machinelearny | 9 points
Ivermectin study outliersI've realized that with so many studies on Ivermectin, it has become quite clear what the expected outcomes should be. Any studies with a RR outside of the overall CI or that doesn't at least have it's CI overlap with the overall CI must have some problem.
I was wondering of anybody can help explain the ones that fall outside (marked in red)
Early Treatment:
https://preview.redd.it/dy4oh5m1d9t61.png?width=1343&format=png&auto=webp&s=2097cb7000876b95d0235778c5307dba6b6dc912
The Espitia-Hernandes study seems to indicate that ivermectin + AZ + cholecalciferol could be much more effective than IVM alone.
Chaccour is single treatment with 28 day symptom resolution, so the single treatment probably reduced the effectiveness. But I have also noticed that 28 day symptom resolution in mild covid has not been a great outcome for Ivermectin.
Any ideas on why the Roy study was such a no-show for IVM?
Prophylaxis:
https://preview.redd.it/iso0rww4d9t61.png?width=1350&format=png&auto=webp&s=257cef158c45cc858ec4943e914679e118199bc4
Behera study seems to report better results than shown in the IVMeta results, at least for those taking more than one dose. Did ivmmeta take the results include that of the single dose only?
Carvallo: I presume the addition of the iota-carrageenan and the oral and nasal spraying etc. made the difference?
Bernigaud: Probably the high dose frequency and the mortality outcome, which we know IVM is very good at preventing death, so that's probably the reason it did so good in this study?
Late:
https://preview.redd.it/ja7fm4ixc9t61.png?width=1343&format=png&auto=webp&s=cd15a17dc6a4fb1ee7824dbe4947e3ff9ab07fd2
Here it seem the is a clear pattern. With mortality and ICU outcomes the results are better and with recovery time or other recovery stats they are not as good (also more subjective...).
The one outlier here seems to be Soto-Becerra, I would not even call Kishoria an outlier, since it actually matches pretty well with the other recovery time outcomes and since it is so underpowered the CI overlaps well with the other recovery based outcomes.
So Soto-Becerrra would be due to it being a retrospective study, with people being treated being more likely to have severe disease and their propensity matching algorithm being a bit shit, with it not even using covid severity as part of the matching criteria... So it should probably be given a very low weight in a meta-analysis. My conclusion is that late treatment mortality RR is probably quite a bit better than .5
Looking at late stage Mortality only:
https://preview.redd.it/h1r5hidtc9t61.png?width=1355&format=png&auto=webp&s=19b0f3e7f503615dfb88a91a86dfcc358c2f0a89
We have the following outliers:
-Sotto-Becerra: Already mentioned, a bit of a crappy study, can't make much from it.
+ Elgazzar: Not really sure why they did so well, but the CI covers the expected range.
+ Budhiraja: This is probably just due to good luck and a small treatment arm. It feels like the CI should be lager than what is shown in the forest plot, if they included 1 more person and that person died, the RR would have reduced to about 27%
- Okumus: Compared with HCQ, so might have influenced results. Also had quite a high death rate, so probably very severe patients. Quite small, so CI overlaps the expected range.
- Beltran: Low single dose in extremely severe disease, probably the reason for the small effect. Again underpowered resulting in CI overlapping with expected range.
Let me know what you guys think.
It seems quite clear that IVM's strength is in preventing death. The most impressive results would probably be found in a very early treatment or prophylaxis study of extremely high risk patients with Mortality as primary outcome giving 0.3mg/kg once a week for prophylaxis and 0.4mg/kg on day 1,2,4,7,14, 21, 28 for early treatment.
If you want to design a study to fail, you would do early treatment in healthy young patients with no comorbidities, single dose treatment as late as possible (delay your consent and randomization process) while still claiming early treatment and take complete resolution of symptoms by day 28 as your primary outcome. To make sure it fails, do it in a region with high IVM use and don't test serum levels so you don't know how much the control group is self medicating with IVM.
[-] SwiftJustice88 | 4 points | Apr 15 2021 04:31:46
Just a small note from me on the Carvallo study, I think nasal sprays and gargling are quite important in COVID prevention. Povidone Iodine seems to be pretty good at killing COVID where it first starts to take really take hold and I bet Iota Carrageenan is very similar. It seems like IVM would act as a solid second line of defense in someone using an antiviral nasal spray multiple times per day which is why the results were astonishing.
permalink
[-] machinelearny | 9 points | Apr 15 2021 11:25:33
Yes, and there has been multiple nasal spray studies showing effect, even simple and very safe nasal sprays. I wonder why that hasn't become a worldwide recommendation? How can an already approved and safe anti-viral nasal spray be spun to have some hidden danger?
If I'd been in government, after the first positive study of nasal sprays, I would have told my constituents, new rules!
Then when ivermectin showing benefit as second line of defence, I would have recommended weekly prophylaxis with .2mg per kg for all higher risk individuals. And I would have done a mass campaign prophylaxing everybody willing for 1 month at the same time to try and stamp out the disease completely.
I would have tried it first in a smaller county or region and then expanded to larger areas if (when) it showed success.
Anyways, those are just pipe dreams :)
permalink
[-] SwiftJustice88 | 2 points | Apr 15 2021 13:58:21
I think this combination would have been extraordinarily effective but like you said the likelihood of mass adoption at this point is minimal. I can’t remember what country it is, but one recently developed a nitric oxide nasal spray that I’m very interested in.
permalink