I agree doctors aren't supposed to uphold a particular conceptualisation of the natural order a priori. To practice Medicine, one must generalise though: Some people can tolerate high salt diets quite well, some smokers can live to 90 without issue - but in the majority of cases, this will be bad for your health and cut a few years off. Simply put, we must create an hypothetical Average Man, with average anatomy and average ranges of electrolytes and such. In this manner, we can pick up when someone deviates from this basic norm - then, if such a deviation is coupled to an obviously undesirable outcome, such as death or disability or decline of function, we can label it pathological or unhealthy, and then taylor treatments in order to alleviate suffering. Some people will be fine with aberrant findings, like physiological bradycardia, but we must play average to see where to look.
So for instance, we see high salt diets cause high blood pressure, which we can conclusively tie to more myocardial infarction and stroke; thus we can advise people against high salt intake, especially if already having high blood pressure. Doesn't mean it will make a difference in this person, but for the average one it will. This aetiology can become quite complex. It is not that we are against salt, but we can tie it to an objectively undesirable outcome.
Here we have population groups with obviously undesirable outcomes by any measure - lower life expectancy, poorer levels of physiological functioning, higher suicide rates - than their peers with similar diet and socio-economic circumstances; Yet I haven't touched idealogy or Science, but merely looked at Empiric Evidence. These are obviously unhealthy in some way therefore, if not perhaps pathologic processes. Now the question is whether this is an environmental process or maladaptive or an innate one. So far no intervention has shown promise to alleviate this suffering, the higher burden of obviously undesirable outcomes, in these populations. In certain cases, such as Transgenders and gender reassignment surgery, we see subjective reports of improved mood, but no drop in suicidality - at the price of complex elective surgery, that bears very high risk of morbidity and mortality itself; or these hormonal treatments that increase cardiovascular risk, and have been shown to dramatically cut life expectancy. The number to treat, or risk/benefit stratefication is not in its favour. With Atheism there are no such strategies that I am aware of, to close the health gap to their religious peers, though. Anecdotal cases may do 'better' on some criteria, as some people with inappropriate sodium secretion may do better on high salt diets, but this doesn't change the average Ideal.
Regardless, for doctors to be able to alleviate suffering at all, it means generalising a 'healthy range' for things. Without touching idealogy, by looking at the most basic factors of mortality and morbidity, these populations are at higher risk. There is an old adage in Medicine of 'treating the pasient, not the xray or blood results', but as I said, I don't see much 'thriving' because of it; more in spite of, if at all. Being asked to ignore the plain evidence in front of us, necessarily averaged out over the population in general, is ignoring suffering. It is ignoring a potential factor in human suffering that we may be able to alleviate by adressing, simply to uphold a specific conceptualisation of the Natural Order and ideology. You can disbelieve in generalising "in that manner", but how are we supposed to decide when we may or may not generalise, if the end result can be shown on mortality statistics? The high suicide rates of Transgenders scream out for intervention strategies outside of ineffective treatment modalities of pandering to it or gender-reassignment for instance, and just ignoring this is highly unethical and against medical principles of beneficence.