Schwartz J
Environmental Epidemiology Program, Harvard School of Public Health, Boston, Massachusetts 02115.
Environ Res. 1994 Jan;64(1):36-52. doi: 10.1006/enrs.1994.1005.
The air pollution disaster in London in 1952 established that very high levels of particulate-based smog can cause dramatic increases in daily mortality. Recently, more than a dozen studies at much lower particle concentrations have reported evidence that exposure to lower levels of airborne particles results in smaller, but nonzero increases in daily mortality. These studies were examined in a meta-analysis. A primary focus of the meta-analysis was to examine effect size estimates across large differences in both the levels of potential confounding factors and in their correlation with airborne particle concentration. In the primary meta-analysis, airborne particle concentration was a significant risk factor for elevated mortality (RR = 1.06, 95% CI = 1.05-1.07). The relative risk is for a 100 micrograms/m3 increase in TSP concentration. While mortality peaked in the cold months in all locations, in the majority of the studies airborne particle concentrations were highest in the warm months, indicating that seasonal patterns were not responsible for the observed associations. The relative risk was 1.06 (95% CI = 1.05-1.07) when the analysis was restricted to cities with summer peaking pollution. The relative risk was identical in cities with above average annual temperatures and cities with colder climates. It was also identical in drier and more humid climates, and similar across a wide range of correlations between temperature and airborne particle concentrations. These results suggest that inadequate weather control was not responsible for the association. A detailed examination of data from Philadelphia showed that control for season and weather was adequate for removing all long-term seasonal and subseasonal patterns from the mortality data, and that using a very flexible nonlinear fit to the weather factors did not disturb the association with TSP. The most reasonable interpretation of this pattern of results is that the association is causal. This is supported by other studies which have reported that particulate air pollution was associated with lung function deficits, increased symptoms, and increased hospitalization.
1952年伦敦的空气污染灾难证实，极高浓度的颗粒烟雾可导致每日死亡率急剧上升。最近，十几项针对低得多的颗粒物浓度开展的研究报告称，有证据表明，接触较低水平的空气中的颗粒物会导致每日死亡率出现较小但非零的上升。这些研究在一项荟萃分析中得到了检验。该荟萃分析的一个主要重点是，在潜在混杂因素水平及其与空气中颗粒物浓度的相关性存在巨大差异的情况下，检验效应量估计值。在主要的荟萃分析中，空气中颗粒物浓度是死亡率升高的一个显著风险因素（风险比=1.06，95%置信区间=1.05-1.07）。该相对风险是针对总悬浮颗粒物（TSP）浓度每增加100微克/立方米而言的。虽然所有地区的死亡率在寒冷月份达到峰值，但在大多数研究中，空气中颗粒物浓度在温暖月份最高，这表明季节性模式并非观察到的这些关联的原因。当分析仅限于夏季污染达到峰值的城市时，相对风险为1.06（95%置信区间=1.05-1.07）。年平均气温高于平均水平的城市和气候较寒冷的城市，相对风险相同。在气候较干燥和较潮湿的地区也是如此，而且在温度与空气中颗粒物浓度之间的广泛相关性范围内情况类似。这些结果表明，天气控制不当并非造成这种关联的原因。对费城数据的详细检查表明，对季节和天气进行控制足以消除死亡率数据中所有长期的季节性和亚季节性模式，而且对天气因素采用非常灵活的非线性拟合并不会干扰与总悬浮颗粒物的关联。对这种结果模式最合理的解释是这种关联具有因果关系。其他研究也支持这一点，这些研究报告称，颗粒物空气污染与肺功能缺陷、症状加重以及住院率增加有关。
