生物识别与生物统计学杂志

Effect of Environmental Factors on Obesity: A Quantile Regression Approach

Abstract

Anthony J Payne, Julia A Knight and Taraneh Abarin

Objectives: This study explored associations of environmental factors with percent trunk fat (PTF) and body mass index (BMI), using quantile regression to explain variability in these traits at percentiles of the distributions.
Methods: Using a sample of 1695 adults from Newfoundland and Labrador, multiple and quantile regression models were used to analyse the significance of environmental factors on the average population and upper percentiles of the BMI and PTF distributions.
Results: Higher physical activity was associated with significantly lower PTF and BMI in the average population and upper percentiles, regardless of age. Both genders in percentiles closer to the median of PTF had more benefit with increased physical activity compared to higher percentiles. Interestingly, adults in higher percentiles of BMI distribution seem to benefit more with increased physical activity compared to percentiles closer to the median.
Conclusion: Using quantile regression as a robust approach toward violation of normality assumptions and outliers, variations in PTF and BMI for individuals across upper percentiles of the distributions based on some lifestyle factors were described. This method may be used to estimate the impact of certain lifestyle on different percentiles of BMI and PTF, rather than average population.