Something that I keep seeing pop up is the idea that meta-analysis somehow eliminates issues with the underlying research
This is just confusingly incorrect
Meta-analysis really is just a simple statistical aggregation of results. Indeed, the most basic way to meta-analyze studies is just to use a mean or median of their point-estimates
But simple means/medians can be misleading - we don't really want a study of 10,000 to have the same weighting as a study of 10, but that's what a simple average provides
When we run a meta-analytic model, what we are actually doing is generating a WEIGHTED mean/median, and confidence interval. Essentially, we take all of the means and SEs, and based on that our model weights them with bigger studies generally contributing more to the model
You'll notice I've said nothing about the underlying quality of the evidence
That's because META-ANALYSIS HAS NOTHING TO DO WITH QUALITY OF EVIDENCE
You can throw anything into a meta-analysis model. Here's a model I just ran on the ratio of hosting to participating in the summer Olympics. This is meaningless!
We tend to put meta-analyses on a pedestal, but the fact is that statistically aggregating evidence is a total waste of time if that evidence is all bad
This recent Cochrane review is a perfect example - they looked at the evidence for ivermectin for COVID-19, but because most of it was terrible they only included a few studies in their model
This is also where the phrase "garbage in, garbage out" comes from. If your meta-analysis includes numbers from studies that are terrible, the final point estimate is as meaningless as my graph above on the Olympics 🤷♂️
Some people think meta-analysis is impressive because it involves fancy statistical software, but it's entirely possible to implement a Dersimonian-Laird inverse-variance model in Excel with a stats textbook and a few hours of time
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Cochrane review on ivermectin just dropped. This is the current gold-standard summary:
"the reliable evidence available does not support the use ivermectin for treatment or prevention of COVID‐19 outside of well‐designed randomized trials."
Worth noting that the main finding of the review is that the available evidence is very low quality, and so low certainty that the impact of ivermectin could be anything from very protective to very detrimental to health (i.e. it could be saving lives or ending them)
Something worth noting - no matter how you look at it, the research community has royally fucked up when it comes to ivermectin
Yes, there are endless terrible trials, but we've had 18 months. WE COULD'VE RUN GOOD TRIALS
We didn't. That is woeful, and frankly inadequate
We spent endless billions researching hydroxychloroquine, and knew by August 2020 that it didn't work, but ivermectin which has been given to 100s OF MILLIONS OF PEOPLE...still no answer. Shameful
Another day, another two ivermectin studies that have impossible values in their tables of results. Both included in meta-analyses (rated as high risk of bias this time)
This is getting truly insane
The entire ivermectin for COVID-19 literature is just...awash with this absolute trash. Studies that are pre-registered after publication. Tables of results that contradict the text. Percentages that are impossible
Endless examples of RCTs where the analysis plan is a single sentence long (we did chi-squared and some t-tests, p<0.05 was significant)
People have claimed that ivermectin is 100% effective as a prophylactic for COVID-19 based on an observational trial of the drug
I'm extremely concerned about this trial. It definitely should not be used as evidence for anything 1/n
2/n The paper is here, and it's actually two studies - a pilot and a larger interventional trial where the authors gave ivermectin+carrageenan to one group of healthcare workers, and compared to a "matched" control media.marinomed.com/8b/7a/c7/nota-…
3/n The authors report that 0% of the people treated with ivermectin+carrageenan got COVID-19, while in the pilot 11% and in the full trial more than 50% of the control group got sick
A fairly well-demonstrated fact about COVID-19 and lockdowns is that the more severe restrictions tend to have smaller marginal benefits once enacted
In practice, this means that the reduction in transmission for, say, limiting travel to 5km radius is probably quite small once you've already implemented other interventions such as stay-at-home orders
In other words, when we talk about 'hard' vs other lockdowns, the language often doesn't match the evidence. There's no strong reason to believe that adding small extra limits on behavior will have a large impact on the spread of COVID-19
The authors used a huge database of people who'd been treated for COVID-19, and looked at the associations between mortality and 84 drugs or drug classes
They found that metformin, nicotine, fluticasone, SNRI, expectorants and ondasetron reduced the risk of death from COVID-19, but given the potential for residual confounding only focused on ondasetron