Hacker News Reader

Comment by sjducb | Hacker News Reader

This is a bit ridiculous. It's well known in medicine that members of different ethnic groups have different risks and complications.

This article describes how members of different ethnic groups have different complications in pregnancy. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7290488/

Here is a review about how race affects Adverse Drug Experiences (In response to commonly prescribed drugs), published in the journal of ethnic and racial health disparities.

https://link.springer.com/article/10.1007/s40615-015-0101-3

Sure it would be better to sequence each person's genome to accurately place them on the spectrum of human genetic diversity, but that's not practical in a hospital or emergency setting.

View on HN

> This is a bit ridiculous. It's well known in medicine that members of different ethnic groups..

Emphasis added. Race != ethnicity; TFA argues against using the former and suggests using the latter (with other things).

> Some of the key recommendations focus on getting rid of the use of race and instead focusing on what the report terms "population descriptors." These can be things like ethnicity, region of residence, and so on.

Image you've got a black patient in front of you and you're choosing between two drugs.

One of the drugs has a greater risk of nasty side affects for the black patient.

Which drug do you give the patient?

Should the research that tells you which drug to give be banned?

Not a great argument. Drug response is based mostly in genetic history, partly in life experience (IE, if you live near a highly polluting plant and develop an allergy while growing up there). While many papers on drug response use race as a variable, that variable is highly confounded and is often not sufficient to make an accurate determination for medication prescription purposes.

Such research is not being banned or proposed to be banned- in fact, there's a lot more interest now in understanding the relationship between genetic history and response to disease, but the work is still fairly primitive.

You're right that race often has a small (but statistically significant) affect, because of all of the confounding variables.

Imagine that you're a hospital who takes predominantly black patients, if you could modify the standard treatment to reduce Adverse Drug Experiences by 5-10% would you do it? Or would you say that it's unacceptable to use race as a variable at all, and continue to give the black people the same treatment as white people, resulting in more Adverse Drug Experiences in your black patients?

You can say that Ancestry is better, but often neither you or the patient actually know their ancestry.

> Or would you say that it's unacceptable to use race as a variable at all,

You're arguing against a claim that neither TFA nor anyone here made. TFA doesn't suggest doctors should ignore past research that examined race - it argues that future research should examine specific identifiers instead of made-up ones.

> often neither you or the patient actually know their ancestry

Sure - which is why using race as a proxy for ancestry is a bad idea. If a drug has known risks to people of West African ancestry, and you have a patient with unknown ancestry, then you don't know whether that drug presents risks to the patient. Whereas, if the research on that drug had only reported "risks to black people" without examining ancestry, then you would believe you could tell by looking at a patient whether the drug presented a risk - and you'd be wrong.

Imagine your patient is Obama.

Which drug do you give him?

Barack Obama has around 50% of his ancestors from Africa, most 'black' americans have a lot less than that. Example: if one of grand parents is black, one of your parents is, so you're black.

This is know as the one drop rule, if one of your ancestors is black you are black. So maybe 95% of your ancestors are from Europe and you are still considered as black.

Latin has even a fuzzier definition. Latin american countries have been build -as the US- with immigration. Immigrants from latin american countries are white, black, asian, etc... Or whatever classification the USA might invent. There is no less diversity in the rest of the Americas than in the USA.

> Should the research that tells you which drug to give be banned?

No, and nobody suggested anything like that, so if you're trolling please don't.

The point of TFA is that a study reporting "drug X has greater risks for black people" is weak science, because any given doctor has little way of determining whether their patient is genetically similar to whoever participated in the study. If the study had instead reported risk levels for, say, people of North African ancestry, then the doctor could ask about the patient's background and make an informed decision, instead of just looking at skin color and hoping for the best.

“Race” is a poor proxy for ancestry. If ancestry is what you care about, look at ancestry. If cultural and environmental factors are what you care about, then look at that.

From your cited link:

> However, few of these studies were specifically designed to evaluate racial or ethnic disparities, lacking a standardized approach to racial/ethnic categorization as well as control for potential confounders.

The NIH study that the article is about provides concrete recommendations that would help alleviate the precise concerns expressed by your article.

You seem like you might be responding to the headline rather than the underlying content of the NIH study.

Reply on news.ycombinator.com