Use of generalized linear mixed models to examine the association between air pollution and health outcomes.

BACKGROUND: Time-series and case-crossover are two techniques that are widely used for assessing the short-term impact of ambient air pollution exposure on health.
MATERIALS AND METHODS: The generalized linear mixed model (GLMM) methodology is proposed here to study the association between ambient air pollution and health outcomes. Poisson random-effects models are applied to analyze the clustered counts, where the groups of days, determined by the triplet <day of week, month, year>, form the clusters. The proposed technique uses a nested structure for the clusters and allows random-effects for hierarchical factors. A random intercept in the models adjusts for different levels of counts among the clusters. A fixed slope represents a common response to the exposure. RESULTS AND
CONCLUSIONS: The obtained results are consistent with those generated by a classical approach (for example the case-crossover technique). The GLMM technique is a valid alternative methodology for studying air health effects.