摘要：Ba c k g r o u n d: Hierarchical Bayesian methods have been used in previous papers to estimate national mean effects of air pollu tants on daily deaths in time-series analy ses.oB j e c t i ve s: We obtained maximum likelihood estimates of the common national effects of the criteria pollu tants on mortality based on time-series data from ≤ 108 metropolitan areas in the United States.Me t h o d s: We used a subsampling bootstrap procedure to obtain the maximum likelihood esti-mates and confidence bounds for common national effects of the criteria pollu tants, as measured by the percentage increase in daily mortality associated with a unit increase in daily 24-hr mean pollu tant concentration on the previous day, while controlling for weather and temporal trends. We considered five pollu tants [PM10, ozone (O3), carbon monoxide (CO), nitrogen dioxide (NO2), and sulfur dioxide (SO2)] in single- and multi pollu tant analy ses. Flexible ambient concentration– response models for the pollu tant effects were considered as well. We performed limited sensitivity analy ses with different degrees of freedom for time trends.re s u l t s: In single- pollu tant models, we observed significant associations of daily deaths with all pollu tants. The O3coefficient was highly sensitive to the degree of smoothing of time trends. Among the gases, SO2and NO2were most strongly associated with mortality. The flexible ambi-ent concentration– response curve for O3showed evidence of non linearity and a threshold at about 30 ppb.co n c l u s i o n s: Differences between the results of our analy ses and those reported from using the Bayesian approach suggest that estimates of the quantitative impact of pollu tants depend on the choice of statistical approach, although results are not directly comparable because they are based on different data. In addition, the estimate of the O3-mortality coefficient depends on the amount of smoothing of time trends.
关键词：criteria pollu ;tants; hierarchical Bayes; multi ;city analy ;ses; spline smoothers; sub ;sampling ;bootstrap.
