Sparklefloof
To start with there are no "races".
Any two black people can be genetically more different from each other than any white person in relation to a black person.
Then, the genetic differences between "races" are smaller than between individuals.
But, it still remains that you simply should not use population data to make decisions to individual patients.
It's fine to start more aggressive treatments if an ethnic group is usually more at risk, but you still do not know if it is necessary for that particular individual. It's just fine to go with a more aggressive treatment because it's not harming anyone.
But, not to be concerned about a person from the other group, because that group is less at risk, is dangerous as you could be missing someone who would need a more aggressive treatment.
And it's certainly not acceptable to dismiss results in one individual just because they are within normal for their ethnic group. Because within each group there is still a range of "normal" for different individuals.
Do look up ecological fallacy.
It is different if you want to run campaigns to fight diabetes, for example, or screen certain genetic diseases.
Population data could also, at best, inform some individual decisions. Say, start by using one treatment that works better in that ethnic group. But there are no guarantees that it works for that individual.
Population data should never determine decisions regarding individual treatment.
I'll give you an example. Blood groups have different frequencies in different world regions. If blood group B is more frequent in ethnic group Y, it would be silly to pick one random person from ethnic group Y and assume that this person had blood group B.
When health systems deal with patients according to their ethnic groups it's mostly for cost-effort/benefit reasons, not for the benefit of the patients.