Hacker News Reader

Comment by based69 | Hacker News Reader

Obese women’s offspring have double the risk of hypospadius, and women having babies after 40 have 4x the risk of hypospadius.

Metformin is a drug that men are probably 100x more likely to be on at 45 vs 25. Therefore men on metformin are far more likely to be reproducing with older (and likely obese) women.

I’m not saying this explains all the correlation, I’m just saying if the study doesn’t really make a good effort at addressing these confounders, its not worth reading.

View on HN

From the artcile: "The numbers were small — 13 metformin-exposed boys were born with genital defects. But after the researchers adjusted for factors including parental ages and maternal smoking status, they found a 3.39-fold rise in the odds of a genital defect. “The rate per se was surprisingly high,” Wensink says."

> after the researchers adjusted for factors including parental ages and maternal smoking status

The exact adjustments made can change the solidity of the finding into anything from "rock solid" to "complete trash". Seemingly solid studies are routinely trashed by third party method reviewers that the initial peer reviewers missed. So we'll know better when they get around to it.

>But after the researchers adjusted for factors including parental ages and maternal smoking status, they found a 3.39-fold rise in the odds of a genital defect

If they didn't adjust for obesity, then it's really pointless to attribute this to the drug. Age and smoking status is not enough.

Why would you assume they didn't? They most likely did.

From the quote it says they only adjusted for age and smoking.

The quote says the factors included age and smoking not that they only included age and smoking.

I did indeed not read that properly. I'd have to check the actual publication. It's at least suspicious that they don't mention it, since obesity is probably the first thing you should think of when saying diabetes.

Adjusting for confounders on an n of 13 is statistically ridiculous.

It's not n=13 it's 13 with genetic defects.

Also not adjusting for confounders is statistically ridiculous. As well as n=13 studies.

i'll call n the total number of patients. if n * 9 / 1000 = 13, you'll find out that n is around 1400, which is a pretty high number in discovery studies. Maybe not for diabetes though, i remember hosting anonymized data from a lot more patients two years ago, i felt this was THE subject all the medical founding was put into (that and air quality and its effect studies).

The 90 day window for taking the drug and seeing an impact is in the studies favour though.

> the researchers saw no effect in offspring of men who took the drug earlier in life or in the year before or after the 90-day window of sperm production. “It really has to do with taking it in that window when the sperm … is being developed,” says senior author Michael Eisenberg, a urologist at Stanford Medicine.

If that is the case (90 days makes a difference in outcome) why even adjust for age and smoking?

The link is probably not causal but it is still important. Meaning there is probably an underlying issue that causes both diabetes and genital abnormalities.

See my other comment where I link it to a functional zinc deficiency.

They obviously control for age, at minimum.

"obviously" is an assumption; it's better to read the paper to confirm this, of course.

Reply on news.ycombinator.com