Assessing Childhood BMI with Alternative Regression Models
Author Information
Author(s): Andreas Beyerlein, Ludwig Fahrmeir, Ulrich Mansmann, André M. Toschke
Primary Institution: Ludwig-Maximilians University of Munich
Hypothesis
Which regression model best predicts childhood BMI and its risk factors?
Conclusion
GAMLSS and quantile regression are more suitable than common GLMs for modeling BMI data and understanding obesity risk factors.
Supporting Evidence
- GAMLSS provided a better fit for BMI data than traditional GLMs.
- Quantile regression allowed for interpretation of specific BMI percentiles.
- TV watching, maternal BMI, and weight gain in the first 2 years were significant predictors of BMI.
Takeaway
This study looked at different ways to understand how things like TV watching and meal frequency affect kids' weight. It found that some methods work better than others.
Methodology
The study compared generalized linear models, GAMLSS, and quantile regression using data from a health examination of preschoolers.
Potential Biases
Underreporting of maternal smoking and high maternal BMI could bias results.
Limitations
The study only included children with complete information, which may have led to underestimation of some effects.
Participant Demographics
Children aged 54 to 88 months, with a near-equal distribution of males and females.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website