4 ms·
Not an expert, but for me it looks more like confirmation bias contaminating meta-analysis. Covid-19 has a lethality around 2%, so a huge fraction of infected p
by marcodiego 5y ago
Not an expert, but for me it looks more like confirmation bias contaminating meta-analysis. Covid-19 has a lethality around 2%, so a huge fraction of infected people can be cured using any placebo. This very low lethality makes data with a bigger number of deaths very easily skewed or noisy.
So, it is very hard to derive any efficacy data except in cases where the improvement is significant, like in the cases of current vaccines.
If a placebo reduces lethality from 2% to 1% that may sound like a 50% reduction, but actually it is just a 1% improvement. This can only be ruled out with higher p-values which demand bigger samples.
There are hundreds of teams in the world right now trying many different drugs to treat Covid-19. It possible to use a meta analysis approach and carefully select works where statistics were favorable by mere chance. This is rare, but considering the enormous number of works, it is possible to find them.
A more trustworthy meta analysis can be derived if peer-reviewed published papers are also chosen randomly. But of course, such meta analysis will also be biased by the kinds of papers that are likely to be published after peer review.
- elliekelly 5y agoIf even half of what is alleged in the linked article is true it’s quite clearly scientific fraud and not anything that could be explained away by “confirmation bias”.
- marcodiego 5y agoYes. I just want to comment that meta analysis is not a panacea. In brazil, where these drugs and treatment became polarized and politicized, some people has been using meta analysis to prove that even cloroquine works against covid.
- tomByrer 5y ago> Covid-19 has a lethality around 2% ?!? I thought it was 0.3%? Citation? Anyhow, all those stats are askew; not everyone who got C19 got tested, but most who died (from any cause) did. > that may sound like a 50% reduction, but actually it is just a 1% improvement Well, you just described how the C19 vax say how they improve survival rates. https://youtu.be/Jkwn5I8tLmE?list=PL2S-_LNXy9Vx24_o9zSTF6MADGr_4upKN https://youtu.be/Jkwn5I8tLmE?list=PL2S-_LNXy9Vx24_o9zSTF6MAD...
- marcodiego 5y ago> ?!? I thought it was 0.3%? Citation? Anyhow, all those stats are askew; not everyone who got C19 got tested, but most who died (from any cause) did. At least in brazil, lethality is under 3% ( https://ourworldindata.org/coronavirus/country/brazil https://ourworldindata.org/coronavirus/country/brazil ). Considering under reporting, you're right, this number can be even lower. > Well, you just described how the C19 vax say how they improve survival rates. In brazil, coronavac has 100% efficacy against death cases during the test phase. Reducing lethality from 3% to 0% is very different than reducing from 2% to 1%.