Time-Series Studies of Particulate Matter

Abstract

Key Words air pollution, public health, epidemiology, regression models s Abstract Studies of air pollution and human health have evolved from descriptive studies of the early phenomena of large increases in adverse health effects following extreme air pollution episodes to time-series analyses based on the use of sophisticated regression models. In fact, advanced statistical methods are necessary to address the challenges inherent in the detection of a relatively small pollution risk in the presence of potential confounders. This paper reviews the history, methods, and findings of the time-series studies estimating health risks associated with short-term exposure to particulate matter (PM), though much of the discussion is applicable to epidemiological studies of air pollution in general. We review the critical role of epidemiological studies in setting regulatory standards and the history of PM epidemiology and time-series analysis. We also summarize recent time-series results and conclude with a discussion of current and future directions of time-series analysis of particulates, including research on mortality displacement and the resolution of results from cohort and time-series studies.

Short-term effects of fine particulate matter pollution on daily health events in Latin America: a systematic review and meta-analysis

“…Cochrane Collaboration recommend tools for randomized controlled trials and non-randomized studies of intervention (Higgins and Green 2011), both not applicable to time-series design. Yet, time-series studies have been playing a relevant role in air pollution regulatory process, because they estimate the burden of disease attributable to air pollution exposure and the related CRF can be used in cost-benefit analysis (Bell et al 2004). Timeseries studies use regression models to assess the effects of short-term changes in PM 2.5 levels on acute health effects by estimating associations between day-to-day variations in both air pollution and in mortality and morbidity counts.…”

“…Statistical specificities (e.g. accounting for serial correlation in the residuals) also play a role (Bell et al 2004). …”

“…exposure characterization and controlling for confounders) (Sheehan and Lan 2015) and of time-series design in particular (e.g. statistical adjustments) (Bell et al 2004). Frailties in at least one of the potential source of bias domains were frequent; underlining that internal validity of time-series study design on EH SRMAs needs further investigations.…”

Bootstrap model averaging in time series studies of particulate matter air pollution and mortality

“…Fitting a model similar to Model (1) is the approach taken in a number of studies including the recent NMMAPS analyses (Daniels et al, 2000;Dominici et al, 2003;Bell et al, 2004;Roberts, 2004a). In these studies, both the confounders and the PM exposure measure included in Model (1) are fixed a priori.…”

“…We used a method previously shown to generate realistic mortality time series (Roberts, 2005), which proceeds by fitting the following Poisson log-linear model similar to those used in previous studies (Bell et al, 2004) to the actual Cook County mortality and meteorological time series data: log ðm t Þ ¼ S t 1 ðtime; 4 df per yearÞ þ S t 2 ðtemp 0 ; 6 dfÞ þ S t 4 ðdew 0 ; 3 dfÞ þ gDOW t ð2Þ…”