Research Projects

Project at a Glance

Title: Spatiotemporal analysis of air pollution and mortality in California based on the American Cancer Society cohort

Principal Investigator / Author(s): Michael Jerrett, Ph.D.

Contractor: University of California, Berkeley

Contract Number: 06-332

Research Program Area: Health & Exposure

Topic Areas: Health Effects of Air Pollution, Vulnerable Populations



Problem: Studies using the American Cancer Society (ACS) Cancer Prevention II (CPS II) cohort to assess the relation between particulate air pollution and mortality rank among the most influential and widely cited. The original study, a reanalysis that introduced new random effects methods and spatial analytic techniques, and recent studies with longer follow-up and improved exposure assignment, have all demonstrated statistically significant and substantively large air pollution effects on all-cause and cause-specific mortality. Due to this robust association and a lack of other large cohort studies on the long-term effects, the ACS studies have proven important to government regulatory interventions and health burden assessments.

At present there are no ACS CPS II statewide studies in California that investigate whether the risks are similar to or different from those reported in the above-mentioned analyses. Existing estimates come from either national-level ACS studies, in which the California subjects comprise less than 15% of the total national sample, or from select metropolitan or county areas of California, where questions remain about their generalizability to the rest of the state. A need therefore exists to investigate whether the results hold across California. In addition, none of the existing ACS studies have used high-resolution exposure assignment or investigated the temporal dimensions of the dose-response relationship. In this study we used advanced exposure modeling to reduce problems of measurement error, and we investigated time windows of exposure.

Description: We identified more than 76,000 California subjects in the ACS cohort to serve as the study population (20,432 deaths with an 18 year follow-up ending in 2000). These subjects were widely distributed across California, giving comprehensive coverage for much of the population of the state (i.e., 54 of 58 California counties have ACS subjects). For the first time in using the ACS CPS-II data, we have geocoded subjects to their home address to refine our exposure assignment.

As a basis for exposure assessment, we utilized interpolation estimates derived by Air Resources Board staff for the California Teachers Cohort Study led by Dr. Michael Lipsett, with Dr. Jerrett as co-investigator. We also implemented geostatistical kriging, advanced remote sensing coupled with atmospheric modeling, land use regression, and Bayesian models capable of assessing space-time patterns in exposure to improve exposure assignment.

We employed a comprehensive set of 20 individual risk factor variables similar to those used in previous ACS studies. These variables control for lifestyle, dietary, demographic, occupational, and educational influences that may confound the air pollution-mortality association. We used ecological variables in the neighborhoods of residence to control for “contextual” neighborhood confounding (e.g., unemployment). Although we used similar variables as in previous analyses to promote comparison to earlier results, we also tested other model specifications.

We assessed the association between air pollution and several causes of death, including cardiovascular (CVD), ischemic heart disease (IHD), respiratory, lung cancer, and other causes. We also evaluated all-cause mortality. There is some debate about the efficacy of evaluating associations between all-cause mortality and air pollution because several causes of deaths in this broad categorization likely have little association with air pollution. We have included the all-cause metric for several reasons. First, the all-cause metric has been used in most of the other published studies to date, and therefore we used this outcome for comparability with previous results. Second, the all-cause measure avoids the potential cross-classification bias between respiratory and CVD deaths. Third, the all-cause metric can be useful in burden of mortality assessments, and it has been used extensively for this purpose. Finally, we use the all-cause metric to compare with the cause-specific effects that we hypothesized should be more strongly related to pollution exposures (i.e., CVD deaths). A related point is the use of the combined "all other" causes of death to serve as a negative "control". The overall results are more compelling if one observes associations only for those causes of deaths for which there exists biological plausibility or where previous results have provided an a priori hypothesis (CVD, IHD, lung cancer), and where the risks for all other effects are null.

We assessed the association between air pollution and death using standard and multilevel Cox proportional hazards models. Control was also applied for residence in the five largest urban conurbations, which potentially have different mortality rates than non-metropolitan areas. We also assessed spatial autocorrelation in the health effect estimates.

Conclusion: Taken together, the results from this investigation indicate consistent and robust effects of PM2.5 − and other pollutants commonly found in the combustion-source mixture with PM2.5 − on deaths from CVD and IHD. We also found significant associations between PM2.5 and all causes of death, although these findings were sensitive to model specification. In Los Angeles, where the monitoring network is capable of detecting intraurban variations in PM2.5, we observed large effects on death from all causes, CVD, IHD, and respiratory disease. These results were consistent with past ACS analyses and with findings from other national or international studies reviewed in this report. Our strongest results were from a land use regression estimate of NO2, which is generally thought to represent traffic sources, where significantly elevated effects were found on deaths from all causes, CVD, IHD, and lung cancer. We therefore conclude that combustion-source air pollution is significantly associated with premature death in this large cohort of Californians.


For questions regarding this research project, including available data and progress status, contact: Heather Choi at (916) 322-3893

Stay involved, sign up with ARB's Research Email Listserver