The risk prediction models for stroke associated pneumonia:a systematic review

doi:10.21203/rs.3.rs-3348358/v1

Download PDF

Research Article

The risk prediction models for stroke associated pneumonia:a systematic review

https://doi.org/10.21203/rs.3.rs-3348358/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Objective:Systematically evaluate the risk prediction models for stroke associated pneumonia.

Method: A systematic search was conducted on the risk prediction models for stroke associated pneumonia in CNKI,Wanfang Database,VIP,PubMed,Embase,Web of Science and Cochrane Library. The search was conducted until June 2023.Two researchers independently screened the literature and extracted data, and used the bias risk assessment tool of predictive model research to analyze the bias risk and applicability of the included literature.

Results: A total of 23 studies and 39 risk prediction models for stroke associated pneumonia were included, with area under the receiver operating characteristic curve (AUC) values ranging from 0.734 to 0.951. All models have good applicability, but the risk of bias is high.

Conclusion: Currently, the construction of risk prediction models for stroke associated pneumonia is not yet complete, and it is still necessary to continue to improve the details of variable screening, missing data processing, and model efficacy evaluation, in order to obtain better and high-precision risk prediction models and provide certain reference and basis for formulating relevant prevention strategies.It is recommended to conduct more prospective studies and conduct external validation of existing models in the future.

Associated pneumonia

Prediction model

system evaluation

Stroke remains the leading cause of death and disability among adults worldwide, with high incidence, mortality, and recurrence rates^{1, 2}. Although stroke is the second leading cause of death worldwide³, many stroke patients do not die from the stroke itself, but from its complications, including lower respiratory tract infections, which are one of the important causes of death among stroke patients⁴. Stroke-associated pneumonia (SAP) refers to the newly emerging lower respiratory tract infection in stroke patients without mechanical ventilation within 7 days of onset.^{5, 6}. Studies have shown that the incidence rate and mortality rate of stroke-associated pneumonia are very high, with an incidence rate of 7.6% -46.5%^7-9, And with a mortality rate of 61.2%¹⁰. Evidence shows that the prolongation of hospitalization time, increase in medical costs, and poor functional outcome of stroke patients are all related to SAP¹¹. Risk prediction models can early identify and screen high-risk stroke associated pneumonia patients, provide targeted nursing measures for patients, and reduce the occurrence of stroke associated pneumonia, and improve patient discharge outcomes. Several prediction models for stroke-associated pneumonia have been constructed at home and abroad, but the predictive performance and clinical applicability of the existing models still need to be further verified. At present, there is no systematic evaluation of the prediction model of stroke-associated pneumonia, so this study systematically evaluates the risk prediction model of stroke-associated pneumonia, to provide a reference basis for the subsequent model establishment and application.

1.1 Search policy

Using subject words combined with free words, we searched the CNKI, Wanfang, VIP, PubMed, EMbase, The Cochrane Library, and web of science databases, and the references of the papers were further traced. The search time limit for database construction is until June 2023. Search term are stroke, strokes, cerebrovascular accident*, CVA, cerebrovascular apoplexy, brain vascular accident*, cerebrovascular stroke*, apoplexy, cerebral stroke, acute stroke, acute cerebrovascular accident, cerebral infarction, cerebral hemorrhage, brain ischemia, pneumonia, SAP, pulmonary infection, lower respiratory infection, Stroke-Associated Pneumonia, Lung Inflammation, Pulmonary Inflammation, prediction model, risk prediction, mode, risk score, risk assessment. Taking the PubMed database as an example, the retrieval formula is as follows:

(((pneumonia[MeSH Terms]) OR (SAP[Title/Abstract] OR pulmonary infection[Title/Abstract] OR lower respiratory infection[Title/Abstract] OR Stroke-Associated Pneumonia[Title/Abstract] OR Lung Inflammation[Title/Abstract] OR Pulmonary Inflammation[Title/Abstract])) AND ((stroke[MeSH Terms]) OR (strokes[Title/Abstract] OR cerebrovascular accident*[Title/Abstract] OR CVA[Title/Abstract] OR cerebrovascular apoplexy[Title/Abstract] OR brain vascular accident*[Title/Abstract] OR cerebrovascular stroke*[Title/Abstract] OR apoplexy[Title/Abstract] OR cerebral stroke[Title/Abstract] OR acute stroke[Title/Abstract] OR acute cerebrovascular accident[Title/Abstract] OR cerebral infarction[Title/Abstract] OR cerebral hemorrhage[Title/Abstract] OR brain ischemia[Title/Abstract]))) AND (prediction model[Title/Abstract] OR risk prediction[Title/Abstract] OR mode[Title/Abstract] OR risk score[Title/Abstract] OR risk assessment[Title/Abstract])

1.2 Inclusion and exclusion criteria

Inclusion criteria:(1)The age of the research subject is ≥ 18 years old;(2)The research content is the construction of a risk prediction model for stroke associated pneumonia;(3)study type included cohort, cross-sectional, and case-control study;(4)outcome measures were SAP.

exclusion criteria:(1)Only for risk factor analysis, no model constructedl;(2)article language is not Chinese or English;(3)Informal publications such as conference abstracts and dissertations;(4)Research materials such as second-class literature, news reports, and case reports;(5)could not obtain full text;(6)only 1 predictor variable for model.

1.3 Literature screening and data extraction

Two researchers independently screened the literature according to the inclusion and exclusion criteria.Endnote X9 software was used to remove duplicates.Browsing the title and abstract, preliminary screening of literature according to inclusion and exclusion criteria, further review of literature that may meet the inclusion criteria, discussion or consultation with a third person in case of disagreement.The extracted data includes:first author, year of publication,country, study of type, study object, sample size,number of models, prevelance of SAP.

1.4 Assessment of the risk of bias and applicability of the included literature

Two researchers independently used the prediction model risk of bias assessment tool (PROBAST) to evaluate the bias risk and applicability of the models included in the literature¹².

1.4.1 Risk of bias assessment

PROBAST divides the potential biases involved in predictive model research into four areas, namely research subjects, predictive factors, outcomes, and analysis, with a total of 20 questions. The evaluator makes judgments on each question, and the answers to each question include "yes/may be yes", "no/may not be", and "lack of information". The bias risk of the prediction model as a whole and each field is classified as low, high, or unclear.Only when all questions in the field are answered with "yes" or "may be" is considered "low risk", and as long as one question is answered with "no" or "may not be", it is considered "high risk".When a problem is judged as "ack of information" and all other problems are considered "low risk", the field is classified as"unclear".

1.4.2 Applicability evaluation

The applicability evaluation of predictive models includes three areas: research subjects, predictive factors, and outcomes.The overall applicability of the predictive model is rated as "low" ,"high" , '"unclear" . Only when all areas are considered"low risk",it is considered "low risk" as a whole. If one or more areas are considered"high risk", it is considered"high risk"as a whole.If a certain area is judged as"unclear"and all other areas are considered"low risk", then the overall classification is"unclear".

1.5 Statistical analysis

Use descriptive analysis to summarize and analyze the basic characteristics of the model and the bias risk and applicability evaluation results included in the study.

2.1 Literature selection

A total of 4707 relevant literature were retrieved, and after screening, 23 were ultimately included(Figure1).

2.2 Basic characteristics of the included study

Finally, 23 articles were included, including 21 from China and 2 from other countries; 14 articles are in English and 9 articles are in Chinese; 19 were retrospective studies and 4 were prospective studies(Table 1).

Table 1 basic characteristics of the studies

First author	year of publication	country	type of study	subject	sample size		Models	Prevalence of SAP (%)
First author	year of publication	country	type of study	subject	exploitation	verify	Models	Prevalence of SAP (%)
Zhang Rui¹³	2022	China	retrospective	stroke	145	-	1	24.1
Wang Meng¹⁴	2020	China	retrospective	Hemorrhagic stroke	1841	462	4	15.7
Duan Xia¹⁵	2022	China	retrospective	Stroke	149	-	1	58.4
Wu Qian¹⁶	2020	China	prospective	Hemorrhagic stroke	173	173	1	25.4
Zeng Ruihuang¹⁷	2023	China	retrospective	Ischemic stroke	265	A:114 B:157	1	11.1
Liu Dun¹⁸	2021	China	retrospective	stroke	195	83	1	-
Wang Qiuyi¹⁹	2021	China	retrospective	Ischemic stroke	257	257	1	14.8
Li Li²⁰	2022	China	retrospective	Ischemic stroke	334	-	1	11.4
Yang Dongdong²¹	2021	China	retrospective	Ischemic stroke	212	212	1	17.0
Ge²²	2019	China	retrospective	Ischemic stroke	11736	2194	5	2.64
Huang²³	2019	China	prospective	Ischemic stroke	643	340	1	12.2
Li²⁴	2022	China	prospective	Ischemic stroke	2366	-	1	19.4
Smith²⁵	2015	Britain	retrospective	stroke	11551	11648	1	6.7
Song²⁶	2022	China	retrospective	Stroke	887	443	1	9.0
Wang²⁷	2021	China	retrospective	Ischemic stroke	1016	-	1	13.6
Yan²⁸	2022	China	prospective	Hemorrhagic stroke	56432	A:14108 B:12523	2	25.52
Yu²⁹	2023	China	retrospective	Hemorrhagic stroke	265	113	2	34.1
Zhang⁹	2021	China	retrospective	Ischemic stroke	258	149	1	46.5
Zheng³⁰	2022	China	retrospective	Hemorrhagic stroke	324	A:144 B:106	7	28.8
Ji³¹	2013	China	retrospective	Ischemic stroke	8820	A:5882 B:3037	1	11.4
Harms³²	2013	Germany	retrospective	Ischemic stroke	223	112	1	31.3
Zhang³³	2023	China	retrospective	Ischemic stroke	295	105	1	11.5
Tsai³⁴	2022	China	retrospective	stroke	4434	1479	2	7.6

*A: number of internal validation cases;B: number of external validation cases;

2.3 Model establishment situation

In this study, 23 articles were included and 39 prediction models were constructed.The total number of samples in each study is 113-56432, and the number of predictive factors for each model is 3-20.The top five repeatedly reported risk factors in the multivariate model are: National Institute of Health Stroke Scale (NIHSS), age,dysphagia,disorders of consciousness, and gender.Fifteen studies converted continuity variables into categorical variables, while six studies retained continuity variables. In the screening of predictor variables, 2 studies used Least absolute shrinkage and selection operator(LASSO) to screen variables; Twelve studies only screened variables through univariate analysis, while five used stepwise regression. The inclusion study involved multiple modeling methods, with 9 studies using the Logistic regression method; Six studies used machine learning methods;Machine learning methods include artificial neural networks, random forest(RF), extreme gradient Boosting(XGBoost), support vector machine(SVM), Gaussian naive Bayes;8 studies use packages from the R language(Table 2).

Table 2 characteristics of models

study	Modeling method	Number of predictors	Verification method	Renditions	Main predictors
Zhang Rui 2022	LR	7	-	-	NIHSS,GCS,Dysphagia,NLR,MLR,PLR, procalcitonin
Wang Meng 2020	ML	4	a	-	NIHSS,age,Dysphagia,leukocyte
Duan Xia 2022	R software	6	-	nomogram	hypertension,smoking,gastric tube,V-VST,NLR,modified Barthel index> 40
Wu Qian 2020	R software	7	a	nomogram	Age≥70years,pulmonary disease,dysphagia,NIHSS≥10,Invasive airway operation,nasal feeding, Preventive use of acid suppressants
Zeng Ruihuang 2023	R software	5	a、b	nomogram	Neutrophils,lymphocytes, prealbumin,fibrinogen, D-dimer
Liu Dun 2021	ML	12	a	-	Age, gender, hypertension, smoking,atrial-fibrillation,use of acid suppressants,visit interval,admission transportation,articulation disorder, disturbance of consciousness,blood pressure,Dysphagia
Wang Qiuyi 2021	R software	8	a	nomogram	Age≥65,dysphagia,consciousness disturbance, diabetes, nasal feeding,Preventive use of antibiotics, Preventive use of gastric mucosal protective agents, NIHSS≥10
Li Li2022	LR	3	-	equation	NIHSS,age,Infarction volume
Yang Dongdong 2021	LR	4	a	Scoring table	NIHSS,heart failure, dysphagia,NLR
Ge2019	ML	20	a	-	Gender, age, hypertension, diabetes, dysphagia, atrial fibrillation, disorders of consciousness,invasive surgery,ADL,Stroke history,marital status, dyslipidemia,coronary artery disease, history of surgery,risk score of pressure ulcer, risk score of fall, Whether laboratory testing is conducted, the frequency of laboratory testing, whether laboratory test values are abnormal, and whether medication is taken
Huang 2019	R software	7	a	nomogram	Age,NIHSS,atrial fibrillation,nasogastric tube,mechanical-ventilation,fibrinogen,leukocyte
Li2022	LR	6	-	equation	Homocysteine,history of COPD,age,dysphagia,INR,walking ability
Smith 2015	LR	4	a	Scoring table	NIHSS, age, gender,MRS
Song2022	R software	5	a	nomogram	Classification of stroke, Elderly Nutrition Risk Index,hemoglobin,fast blood glucose,A2DS2 score
Wang2021	LR	7	-	Scoring table	Age,NIHSS,GCS,dysphagia, heart failure, creatinine, use of proton pump inhibitors
Yan2022	ML	9	a、b	-	Age,mRS,fasting blood glucose, NIHSS,GCS, C-reactive protein, dysphagia,copd, smoking，
Yu2023	R software	7	a	nomogram	Intraventricular hemorrhage,hypertension, dysphagia,GCS, NIHSS, SIRI, PLR
Zhang 2021	LR	3	a	Scoring table	History of stroke, lymphocyte,dysphagia
Zheng 2022	ML	6	a、b	-	Nasogastric tube feeding, airway support, disturbance of consciousness,Ventricular drainage surgery, bleeding volume,Hospitalization in the intensive care unit
Ji2013	LR	12	a、b	Scoring table	NIHSS,age,atrial fibrillation,congestive heart failure,dysphagia, MRS, COPD, smoking,GCS, blood glucose,subtype of stroke
Harms2013	LR	4	a	Scoring table	leukocyte,GCS,blood,age pressure
Zhang2023	R software	4	b	nomogram	NIHSS,NLR,swallowing function,viscosity swallowing test score
Tsai2022	random forest	17	a	-	NIHSS, age, dysphagia, male,blood glucose, smoking,GCS,dysarthria, body mass index, leukocyte , platelets,smoking,heart rate,INR,aspartate aminotransferase values, glucose, triglycerides

*ML:machine learning; NLR:Neutrophil to lymphocyte ratio; MLR:monocyte -lymphocyte ratio; PLR:platelet to lymphocyte ratio; SIRI:Systemic infammactery response syndrome; mRS:Modified Rankin Scale; INR:International Standardized Ratio; GCS:Glasgow Coma Scale; NIHSS:National institute of Health stroke scale;

a: Internal verification; b: external validation

2.4 Model performance

The area under the receiver operating characteristic curve (AUC) values of 39 predictive models ranging from 0.734 to 0.951. 11 studies reported H-L goodness of fit tests, of which 5 reported P-values and P≥0.05, indicating Good consistency of the model. In terms of model validation, 1 model underwent external validation, 4 models underwent both external and internal validation, 5 models were not validated, and the rest were all validated internally. The main methods for internal verification are bootstrap sampling, K-fold test, and cross validation. Five studies reported a clinical decision curve. In terms of model presentation, 7 models were presented using column charts, 2 models were presented using equations, 6 models were presented using rating tables, and the remaining models were not presented.

2.5 Risk of bias and applicability evaluation results

The risk of bias assessment results of all studies were high risk, and the applicability evaluation results of the included studies showed that all studies had a low risk of applicability, in terms of study subjects, predictors, and study results. The evaluation results of bias risk and applicability risk included in the study are shown in Table 3.

Table 3 Results of risk of bias and applicability evaluation

study	Risk of bias				Applicability			Total
study	Study subject	Predictive factors	outcom	analyse	Study subject	Predictive factors	outcom	Risk of bias	Applicability
Zhang Rui 2022	-	+	+	+	+	+	+	-	+
Wang Meng 2020	-	-	-	-	+	+	+	-	+
Duan Xia 2022	-	-	+	-	+	+	+	-	+
Wu Qian 2020	+	-	U	-	+	+	+	-	+
Zeng Ruihuang 2023	-	-	U	-	+	+	+	-	+
Liu Dun 2021	-	-	-	U	+	+	+	-	+
Wang Qiuyi 2021	+	-	U	-	+	+	+	-	+
Li Li 2022	+	-	U	-	+	+	+	-	+
Yang Dongdong 2021	+	-	U	-	+	+	+	-	+
Duan Xia 2022	+	-	U	-	+	+	+	-	+
Ge 2019	-	-	-	-	+	+	+	-	+
Huang 2019	+	+	-	-	+	+	+	-	+
Li 2022	+	+	+	-	+	+	+	-	+
Smith 2015	+	-	-	-	+	+	+	-	+
Song 2022	+	-	+	-	+	+	+	-	+
Wang 2021	+	-	-	-	+	+	+	-	+
Yan 2022	+	-	+	-	+	+	+	-	+
Yu 2023	+	+	+	-	+	+	+	-	+
Zhang 2021	+	-	+	-	+	+	+	-	+
Zheng 2022	+	-	-	-	+	+	+	-	+
Ji 2013	+	-	+	-	+	+	+	-	+
Harms2013	+	-	+	-	+	+	+	-	+
Zhang2023	+	-	+	-	+	+	+	-	+
Tsai 2022	+	-	-	-	+	+	+	-	+

*"+" indicates low bias risk/high applicability; "-"indicates high bias risk/low applicability; U: Not clear

3.1 the overall bias risk of the stroke associated pneumonia prediction model is relatively high, and further improvement is needed

Through reading the included literature, it can be seen that the main bias risks come from the following aspects.(1) In terms of study objects: The main bias comes from the fact that most research types are retrospective and case-control studies. Firstly, the initial purpose of the data of retrospective research is not to develop or validate predictive models, and there may be important predictive factors that cannot be obtained in the case;secondly, During the research process, it cannot be guaranteed that evaluators have received unified training, resulting in biased results.(2) in terms of predictors:The main biases that exist are multicenter research and subjective biases. If the predictive factors that require subjective interpretation are measured by measurement personnel with different clinical experiences, it will lead to the risk of bias in the predictive factors. When using data from multiple medical institutions for research, differences in the evaluation methods of the predictive variables in each center may also lead to bias³⁵. Regarding the issue of whether the definitions and measurements of predictive factors for all research subjects are consistent or similar, the literature included in this study is mostly retrospective studies. The article does not specifically state that the predictive factors were evaluated by the researchers themselves or trained researchers, so this issue is mostly biased and high-risk; (3) In terms of study outcomes: Since stroke-related pneumonia is determined by imaging examination and laboratory results, Will involve subjective judgment, If the outcome indicators are measured by untrained or inexperienced surveyors, it may lead to potential bias. Only 5 studies (Duan Xia, Hui, Li, Yu, Yan) pointed out that the outcome was determined by professional personnel.More than half of the studies did not explain the measurement methods for outcomes in the text, resulting in a lack of judgment on this information;(4) In terms of analysis: PROBAST requires that when the number of events (EPV) per variable is ≥ 20, it is considered low risk, but more than half of the included studies have an EPV<20.The remaining risk of bias mainly comes from the single selection method of predictors, and most of the studies selected the predictors of statistical significant related to univariate to develop or verify the model, thus increasing the possibility of omission of important variables;Secondly, inappropriate conversions between continuous variables and categorical variables, as well as a lack of description of missing data processing, may also lead to bias.

3.2 The predictive model lacks clinical practicality

In clinical work, medical staff mainly focus on the availability of predictive factors and the convenience of model use.39 models were included in this study, covering various predictive factors such as demographics, clinical features, and biomarkers. This indicates that stroke associated pneumonia is influenced by numerous factors.However, having too many predictive factors in a model can easily lead to overfitting and increase the time burden on healthcare professionals in practical applications. At present, there is evidence to suggest that simplified models are easier to apply in clinical practice³⁶. Most of the predictive models included in this study have more than 5 predictive factors, with a maximum of 20. Future research can adjust the number of predictive factors to consider clinical practicality.From the perspective of predictors involved in each model, there are many non-interventible predictors, such as age, gender, NIHSS at baseline, history of hypertension, history of diabetes, stroke classification, etc.Assuming that healthcare professionals have screened high-risk SAP patients through a predictive model, intervention measures must be taken, but the predictive factors in the model are all non-interventible factors, which will increase the difficulty and workload of nursing staff. Multiple studies have shown that modifiable factors^37-41, such as low serum iron levels, low triiodothyronine (T3) syndrome, vitamin D deficiency, prolonged hospitalization, use of nasogastric tubes, and a history of alcohol abuse, are all associated with stroke associated pneumonia.Therefore, it is recommended to increase the proportion of modifiable predictive factors in subsequent studies and construct a more suitable prediction model for stroke associated pneumonia for clinical healthcare professionals.

3.3 The construction of a prediction model for stroke-associated pneumonia is still in the development stage

SAP was proposed by Hilker in 2003⁴².Stroke associated pneumonia is a type of pneumonia that refers to a new lower respiratory tract infection that occurs within 7 days after a stroke.There is heterogeneity in the definition of stroke associated pneumonia in existing studies. This study found that some researchers confuse the concepts of stroke associated pneumonia and post stroke pneumonia, roughly categorizing stroke associated pneumonia as hospital acquired pneumonia after stroke.This may lead to a decrease in the applicability of the model. In subsequent research, researchers should pay more attention to the definition of outcomes and strictly follow the standard definition of outcomes when including research subjects to reduce bias. This study included a total of 39 predictive models. Except for predictive models that use the same set of predictive factors and different modeling methods, no two models have the same set of predictive factors. This indicates that this field is still in the development stage and needs further exploration. The repeated predictors involved in all models : NIHSS, age, dysphagia, consciousness disorders,lymphocyte,NLR,hypertension,diabetes,atrial-fibrillation,smoking,procalcitonin,pulmonary diseas, Preventive use of acid suppressants,nasal feeding, heart failure,leukocyte, blood glucose. It can be seen that some prediction models have already included biological markers as predictive factors. With continued research, scholars both domestic and overseas have discovered the correlation between biomarkers and stroke-associated pneumonia. Kwan⁴³Found that IL-6 can independently predict post-stroke infection, especially post-stroke pneumonia. Yang⁴⁴ has also found that adding serum IL-6 to the A2DS2 score significantly improved the AUC of A2DS2. The detection of serum IL-6 is convenient and inexpensive, and it is expected to be promoted in clinical practice in the future. Du ⁴⁵found that It was found that the A2DS2 score combined with procalcitonin and C-reactive protein had a higher ability to predict SAP than the three alone. At present, the methods for predicting risk of SAP are relatively single, mostly relying solely on the prediction scales of SAP , biological markers, inflammatory factors.However, research has found that combining biomarkers with predictive scoring scales can improve predictive performance.Future research directions can consider adding more easily available biological biomarkers to predictive models.

Ethics approval and consent to participate

Not applicable

Consent for publication

Not applicable

Availability of data and materials

All data generated or analysed during this study were included in 23 published articles: References IX and 13-34.

Competing interests

The authors declare that they have no competing interests.

Funding

Not applicable

Authors' contributions

KY is responsible for the conception and design of the article, and the writing of the paper; ZJ is responsible for the collection and collation of research materials; CZ is responsible for editing table1-table3 and figure1; ZKL is responsible for statistical analysis; AXM is responsible for the revision of the final version and is responsible for the paper.All authors read and approved the final manuscript.

Acknowledgements

Not applicable

Virani SS, Alonso A, Benjamin EJ, et al. Heart Disease and Stroke Statistics-2020 Update: A Report From the American Heart Association. Circulation 2020; 141: e139-e596. DOI: 10.1161/CIR.0000000000000757.
Wang W, Jiang B, Sun H, et al. Prevalence, Incidence, and Mortality of Stroke in China: Results from a Nationwide Population-Based Survey of 480 687 Adults. Circulation 2017; 135: 759-771. DOI: 10.1161/CIRCULATIONAHA.116.025250.
Collaborators GS. Global, regional, and national burden of stroke and its risk factors, 1990-2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Neurol 2021; 20: 795-820. 2021/09/07. DOI: 10.1016/s1474-4422(21)00252-0.
Koennecke H-C, Belz W, Berfelde D, et al. Factors influencing in-hospital mortality and morbidity in patients treated on a stroke unit. Neurology 2011; 77: 965-972. DOI: 10.1212/WNL.0b013e31822dc795.
Wang Yongjun, Yuguo C, Chuanzhu L, et al. An Updated Chinese Consensus Statement on Stroke-associated Pneumonia 2019. Chinese Journal of Critical Care Medicine 2019; 39: 1135-1143.
Kishore AK, Jeans AR, Garau J, et al. Antibiotic treatment for pneumonia complicating stroke: Recommendations from the pneumonia in stroke consensus (PISCES) group. European Stroke Journal 2019; 4: 318-328. DOI: 10.1177/2396987319851335.
Tsai H-C, Hsieh C-Y and Sung S-F. Application of machine learning and natural language processing for predicting stroke-associated pneumonia. Frontiers In Public Health 2022; 10: 1009164. DOI: 10.3389/fpubh.2022.1009164.
Zapata-Arriaza E, Moniche F, Blanca P-G, et al. External Validation of the ISAN, A2DS2, and AIS-APS Scores for Predicting Stroke-Associated Pneumonia. Journal of Stroke and Cerebrovascular Diseases : the Official Journal of National Stroke Association 2018; 27: 673-676. DOI: 10.1016/j.jstrokecerebrovasdis.2017.09.059.
Zhang BW, Zhao WB, Wu CJ, et al. SDL Index Predicts Stroke-Associated Pneumonia in Patients After Endovascular Therapy. Frontiers in Neurology 2021; 12. DOI: 10.3389/fneur.2021.622272.
Tinker RJ, Smith CJ, Heal C, et al. Predictors of mortality and disability in stroke-associated pneumonia. Acta Neurol Belg 2021; 121: 379-385. DOI: 10.1007/s13760-019-01148-w.
Teh WH, Smith CJ, Barlas RS, et al. Impact of stroke-associated pneumonia on mortality, length of hospitalization, and functional outcome. Acta Neurologica Scandinavica 2018; 138: 293-300. DOI: 10.1111/ane.12956.
Wolff RF, Moons KGM, Riley RD, et al. PROBAST: A Tool to Assess the Risk of Bias and Applicability of Prediction Model Studies. Ann Intern Med 2019; 170: 51-58. DOI: 10.7326/M18-1376.
Zhang R, Huang R, Tan D, et al. Establishment of a model of influencing factors for stroke-associated pneumonia and itscomparison with 3 prediction scoring system. Chinese Journal of Nosocomiology 2022; 32: 2767-2771.
Wang M, Qin L, Wang C, et al. Machine Learning-based Models for Prediction of lntracerebral Hemorrhage Associated Pneumonia. Chinese Journal of Stroke 2020; 15: 243-249.
Duan X, Hou J, Jian H, et al. Risk factors and predictive model of stroke-associated pneumonia in patients with dysphagia. Chinese Journal of Rehabilitation Medicine 2022; 37: 616-622.
Wu Q and Lu W. Analysis of Risk Factors of Senile Acute Cerebral Hemorrhage Complicated with AssociatedPneumonia and Establishment of Risk Prediction Model. Journal of Guizhou Medical University 2020; 45: 616-622.
Zeng R, Chen Z, Wang X, et al. Predictive value of routine testing data mining for acute ischemic strokecomplicated with stroke-associated pneumonia. Chinese Journal of Infection Control 2023; 22: 142-149.
Liu D and Wang T. Study on the Prediction Model of Stroke-related Pneumonia Based on K-means. China Digital Medicine 2021; 16: 86-89.
Wang Q and You P. Analysis of the characteristic of stroke-related pneumonia pathogensand establishment of a risk mapping model. Chinese Journal of Nosocomiology 2021; 31: 967-971.
Li L, Su Y, Yuan G, et al. Constructing a predictive model for stroke associated pneumonia to improve prognosis. Chinese Journal for Clinicians 2022; 50: 59-61.
Yang D and Zhang Z. Analysis of risk factors of acute ischemic stroke-associated pneumonia and construction of its predictive scoring model. Journal of Neuroscience and Mental Health 2021; 21: 121-127.
Ge YQ, Wang QH, Wang L, et al. Predicting post-stroke pneumonia using deep neural network approaches. International Journal of Medical Informatics 2019; 132. DOI: 10.1016/j.ijmedinf.2019.103986.
Huang GQ, Lin YT, Wu YM, et al. Individualized Prediction Of Stroke-Associated Pneumonia For Patients With Acute Ischemic Stroke. Clinical Interventions in Aging 2019; 14: 1951-1962. DOI: 10.2147/cia.S225039.
Li YM, Zhao L, Liu YG, et al. Novel Predictors of Stroke-Associated Pneumonia: A Single Center Analysis. Frontiers in Neurology 2022; 13. DOI: 10.3389/fneur.2022.857420.
Smith CJ, Bray BD, Hoffman A, et al. Can a Novel Clinical Risk Score Improve Pneumonia Prediction in Acute Stroke Care? A UK Multicenter Cohort Study. Journal of the American Heart Association 2015; 4. DOI: 10.1161/jaha.114.001307.
Song XD, He Y, Bai J, et al. A nomogram based on nutritional status and A(2)DS(2) score for predicting stroke-associated pneumonia in acute ischemic stroke patients with type 2 diabetes mellitus: A retrospective study. Frontiers in Nutrition 2022; 9. DOI: 10.3389/fnut.2022.1009041.
Wang Y and Wu ZS. The establishment of a stroke-associated pneumonia predictive scoring system. Neurology Asia 2021; 26: 485-490. DOI: 10.54029/2021xdx.
Yan J, Zhai WQ, Li ZX, et al. ICH-LR2S2: a new risk score for predicting stroke-associated pneumonia from spontaneous intracerebral hemorrhage. Journal of Translational Medicine 2022; 20. DOI: 10.1186/s12967-022-03389-5.
Yu TT, Liu HM, Liu Y, et al. Inflammatory response biomarkers nomogram for predicting pneumonia in patients with spontaneous intracerebral hemorrhage. Frontiers in Neurology 2023; 13. DOI: 10.3389/fneur.2022.1084616.
Zheng Y, Lin YX, He Q, et al. Novel machine learning models to predict pneumonia events in supratentorial intracerebral hemorrhage populations: An analysis of the Risa-MIS-ICH study. Frontiers in Neurology 2022; 13. DOI: 10.3389/fneur.2022.955271.
Ji RJ, Shen HP, Pan YS, et al. Novel Risk Score to Predict Pneumonia After Acute Ischemic Stroke. Stroke 2013; 44: 1303-+. DOI: 10.1161/strokeaha.111.000598.
Harms H, Grittner U, Dröge H, et al. Predicting post-stroke pneumonia: the PANTHERIS score. Acta Neurol Scand 2013; 128: 178-184. 2013/03/07. DOI: 10.1111/ane.12095.
Zhang WB, Tang TC, Zhang AK, et al. A Clinical Prediction Model Based on Post Large Artery Atherosclerosis Infarction Pneumonia. Neurologist 2023; 28: 19-24. DOI: 10.1097/nrl.0000000000000434.
Tsai HC, Hsieh CY and Sung SF. Application of machine learning and natural language processing for predicting stroke-associated pneumonia. Frontiers in Public Health 2022; 10. DOI: 10.3389/fpubh.2022.1009164.
Du M, Haag D, Song Y, et al. Examining Bias and Reporting in Oral Health Prediction Modeling Studies. J Dent Res 2020; 99: 374-387. 2020/02/08. DOI: 10.1177/0022034520903725.
Wong J, Taljaard M, Forster AJ, et al. Addition of time-dependent covariates to a survival model significantly improved predictions for daily risk of hospital death. J Eval Clin Pract 2013; 19: 351-357. DOI: 10.1111/j.1365-2753.2012.01832.x.
Li J, Feng L, Huang Q, et al. An L-Shaped Relationship Between Serum Iron and Stroke-Associated Pneumonia. Clin Interv Aging 2021; 16: 505-511. 2021/04/02. DOI: 10.2147/cia.S301480.
Chen H, Xu M, Huang Y, et al. Low triiodothyronine syndrome is associated with stroke-associated pneumonia. European Journal of Clinical Investigation 2022; 52: e13840. DOI: 10.1111/eci.13840.
Huang G-Q, Cheng H-R, Wu Y-M, et al. Reduced Vitamin D Levels are Associated with Stroke-Associated Pneumonia in Patients with Acute Ischemic Stroke. Clinical Interventions In Aging 2019; 14: 2305-2314. DOI: 10.2147/CIA.S230255.
Almeida SRM, Bahia MM, Lima FO, et al. Predictors of pneumonia in acute stroke in patients in an emergency unit. Arq Neuropsiquiatr 2015; 73: 415-419. DOI: 10.1590/0004-282X20150046.
Patel UK, Kodumuri N, Dave M, et al. Stroke-Associated Pneumonia: A Retrospective Study of Risk Factors and Outcomes. The Neurologist 2020; 25: 39-48. DOI: 10.1097/NRL.0000000000000269.
Hilker R, Poetter C, Findeisen N, et al. Nosocomial pneumonia after acute stroke: implications for neurological intensive care medicine. Stroke 2003; 34: 975-981.
Kwan J, Horsfield G, Bryant T, et al. IL-6 is a predictive biomarker for stroke associated infection and future mortality in the elderly after an ischemic stroke. Exp Gerontol 2013; 48: 960-965. DOI: 10.1016/j.exger.2013.07.003.
Yang J, Dai Y, Zhang Z, et al. Value of Combination of the A2DS2 Score and IL-6 in Predicting Stroke-Associated Pneumonia. Neuropsychiatr Dis Treat 2020; 16: 2353-2359. DOI: 10.2147/NDT.S268878.
Hebin D, Lianbao X, Wei S, et al. Combination of A2DS2 score, procalcitonin and C-reactive protein predicts stroke-associated pneumonia in patients with acute ischemic stroke. International Journal of Cerebrovascular Diseases 2019; 27: 731-735.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

The risk prediction models for stroke associated pneumonia:a systematic review

Status:

Version 1

Abstract

Figures

Introduction

Data and methods

Results

Discussion

Declarations

References

Additional Declarations

Status:

Version 1