"Wait! You're both right!" ;-)
Part of the problem is that "race" is used in both colloquial English and in technical speech with rather different definitions. It is used technically as an informal synonym of "subspecies", and is similar in meaning to "breeding population"; the rough meaning is a group that has small but significant (for recognition, habitat choice, etc.) genetic differences from other populations.
In general, colloquial English, it generally just means any recognizable group, often but not always with superficial but visible differences in appearance. But sometimes it's even less meaningful that that. In the US, one of the commonly-recognized "races" is "Hispanic", which of course is based on language and culture, and can refer to a person of any subspecies who speaks with a Spanish-sounding accent. This is a good illustration of how meaningless the term is in its general usage.
There has been a general effort by biologists to avoid the use of the term, and it's not really officially part of standard terminology. But "subspecies" isn't understood by the general population, and "race" is one syllable rather than three, so the informal usage continues, especially when talking to the general public.
In the case of medical discussions of problems with a genetic component, it's common to use "population" instead, but that has four syllables. ;-) It's mostly useful if the population can be given a name. Thus, Tay-Sachs disease seems to be from a mutation that appeared first in a small eastern-European Jewish population, which of course isn't a "race", but is a recognizable and namable sub-sub-population. Knowing this is medically useful in diagnosis, since it gives a good hint at which tests might be useful when the appropriate symptoms are seen.
Sickle-cell disease is an interesting case, because its population is semi-genetic: It developed in malaria-plagued areas of central Africa, and partly defends against that disease. The population isn't actually a subspecies, since it includes people who are genetically quite different from each other. But the mutation(s) involved spread from group to group because it gives some immunity to malaria (at the cost of many early deaths from sickle-cell disease). Again, knowing about patients' central-African ancestry is medically useful in diagnosis, since it hints at what tests to try first when the symptoms are seen. In the US, the term "African-American" is a racial classification that correlates well with this disease, so it's medically useful despite being biologically somewhat bogus.
The poison-ivy/oak/sumac/etc. problem is actually a lot more complicated, and much of the information about it is of poor scientific/medical quality. Interested readers might try googling for info about the Asian lacquer tree to find some of the most useful information. There is a wide variety of sensitivity to this toxin, and different human populations differ statistically in their reaction to it. There are sensitive and resistant people in all populations, and there are populations that are mostly at one or the other end of the spectrum. It's often medically classified as a "white" problem, but this is an extreme over-simplification. East Asians are often very resistant to it, but they generally know to give lacquered things a number of good washings before using them in ways that involve close physical contact, such as in tableware. Asian artists who work with this lacquer understand its toxicity fairly well, and know how to deal with it. But it's not a simple story, and there's a lot of mythology involved.