Academic journal article Environmental Health Perspectives

Using Supervised Principal Components Analysis to Assess Multiple Pollutant Effects

Academic journal article Environmental Health Perspectives

Using Supervised Principal Components Analysis to Assess Multiple Pollutant Effects

Article excerpt

BACKGROUND: Many investigations of the adverse health effects of multiple air pollutants analyze the time series involved by simultaneously entering the multiple pollutants into a Poisson log-linear model. This method can yield unstable parameter estimates when the pollutants involved suffer high intercorrelation; therefore, traditional approaches to dealing with multicollinearity, such as principal component analysis (PCA), have been promoted in this context.

OBJECTIVES: A characteristic of PCA is that its construction does not consider the relationship between the covariates and the adverse health outcomes. A refined version of PCA, supervised principal components analysis (SPCA), is proposed that specifically addresses this issue.

METHODS: Models controlling for long-term trends and weather effects were used in conjunction with each SPCA and PCA to estimate the association between multiple air pollutants and mortality for U.S. cities. The methods were compared further via a simulation study.

RESULTS: Simulation studies demonstrated that SPCA, unlike PCA, was successful in identifying the correct subset of multiple pollutants associated with mortality. Because of this property, SPCA and PCA returned different estimates for the relationship between air pollution and mortality.

CONCLUSIONS: Although a number of methods for assessing the effects of multiple pollutants have been proposed, such methods can falter in the presence of high correlation among pollutants. Both PCA and SPCA address this issue. By allowing the exclusion of pollutants that are not associated with the adverse health outcomes from the mixture of pollutants selected, SPCA offers a critical improvement over PCA.

KEY WORDS: air pollution, mortality, multiple pollutants, principal components analysis, time series. Environ Health Perspect 114:1877-1882 (2006). doi:10.1289/ehp.9226 available via [Online 24 August 2006]


Numerous time-series studies have investigated the association between daily adverse health outcomes and daily ambient air pollution concentrations (Chock et al. 2000; Cifuentes et al. 2000; Goldberg et al. 2003; Kelsall et al. 1997; Kwon et al. 2001; Moolgavkar 2000; Ostro et al. 1999; Smith et al. 2000; Stieb et al. 2002). These studies typically fit a Poisson log-linear model to concurrent time series of daily mortality or morbidity, ambient air pollution, and meteorologic covariates. The fitted models are then used to quantify the adverse health effects of ambient air pollution. Because the U.S. Environmental Protection Agency regulates pollutants independently, much of the current time-series research on the adverse health effects of air pollution has focused on estimating the effect of an individual pollutant (Dominici and Burnett 2003). However, because of the potential for high correlations to exist between ambient air pollutants, the results from studies that focus on a single pollutant can be difficult to interpret in practice (Vedal et al. 2003). For example, an observed positive association could occur because the single air pollutant is a proxy for another air pollutant or for a mixture of air pollutants.

To overcome the limitations of single-pollutant time-series studies, a number of studies have investigated the concurrent adverse health effects of multiple air pollutants (Moolgavkar 2000; Wong et al. 2002). In the majority of these studies, the multiple air pollutants are simultaneously entered into a single Poisson log-linear model. The results from these studies are then used to isolate the adverse health effects of the individual pollutants. However, one important question that these multiple pollutant studies fail to answer is whether there is a specific mixture of pollutants associated with adverse health outcomes. Moreover, it has recently been stated that it may be more reasonable to assume that there is a mixture of pollutants that is considered harmful to health (Dominici and Burnett 2003; Moolgavkar 2003; Stieb et al. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.