Waist-to-height ratio or waist circumference, adjusted for height, as outcomes

Based on what @f2harrell has discussed regarding issues with BMI, I was wondering whether the same issues would apply to other ratio measures such as waist-to-height ratio. In my mind, it would make more sense to model the outcome as log waist circumference, adjusted for log height (potentially using restricted cubic splines) as this allows for more flexibility in the waist circumference to height relationship.

Are you aware of any articles I could use to justify such a decision, either specifically for W2H ratio or any other ratio measures?

Please see this paper.

1 Like

Thank you for an interesting reference paper.

I have a question. For this flexible model rcs(log_ weight)*rcs(log_ height), what is the reason for logarithmic transformation of height and weight?

Also, my understanding is that the number of knots can be changed according to the actual situation. Is my understanding correct?

Thank you very much.

Logs are used there in the expectation of log-like relationships, allowing one to slightly reduce the number of knots.