higher INR and no WA could predict SAP in patients with acute ischemic stroke. In addition, we designed a simple and practical predictive model for SAP, which showed relatively good accuracy. These findings might help identify high-risk patients with SAP and provide a reference for the timely use of preventive antibiotics.

Risk factors and pathogenic microorganism characteristics for pneumonia in convalescent patients with stroke: A retrospective study of 380 patients from a rehabilitation hospital

BACKGROUND

Pneumonia is a major complication leading to death after stroke. The risk factors of pneumonia in convalescent patients who have experienced stroke remain poorly defined.

METHODS

To identify the risk factors of pneumonia, we applied logistic regression as a statistical method using SPSS23.0 statistical software, based on a sample of 380 patients. And statistical description method was used to analyze pathogens' characteristics and drug resistance.

RESULTS

Ultimately, the obtained logistic model has statistical significance (χ2(13) = 91.560, P <0.0005). The sensitivity of the model is 41.7%, the specificity is 97.6%, the positive predictive value is 76.9%, and the negative predictive value is 89.8%. The Barthel index (BI) (OR=1.97, 95% CI: 1.01-3.87), basic lung diseases (OR=4.24, 95% CI: 1.02-17.61), trachea ventilation (OR=6.56, 95% CI: 1.18-36.34), feeding tube (OR=6.06, 95% CI: 2.59-14.18), and hypoproteinemia (OR=3.97, 95% CI: 1.56-10.10) were statistically significant (P<0.05). Among patients who have pneumonia, the proportion of gram-positive bacteria, gram-negative bacteria and fungal infection is 10.00%, 54.29%, 5.71% respectively. The study most frequently isolated Pseudomonas aeruginosa (18.57%), followed by Acinetobacter baumannii (10.00%,) and Klebsiella pneumoniae (10.00%). The drug resistance rate of Pseudomonas aeruginosa, Acinetobacter baumannii and Klebsiella pneumoniae to different antibiotics ranged from 0.00-37.77%, 0.00-85.71% and 0.00-57.14%, respectively.

CONCLUSIONS

The lower BI scores, basic lung diseases, trachea ventilation, tube feeding, and hypoproteinemia are independent risk factors of pneumonia among convalescent patients with stroke. The main pathogens that caused pneumonia were gram-negative bacteria, and such organisms have different degrees of resistance to drugs.

Risk Factors for Stroke Associated Pneumonia

BACKGROUND:

Stroke patients are at high risk for stroke-associated pneumonia (SAP). If patients suffer from pneumonia their prognosis will worsen.

AIM:

To identify factors that increases the risk of SAP in stroke patients.

METHODS:

A group of 508 patients hospitalized within 5 days after the onset of stroke were enrolled prospectively.

RESULTS:

The incidence of SAP was 13.4%. Some major risk factors for SAP are: mechanical ventilation (MV) had odds ratio (OR) 16.4 (p <0.01); the National Institutes of Health Stroke Scale (NIHSS) > 15 OR 9.1 (p <0.01); the Gugging Swallowing Screen (GUSS) 0-14 OR 11.7 (p <0.01).

CONCLUSION:

SAP is a frequent complication. We identified some risk factors of SAP, especially stroke severity (NIHSS > 15), swallowing disorder (GUSS < 15) and mechanical ventilation.

AND score: a simple tool for predicting infection in acute ischemic stroke patients without a ventilator in the Chinese population

Objective

We aimed to develop a simple and user-friendly scoring system to predict all-cause hospital-acquired infections (HAIs) after acute ischemic stroke (AIS) in the Chinese population.

Methods

AIS patients from a retrospective cohort study at our center were included from January 2016 to December 2018. HAIs were diagnosed based on the current criteria from Ministry of Health of the People’s Republic of China. Stepwise logistic regression models were performed to screen independent predictors of HAI after AIS. A scoring system was developed by including each of the above significant predictors.

Results

Among 1211 patients, 76 patients (6.28%) developed HAI. Age, baseline National Institute of Health stroke scale (NIHSS) score, and dysphagia were independent predictors of HAI. For the AND score, A refers to age, N refers to NIHSS, and D refers to dysphagia. The AND score showed a high area under the receiver operating characteristics (AUROC) curve (0.679), which comprised age (65–74 years was 4 points, 75–84 years was 6 points, ≥85 years was 8 points), NIHSS score ≥10 (5 points), and dysphagia (6 points).

Conclusions

We developed a simple scoring system to predict all-cause infections after AIS patients without a ventilator in the Chinese population.

Braden scale for assessing pneumonia after acute ischaemic stroke

Background

The prevention of pneumonia is critical for patients with acute ischaemic stroke (AIS). The six subscales in the Braden Scale seem to be related to the occurrence of pneumonia. We aimed to evaluate the feasibility of using the Braden Scale to predict the occurrence of pneumonia after AIS.

Methods

We studied a series of consecutive patients with AIS who were admitted to the hospital. The cohort was subdivided into pneumonia and no pneumonia groups. The scores on the Braden Scale, demographic characteristics and clinical characteristics were obtained and analysed by statistical comparisons between the two groups. We investigated the predictive validity of the Braden Scale by receiver operating characteristic (ROC) curve analysis.

Results

A total of 414 patients with AIS were included in this study. Of those 414 patients, 57 (13.8%) patients fulfilled the criteria for post-stroke pneumonia. There were significant differences in age and histories of chronic obstructive pulmonary disease (COPD), dysphagia and Glasgow Coma Scale (GCS) score between the two groups, and the National Institutes of Health Stroke Scale (NIHSS) score in the pneumonia group was significantly higher than that in the no pneumonia group (P < 0.01). The mean score on the Braden Scale in the pneumonia group was significantly lower than that in the no pneumonia group (P < 0.01). The six subscale scores on the Braden Scale were all significantly different between the two groups. The area under the curve (AUC) for the Braden Scale for the prediction of pneumonia after AIS was 0.883 (95% CI = 0.828–0.937). With 18 points as the cutoff point, the sensitivity was 83.2%, and the specificity was 84.2%.

Conclusion

The Braden Scale with 18 points as the cutoff point is likely a valid clinical grading scale for predicting pneumonia after AIS at presentation. Further studies on the association of the Braden Scale score with stroke outcomes are needed.

Hospital-Wide Surveillance of Healthcare-Associated Infections as a Source of Information about Specific Hospital Needs. A 5-Year Observation in a Multiprofile Provincial Hospital in the South of Poland

Healthcare-associated infections (HAIs) are adverse complications of hospitalisation resulting in delayed recovery and increased costs. The aim of this study was an analysis of epidemiological factors obtained in the framework of constant, comprehensive (hospital-wide) infection registration, and identification of priorities and needs in infection control, both with regard to targeted surveillance, as well as preventative actions. The study was carried out according to the methodology recommended by the HAI-Net (Surveillance Network) coordinated by the European Centre for Disease Prevention and Control, in the multiprofile hospital in Southern Poland, between 2012 and 2016. A total of 159,028 patients were under observation and 2184 HAIs were detected. The incidence was 1.4/100 admissions (2.7/1000 patient-das of hospitalisation) and significantly differed depending on the type of the patient care: in intensive care units (ICU) 16.9%; in surgical units, 1.3%; non-surgical units, 1.0%; and paediatric units, 1.8%. The most common HAI was gastrointestinal infections (GIs, 28.9%), followed by surgical site infections (SSIs, 23.0%) and bloodstream infections (BSIs, 16.1%). The vast majority of GIs, BSIs, urinary tract infections, and incidents of pneumonia (PN) were detected in non-ICUs. As many as 33.2% of cases of HAI were not confirmed microbiologically. The most frequently detected etiologic agent of infections was Clostridium difficile-globally and in GI (49%). Comprehensive analysis of the results allowed to identify important elements of surveillance of infections, i.e., surveillance of GI, PN, and BSI not only in ICU, but also in non-ICU wards, indicating a need for implementing rapid actions to improve compliance with HAI prevention procedures.

Development and internal validation of a prediction rule for post-stroke infection and post-stroke pneumonia in acute stroke patients

Introduction

Patients with acute stroke are at high risk for infection. These infections are associated with unfavourable outcome after stroke. A prediction rule can identify the patients at the highest risk for strategies to prevent infection. We aim to develop a prediction rule for post-stroke pneumonia and other infections in patients with acute stroke.

Patients and methods

We used data from the Preventive Antibiotics in Stroke Study, a multicentre randomised trial comparing preventive ceftriaxone vs. standard stroke care in patients with acute stroke. Possible predictors for post-stroke pneumonia or infection were selected from the literature. Backward elimination logistic regression analysis was used to construct prediction rules for pneumonia or infection. Internal validation was performed and a risk chart was constructed. We adjusted for preventive antibiotic use.

Results

Pneumonia was diagnosed in 159 of the 2538 included patients, and infection in 348. Pneumonia was predicted by higher age, male sex, pre-stroke disability, medical history of chronic obstructive pulmonary disease, more severe stroke, dysphagia and intracerebral haemorrhage (rather than ischaemic stroke). Infections were predicted by higher age, male sex, history of diabetes, chronic obstructive pulmonary disease, more severe stroke, dysphagia, use of bladder catheter, preventive antibiotic use and intracerebral haemorrhage. With the prediction rule developed, risks for pneumonia ranged from 0.4% to 56.2% and from 1.8% to 88.0% for infection. Discrimination of the score was good (C-statistic, 0.84; 95% CI: 0.81–0.87 and 0.82; 95% CI: 0.79–0.84 for pneumonia and infection).

Conclusions

The Preventive Antibiotics in Stroke Study pneumonia and infection rule identify patients at the highest risk for post-stroke pneumonia or infection and may be used for future studies and novel therapies, after confirmation in an external population.