Closed steven-moore closed 5 years ago
Note that this runs when adding race_grp as another adjustment (screenshot below). To me this suggests a mistake in array processing somewhere, possibly triggered if eigenvalues suggest dropping BMI when there is only one variable in the array.
If this is indeed the issue, it likely did not come up before because all models (except one) used multiple covariates. The only exception was a gender-adjusted model, and gender is not likely to be dropped based on eigenvalues.
One more thing: This error also occurs on the production site, and so is not an artifact of recent changes.
Fixed in recent tests: issue now closed.
I am able to model BMI in relation to metabolites (see screenshot immediately below) but I am not able to adjust for BMI (per second screenshot). Perhaps this is an error in matching up covariate names with the translated name used for display: "BMI as a continuous measure"?