The study itself is interesting - sleep duration and risk of dementia, lots of follow-up, decent sized sample (although relatively few events) nature.com/articles/s4146…
But the headline is super misleading for so many reasons. My faves:
1. absolute risk is really small (~1 case per 1,000 person-years) 2. The authors acknowledge later in the article that they don't know if this is causal or not
Also, tons of previous research has suggested that there is a bi-directional association between sleep and dementia risk (i.e. too little or too much is bad) which the authors also acknowledge
Better headline might be "Lack of SLEEP may increase your dementia risk by 0.1%, however other lifestyle factors cannot be excluded as the real cause"
• • •
Missing some Tweet in this thread? You can try to
force a refresh
2021 will hopefully be the year that the armchair epidemiologists stop being wrong about infectious disease, excess mortality, etc, and move on to being wrong about something else
Maybe economics?
To clarify, because of course I need to (sigh) this is a joke about the twitter randoms who have deemed themselves experts not a critique of interdisciplinary work
I'm currently working on a paper with 3 economists, an immunologist, a demographer, and 2 statisticians on COVID-19. Non-epis have great and valuable insight!
One interesting point is that this article gets several facts wrong. Whether that detracts from the commentary on science or not is I suppose up to the reader
This statement, for example, isn't really true. The U.S. has had school closures much less severe than (for example) South Korea, or a dozen other places. The reference only talks about Europe!
"Studies have repeatedly concluded" - links to a tweet, and two articles on teacher's unions. There are many studies that have concluded precisely the opposite
"Chronic disease has caused COVID-19 deaths, if we didn't have so much diabetes fewer people would've died" - incredibly dumb argument for many reasons, not least that it is true of LITERALLY ALL HUMAN DEATHS
Yes, if we had solved the biggest medical issue of the modern age fewer people would've died of COVID-19
What of it?
I mean, seriously, we've been trying to 'fix' NCDs for decades, and while they are in theory somewhat preventable they are still a large and growing problem in most places in the world
The basic idea here is that we could be either undercounting or overcounting COVID-19 deaths
I think the most likely explanation is some combination of the two
Based on some very careful examinations of death reporting systems, we can say that there are probably some portion of deaths that are recorded as due to COVID-19 but were not caused by the virus
2/n The article in question is a review of face masks. At face value, it's essentially an opinion piece arguing that masks are ineffective ncbi.nlm.nih.gov/pmc/articles/P…
3/n Digging a little bit deeper, some of the stuff in here is pretty obviously wrong. For example, this incorrect statement about 99% mild/asymptomatic is referenced to Worldometers (not a specific graph, just the site)
Many places in India had very large initial epidemics, and it has been proposed that some cities (i.e. Mumbai) might have been at or near herd immunity towards the end of last year
The new massive waves even in areas with many infections before raise several possibilities, none of them great
1. reinfections 2. variants 3. large over-ascertainment of past infection