Search

SUMMARY Motivated by a study of the association between counts of daily mortality and air pollution, we present a frequency domain estimation approach for log-linear models that accounts for both overdispersion and autocorrelation. The methods also allow for the discounting or downweighting of information at particular frequencies at which, for example, confounding variables are likely to have greatest influence. This allows flexible sensitivity analyses to be carried out to assess the possible effect of confounders on the estimated effect. We apply the methods to estimate the association between counts of mortality and the concentration of airborne particles in Philadelphia, USA, for the years 1974–1988. We obtain an estimated effect of particulate air pollution on mortality that is significantly greater than zero but less than that obtained by a standard log-linear analysis.

SUMMARY A common problem in environmental epidemiology is the estimation and mapping of spatial variation in disease risk. In this paper we analyse data from the Walsall District Health Authority, UK, concerning the spatial distributions of cancer cases compared with controls sampled from the population register. We formulate the risk estimation problem as a nonparametric binary regression problem and consider two different methods of estimation. The first uses a standard kernel method with a cross-validation criterion for choosing the associated bandwidth parameter. The second uses the framework of the generalized additive model (GAM) which has the advantage that it can allow for additional explanatory variables, but is computationally more demanding. For the Walsall data, we obtain similar results using either the kernel method with controls stratified by age and sex to match the age–sex distribution of the cases or the GAM method with random controls but incorporating age and sex as additional explanatory variables. For cancers of the lung or stomach, the analysis shows highly statistically significant spatial variation in risk. For the less common cancers of the pancreas, the spatial variation in risk is not statistically significant.

The second colloquium on particulate air pollution marks the reemergence of air pollution, generated by combustion sources, as a serious public health concern throughout the developed world. This article provides a perspective on the epidemiologic evidence on particulate air pollution and health. It addresses the evolution of methods for investigating air pollution and mortality and morbidity and the inter-pretational framework for the findings of the studies. Our work at the Johns Hopkins University School of Hygiene and Public Health, funded by the Health Effects Institute as part of its Particle Epidemiology Evaluation Project, provides the foundation; data are from public-use resources and are analyzed using Poisson regression. Data from previous analyses were replicated, and a detailed analysis for Philadelphia 1974–1988 was conducted to explore the sensitivity of the findings to key modeling assumptions. Our findings show that the association of total suspended particles with mortality pers...

Because weather has the potential to confound or modify the pollution–mortality relationship, researchers have developed several approaches for controlling it in estimating the independent effect of air pollution on mortality. This report considers the consequences of using alternative approaches to controlling for weather and explores modification of air pollution effects by weather, as weather patterns could plausibly alter air pollution's effect on health. We analyzed 1973–1980 total mortality data for Philadelphia using four weather models and compared estimates of the effects of TSP and SO2on mortality using a Poisson regression model. Two synoptic categories developed by Kalkstein were selected—the Temporal Synoptic Index (TSI) and the Spatial Synoptic Classification (SSC)—and compared with (1) descriptive models developed by Schwartz and Dockery (S-D); and (2) LOESS, a nonparametric function of the previous day's temperature and dew point. We considered model fit using Akaike's Information Criterion (AIC) and changes in the estimated effects of TSP and SO2. In the full-year analysis, S-D is better than LOESS at prediciting mortality, and S-D and LOESS are better than TSI, as measured by AIC. When TSP or SO2was fit alone, the results were qualitatively similar, regardless of how weather was controlled; when TSP and SO2were fit simultaneously, the S-D and LOESS models give qualitatively different results than TSI, which attributes more of the pollution effect to SO2than to TSP. Model fit is substantially poorer with TSI. This pattern was repeated in analyses of summer and winter months, which included SSC. In summary, using synoptic weather categories in regression models does not meaningfully change the association between mortality and air pollution indexes. We also found little evidence that weather conditions modified the effect of pollution, regardless of the approach used to represent weather.

Analyses involving data from many locations throughout the world have now been conducted to assess the association between air pollution and mortality. To date, six independent analyses of mortality data for Philadelphia, Pennsylvania, have been reported. In this new analysis of Philadelphia data for 1974-1988, Poisson regression models were developed to estimate the increased risk of daily mortality associated with air pollution while controlling for longer-term time trends and season and for weather. Model development was based on prior understanding of the effects of these factors on mortality and on consideration of model fit. The authors found moderate correlations of daily concentrations of total suspended particles (TSP), sulfur dioxide (SO2), nitrogen dioxide (NO2), and carbon monoxide (CO), and only slight correlations of ozone (O3) with other pollutants. When included individually in the model, the means of current and previous days' levels of TSP, SO2, and O3 had statistically significant effects on total mortality; pollutant increases of an interquartile range (34.5 micrograms/m3, 12.9 ppb, and 20.2 ppb, respectively) were associated with increases in mortality of around 1% for TSP and SO2, and of around 2% for O3. The effects of TSP and SO2 were diminished when both pollutants were simultaneously included in the model, whether pairwise or in the full multi-pollutant model. These analyses confirm the association between TSP and mortality found in previous studies in Philadelphia and show that the association is robust to consideration of other pollutants in the model.Data from various studies on the association between air pollution and mortality in Philadelphia, Pennsylvania, undertaken in the period 1974-1988 are used in a new analysis in which the authors attempt to control for the effects of temporal factors, weather, and other pollutants in order to estimate the independent effect of total suspended particles (TSP) on mortality. The results confirm the association between TSP and mortality found in previous studies in Philadelphia and show that the association is robust to consideration of other pollutants.

We consider the problem of estimating the spatial variation in relative risks of two diseases, say, over a geographical region. Using an underlying Poisson point process model, we approach the problem as one of density ratio estimation implemented with a non-parametric kernel smoothing method. In order to assess the significance of any local peaks or troughs in the estimated risk surface, we introduce pointwise tolerance contours which can enhance a greyscale image plot of the estimate. We also propose a Monte Carlo test of the null hypothesis of constant risk over the whole region, to avoid possible over-interpretation of the estimated risk surface. We illustrate the capabilities of the methodology with two epidemiological examples.