Read on Twitter
Inbox is filled with this a metaanalysis of mortality by Dr. Andrew Hill at Liverpool re: ivermectin. Easily accessible drug, safety profile, cheap. Some data at beginning at pandemic looking at in-vitro BUT concentrations needed to replicate would be fatal. Major issues here:
One big caveat to all of these trials - time to clearance of RT-PCR is a terrible outcome. Yes high viral loads are probably not a great prognostic sign, but their clearance is very controversial. Many healthy or even asymptomatic individuals shed for > 3-6 months
1) Study 1: Mahmud et al. Only printed here: https://clinicaltrials.gov/ct2/show/results/NCT04523831
Issues: Mortality is literally 0 vs. 3 out of 180 in each arm. No baseline risk / characteristics. Very well could be randomization alone, and point estimate leaves a lot to the imagination.
Issues: Mortality is literally 0 vs. 3 out of 180 in each arm. No baseline risk / characteristics. Very well could be randomization alone, and point estimate leaves a lot to the imagination.
2) Study 2: https://www.researchsquare.com/article/rs-100956/v1
Issues: No real data on how randomized, and significant imbalances in underlying comorbidities (IHD in the highest risk of death group). Control is HQ so unclear if the drug helped or HQ increased mortality. So could be all balance/control fx
Issues: No real data on how randomized, and significant imbalances in underlying comorbidities (IHD in the highest risk of death group). Control is HQ so unclear if the drug helped or HQ increased mortality. So could be all balance/control fx
3) Study 3: https://www.researchsquare.com/article/rs-109670/v1
Issues: No PCR in 40-50% of controls. I'm sorry, but having GGO on CT and a fever/cough is not always COVID, and may be a prognostically very different (aspiration, other viral, brucella, tb etc). No baseline comorbidites, small death no
Issues: No PCR in 40-50% of controls. I'm sorry, but having GGO on CT and a fever/cough is not always COVID, and may be a prognostically very different (aspiration, other viral, brucella, tb etc). No baseline comorbidites, small death no
4) Study 4: https://www.medrxiv.org/content/10.1101/2020.10.26.20219345v1
Issues: randomization (odds and even days), signal was seen in severe (0/11 vs 6/22) not significant/small numbers. Could all be doxy effect. Still could all be statistical/balance between who was offered therapy/enrolled
Issues: randomization (odds and even days), signal was seen in severe (0/11 vs 6/22) not significant/small numbers. Could all be doxy effect. Still could all be statistical/balance between who was offered therapy/enrolled
So if you put it all together - a brilliant effect. But the power of meta-analysis means more cases. But using mortality as a hard outcome here (which given the heterogeneity of these studies and outcomes the only choice) shows significant gaps. This evidence is very low grade
I would love nothing else for this pandemic to end with ivermectin, seriously- it's a cheap, safe, and globally available drug. And i'm looking forward to the many trials coming about, hopefully as a real RCT, and if there is a signal incorporating it into SOLIDARITY
But for now - I'd make sure everyone does their own appraisal. This is complete echoes of what happened in hydroxychloroquine - where people raced to prescribe it offline, rather than study it in trials. There is a higher standard here for all.
https://jammi.utpjournals.press/doi/10.3138/jammi-2020-06-22
https://jammi.utpjournals.press/doi/10.3138/jammi-2020-06-22
