Anthropometric factors insufficient to detect prediabetes in kids

Anthropometric measures are not sufficiently accurate for predicting elevated fasting plasma glucose (FPG) and glycated haemoglobin (HbA1c) in children, a new study reports.

Researchers conducted a cross-sectional survey of 5,556 students aged 7–9 years. Anthropometric measures collected included standing height, weight, waist circumference (WC), and chest circumference (CC), among others. Blood samples were collected after fasting for at least 10 hours and were used for the measurement of FPG and serum HbA1c.

The median FPG and HbA1c levels were 4.5 mmol/L and 5.3 percent, respectively. Children who had higher FPG also showed greater anthropometric measures than their normal-FPG comparators (p<0.05 for all). Similarly, those with elevated HbA1c levels had higher waist-to-height, waist-to-hip, and waist-to-chest ratios (p<0.05 for all).

However, receiver operator characteristic (ROC) curve analysis found that anthropometric measures were only moderately predictive of FPG and HbA1c, at best. Weight and WC, for instance, had the highest areas under the curve (AUC) for the prediction of elevated FPG, with corresponding values of 0.651 (95 percent confidence interval [CI], 0.583–0.719) and 0.650 (95 percent CI, 0.584–0.717).

Body mass index and its z-score both also showed some predictive power, but which was weaker in magnitude. The other anthropometric measures were underpowered to assess elevations in FPG.

For HbA1c, only conicity index (AUC, 0.547, 95 percent CI, 0.519–0.574) showed some predictive efficacy. All other anthropometric indices were weaker or were completely null.

“We do not recommend anthropometric indicators to screen for prediabetes in young children, for the poor diagnostic performance,” researchers said.