Steven Stovitz retweeted
#TWICPodcast update
You may have seen one of >3000k journal articles describing #obesity paradox: high BMI is protective in, say, AF or Heart failure
That is so weird, b/c exactly zero docs tell normal-weight pts with AF or HF to eat Cheetos or donuts
How can this be?
🧵
Steven Stovitz retweeted
EBM learning
Does my patient have SARS-CoV-2 infection? A reminder of clinical probability formulas
ow.ly/Yjim50EpbAz
#Coronavirus #Covid19
Steven Stovitz retweeted
"Causal inference aims to determine when we can infer that associations are or are not due to causal effects." Causal inference for clinicians buff.ly/2Xjrpze #ebm @BMJ_EBM @bmj_latest
Steven Stovitz retweeted
Causal inference for clinicians ebm.bmj.com/content/early/20…
Steven Stovitz retweeted
Does reluctance to use the word “causal” lead to scientists misrepresenting the goals & findings of their research? A new commentary calls for greater clarity of purpose & language in scientific studies goo.gl/sxUHwt
Steven Stovitz retweeted
This should be required reading for anyone trying to do research derived from retrospective EMR data. The possible biases HAVE to be considered & understood. xitter.cf/statsepi/status/965904…
Steven Stovitz retweeted
This may be the most important “study” of 2018. Seriously.
If you believe in the obesity paradox (or other paradoxes), read this…again and again>
ncbi.nlm.nih.gov/pubmed/2937…
#SelectionBias
h/t @statsepi
cc @PrashSanders
#HealthJournalism
@VinayPrasad82
Steven Stovitz retweeted
Thanks to Hailey Banack for a great workshop on quantitative bias analysis @UCSF_Epibiostat yesterday. Critical not just to recognize limitations but assess plausible magnitude. Check out Hailey & @ER_Mayeda's workshop at 2018 @societyforepi conference: epiresearch.org/annual-meeti…
Steven Stovitz retweeted
Healthy #weight was associated with 4–13 times higher relative risk of #diabetes compared with being #obese!
While apparently shocking, the study’s fatal flaw (selection bias) explains the ‘paradoxical’ finding.buff.ly/2FEAFVv
#obesity #epidemiology #Science #important
Steven Stovitz retweeted
New finding: "Pediatric obesity is clearly associated with a lower diabetes risk." ncbi.nlm.nih.gov/pubmed/2937…
Yet another paradoxical association explained by #selectionbias due to selection on a #collider (diabetes test).
Highly recommended paper by Stovitz, Banack, and Kaufman.
