While many commentators who supported the approval of BiDil for black patients state that “race” is not a scientifically precise term for identifying relevant genomic or physiological characteristics that differentiate population groups, nevertheless, they argue that “self-identified race” is a useful proxy for those characteristics. However, what is the evidence that the proxy “self-identified race” is a reliable surrogate? The best evidence derives from the fact that genetic variation conferring disease susceptibility is not equally distributed among ancestral populations. For example, sickle cell anaemia is more prevalent in populations whose ancestry can be traced to sub-Saharan Africa. However, “self-identified race” is a subjective term, influenced by cultural factors, and not even grounded in the ancestral genomics of, for example, the International HapMap Project. For the purpose of the clinical trials, “self-identified race” is interpreted as a dichotomous variable (black or non-black). If race were used as a proxy for ancestral African genomics it should be a continuous function (10%, 30%, 70%, etc). It makes no scientific sense to map a continuous function onto a dichotomous variable…
Sheldon Krimsky, “The short life of a race drug,” The Lancet, Volume 379, Issue 9811 (2012-01-14 through 2012-01-20): 114. http://www.thelancet.com/journals/lancet/article/PIIS0140-6736(12)60052-X/fulltext.