• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>Annual conference of the International Society of Exposure Science >Influence of seasonal adjustment in regression models on mortality risk estimates for multiple ambient air pollutants
【24h】

Influence of seasonal adjustment in regression models on mortality risk estimates for multiple ambient air pollutants

机译：回归模型中的季节调整对多种环境空气污染物的死亡风险估计的影响

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Background: The influence of model specification on risk estimates from time-series studies of air pollution is an important issue. Recent studies reported that ozone risk estimates are sensitive to the extent of adjustment for seasonal/temporal trends, but they have not identified the appropriate amount of adjustment. Aims: To identify the extent of temporal adjustment required to control for influenza epidemics and to characterize the sensitivity of fine particle (PM2.5) and gaseous pollutant mortality risk estimates to alternative model specifications. Methods: We characterized the effect of influenza epidemics on air pollution-mortality risk estimates for 33 U.S. cities with year-round PM2.5, ozone, nitrogen dioxide (NO2), sulfur dioxide, and carbon monoxide (CO) data for 2001-2006. Residuals from Poisson models were examined that considered 3 to 12 degrees of freedom per year (df/yr) to fit temporal trends with natural splines. Excess mortality risk was calculated for an interquartile range increase in the average of 0- and 1-day lags for all pollutants for all-year, warm (April-September) and cold (October-March) seasons by varying the extent of temporal adjustment in three models with alternative weather specifications similar to those used in previous multi-city studies. Results: Influenza peaks were not adequately fitted with small temporal adjustment, such as 3 df/yr, as reflected in residual diagnosis. The impact of the extent of temporal adjustment on estimated excess mortality risks for pollutants varied across the alternative weather models and by seasonal subsets. The risk estimates for CO and NO2 were particularly sensitive to alternative weather model specifications. Conclusions: In multi-city mortality time-series studies that include large cities where influenza peaks are prominent, small df/yr for temporal adjustment is not recommended. Therefore, sensitivity analyses should only be conducted for models that adequately control for temporal trends.
机译：背景：模型规范对空气污染的时间序列研究中的风险估计的影响是一个重要的问题。最近的研究报告说，臭氧风险估计值对季节性/时间趋势的调整程度很敏感，但尚未确定适当的调整量。目的：确定控制流感流行所需的时间调整程度，并表征细颗粒物（PM2.5）和气态污染物死亡率风险评估对替代模型规范的敏感性。方法：利用2001-2006年全年的PM2.5，臭氧，二氧化氮（NO2），二氧化硫和一氧化碳（CO）数据，我们表征了流感流行对33个美国城市空气污染死亡率风险估计的影响。考察了Poisson模型中的残差，这些残差考虑了每年3至12个自由度（df / yr），以使自然趋势符合自然趋势。通过改变时间调整的程度，计算出全年，温暖（4月至9月）和寒冷（10月至3月）所有污染物在0和1天滞后的平均四分位数间距增加中的死亡风险，在三个模型中，它们的替代天气规格与之前的多城市研究中使用的相似。结果：流感峰值不能通过较小的时间调整（例如3 df / yr）进行适当调整，这在残留诊断中得到了反映。时间调整程度对估计的污染物超标死亡风险的影响因替代天气模型和季节子集而异。一氧化碳和二氧化氮的风险估计对替代天气模型规范特别敏感。结论：在多城市死亡率时间序列研究中，包括流感高峰最为突出的大城市，不建议将小df / yr用于时间调整。因此，仅应针对可充分控制时间趋势的模型进行敏感性分析。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号