TEMPERATURE AND MORTALITY ASSOCIATIONS BY CAUSE AND AGE IN TWO MEXICAN CITIES

Abstract

Studies conducted in diverse regions of the world have shown mortality and morbidity increases at temperature extremes, and global climate change may worsen the toll in certain areas. We evaluated associations between apparent temperature (a combination of temperature and relative humidity) and daily mortality in two Mexican cities (Mexico City and Monterrey), and examined whether associations differed by age and cause-of-death. Monterrey is Mexico's third largest city and is in the north (altitude 540 M); Mexico City, the capital and Mexico's largest city (altitude 2,240) is in central Mexico. Both have semi-arid climates. We used robust Poisson regression to model time series from each city: 1994–1998 in Mexico City, and 1996–1999 in Monterrey, fitting separate models for total, and cause- and age-specific mortality. We used natural spline terms to control for possible confounding by respiratory epidemics, defined as periods where respiratory-cause mortality exceeded the 90th percentile for 10 or more consecutive days. Temperature was modeled as the natural spline of the mean apparent temperature lags 0 to 13, and covariates in the model were indicator variables for day of week and public holidays, a linear term for PM10 (average lags 0–3), and natural splines for day of study and the respiratory epidemics. Autoregressive terms were added if necessary and degrees of freedom for day of study adjusted to minimize the sum of the partial autocorrelation function. Percent changes in mortality and 95% confidence intervals (CI's) associated with ‘heat’ and ‘cold’ were calculated for 1 degree increments enclosing the most extreme temperature reached during 50 or more days, relative to the increment enclosing the mean temperature (15.5o for Mexico City, and 25.3o for Monterrey). Because the ‘extreme’ heat increment for Mexico City was 19–20 o and the cold increment for Monterrey was 17–18 o, neither of which is likely to be physiologically stressful, we report only the results for the Mexico City cold effect (10–11o) and Monterrey hot effect (35–36 o). For total mortality, the cold effect in Mexico City was 13.1% (95% CI: 10.9%, 15.3%) and the heat effect in Monterrey, 16.9% (95% CI: 8.1%, 26.5%). In Monterrey, the heat effect was stronger among those over age 65: 30.2% excess mortality (95% CI: 17.7%, 44.1%). Cardiovascular and respiratory mortality had lower effects compared to all-cause. In Mexico City, the cold association was higher for respiratory mortality (26.2%, 95% CI: 19.7%, 33.0%), and for those over 65 (15.6%, 95% CI:12.6%, 18.6%). Cardiovascular mortality was slightly higher than for total. The elderly are at greatest risk for heat-related mortality in Monterrey, and the elderly and those with respiratory conditions are most vulnerable to cold in Mexico City.