Roberts Steven
School of Finance and Applied Statistics, Australian National University, Crisp Building 026, ACT 0200, Australia.
Environ Res. 2004 Nov;96(3):328-37. doi: 10.1016/j.envres.2004.01.015.
In many community time series studies on the effect of particulate air pollution on mortality, particulate air pollution is modeled additively. In this study, we investigated the interaction between daily particulate air pollution and daily mean temperature in Cook County, Illinois and Allegheny County, Pennsylvania, using data for the period 1987-1994. This was done through the use of joint particulate air pollution-temperature response surfaces and by stratifying the effect of particulate air pollution on mortality by temperature. Evidence that the effect of particulate air pollution on mortality may depend on temperature is found. However, the results were sensitive to the number of degrees of freedom used in the confounder adjustments, the particulate air pollution exposure measure, and how the effects of temperature on mortality are modeled. The results were less sensitive to the estimation method used--generalized linear models and natural cubic splines or generalized additive models and smoothing splines. The results of this study suggest that in community particulate air pollution mortality time series studies the possibility of an interaction between daily particulate air pollution and daily mean temperature should be considered.
在许多关于颗粒物空气污染对死亡率影响的社区时间序列研究中，颗粒物空气污染采用相加模型。在本研究中，我们利用1987 - 1994年期间的数据，调查了伊利诺伊州库克县和宾夕法尼亚州阿勒格尼县每日颗粒物空气污染与日平均温度之间的相互作用。这是通过使用联合颗粒物空气污染 - 温度响应面以及按温度对颗粒物空气污染对死亡率的影响进行分层来实现的。发现了颗粒物空气污染对死亡率的影响可能取决于温度的证据。然而，结果对混杂因素调整中使用的自由度数量、颗粒物空气污染暴露测量方法以及温度对死亡率影响的建模方式敏感。结果对所使用的估计方法（广义线性模型和自然立方样条或广义相加模型和平滑样条）不太敏感。本研究结果表明，在社区颗粒物空气污染死亡率时间序列研究中，应考虑每日颗粒物空气污染与日平均温度之间相互作用的可能性。
