Alternative waist-to-height ratios associated with risk biomarkers in youth with diabetes: comparative models in the SEARCH for Diabetes in Youth Study

Abstract

Background/objectives: The waist-to-height ratio (WHtR) estimates cardiometabolic risk in youth without need for growth charts by sex and age. Questions remain about whether waist circumference measured per protocol of the National Health and Nutrition Examination Survey (W_NHAHtR) or World Health Organization (W_WHOHtR) can better predict blood pressures and lipid parameters in youth.

Participants/methods: WHtR was measured under both anthropometric protocols among participants in the SEARCH Study, who were recently diagnosed with diabetes (ages 5-19 years; N = 2 773). Biomarkers were documented concurrently with baseline anthropometry and again ~7 years later (ages 10-30 years; N = 1 712). For prediction of continuous biomarker outcomes, baseline W_NHAHtR or W_WHOHtR entered semiparametric regression models employing restricted cubic splines. To predict binary biomarkers (high-risk group defined as the most adverse quartile) linear W_NHAHtR or W_WHOHtR terms entered logistic models. Model covariates included demographic characteristics, pertinent medication use, and (for prospective predictions) the follow-up time since baseline. We used measures of model fit, including the adjusted-R² and the area under the receiver operator curves (AUC) to compare W_NHAHtR and W_WHOHtR.

Results: For the concurrent biomarkers, the proportion of variation in each outcome explained by full regression models ranged from 23 to 46%; for the prospective biomarkers, the proportions varied from 11 to 30%. Nonlinear relationships were recognized with the lipid outcomes, both at baseline and at follow-up. In full logistic models, the AUCs ranged from 0.75 (diastolic pressure) to 0.85 (systolic pressure) at baseline, and from 0.69 (triglycerides) to 0.78 (systolic pressure) at the prospective follow-up. To predict baseline elevations of the triglycerides/HDL cholesterol ratio, the AUC was 0.816 for W_WHOHtR compared with 0.810 for W_NHAHtR (p = 0.003), but otherwise comparisons between alternative WHtR protocols were not significantly different.

Conclusions: Among youth with recently diagnosed diabetes, measurements of WHtR by either waist circumference protocol similarly helped estimate current and prospective cardiometabolic risk biomarkers.