【作者】 叶瑜；

【导师】 陈坤；

【作者基本信息】 浙江大学 ， 流行病与卫生统计学， 2010， 硕士

【摘要】 研究背景及目的心血管疾病（Cardiovascular disease,CVD）是一组以心脏和血管异常为主的循环系统疾病,包括心脏和血管疾病、肺循环疾病以及脑血管疾病。随着经济的飞速发展和人们生活水平的不断提高,心血管疾病对居民健康的影响日益加重。在我国,心血管疾病两周患病率已位居疾病别两周患病率第1位;在发达国家,心血管疾病列首位死因。因此心血管疾病已成为危害居民健康的重大公共卫生问题。随着工业化进程的加快和经济的迅猛发展,大气污染正日趋严重,WHO估计大气污染导致每年约三百万早产儿死亡。流行病学调查也显示,大气污染对心血管、呼吸系统等疾病的发病、死亡存在一定的影响。因此,众多流行病学家致力于大气污染与人体健康关系的研究,以期为大气污染引发机体反应的机制、机理研究提供依据。目前,大气污染的心脑血管效应逐渐成为新的研究热点。一些研究表明在大气污染的急性效应下,引起心肌梗死、缺血性脑卒中、缺血性心脏病、心脑血管疾病等急诊、门诊、死亡的增加;然而也有相关研究未发现大气污染对心脑血管疾病的急性效应。随着研究的不断深入,大气污染急性效应易感人群的探索也已开展,但各研究结果存在差异。由于不同国家和地区的大气污染水平、人群构成和社会经济因素都存在一定的差异,且随着大气污染流行病学研究方法的不断完善,混杂和偏倚得到了更加有效的控制,从而可以更加准确地评价大气污染暴露的健康效应。本研究采用时间分层的病例交叉设计,评价杭州市大气污染物(PM₁₀、SO₂、NO₂)对居民心脑血管疾病急性发作的影响,并进一步按季节（春季、夏季、秋季、冬季）进行分层分析。同时,为了探讨在大气污染暴露下哪些居民更易急性发作心脑血管疾病,特将研究对象分为男性和女性、65岁以上和以下、有慢性缺血性心脏病史和无慢性缺血性心脏病史、有高血压史和无高血压史、有糖尿病史和无糖尿病史、有高血脂症史和无高血脂症史、嗜烟对象和非嗜烟对象、嗜酒对象和非嗜酒对象等不同特征的研究对象,以期更全面、更细致地探讨大气污染与居民心脑血管疾病急性发作的关系,为心脑血管疾病的预防控制提供科学依据。材料与方法本研究以杭州市居民为研究对象,以杭州市CDC提供的2002年～2004年居民心脑血管疾病急性发作数据为基础,该数据来源于上城区、下城区、建德市、富阳市四个监测点,每个监测点监测人口数为10万人。对于监测点内急性发作后28日内再次急性发作的病例则仅记为一个病例,归入其28日内第一次发作的日期。大气污染物资料为杭州市环境监测中心提供的2002年～2004年各污染物日平均浓度监测数据,变量包括PM₁₀、SO₂和NO₂。同期的气象资料温度（T）和相对湿度（RH）由杭州市中心气象台提供。本研究采用时间分层的病例交叉设计,即比较同一研究对象在急性事件发生前一段时间（病例期）的暴露情况与未发生事件的某段时间内（对照期）的暴露情况,以评价暴露与事件的关联。病例交叉设计同时间序列Poisson回归模型都是研究大气污染短期健康效应的方法,总体上说,两者得出的结果一致。时间分层的对照选择,指在一个时间层内,病例期和对照期处于同一年、同一个月和同一个星期几。本研究以心脑血管疾病急性发作为事件,纳入滞后0～3日及平均累积滞后1～3日的大气污染物浓度进入模型,对照期的暴露水平则不考虑此效应;其中滞后0日即为心脑血管疾病急性发作当日的污染物浓度（lag0）,平均累积滞后1日的大气污染物浓度为当日与滞后1日大气污染物的平均浓度,平均累积滞后2日为当日、滞后1日、滞后2日的平均浓度,依次类推。单污染物模型的最佳滞后期根据各污染物最大效应值（OR值）确定。统计分析过程主要分为两部分,描述性分析和病例交叉设计的模型拟合。描述性分析包括心脑血管疾病急性发作资料、大气污染物资料、气象资料的统计描述,并进行差异性分析及相关分析。病例交叉设计的模型拟合采用条件logistic回归模型,以每日心脑血管疾病急性发作人数为权重,分析大气污染物对心脑血管疾病急性发作的效应。同时进一步根据季节和研究对象的不同特征进行分层,一是探讨不同季节下大气污染物对心脑血管疾病急性发作影响的差异,二是探讨在大气污染暴露下具有哪些特征的研究对象更易引发心脑血管疾病的急性发作,为大气污染心脑血管疾病急性效应易感人群的研究提供线索。所有的数据整理、管理和描述性分析均在SPSS for Windows 13.0中完成,以双侧P值小于0.05视为有统计学意义。时间分层的病例交叉设计中对照的选择和匹配通过Visual Basic 6.0编写的程序实现,病例交叉设计的条件logistic回归模型及大气污染物心脑血管疾病急性效应的参数估计则通过R 2.9.2软件中的survival程序包实现。结果1.杭州市2002年1月1日～2004年12月29日,四个监测点共报告病例1618例,5例病例因发病日期缺失,故最后纳入1613例病例进入本次研究。研究期间,报告的病例数呈逐年增加趋势。其中男性病例数略多于女性病例数,65岁以上老年人占74.89%;病例中具有高血压史的病例较多,具有慢性缺血性心脏病史、糖尿病史、高血脂症史和嗜烟、嗜酒的病例较少。2.大气污染监测资料基本上能较为准确地反映杭州市CDC监测的40万人口的大气污染暴露水平。各变量呈偏态分布,其中PM₁₀为113.00μg/m³（IQR,14.00μg/m³～476.00μg/m³）,超过国家环境空气质量二级标准的天数达20.38%;SO₂为46.00μg/m³（14.00μg/m³～132.00μg/m³）,未超过国家二级标准;NO₂为53.00μg/m³（18.0μg/m³～135.00μg/m³）,超过国家二级标准的天数为10.69%。各污染物的季节日平均浓度存在统计学差异,从数值上看,夏季各污染物浓度的中位数偏低,冬季偏高。PM₁₀日平均浓度的各年中位数在研究期间存在差异,呈下降趋势:SO₂和NO₂在统计学上未见此差异。3.调整气象因素后,PM₁₀、SO₂、NO₂单污染物模型的最佳滞后期均为lag0。此时PM₁₀、SO₂、NO₂日平均浓度每升高10μg/m³,居民心脑血管疾病急性发作的OR值分别为0.998（0.986～1.009）、1.017（0.981～1.054）、1.006（0.969～1.045）。4.按季节分层,春季、夏季、秋季各污染物在最佳滞后期时与心脑血管疾病急性发作的关联均未见统计学意义,冬季时PM₁₀、SO₂、NO₂浓度增高引起的OR值分别为1.018（0.999～1.036）、1.078（1.022～1.138）、1.028（0.970～1.090）。按性别分层,男性的结果未发现统计学意义,SO₂对女性的效应值为1.066（1.006～1.130）,且与男性的效应值存在统计学差异。按年龄分层,65岁以上和以下居民的分析结果均未见统计学意义。结果提示冬季时大气污染物对居民心脑血管疾病急性发作的危害可能较大,女性比男性在大气污染暴露下更加易发心脑血管疾病。5.按研究对象疾病史和生活习惯分层,得到有慢性缺血性心脏病史的对象和无慢性缺血性心脏病史的对象、有高血压史的对象和无高血压史的对象、有糖尿病史的对象和无糖尿病史的对象、有高血脂症史的对象和无高血脂症史的对象、嗜烟对象和非嗜烟对象、嗜酒对象和非嗜酒对象等不同特征的研究对象在最佳滞后期时心脑血管疾病急性发作与大气污染的关联均未见统计学意义。比较不同组效应值异质性,嗜烟对象和非嗜烟对象在lag2时,SO₂的效应值存在统计学差异,提示非嗜烟对象在大气污染暴露下更易引发心脑血管疾病急性发作。结论本研究采用时间分层的病例交叉设计,基于杭州市四城区提供的心脑血管疾病急性发作资料以及收集的大气污染物和气象因素数据,探讨大气污染与居民心脑血管疾病急性发作的关系。结果表明,大气污染物与心脑血管疾病急性发作存在一定的关联,且随着季节变化其健康效应大小有所差异,冬季最为显著。女性、非嗜烟对象在大气污染浓度增高的情况下可能更易导致心脑血管疾病的急性发作。更多还原

【Abstract】 Backgrounds and ObjectivesCardiovascular disease （CVD） refers to the class of heart and vascular diseases including myocardial infarction, stroke, hypertension, disease of pulmonary circulation, peripheral vascular disease and cerebrovascular disease. With the developing of economy and the improving of the living standard, CVD is common now and constitutes a major threat to public health. In China, the first disease with the highest proportion to be suffered is CVD, and it was also the top death cause in developed countries. The etiology of CVD is complicated. Besides the biological factors, the psychological and environmental factors also play an important role.Along with the industrialization speeding up, the environmental pollution is day by day serious. The WHO estimates that air pollution is responsible for 3 million premature deaths each year. Numerous epidemiologic studies also demonstrated associations between increased concentrations of air pollutants and CVD. Thus many epidemiologists focused on the study of the relationship of air pollution and human health, with the purpose of providing evidence for the main mechanisms that link air pollution with CVD. Numerous studies showed an increased risk of mortality, hospital admissions and emergency room （ER） visits for myocardial infarction, ischemic stroke, ischemic heart disease and cardio-cerebrovascular disease in relation to increased air pollution levels, but some studies didn’t find the associations. Along with the research went through, the research of subgroup that is sensitive to the effects of air pollution is also carried out, but the results were different.Because several factors differ among countries or districts such as the level of environmental pollution, the composition of the population and socioeconomics, and with the development of epidemiological study design, confounding factors and bias can be controlled better, the acute effect of air pollution can be evaluated more accurate.Time-stratified case-crossover study was used to analyze the effect of air pollutants (PM₁₀, SO₂, NO₂) on the acute onset of cardio-cerebrovascular disease （ACD） in our study. Additionally, similar analyses were undertaken to examine whether associations between air pollution and ACD were similar between men and women, people above 65 years old and below, and by seasons （spring, summer, autumn, winter）. To examine potential effect modification of the influence of air pollutants on the risk of ACD, analyses stratified by the presence or absence of a secondary diagnosis （chronic ischemic heart disease, hypertension, diabetes, hyperlipemia） were conducted, for exploring the subgroups that are at higher risk for ACD due to increased air pollutant exposure. The subgroups with or without drinking and smoking are also considered.Material and methodsThe residents in Hangzhou are taken as study’s population. The information of ACD cases were extracted from the center of disease prevention and control in Hangzhou city （HZ CDC） from 2002～2004. And the monitoring data was collected from four monitoring points that set in Shangcheng district, Xiacheng district, Fuyang city and Jiande city, every monitoring point monitors 100 thousand people. One case is only denoted by one time within-28days, classified into the time of the first acute onset of disease in the cycle.The data about the mean daily concentration of air pollutions （PM10, SO2, NO2） during the study’s period were provided bythe center for environmental surveillance in Hangzhou city. The meteorological factors including mean daily temperature and daily relative humidity in the same period were collected from the center for meteorological agency in Hangzhou city.The time-stratified case-crossover design was used. In this design, a specified referent window is identified for each subject that includes the case event matched to a set of control, or nonevent, periods for the same individual. The analytic approach compares exposures occurring during case events, defined as the "hazard" or "at-risk" period, with exposures for control periods, in which the outcome of interest did not occur. The case-crossover design and the time-series model are all the method to evaluate the acute effect of air pollutants on human health, generally speaking, they have the similar results. To avoid bias, we used the time-stratified approach, which divides time into strata and uses the remainder of the days in each stratum as the referents for a case in that stratum, so there was several control times before or after the case time in a matched set. Concentrations on the concurrent day and previous 1 to 3 days （lag0～lag3） and 2- （lag01） to 4-day （lag03） lagged moving average concentrations were evaluated. The concentrations of lag01 were the average of lag0 and lag1, and so forth. But the control period didn’t consider the effect. Single-pollutant model was fitted, and the lag time was selected according the maximum OR.The statistical analysis was composed of two parts, descriptive analysis and model fitting of case-crossover design. The descriptive analysis included the description of ACD, air pollutants and meteorological factors, difference analysis and correlation analysis were also carried out. The associations between ACD cases and levels of air pollutants concentrations were estimated using the odds ratios （OR） and their 95% confidence intervals （CI）, which were produced using conditional logistic regression with weights equal to the number of ACD cases on that day. We also conducted separate analyses for season and subgroups with different characters, for the purpose of studying the relationship between air pollutants and ACD, and exploring the potentially sensitive groups to air pollutants.All the data management and descriptive analysis were performed by SPSS 13.0 for windows software package, and all reported p values were based on two-sided tests, a p value of <0.05 was considered statistically significant. Time-stratified referent selection was realized with procedure written by Visual Basic 6.0 software, and the acute effect of air pollutants on ACD were performed by survival package of R 2.9.2 software. Results1. January 1, 2002 to December 29, 2004, a total of 1618 cases were collected from the four monitoring points, as 5 cases with the date of ACD were missing, 1613 cases were brought into our study at last. During the study period, the total cases of every year showed an increasing tendency. The proportion of male was somewhat more than that of female, and the value of the people above 65 years old was 74.89%. The number of the patients with the second diagnosis of hypertension was more than that without, and the numbers of the patients with chronic ischemic heart disease, diabetes, hyperlipemia, the habit of drinking or smoking were less than the opposite subgroup.2. The data of air pollutants could relatively substituted population exposure level for individual. Each variable were presented as abnormal distribution. The median daily concentration of PM₁₀, SO₂ and NO₂ were 113.00μg/m³ （14.00μg/m³ to 476.00μg/m³）, 46.00μg/m³ （14.00μg/m³ to 132.00μg/m³） and 53.00μg/m³ （18.00μg/m³ to 135.00μg/m³）, respectively. The days that PM₁₀ levels exceeded current limits （150μg/m³） accounted for 20.38% during study period, and the days （=80μg/m³） of NO₂ was 10.69%, but SO₂ level was under the criteria. The median daily concentrations of air pollutants separated by season were found statistically significance, lower in summer and higher in winter according to the median values. The values of PM₁₀ were decreased with year, however, SO₂ and NO₂ weren’t observed significant difference separated by year.3. After adjusted for meteorological factors, the OR of ACD by a 10μg/m³ increase in PM₁₀ at same day was 0.998 （0.986～1.009）. For a 10μg/m³ increase in SO₂, the OR was 1.017 （0.981～1.054） in the whole year （lag0）, and NO₂ exposure （lag0） was associated with ACD with OR was 1.006 （0.969～1.045）.4. Separated by season, no association was founded between air pollutants and ACD in spring, summer and autumn. In winter, the ORs of ACD by a 10μg/m³ increase in PM₁₀, SO₂ and NO₂ were 1.018（0.999～1.036）, 1.078（1.022～1.138）, 1.028（0.970～1.090）, respectively. Besides, no associations were observed between air pollutants and ACD of men, subgroup with above 65 years old or below, but a significant result was found between SO₂ and ACD of female. The results showed that residents may at higher risk for ACD due to increased air pollutant exposure in winter, and female may more sensitive to air pollutants than male.5. No associations were observed between air pollutants and ACD of subgroup with or without chronic ischemic heart disease, hypertension, diabetes, hyperlipemia, habit of drinking or habit of smoking, as same as the results of the subgroup without these diseases or habits. Compared the OR values of these subgroups, no statistical heterogeneity were observed except between the patients with and without smoking habits at lag2.ConclusionBased on the time-stratified case-crossover design, short-term effect of daily mean concentration of air pollutants adjusted for meteorological factors on the acute onset of cardio-cerebrovascular disease was evaluated. There were adverse effects of air pollution on ACD, and the effect values varied with season, most serious in winter. Female might especially sensitive to the adverse effects of air pollution, and the results of single-factor subgroup analyses showed that people without smoking habit was at greater risk of ACD associated with air pollution. In a word, female, the people without smoking habit might the sensitive group to the effects of air pollution on ACD.更多还原