So what would you add to better detect asymptomatic cases? Also they make a similar statement in the discussion.
-
-
Replying to @MoNscience @GidMK
Also their analysis included people with low symptomatologypic.twitter.com/f2aTy5F8E8
1 reply 0 retweets 1 like -
Replying to @MoNscience
Yes I'm not saying this is the fault of the researchers, nor that they're wrong per se, but the inherent issue is that using retrospective analyses of routinely collected data you're always at risk of selection bias
2 replies 0 retweets 1 like -
Replying to @GidMK @MoNscience
What we really need is a longitudinal cohort of the same individuals that has been followed over time, with a non-probabilistic sampling method to minimize selection bias, and who are tested (say) twice a month regardless of symptoms. But that is a really hard ask
1 reply 0 retweets 0 likes -
Replying to @GidMK @MoNscience
So I'm not saying the researchers could've done a better study using the data at hand, because I love this piece of research, but that to really answer the question we kind of need a different study design altogether
1 reply 0 retweets 0 likes -
Replying to @GidMK
I’m not seeing the selection bias in free-for-all testing but agree on looking longitudinal. But “they need more longitudinal data” can apply to most studies.
1 reply 0 retweets 0 likes -
Replying to @MoNscience
Because people choose themselves whether to get tested. While there may not have been a direct monetary cost, testing is certainly not without secondary costs including time, effort, discomfort, etc...
1 reply 0 retweets 0 likes -
Replying to @GidMK
But if the barriers to access testing are extremely low, people will seek tests after close calls/suspected exposures even without symptoms?
1 reply 0 retweets 0 likes -
Replying to @MoNscience
They will be more likely to seek testing than if there's a monetary cost, but the potential for bias is certainly there. Even in vast studies that offer free tests (i.e. ENE-COVID) AND follow people up with multiple phone calls 20-30% of people don't do it
2 replies 0 retweets 0 likes -
Replying to @GidMK
So the alternative is active regular surveillance...of the whole population of Denmark? That doesn’t seem plausible. And if a team went with a subset of Denmark, then they’d be introducing bias.
1 reply 0 retweets 0 likes
No, you could definitely conduct active surveillance without such a massive sample. The REACT-2 study is a beautiful example, although unfortunately they do not always re-test the same individuals over time
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.