The genetic "races" do not correspond to the social ones, however.
There is no "concept" of genetic race in which all the white people are one race, all the yellow people are one race, all the red people are one race, all the brown people are one race, and all the black people are one race.
That is very true. And it sort of goes back to the earlier point about the boundaries that define race being fuzzy and more or less arbitrary. However, it's also true that while the correspondence between the genetics and the social constructs is highly imperfect, there
is still a certain degree of correlation. I'll expand on the implications of this below.
And that is one of the major difficulties facing attempts to correlate things like disease with genetic race, using data classified by social race. The correlation between the genetics and the sociology is difficult to make rigorous.
If, as a doctor, you commonly suspect (and routinely test for) sickle cell among the children of immigrants from Micronesia or Australia, because they are black, and overlook the symptoms in children of immigrants from northern India or Turkey, because they are not black, you will be making mistakes.
It is indeed difficult to make the correlation completely rigorous -- however, that doesn't mean the social constructs aren't useful in medical situations. For example, let's consider the situation in the United States, which has a very diverse ethnic composition. Based on 2003 statistics (
1,
2), 1 in 12 African Americans has at least one of the sickle cell allele (being a recessive trait, it takes 2 alleles to develop the full disease). However, in the general US population the prevalence of the allele is 1 in 147. This means that, without any other knowledge, we know that a patient is over 12 times more likely than the average patient to have the sickle cell allele simply by virtue of being African American. I couldn't find precise statistics on the full blown disease, but presumably the relative proportions are about the same -- the image below suggests that this is indeed the case, although I couldn't locate the statistical source.
This is important for medical diagnoses
not because it determines who gets tested; a doctor who neglects to test certain groups in the way that you described will indeed make many errors. Rather, it affects medical diagnoses because knowing the base prevalence rates in the population to which a member belongs is very important for interpreting the results of medical tests. This is because medical tests are not perfectly accurate -- they sometimes give false positives and false negatives, and these accuracy rates can be and are regularly quantified.
Minding the relevant base rate in interpreting a medical test is a very unintuitive concept, so to wrap our heads around it, consider the following hypothetical problem.
If a test to detect a disease whose prevalence is 1/1000 has a false-positive rate of 5 percent, what is the chance that a person found to have a positive result actually has the disease, assuming that you know nothing about the person’s symptoms or signs?
The most common answer -- given by the majority of respondents when this problem was first presented by researchers Cascells, Schoenberger and Graboys in 1978 -- is 95%. The correct answer, pursuant to
Bayes' Rule, is 1.96%. This is most easily explained by imagining that there are 1000 people, of whom 51 will test positive (50 false positives and one true positive) and of whom only one actually has the disease (prevalence is 1/1000). 1 out of 51 equals a probability of 1.96%.
There are two reasons why this is important in the current context. One is that the correct probability of having the disease is sensitive to the base prevalence rate one uses in the calculation. In the case of sickle cell, the calculation for an African American patient would use a much higher base rate, and the output would be a much higher probability of disease. But wait -- why wouldn't you just begin treatment on anyone who receives a positive test result, just in case? This is the other reason why base rates are important. Treating a disease that one doesn't actually have can be costly, dangerous, or both. It is not a decision to be made lightly, and treatment should be pursued only if the probability of disease exceeds a certain threshold which is to be determined by the patient and/or doctor.
Using the base rates ascribed to various socially-constructed race groups is entirely prudent. It is, in fact, the doctors who ignore race-based differences in disease prevalence who will make the most errors.