01-06-2012 | Original Article
A quantile regression approach to re-investigate the relationship between sleep duration and body mass index in Taiwan
Published in: International Journal of Public Health | Issue 3/2012
Login to get accessAbstract
Objective
Previous studies on the relationship between sleep duration and body mass index (BMI) have shown inconsistent results by using estimation strategies within the framework of ordinary least squares (OLS). This study examined the relationship between sleep duration and BMI by using quantile regression to account for the potential heterogeneous effect of sleep duration on BMI in different BMI categories.
Methods
The data of 2,392 adults were from the 2005 Panel Study of Family Dynamics in Taiwan. The dependent variable was BMI of the respondents. Both OLS and quantile regression models were used for comparison.
Results
The OLS model does not show significant relationship, while the quantile regression model shows a U-shaped relationship between sleep duration and BMI beyond the 90th percentile in men (BMI = 28.69) and an inverse U-shaped relationship at the 30th percentile of BMI in women (BMI = 21.37).
Conclusions
Quantile regression can provide information that may be masked by OLS in analyzing the relationship between sleep duration and BMI. Sleep modification with the aim to obtain the optimal sleep duration may help to reduce BMI in obese men.