Age-related considerations in the relationship between educational level and metabolic syndrome prevalence.