Study On Risk Factors And Risk Assessment Model For Stroke In Community Residents In Shenzhen

Objectives1.To analyses stroke prevalence and risk factors among Chinese adults aged>40 years old in Shenzhen,providing clues for the study of etiology and the risk assessment model for stroke.2.To establish a risk assessment model for stroke specific to middle-aged and older adults,providing an assessment instrument for the stroke prevention and control.3.To investigate the application of system dynamics(SD)in the construction of the risk assessment model for stroke,and it not only could provide basis for quantitative predication of stroke risk,but also provide a new idea for risk assessment for chronic diseases.MethodsThis study was supported by the special fund for twelfth Five-year healthcare reform project of National Health and Family Planning Commission between 2012 and 2014,named"Screening and intervention project for high risk population of stroke".According to the selection requirements of the national screening project,adults aged>40 years old was considered as screening population,and three communities were selected using the cluster sampling method in Shenzhen.A total of 12,908 participants were included in this study.Demographic characteristics,socio-economic status,behavior life style,history of chronic diseases,family history of chronic diseases,physiological and biochemical indicators were interviewed.Frequency and proportion were used to describe the distribution characteristics and risk factors for stroke,the chi-square test and Fisher's exact probability method were conducted to compare the differences of the stroke prevalence in different groups.Multivariable logistic regression,Back Propagation(BP)neural network model and Decision Tree(DT)model were used to investigate the determinant factors associated with stroke risk.Based on the evidence-based medicine,Delphi method and relevant research data,Harvard Cancer Risk Index was carried out to establish the risk assessment model for stroke.According to the relevant literature,Delphi method and system analysis,rate variable fundamental in-tree model was performed to explore SD structural model for stroke risk assessment.Descriptive analysis,univariate analysis and multivariable analysis were carried out using SAS 9.2 and STATA 12.0.The complex,dynamics,feedback analysis of system was conducted using Venple5.1.Receiver Operating Characteristic(ROC)curve was performed with SPSS 20.0.Results1.Basic characteristics of study population.A total of 12,908 residents aged>40 years old in Shenzhen were investigated in this study.The study participants had a mean age of 57.11(standard deviation(SD)=10.71)years,and the mean age was 57.28(SD=11.27)years for male.The majority(99.91%)of participants was Han nationality and 42.40%of participants were engaged in physical labor.Over three-quarter of participants(78.28%)had high school degree or below,and most people(80.91%)were with at least one kind of health insurance.2.The analysis of stroke prevalence among community residents in Shenzhen.There was significant difference in sexes for stroke prevalence among Chinese adults aged>40 years old in Shenzhen.The crude prevalence of stroke was 4.06%for study population,5.46%for male and 2.89%for female,respectively.The age-standardized prevalence of stroke was 3.70%,and it was much higher in male than female(4.79 vs.2.76%,P<0.0001).The results showed that there were more than half of stroke patients aged 40-64 years(50.57%).3.Study on the influencing factors for stroke among community residents in Shenzhen.In the multivariable logistic regression analysis,age,sex,excessive salt intake,medical insurance,alcohol consumption,smoking status,inadequate intake of fruit,marital status,inadequate intake of milk and dairy products,family history of stroke,physical inactivity,hypertension and high homocysteine(HCY)were significantly associated with the risk of stroke among middle-aged and older Chinese.The results of BP neural network model showed that the most important influencing factor associated with stroke was age,following by smoking status,educational level,physical inactivity,marital status,excessive salt intake,high HCY,atrial fibrillation,inadequate intake of dairy products and hypertension.The results of DT model showed that the age was the most significant factor associated with the risk of stroke,followed by the excessive salt intake and family history of stroke,hypertension,atrial fibrillation,high HCY,smoking status,medical insurance,marital status and physical inactivity.In addition,the BP neural network model found that educational level,one of important socio-economic status indicators,was associated with stroke in middle-aged and older adults,but did not be identified in other two models.4.Study on the risk assessment for stroke in community residents based on Harvard Cancer Risk Index.Age,smoking status,gender,educational level,family history of stroke,overweight or obesity,hypertension,diabetes and dyslipidemia were included in the risk assessment model for stroke according to the Harvard Cancer Risk Index method.The results showed that the risk score values of age,gender,educational level,family history of stroke,smoking status,overweight or obesity,diabetes and dyslipidemia were 10 points except for 25 points for hypertension,and the average score of population stroke risk was 34 points.The study population was evaluated with the stroke risk assessment model established.The results showed that 1,713,3,870,443,5,787 and 636 individuals were identified as much lower,lower,average,high,and much higher risk,respectively.The incidence of stroke was different in the different risk levels(?3=85.4190,P<0.0001).The assessment result of predictive performance of the model showed that the AUC was 0.660(95%CI:0.636-0.684).The ratio of X(X was equal to the ratio of individual stroke risk scores and population average stroke risk scores)was at 1.40,which was considered as the best predictive cut-off point of stroke risk in individuals.This cut-off point at 1.4 was with the highest value of the Youden's index,and the sensitivity and specificity were 53.00%and 70.00%,respectively.5.Study on the risk assessment for stroke in community residents based on SD model.The rate variable fundamental in-tree model,a standardized SD modeling method,was conducted to establish 10 pairs of flows and flow rates dominated by the incidence of stroke.The relationships among these flows were analyzed systematically,and the interaction graphs of 10 pairs of flows and flow rates were drawn.Based on the two branch diagrams,ten in-trees models were established.We have qualitatively investigated the dynamic,complex,and feedback relationships among influencing factors,and finally developed the SD structural model of stroke risk assessment.Conclusions1.The prevalence of stroke was higher among middle-aged and older adults in Shenzhen than national average,and there were obvious differences in demographic characteristics,which were consistent with the basic characteristics of stroke epidemiology in China.2.Age,sex,alcohol consumption,excessive salt intake,smoking status,inadequate intake of fruit,marital status,inadequate intake of milk and dairy products,family history of stroke,physical inactivity,medical insurance,hypertension and high HCY were significantly associated with the risk of stroke among middle-aged and older Chinese.3.The risk assessment model for stroke based on Harvard Cancer Risk Index can effectively conduct the risk stratification and quantification of disease.The predictive performance of model was medium;however,the accuracy of model needed to be improved,which can be used for a risk tool for the screening the high risk population of stroke.However,as data source was limited,this model needed further verification and improvement before they could be generalized and applied.4.The SD structural model for stroke risk assessment can profoundly reveal the functional mechanism and interaction effects among influencing factors,and can clearly describe the complexity,feedback and dynamic of the model.