Application of nonparametric quantile regression to body mass index percentile curves from survey data

Area of Science:

• Pediatrics
• Biostatistics
• Public Health

Background:

• Rising childhood overweight rates in the U.S. necessitate accurate growth assessment tools.
• Previous Body Mass Index (BMI)-for-age curves (CDC, 2000) had limitations in their statistical methodology.
• These limitations may affect the precise estimation of BMI patterns and overweight prevalence in children.

Purpose of the Study:

• To develop an improved, nonparametric statistical method for estimating national BMI-for-age percentile curves.
• To re-estimate U.S. national BMI-for-age percentile curves using the new method.
• To provide a more accurate assessment of childhood overweight prevalence.

Main Methods:

• Developed a nonparametric double-kernel-based method with automatic bandwidth selection.
• Incorporated sample weights into bandwidth selection and applied median correction for bias reduction.
• Rescaled bandwidth for scale invariance to enhance curve estimation accuracy.

Main Results:

• Re-estimated national BMI-for-age percentile curves using the advanced nonparametric approach.
• Provided updated prevalence estimates for high-BMI children in the U.S.
• The new method offers a more robust statistical foundation for growth assessment.

Conclusions:

• The developed nonparametric method provides more accurate BMI-for-age percentile curves compared to previous models.
• These enhanced curves offer a superior benchmark for evaluating child growth and overweight status.
• Accurate national curves are crucial for effective clinical and public health interventions regarding childhood obesity.