Based on published collaborative research co-authored by Dr Arash Ardalan.
The research question
Is the association between body composition and bone density adequately represented by a straight line?
Study context
The study included 478 Iranian children and adolescents aged 9–18 with DXA measurements. It examined overall body mass and its components in relation to bone-density Z-scores.
Statistical approach
- Use generalised additive models to assess the shape of the relationships.
- Examine overall body mass, fat mass and fat-free mass in relation to the outcome.
- Describe where associations change across the observed measurement range.
What the analysis provided
The paper reported non-linear patterns in the relationships between body composition and bone density, illustrating why one constant slope can be an incomplete description.
The measurements were cross-sectional and came from a specific study population. The reported associations do not establish the effects of changing body composition.
Practical lessons for your project
- Inspect the shape
- Explore the data and ask whether a linear relationship is a reasonable working assumption.
- Justify model flexibility
- Choose complexity using the study question, data support and model checks.
- Make curves interpretable
- Explain the fitted relationship over a meaningful range and show its uncertainty.
Original publication
Marjan Jeddi; Arash Ardalan; Seyed Taghi Heydari; Mohammad Hossein Dabbaghmanesh
Non-linear association of body composition and its components with bone density in Iranian children and adolescents
Archives of Osteoporosis · 16(1), 77 · 2021