I'm trying to study how age as a confounder may affect the relationship between commute mode and BMI (and design a research proposal).

To explain age as a potential confounder, I compared age to commute mode and then age to BMI. I found a weak association between age and BMI in my data.

In the literature does anyone know if BMI is expected to change with age? Is there a linear/ other relationship form? Please could someone attach any papers/ links for help.

Also, does anyone know what a clinically meaningful BMI is too? Would a change of 1-2 be important?

  • 1
    $\begingroup$ interesting - well thought! whether "important" could depend on what interests you, and may could only be answered after your model is complete. from the perspective of study design, you might run into the question of how sensitive your study / model should be (e.g.: should it be able to possibly be able to measure whether a BMI difference of 1-2 should have an impact on the entity you are interested). The relationship between BMI and age might depend on country etc, but at least in the USA is not linear (dqydj.com/… ) $\endgroup$
    – tsttst
    Mar 7, 2021 at 18:51
  • $\begingroup$ In (US) NHANES 2011-2012 data there is a significant quadratic relationship for 16+year olds: males: E(BMI) ~ 21.26 + 0.03age - 0.003age^2, females: E(BMI) ~ 18.2 + 0.47age - 0.004age^2, highly statistically significant. But the R^2 are not large (3% and 5%). This is with around n=3000 in each group. (I just pulled the NHANES data and ran the regression, so I don't have a citation). $\endgroup$
    – user11599
    Mar 9, 2021 at 1:18


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy