8 ms·

Without randomized trials you cannot recommend a drug because you cannot exclude confounders. It seems wrong to me to even speak of "overwhelming evidence" with

by 13415 5y ago

Without randomized trials you cannot recommend a drug because you cannot exclude confounders. It seems wrong to me to even speak of "overwhelming evidence" without a randomized trial or other sound ways to control or identify for confounders (e.g. causal models).

mgamache 5y ago
Actually a meta analysis can offer better evidence then a single RCT. https://ascopost.com/issues/june-10-2014/randomized-trials-vs-meta-analyses-which-is-the-better-bet/ https://ascopost.com/issues/june-10-2014/randomized-trials-v...
- Tenoke 5y ago
  The best meta-analyses are those made out of RCTs at any rate.
- MilnerRoute 5y ago
  Wait a minute... This post is a debate between two opposing viewpoints. So it also makes the exact opposite point: The primary difference between a randomized controlled trial and meta-analysis is that the former “provide the highest level of evidence because they contain the least amount of bias. Randomized controlled trials reduce bias, while meta-analyses increase bias." This Hacker News story is about a recent meta-analysis. But an actual randomized control trial of Ivermectin in March (on 476 patients) found the duration of symptoms on treated and untreated patients "was not significantly different." https://jamanetwork.com/journals/jama/fullarticle/2777389 https://jamanetwork.com/journals/jama/fullarticle/2777389
  - ALittleLight 5y ago
    Important thing to know about that study is they are testing time to recovery for people with mild covid. I'm not anywhere near an expert, but two things stood out to me reading this study. First, time to recover in the ivermectin group was 10 days versus 12 in the placebo. The paper calls this "not significant" but it's not clear to me whether that's statistically not significant, and if so, why wouldn't it be, or if knocking two days off recovery isn't that meaningful. The other thing I noticed reading it was that only one person out of 400 died and that person was in the placebo group. I think the Brett Weinstein response would be something like - A. Ivermectin's best benefit is as a prophylactic not a treatment. B. Ivermectin should be administered as early after onset as possible to treat, whereas here there was some delay to get people organized, enrolled in trial, etc. C. It's hard to measure effect when the disease is mild.
    - bonzini 5y ago
      > why wouldn't it be Because you only gave the average time of recovery but not the standard deviation.
      - ALittleLight 5y ago
        I didn't see the standard deviation reported in the paper. Is it? I did see the innerquartile range and it was the same for both the ivermectin group and the placebo. To me that suggests that variation in outcomes is probably similar between the two groups.
        Fomite 5y ago
        They provide it in the form of a hazard ratio: "hazard ratio for resolution of symptoms, 1.07 [95% CI, 0.87 to 1.32]" So statistically, those on ivermectin recovered 1.07 times as quickly as those on placebo, but the uncertainty bounds are from 0.87 times as fact to 1.32 times as fast.
  - spadez 5y ago
    > But an actual randomized control trial of Ivermectin in March Please see this https://news.ycombinator.com/item?id=27566123 https://news.ycombinator.com/item?id=27566123
goldemerald 5y ago
While randomized trials are certainly the gold-standard for determining if you should give a new drug to treat an arbitrary disease, that process is far too conservative during a pandemic where a huge majority of the world cannot access proper medical care. Ivermectin is already massively produced, used throughout the world, and cheap to manufacture (compared to new anti-Covid drugs). Even months ago before the benefits were known, the risks of taking the drug were very small, especially when the treatment guidelines were to do practically nothing. Still, the many pieces of non-randomized evidence can still clearly point towards this drugs efficacy. The sudden drop in Covid cases in India as the clearest indicator to me.
- jsnell 5y ago
  > The sudden drop in Covid cases in India as the clearest indicator to me. What's the alleged connection here? Are you claiming that some non-trivial number of patients in India were treated with this? How many, and at what point after being infected? And just how did the treatment affect the case numbers? Generally you'd expect the vast majority of secondary infections to happen before the diagnosis, not after, so a treatment seems totally irrelevant to the case numbers.
  - spadez 5y ago
    >Are you claiming that some non-trivial number of patients in India were treated with this Yes, example from two states. https://indianexpress.com/article/cities/lucknow/uttar-pradesh-government-says-ivermectin-helped-to-keep-deaths-low-7311786/ https://indianexpress.com/article/cities/lucknow/uttar-prade... https://www.news18.com/news/india/mortality-rate-much-lower-goa-govt-defends-use-of-ivermectin-for-covid-19-treatment-3786590.html https://www.news18.com/news/india/mortality-rate-much-lower-... The testimony video in this page cites examples in many other countries https://covid19criticalcare.com/ https://covid19criticalcare.com/
    - jsnell 5y ago
      I see no numbers on how many people were treated on either of the first two links, and certainly don't intend to comb through the dross on the third link. So do you actually have a number? It doesn't need to be exact, just a credible source on the order of magnitude will be enough. Are we talking a thousand, a million, or a billion here? To reduce R by a factor of 2 by the use of a prophylactic drug, you'd need to have half of the population on a regime of the drug. It seems pretty obvious that did not happen if this kind of reporting is the most impactful there is. Just think of the logistics of trying to do that! India has a population of 1.4 billion. On the other hand, if they only gave the snakeoil to e.g. a single digit number of millions of people the reduction in the number of infections would be imperceptible.
wrycoder 5y ago
https://sebastianrushworth.com/2021/05/09/update-on-ivermectin-for-covid-19/ https://sebastianrushworth.com/2021/05/09/update-on-ivermect... I’ve included every double-blind randomized placebo-controlled trial I could find of ivermectin as a treatment for covid. Using only double-blind placebo-controlled trials means that only the highest quality studies are included in this meta-analysis, which minimizes the risk of biases messing up the results as far as possible. In order to be included, a study also had to provide mortality data, since the goal of the meta-analysis is to see if there is any difference in mortality .... What we see is a 62% reduction in the relative risk of dying among covid patients treated with ivermectin. That would mean that ivermectin prevents roughly three out of five covid deaths. The reduction is statistically significant (p-value 0,004). In other words, the weight of evidence supporting ivermectin continues to pile up. It is now far stronger than the evidence that led to widespred use of remdesivir earlier in the pandemic, and the effect is much larger and more important (remdesivir was only ever shown to marginally decrease length of hospital stay, it was never shown to have any effect on risk of dying). I understand why pharmaceutical companies don’t like ivermectin. It’s a cheap generic drug. Even Merck, the company that invented ivermectin, is doing it’s best to destroy the drug’s reputation at the moment. This can only be explained by the fact that Merck is currently developing two expensive new covid drugs, and doesn’t want an off-patent drug, which it can no longer make any profit from, competing with them.
Vinceo 5y ago
Does that mean that the covid vaccines shouldn't be recommended as there's no randomized trials on their long-term effects? Just curious as I'm not well versed on this topic.
- 13415 5y ago
  Covid vaccines have been tested with randomized trials (Phase 3 of the testing phase). Regarding "long-term", that obviously depends on how you define "long-term but there is no need to move the goalpost. Generally speaking, in evidence-based medicine drugs and vaccines are not recommended based on incomplete statistics or hunches.

by 13415 5y ago

mgamache 5y ago

Actually a meta analysis can offer better evidence then a single RCT. https://ascopost.com/issues/june-10-2014/randomized-trials-vs-meta-analyses-which-is-the-better-bet/ https://ascopost.com/issues/june-10-2014/randomized-trials-v...

Tenoke 5y ago
The best meta-analyses are those made out of RCTs at any rate.
MilnerRoute 5y ago
Wait a minute... This post is a debate between two opposing viewpoints. So it also makes the exact opposite point: The primary difference between a randomized controlled trial and meta-analysis is that the former “provide the highest level of evidence because they contain the least amount of bias. Randomized controlled trials reduce bias, while meta-analyses increase bias." This Hacker News story is about a recent meta-analysis. But an actual randomized control trial of Ivermectin in March (on 476 patients) found the duration of symptoms on treated and untreated patients "was not significantly different." https://jamanetwork.com/journals/jama/fullarticle/2777389 https://jamanetwork.com/journals/jama/fullarticle/2777389
- ALittleLight 5y ago
  Important thing to know about that study is they are testing time to recovery for people with mild covid. I'm not anywhere near an expert, but two things stood out to me reading this study. First, time to recover in the ivermectin group was 10 days versus 12 in the placebo. The paper calls this "not significant" but it's not clear to me whether that's statistically not significant, and if so, why wouldn't it be, or if knocking two days off recovery isn't that meaningful. The other thing I noticed reading it was that only one person out of 400 died and that person was in the placebo group. I think the Brett Weinstein response would be something like - A. Ivermectin's best benefit is as a prophylactic not a treatment. B. Ivermectin should be administered as early after onset as possible to treat, whereas here there was some delay to get people organized, enrolled in trial, etc. C. It's hard to measure effect when the disease is mild.
  - bonzini 5y ago
    > why wouldn't it be Because you only gave the average time of recovery but not the standard deviation.
    - ALittleLight 5y ago
      I didn't see the standard deviation reported in the paper. Is it? I did see the innerquartile range and it was the same for both the ivermectin group and the placebo. To me that suggests that variation in outcomes is probably similar between the two groups.
      - Fomite 5y ago
        They provide it in the form of a hazard ratio: "hazard ratio for resolution of symptoms, 1.07 [95% CI, 0.87 to 1.32]" So statistically, those on ivermectin recovered 1.07 times as quickly as those on placebo, but the uncertainty bounds are from 0.87 times as fact to 1.32 times as fast.
- spadez 5y ago
  > But an actual randomized control trial of Ivermectin in March Please see this https://news.ycombinator.com/item?id=27566123 https://news.ycombinator.com/item?id=27566123

goldemerald 5y ago

While randomized trials are certainly the gold-standard for determining if you should give a new drug to treat an arbitrary disease, that process is far too conservative during a pandemic where a huge majority of the world cannot access proper medical care. Ivermectin is already massively produced, used throughout the world, and cheap to manufacture (compared to new anti-Covid drugs). Even months ago before the benefits were known, the risks of taking the drug were very small, especially when the treatment guidelines were to do practically nothing. Still, the many pieces of non-randomized evidence can still clearly point towards this drugs efficacy. The sudden drop in Covid cases in India as the clearest indicator to me.

jsnell 5y ago
> The sudden drop in Covid cases in India as the clearest indicator to me. What's the alleged connection here? Are you claiming that some non-trivial number of patients in India were treated with this? How many, and at what point after being infected? And just how did the treatment affect the case numbers? Generally you'd expect the vast majority of secondary infections to happen before the diagnosis, not after, so a treatment seems totally irrelevant to the case numbers.
- spadez 5y ago
  >Are you claiming that some non-trivial number of patients in India were treated with this Yes, example from two states. https://indianexpress.com/article/cities/lucknow/uttar-pradesh-government-says-ivermectin-helped-to-keep-deaths-low-7311786/ https://indianexpress.com/article/cities/lucknow/uttar-prade... https://www.news18.com/news/india/mortality-rate-much-lower-goa-govt-defends-use-of-ivermectin-for-covid-19-treatment-3786590.html https://www.news18.com/news/india/mortality-rate-much-lower-... The testimony video in this page cites examples in many other countries https://covid19criticalcare.com/ https://covid19criticalcare.com/
  - jsnell 5y ago
    I see no numbers on how many people were treated on either of the first two links, and certainly don't intend to comb through the dross on the third link. So do you actually have a number? It doesn't need to be exact, just a credible source on the order of magnitude will be enough. Are we talking a thousand, a million, or a billion here? To reduce R by a factor of 2 by the use of a prophylactic drug, you'd need to have half of the population on a regime of the drug. It seems pretty obvious that did not happen if this kind of reporting is the most impactful there is. Just think of the logistics of trying to do that! India has a population of 1.4 billion. On the other hand, if they only gave the snakeoil to e.g. a single digit number of millions of people the reduction in the number of infections would be imperceptible.

wrycoder 5y ago

https://sebastianrushworth.com/2021/05/09/update-on-ivermectin-for-covid-19/ https://sebastianrushworth.com/2021/05/09/update-on-ivermect... I’ve included every double-blind randomized placebo-controlled trial I could find of ivermectin as a treatment for covid. Using only double-blind placebo-controlled trials means that only the highest quality studies are included in this meta-analysis, which minimizes the risk of biases messing up the results as far as possible. In order to be included, a study also had to provide mortality data, since the goal of the meta-analysis is to see if there is any difference in mortality .... What we see is a 62% reduction in the relative risk of dying among covid patients treated with ivermectin. That would mean that ivermectin prevents roughly three out of five covid deaths. The reduction is statistically significant (p-value 0,004). In other words, the weight of evidence supporting ivermectin continues to pile up. It is now far stronger than the evidence that led to widespred use of remdesivir earlier in the pandemic, and the effect is much larger and more important (remdesivir was only ever shown to marginally decrease length of hospital stay, it was never shown to have any effect on risk of dying). I understand why pharmaceutical companies don’t like ivermectin. It’s a cheap generic drug. Even Merck, the company that invented ivermectin, is doing it’s best to destroy the drug’s reputation at the moment. This can only be explained by the fact that Merck is currently developing two expensive new covid drugs, and doesn’t want an off-patent drug, which it can no longer make any profit from, competing with them.

Vinceo 5y ago

Does that mean that the covid vaccines shouldn't be recommended as there's no randomized trials on their long-term effects? Just curious as I'm not well versed on this topic.

13415 5y ago
Covid vaccines have been tested with randomized trials (Phase 3 of the testing phase). Regarding "long-term", that obviously depends on how you define "long-term but there is no need to move the goalpost. Generally speaking, in evidence-based medicine drugs and vaccines are not recommended based on incomplete statistics or hunches.