ABSTRACT
Background: The availability of measures to operationalize allostatic load - the cumulative toll on the body of responding to stressor demands - in population health surveys may differ across years or surveys, hampering analyses on the entire sampled population. Here, impacts of variable selection and calculation method were evaluated to generate an allostatic load index applicable across all cycles of the Canadian Health Measures Survey (CHMS). Methods: Data from CHMS cycles 1 to 4 were used to compare allostatic load scores when replacing the most prevalent risk factor, waist-to-hip ratio - available in cycles 1 to 4 but not 5 and 6 - with body mass index (BMI), waist circumference, waist circumference within BMI groups (classified as normal, overweight, or obese), or waist-to-height ratio. Indexes were generated using clinical or sex-specific empirically defined risk thresholds and as count-based or continuous scores. Logistic regression models that included age and sex were used to relate each potential index to socioeconomic indicators (educational attainment, household income). Results: Of the variables assessed, waist-to-height ratio and waist circumference were closest to waist-to-hip ratio according to an individual's percentile ranking and in classifying "at risk" using either clinical or empirically defined cut-offs. Allostatic load profiles generated using waist-to-height ratios most closely resembled profiles constructed using waist-to-hip ratios. Sex-dependent associations with educational attainment and household income were maintained across constructs whether indexes were count-based or continuous. Interpretation: Allostatic load profiles and associations with socioeconomic indicators were robust to variable substitution and method of calculation, supporting the use of a harmonized index across survey cycles to assess the cumulative toll on health of stressor exposure.
