Study On The Incidence And Related Factors Of Stroke And Its Subtypes Based On A Cohort Population From 11 Provinces In China

Background and PurposeStroke has become the second leading cause of death worldwide.In China,stroke has become the disease with the highest burden of disease.At present,the number of stroke patients in China is 13 million,accounting for about one-fifth of stroke patients worldwide.The situation of stroke prevention and control is still severe in China.There are differences in the incidence and prevalence of stroke among people with different characteristics,such as region and gender.Therefore,understanding the incidence of stroke in people with different characteristics and early identification of stroke and its subtype related risk factors are the basis and key to the prevention and control of stroke.A large number of studies have confirmed that age,gender,and blood pressure are risk factors for stroke.However,there are still some risk factors,such as hypertension subtypes,prediabetes,self-rated health,etc.,and the relationship between the interaction of risk factors and the incidence of stroke and its subtypes needs to be analyzed in-depth due to insufficient research evidence.Stroke risk prediction model can comprehensively consider multiple risk factors,effectively identify high-risk groups of stroke,help clarify prevention priorities.However,there are still few cohort-based stroke risk prediction models in China,and more cohort studies are needed.The objectives of the study:(1)To analyze the incidence of stroke and its subtypes in people with different baseline characteristics in China using data from an 11-province cohort;(2)To explore the relationship between obesity,pre-diabetes and diabetes,hypertension subtype,dyslipidemia,and self-rated health,and the interaction between diabetes,hypertension,and dyslipidemia,and the onset of stroke and its subtypes;(3)To develop a suitable risk prediction model for stroke,which can be used to efficiently identify high-risk groups for stroke and provide key technology for establishing a stroke prevention and control model in China.MethodsThis study used data from the Chinese Adults Major Chronic Diseases Risk Assessment System(MCDRAS)project.The project is a prospective cohort study.Baseline data were obtained from the 2010 China Chronic Disease Risk Factors Surveillance(CCDRFS).The baseline survey included questionnaires,physical measurements,and laboratory tests.The MCDRAS constructed a follow-up cohort of 36,195 non-stroke individuals from 60 surveillance sites of 11 provinces(35 rural sites and 25 urban sites)in the CCDRFS.This baseline population was followed up with a household questionnaire and an individual questionnaire in 2016-2017.The outcome event was the first-ever stroke,stroke subtypes include ischemic stroke and hemorrhagic stroke.Continuous data were described using Mean±SD and t-test or ANOVA were used for comparison between groups.Categorical data were described using n(%),and differences between groups were compared using the chi-square test or rank-sum test.Incidence density was used to describe the incidence rate,and confidence intervals for incidence rates were estimated using Poisson distribution.Comparisons between rates were using the χ2 test or the Cochrane-Armitage method.Multifactorial analysis was performed using Cox proportional risk regression models.Dies and stroke patients diagnosed within 1 year were excluded in sensitivity analysis,and the model adjustment factors were consistent with Multifactorial analysis.Likelihood ratio test was used for interaction analysis of subgroups to compare whether the differences between models with and without interaction terms were statistically significant.Analysis of interactions between factors included analysis of multiplicative and additive interactions.A Cox regression model was used for fitting the risk prediction model of stroke incidence risk,and the risk prediction model life is set at six years.Risk classes were assigned according to the risk prediction probability values of the non-outcome people.Model discrimination and calibration were evaluated using AUC and Brier scores.AUC>0.7 for good model discrimination and Brier score between 0 and 0.25 for good model calibration.Ten-fold cross-validation was used for internal validation.Nomogram was used for model presentation.The above analyses were performed in SAS 9.4(SAS Institute Inc.)and R4.0.2(2020-06-22)software.A two-sided test with P<0.05 was considered statistically significant.Results1.Basic information of the study subjects36195 study subjects were surveyed at baseline in 2010 and 27436 were actual followed up in 2016-2017.Subjects with baseline cancer,those with missing baseline waist circumference,those with missing follow-up outcomes,those with incorrect outcome diagnosis dates,and those with abnormal baseline BMI and waist circumference were excluded,and 27112 study subjects were ultimately included in the analysis.Of these,576(2.1%)died,12259(45.2%)were men and 14853(54.8%)were women.2.Stroke and its subtype incidence statusA total of 1333 first-ever stroke events occurred in this study,including 1128 ischemic strokes and 205 hemorrhagic strokes.The incidence of stroke in the total subjects was 7.78/1000 person-years,with 6.58/1000 person-years for ischemic stroke and 1.20/1000 person-years for hemorrhagic stroke.In men,the stroke incidence was 7.73/1000 person-years,ischemic stroke was 6.36/1000 person-years,and hemorrhagic stroke was 1.37/1000 person-years.In women,the stroke incidence was 7.82/1000 person-years,ischemic stroke was 6.77/1000 person-years,and hemorrhagic stroke was 1.05/1000 person-years.There were no statistically significant difference in stroke and its subtypes incidence between men and women(all for P>0.05).The incidence of stroke and its subtypes was highest in people aged ≥60 years;the incidence of stroke and its subtypes was higher in rural areas than in urban areas;the incidence of stroke and ischemic stroke was higher in eastern areas than in central and western areas(all for P<0.05).3.Analysis of risk factors for stroke and its subtypesAnalysis of the relationship between obesity status and stroke incidence showed that obesity status was evaluated using three indicators,BMI,waist circumference,and waist-height ratio,respectively.The risk of stroke increased in overweight and obese(BMI<24 kg/m2 as reference),centrally obese stage Ⅰ and Ⅱ(normal as reference),and waist-height ratio in the range of 0.46～and ≥0.50(<0.46 as reference)populations 17%(95%CI:3%-32%)and 21%(95%CI:3%-42%),19%(95%CI:2%-38%)and 25%(95%CI:10%-42%),18%(95%CI:-2%-42%)and 33%(95%CI:13%-57%),respectively,with no change in the sensitivity analysis.Analysis of the relationship between obesity status and ischemic stroke incidence showed an increased risk of ischemic stroke incidence in the obese population evaluated by all three indicators,with no change in the results of sensitivity analysis and subgroup analysis.The analysis of the relationship between obesity status and hemorrhagic stroke incidence showed that no increased risk of hemorrhagic stroke was found in the obese population,and the results of sensitivity analysis and subgroup analysis remained unchanged.Analysis of the relationship between hypertension subtypes and stroke incidence showed that the risk of stroke was increased by 50%(95%CI:28%-75%),45%(95%CI:6%-97%),and 125%(95%CI:97%-158%)in isolated systolic hypertension,isolated diastolic hypertension,and systolic diastolic hypertension groups,respectively.The sensitivity analysis showed an increased risk of stroke incidence in isolated systolic hypertension(43%,95%CI:22%-68%),and systolic diastolic hypertension groups(119%,95%CI:91%-151%).Analysis of the relationship of hypertension subtypes and ischemic stroke showed an increased risk of stroke in the isolated systolic hypertension group(39%,95%CI:17%-65%)and the systolic diastolic hypertension group(108%,95%CI:80%-140%),with no change in a sensitivity analysis.Analysis of the relationship of hypertension subtypes and hemorrhagic stroke outcome showed an increased risk in isolated systolic hypertension(135%,95%CI:54%-257%),isolated diastolic hypertension(223%,95%CI:72%-506%),and systolic diastolic hypertension groups(281%,95%CI:166%-445%),with no change in a sensitivity analysis.Only an effect modification of the relationship between high blood pressure and ischemic stroke by age was found in the subgroup analysis(P for interaction<0.001).Analysis of the association between pre-diabetes and diabetes and stroke onset showed 4%(95%CI:-10%-20%)and 20%(95%CI:4%-40%)increased risk of stroke onset pre-diabetes and diabetes groups,respectively,with no change in the sensitivity analysis.Analysis of the relationship between pre-diabetes and diabetes and ischemic stroke onset showed an increased risk of 6%(95%CI:-9%-23%)and 24%(95%CI:6%-46%)in the pre-diabetes and diabetes groups,respectively,with no change in the sensitivity analysis,and subgroup analysis only revealed an effect of age(P for interaction=0.017)and sex(P for interaction=0.030)on the relationship between diabetes and ischemic stroke onset modifying effects.Analysis of the relationship between pre-diabetes and diabetes and the development of hemorrhagic stroke showed no increased risk of development in either the pre-diabetes or diabetes groups,and the results of sensitivity and subgroup analyses were unchanged.Analysis of the relationship between dyslipidemia and stroke and its subtypes showed that no increased risk of stroke and its subtypes was found in the hypercholesterolemic and high LDL cholesterol groups,respectively,using the total cholesterol and LDL cholesterol normal groups as reference;and no decreased risk of stroke and its subtypes was found in the HDL cholesterol normal group,using the low HDL cholesterol group as reference.The risk of stroke and ischemic stroke was found to be increased by 20%(95%CI:6%-35%)and 18%(95%CI:4%-35%)in the triglyceride 1.13～mmol/L group(<1.13 mmol/L as reference),respectively,with no change in the sensitivity and subgroup analyses.Analysis of the association between self-rated health and stroke and its subtypes showed an increased risk of stroke,ischemic stroke,and hemorrhagic stroke only in the poor self-rated health group(excellent self-rated health group as reference)by 61%(95%CI:17%-121%),42%(95%CI:2%-97%),and 404%(95%CI:52%-1570%),respectively,with no change in a sensitivity analysis.The subgroup analysis only found an effect modification of the relationship between self-rated health and ischemic stroke incidence by age(P for interaction=0.005).Analysis of the interaction between diabetes mellitus,hypertension,and dyslipidemia with the development of stroke showed that a multiplicative(P<0.05)and additive interaction for diabetes and hypercholesterolemia,with RERI=1.39(95%CI:0.32-2.47),AP=0.52(95%CI:0.29-0.74),and S=5.50(95%CI:1.36-22.36).Analysis of the interaction between diabetes mellitus,hypertension,and dyslipidemia with the development of ischemic stroke showed that multiplicative(P<0.05)and additive interactions for diabetes and hypercholesterolemia,with RERI=1.50(95%CI.0.30-2.69),AP=0.53(95%CI:0.29-0.76)and S=5.22(95%CI:1.35-20.15).Analysis of the interaction between diabetes mellitus,hypertension,and dyslipidemia with the development of hemorrhagic stroke did not reveal multiplicative and additive interactions.4.Stroke risk prediction model development and validationA 6-year stroke risk prediction model was developed using Cox proportional risk regression,and the final fitted model of this study was:ln[λ(t|Xi)/λ0(t)]=0.073492934*age+0.168407120*education level+0.007396615*systolic blood pressure+0.013696342*diastolic blood pressure+0.150932388*diabetes+0.006669641*waist circumference+0.432685727*poor self-rated health.6-year stroke incidence risk prediction probability:P=1-0.999947^exp(0.073492934*age+0.168407120*education level+0.007396615*systolic blood pressure+0.013696342*diastolic blood pressure+0.150932388*diabetes mellitus+0.006669641*waist circumference+0.432685727*poor self-rated health).Risk classes:low risk:<1.5%,normal risk:1.5%-7.9%,intermediate risk:8%-17.9%,and high risk:≥18%.The internal validation of the model was performed by 10 fold cross-validation,and the mean of AUC for the discrimination index and Brier score for the calibration index were 0.7415 and 0.0475,respectively.Conclusions1.The incidence of stroke in China is still high,and ischemic stroke is the primary stroke style.The incidence of stroke is higher in rural areas than in urban areas.The difference in incidence between men and women showed no statistically significant.The incidence is higher in the elderly population than in the young population and higher incidence were seen in the east than in the central and western regions.Stroke prevention and control measures should be developed according to the actual situation in the region.2.People with obesity have an increased risk of stroke and ischemic stroke.The risk of stroke and ischemic stroke were increased in patients with isolated systolic hypertension and systolic diastolic hypertension.The risk of hemorrhagic stroke increased in isolated systolic hypertension,isolated diastolic hypertension,and systolic diastolic hypertension.The risk of both stroke and ischemic stroke were increased in diabetics.People with poor self-rated health have an increased risk of stroke,ischemic stroke,and hemorrhagic stroke.Diabetes mellitus and hypercholesterolemia showed a positive additive interaction with the risk of stroke and ischemic stroke.3.A 6-year stroke risk prediction model was established with good discrimination and calibration,which can be used as a primary screening tool for people at high risk of stroke.