Determinants and predictors of dengue infection in Mexico.

A national serosurvey was conducted in Mexico from March to October 1986 to identify predictors of dengue transmission and target areas at high risk of severe annual epidemics. A total of 3,408 households in 70 localities with populations less than 50,000 were randomly sampled, and serology was obtained from one subject under age 25 years in each household. When comparing exposure and infection frequencies across the 70 communities, the authors found that median temperature during the rainy season was the strongest predictor of dengue infection, with an adjusted fourfold risk in the comparison of 30 degrees C with 17 degrees C. High temperatures increase vector efficiency by reducing the period of viral replication in mosquitoes. The proportion of houses in a community with larva on the premises was significantly associated with the community proportion infected (odds ratio (OR)adj = 1.9; 95% confidence interval (CI) 1.4-2.5), as was the proportion of households with uncovered water containers present (ORadj = 1.9; 95% CI 1.4-2.7). Because these factors have effects beyond the individual household and subjects infected from them create a risk for other subjects, both analyses of effects and organization of control efforts must be at the community level. A predictive model was constructed using the community level risk factors to classify communities as being at high, medium, or low risk of experiencing an epidemic; 57% of these communities were correctly classified using this model.