School of Public Health, Institute of Health and Biomedical Innovation, Queensland University of Technology, Kelvin Grove, Brisbane, Queensland 4059, Australia.
Sci Total Environ. 2010 Dec 15;409(2):300-6. doi: 10.1016/j.scitotenv.2010.10.013. Epub 2010 Nov 4.
Many studies have illustrated that ambient air pollution negatively impacts on health. However, little evidence is available for the effects of air pollution on cardiovascular mortality (CVM) in Tianjin, China. Also, no study has examined which strata length for the time-stratified case-crossover analysis gives estimates that most closely match the estimates from time series analysis.
The purpose of this study was to estimate the effects of air pollutants on CVM in Tianjin, China, and compare time-stratified case-crossover and time series analyses.
A time-stratified case-crossover and generalized additive model (time series) were applied to examine the impact of air pollution on CVM from 2005 to 2007. Four time-stratified case-crossover analyses were used by varying the stratum length (Calendar month, 28, 21 or 14 days). Jackknifing was used to compare the methods. Residual analysis was used to check whether the models fitted well.
Both case-crossover and time series analyses show that air pollutants (PM(10), SO(2) and NO(2)) were positively associated with CVM. The estimates from the time-stratified case-crossover varied greatly with changing strata length. The estimates from the time series analyses varied slightly with changing degrees of freedom per year for time. The residuals from the time series analyses had less autocorrelation than those from the case-crossover analyses indicating a better fit.
Air pollution was associated with an increased risk of CVM in Tianjin, China. Time series analyses performed better than the time-stratified case-crossover analyses in terms of residual checking.
许多研究表明，环境空气污染对健康有负面影响。然而，在中国天津，关于空气污染对心血管疾病死亡率（CVM）影响的证据很少。此外，尚无研究探讨时间分层病例交叉分析的哪个时间分层长度可以得出与时间序列分析最接近的估计值。
本研究旨在评估中国天津空气污染对 CVM 的影响，并比较时间分层病例交叉和时间序列分析。
应用时间分层病例交叉和广义相加模型（时间序列）来研究 2005 年至 2007 年期间空气污染对 CVM 的影响。通过改变分层长度（日历月、28 天、21 天或 14 天），进行了 4 次时间分层病例交叉分析。使用 Jackknifing 来比较方法。残差分析用于检查模型拟合情况。
病例交叉和时间序列分析均显示，空气污染物（PM（10）、SO（2）和 NO（2））与 CVM 呈正相关。时间分层病例交叉分析的估计值随分层长度的变化而变化很大。时间序列分析的估计值随每年时间自由度的变化而略有变化。时间序列分析的残差的自相关性小于病例交叉分析的残差，表明拟合效果更好。
在中国天津，空气污染与 CVM 风险增加有关。就残差检查而言，时间序列分析的表现优于时间分层病例交叉分析。
