Abstract: Background　Baseline total cholesterol（TC） is an influencing factor of new-onset stroke，but no studies have been conducted on the effect of cumulative total cholesterol exposure（cumTC） on new-onset stroke.Objective　To explore the effect of cumTC on new-onset stroke.Methods　Using prospective cohort study method，52 427 working and retired employees of Kailuan Group who had physical examination in 11 hospitals including KaiLuan General Hospital from 2006 to 2010 were selected as the study objects.The starting point of the follow-up was when the objects completed the physical examination in 2010，and the end point was when new-onset stroke or deaths happened or the end of the follow-up（2017-12-31）.Baseline data and the incidence of new-onset stroke in each group were observed after all objects were grouped according to the cumTC quartile.The cumulative incidence was calculated by Kaplan-Meier method and tested by Log-Rank method.Cox regression model was used to analyze the effect of different cumTC on new-onset stroke.The fitting degree of different models was compared by likelihood ratio test.Results　There were significant differences in age，male proportion，systolic blood pressure（SBP），diastolic blood pressure（DBP），BMI，cumTC，TC，high-density lipoprotein cholesterol（HDL-C），fasting blood glucose（FBG），stroke，ischemic stroke，smoking，drinking，hypertension and diabetes detection rates，taking lipid-lowering drugs proportion among groups with different levels of cumTC（P<0.05）.During a mean follow-up year of（6.89±0.72），there were 1 611 new cases of stroke（1 415 cases of ischemic stroke and 229 cases of hemorrhagic stroke）.The cumulative incidence of stroke was 2.53%（331/13 089），2.92%（384/13 148），4.18%（547/13 085） and 4.36%（571/13 105） in the first to the fourth quartile group，respectively，and the cumulative incidence of ischemic stroke was 2.17%（284/13 089），2.51%（330
/13 148），3.71%（485/13 085） and 3.99%（523/13 105），respectively.The difference was statistically significant by Log-Rank test（χ2=73.15，P<0.001；χ2=109.13，P<0.001）.Cox regression analysis showed that after adjusting age，gender，BMI，hypertension，diabetes，smoking，drinking，physical exercise，HR（95%CI） of stroke and ischemic stroke in the fourth quartile group were 1.27（1.10，1.48） and 1.37（1.16，1.60） and those in the third quartile group were 1.22（1.05，1.41） and 1.27（1.08，1.50） compared with that in the first quartile group of cumTC.Increasing the factors of baseline TC level and taking lipid-lowering drugs，after likelihood ratio test，the fitting degree of the model was not optimized（P>0.05）.After adjusting age，gender，BMI，hypertension，diabetes，smoking，drinking，physical exercise，HR（95%CI） of stroke and ischemic stroke in the group with cumulative exposure for more than 4 years were 1.20（1.02，1.40） and 1.28（1.08，1.50），compared with the group with cumulative exposure to hypercholesterolemia for 0 years.Increasing the factors of baseline TC level and taking lipid-lowering drugs，the cumulative exposure of hypercholesterolemia had no statistical correlation with the new-onset stroke，ischemic stroke and hemorrhagic stroke.After the likelihood ratio test，the fitting degree of the model was not optimized（P>0.05）.Conclusion　High cumTC is a risk factor of new-onset stroke and ischemic stroke.

Key words: Total cholesterol, Cumulative exposure, Stroke, Root cause analysis, Prospective studies, Cohort studies

摘要： 背景　基线总胆固醇（TC）是新发脑卒中的影响因素，而国内外尚无关于累积总胆固醇暴露值（cumTC）对新发脑卒中影响的研究。目的　探究cumTC对新发脑卒中的影响。方法　采用前瞻性队列研究方法，选取2006—2010年在开滦总医院等共计11家医院进行健康体检的开滦集团在职及离退休职工52 427例为研究对象。以研究对象完成2010年度健康体检时点为随访起点，以新发脑卒中事件、死亡或至随访结束（2017-12-31）为随访终点。依据cumTC四分位进行分组，观察各组人群基线资料和新发脑卒中发病率。经Kaplan-Meier法计算累积发病率，以Log-Rank法进行检验，采用Cox回归模型分析不同cumTC组对新发脑卒中事件的影响，采用似然比检验比较不同模型的拟合程度。结果　不同cumTC水平组间年龄、男性占比、收缩压（SBP）、舒张压（DBP）、体质指数（BMI）、cumTC、TC、HDL-C、FBG、脑卒中、缺血性脑卒中、吸烟、饮酒、高血压病及糖尿病检出率、服调脂药占比比较，差异有统计学意义（P<0.05）。平均随访（6.89±0.72）年，共新发脑卒中1 611例（缺血性脑卒中1 415例、出血性脑卒中229例）。第一分位组~第四分位组脑卒中累积发病率分别为2.53%（331/13 089）、2.92%（384/13 148）、4.18%（547/13 085）、4.36%（571/13 105）；缺血性脑卒中累积发病率分别为2.17%（284/13 089）、2.51%（330/13 148）、3.71%（485/13 085）、3.99%（523/13 105），经Log-Rank检验，差异均有统计学意义（χ2=73.15，P<0.001；χ2=109.13，P<0.001）。Cox 回归分析显示：校正年龄、性别、BMI、高血压、糖尿病、吸烟、饮酒、体育锻炼变量后，与 cumTC 第一分位组相比，第四分位组发生脑卒中、缺血性脑卒中的 HR （95%CI）分别是1.27（1.10，1.48）、1.37（1.16，1.60），第三分位组发生脑卒中、缺血性脑卒中的HR（95%CI）分别是1.22（1.05，1.41）、1.27（1.08，1.50），增加校正基线 TC 水平、服调脂药后，经似然比检验后，模型的拟合程度并未得到优化（P>0.05）。校正年龄、性别、BMI、高血压、糖尿病、吸烟、饮酒、体育锻炼变量后，与高胆固醇血症累积暴露 0 年相比，累积暴露≥ 4 年发生脑卒中、缺血性脑卒中的 HR（95%CI）分别是 1.20（1.02，1.40）、1.28（1.08，1.50），增加校正基线 TC 水平、服调脂药后，高胆固醇血症累积暴露对新发生脑卒中、缺血性脑卒中、出血性脑卒中的影响无统计学意义，经似然比检验后，模型的拟合程度并未得到优化（P>0.05）。结论　高TC暴露是新发脑卒中、缺血性脑卒中的危险因素。

关键词: 总胆固醇, 累积暴露值, 卒中, 影响因素分析, 前瞻性研究, 队列研究