An earlier study looked at BF% vs BMI-based obesity definitions, when you adjust for fitness (as measured with a treadmill test). They compare the all-cause-mortality risk for normal weight + metabolically healthy vs obese (for each definition) metabolically healthy. When adjusting for fitness, the risks / hazard ratios are not significantly different.
I.e. if you do someone's bloodwork and a treadmill test, and they look metabolically normal, knowing whether their BF% is high (or if their BMI is high) doesn't seem to tell you more about whether they're going to die. If they're fat but fit, trying to lower their BMI or BF% doesn't seem to be about making them healthier.
If we're talking about interventions which might involve putting someone on medication with side effects for the rest of their lives, just do the extra tests! There's no reason to reduce the decision input variables to just "are they obese?" when better indicators are known.
I.e. if you do someone's bloodwork and a treadmill test, and they look metabolically normal, knowing whether their BF% is high (or if their BMI is high) doesn't seem to tell you more about whether they're going to die. If they're fat but fit, trying to lower their BMI or BF% doesn't seem to be about making them healthier.
If we're talking about interventions which might involve putting someone on medication with side effects for the rest of their lives, just do the extra tests! There's no reason to reduce the decision input variables to just "are they obese?" when better indicators are known.
https://pubmed.ncbi.nlm.nih.gov/22947612/