-
-
-
The study was a very simple cohort study. They took 3600 kids who were born in 1975/76, followed them for 40 years, and split them up based on how stressed their mothers reported being before they were bornpic.twitter.com/iILZbYkTvB
Show this thread -
After correcting for a modest number of confounders, they found that kids whose mothers were stressed when they were pregnant were more likely to have mental health problems when they grew uppic.twitter.com/jLgOLhB9BZ
Show this thread -
Basically, when mothers were stressed their kids had worse mental health To be honest, that's not a huge finding in my opinionpic.twitter.com/gwj1IdxgkJ
Show this thread -
Thing is, this study also had some ~serious~ problems For one thing, the sample size seems big, but the number of kids who developed mental health issues was just 40 out of 3600!pic.twitter.com/J6ixcqTm9S
Show this thread -
They also really didn't control for many things - while this may have been due to sample size, it seriously limited the conclusions
Show this thread -
They didn't, for example, control for maternal alcohol or drug intake, which can both have a HUGE impact on childrens' mental health
Show this thread -
The absolute risk increase was also pretty miniscule Kids exposed to no stress had a 0.5% risk of developing personality disorders Kids exposed to some stress had a 1.6% risk Triple the odds, but only 1.1% risk increase!pic.twitter.com/qTpdRcW2oP
Show this thread -
Also, the study assessed stress using one questionnaire given partway through the pregnancy in 1975, which I'd argue is...problematic
Show this thread -
On top of all of this, the study used a statistical technique called logistic regression Without going into the details, this is very questionable for a study like this
Show this thread -
Basically, if you have too few outcomes (i.e. only 40) in your sample, and include a bunch of covariates in the statistical model, you end up with a bit of a meaningless result I mean, just look at those confidence intervals!pic.twitter.com/kIKGKcoSov
Show this thread -
(As a side note, there appear to be a few numerical errors in the paper, such as this one in the table above)pic.twitter.com/XB7m1aiQ20
Show this thread -
If you read that table, kids exposed to "any" stress have 3.37x the odds of having any diagnosis of personality disorders, but the confidence interval goes from 1.61x to 7.07x That's ~huge~
Show this thread -
TL:DR: - association between stress in pregnancy and kids getting personality disorders - very weak - risk increase is tiny - statistical analysis is problematic - HUGE potential for residual confounding - ALMOST CERTAINLY NOT SOMETHING FOR MUMS TO WORRY ABOUT
Show this thread -
Now, while the media certainly had a role to play in the misinterpretation of this study, I think the problem started in the conclusions This is not well worded at allpic.twitter.com/cIm8jvpLvi
Show this thread -
The authors have been quoted elsewhere saying that this definitely isn't causal, but if you read the conclusions of the study it sounds like it absolutely is That's a huge problem
Show this thread -
Show this thread
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.