Refined ambient PM2.5 exposure surrogates and the risk of myocardial infarction

Abstract

Using a case-crossover study design and conditional logistic regression, we compared the relative odds of transmural (full-wall) myocardial infarction (MI) calculated using exposure surrogates that account for human activity patterns and the indoor transport of ambient PM2.5 with those calculated using central-site PM2.5 concentrations to estimate exposure to PM2.5 of outdoor origin (exposure to ambient PM2.5). Because variability in human activity and indoor PM2.5 transport contributes exposure error in epidemiologic analyses when central-site concentrations are used as exposure surrogates, we refer to surrogates that account for this variability as “refined” surrogates. As an alternative analysis, we evaluated whether the relative odds of transmural MI associated with increases in ambient PM2.5 is modified by residential air exchange rate (AER), a variable that influences the fraction of ambient PM2.5 that penetrates and persists indoors. Use of refined exposure surrogates did not result in larger health effect estimates (ORs = 1.10 – 1.11 with each interquartile range increase.), narrower confidence intervals, or better model fits compared to the analysis that used central-site PM2.5. We did observe evidence for heterogeneity in the relative odds of transmural MI with residential AER (effect-modification), with residents of homes with higher AERs having larger ORs than homes in lower AER tertiles. For the level of exposure-estimate refinement considered here, our findings add support to the use of central-site PM2.5 concentrations for epidemiological studies that employ similar case-crossover study designs. In such designs, each subject serves as his or her own matched control. Thus, exposure error related to factors that vary spatially or across subjects should only minimally impact effect estimates. These findings also illustrate that variability in factors that influence the fraction of ambient PM2.5 in indoor air (e.g., AER) could possibly bias health effects estimates in study designs for which a spatio-temporal comparison of exposure effects across subjects is conducted.

The importance of the exposure metric in air pollution epidemiology studies: When does it matter, and why?

“…Health risk CIs utilizing human exposure model estimates which include only exposures to ambient generated pollution decreased with mixed impact on the health risk estimates compared to those obtained using ambient monitor measurements (Mannshardt et al 2013;Sarnat et al 2013b). However, posterior intervals increased in one study (Chang et al 2012), and no change was seen in either the health risk estimate or the CIs in two studies (Hodas et al 2013;Jones et al 2013). The smaller confidence intervals are potentially due to less exposure error, which decreases the uncertainty in the health risk estimates and increases the chances for detecting their statistical significance.…”

“…Nevertheless, analysis by SE Sarnat et al in a time-series study at the finer ZIP code level showed some impact on effect estimates for cardiovascular and respiratory outcomes from exposure to traffic-related pollutants (EC, NO x , CO) when dispersion/statistical models were used compared to CS measurements (Baxter et al 2013b;Sarnat et al 2013b). A case-crossover study of exposure to PM 2.5 , using estimates from ambient monitors and from a semiempirical air exchange rate (AER)/mass balance model, showed little to no difference in odds ratios (ORs) or in confidence intervals (CIs) for transmural myocardial infarction outcomes (Hodas et al 2013). However, this same study, in addition to a second case-crossover study (Jones et al 2013) of respiratory hospitalizations in New York City, found evidence of effect modification by classifications of AER categories, which serve as indicators of infiltration rate, suggesting spatial consideration of housing factors could modify the health effects estimates for different population sub-groups.…”

“…This is particularly true when conventional time-series or case-crossover statistical methods are applied using aggregate heath data. In addition, in the case-crossover design where each household is used as its own control and within-season AERs are less variable, effect modification because of AER or other enhanced exposure predictors is not likely to be detected unless prior stratification of the data is performed on key household, behavioral or locational factors (Hodas et al 2013;Jones et al 2013;Sarnat et al 2013a).…”

“…While some factors that are not temporally varying (e.g., location of a subject's home) are likely to be less important for time-series studies (Hodas et al 2013;Jones et al 2013;Mannshardt et al 2013;Sarnat et al 2013b), other timevarying factors must still be taken into consideration. While certain time-varying factors such as behavioral changes due to weather are accounted for in human exposure models, these changes will only be reflected when human exposure models are used to obtain exposure estimates, and may not be as influential on exposure estimates as air concentration changes due to greater fluctuations in meteorology.…”

“…The statistical methodologies applied when CS measurements are used are more prone to estimation biases and/or higher imprecision in these multi-geographic epidemiologic applications, if the spatial distribution of pollutants or household characteristics (e.g., AER, infiltration rates, other exposure predictors) vary between the study sites and exhibit a different pattern from the ambient pollution concentrations over time and space. These types of studies often benefit from using more refined exposure metrics than the traditionally used ambient monitoring data (Baxter et al 2013b;Dai et al 2014;Hodas et al 2013;Özkaynak et al 2013). …”