4 ms·

Why is everyone so confident with their conclusions without RCTs? This article points out how deeply complicated ethanol consumption is with our biochemical sys

Why is everyone so confident with their conclusions without RCTs? This article points out how deeply complicated ethanol consumption is with our biochemical systems. Without RCTs we really can't measure the overall effect of ethanol without quantifying the effects of ethanol on each of those systems. Can't we just reserve judgement here, and use some humbleness in the face of our ignorance to drive for a real RCT funded by the public? Can we try to avoid the trap of becoming confidence-men?

nostromo 5y ago
This is an impossible ask and exactly what I mean about moving the goalposts. For example: there are zero RCTs that prove to us cigarettes cause cancer. But we're still comfortable in saying cigarettes cause cancer. It's entirely ridiculous to suggest we do a RCT and ask someone to smoke for 30 years to remove all doubt that cigarettes cause cancer. And who would pay for it? So all we have are mountains of data showing that cigarettes shorten lifespan. The same is true here, but the opposite result. Doing a RCT on alcohol for decades will never happen. All we have are mountains of data showing either no harm or a small benefit on lifespan. So that should be our null hypothesis: that moderate drinking is either harmless or slightly beneficial.
- bena 5y ago
  This is a fun one because it's a good example of how to bend the truth with the truth. About 80 to 90 percent of people who have lung cancer were smokers. Only about 10 to 15 percent of smokers will develop lung cancer. So it's not a guarantee. It increases the risk. Just like alcohol (https://www.cancer.gov/about-cancer/causes-prevention/risk/alcohol/alcohol-fact-sheet https://www.cancer.gov/about-cancer/causes-prevention/risk/a...) The problem with doing RCT with alcohol and with smoking is that we're talking about lifestyle studies. You can't really do a study with the goal being "see if we can give this dude cancer during his lifetime". And let's also remember that the study that people who drink moderately had longer lifespans just really notes correlation. It does not prove a causation. Drinking is ubiquitous. Most people drink. Drinking is also a luxury. So if you're poor, you either drink a lot or very little. Because you either have a problem or you're too broke to afford it. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3185179/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3185179/) Guess which income bracket engages in moderate drinking the most? That's right, the income bracket that also allows you to afford better health outcomes overall. So while it may be true that moderate drinking correlates with longer lifespans, it's a knock-on effect to the fact that being poor sucks.
- Scoundreller 5y ago
  The issue is, the difficulty of doing an randomized study doesn’t nullify the benefits of randomization (a big one being eliminating confounding variables, for which there can be a lot for who decides to take up an addiction and why). The control part would be particularly difficult (unless maybe you gave everyone a real or sham nicotine patch? But then you can’t exclude some other benefit from straight nicotine itself). One could control for all the confounders they want, but you still risk missing some that are unknown or undervalued.
- beebmam 5y ago
  So your argument here is that it is ridiculous to suggest that an RCT is possible in this circumstance, so we should therefore accept observational studies which derive correlations as truth. Not only does that conclusion not follow from your premise (a logical fallacy), but it's an absurd suggestion as a process for deriving truth. Observational studies, at best, offer an insight into possible hypotheses, and should by no means be considered persuasive unless all interacting systems have been controlled for. In the case of ethanol, there are an enormous amount of interacting systems that need to be controlled for. Your premise is also strictly false, as it's clear that not only is an RCT for ethanol consumption possible, but it was planned and partially in progress before it was terminated. It's truly incredible to see the kind of language that you use here, how confident you are about your fallacious argument. I kindly request that you turn down your confident language. Science is a long and careful process of pushing back the fog of ignorance, and if you are serious about the search for truth you shouldn't use such confident language.
abra0 5y ago
Because only people with a high enough conviction of benefits of moderate drinking are commenting here and arguing for. However I have no idea how they do become so strongly convinced in that, given the low quality of the data that exists, the general reproducibility crisis in bio-sciences, precedent of complete compromise of a large study, things like https://slatestarcodex.com/2014/12/12/beware-the-man-of-one-study/ https://slatestarcodex.com/2014/12/12/beware-the-man-of-one-..., the argument that if moderate amounts of ethanol were beneficial we would have just evolved to synthesize it, etc. I think the most likely outcome is boring: moderate amounts of alcohol are moderately harmful, but not harmful enough to be immensely obvious and not harmful enough to outweigh the non-biological benefits some people extract from it.

nostromo 5y ago

This is an impossible ask and exactly what I mean about moving the goalposts. For example: there are zero RCTs that prove to us cigarettes cause cancer. But we're still comfortable in saying cigarettes cause cancer. It's entirely ridiculous to suggest we do a RCT and ask someone to smoke for 30 years to remove all doubt that cigarettes cause cancer. And who would pay for it? So all we have are mountains of data showing that cigarettes shorten lifespan. The same is true here, but the opposite result. Doing a RCT on alcohol for decades will never happen. All we have are mountains of data showing either no harm or a small benefit on lifespan. So that should be our null hypothesis: that moderate drinking is either harmless or slightly beneficial.

bena 5y ago
This is a fun one because it's a good example of how to bend the truth with the truth. About 80 to 90 percent of people who have lung cancer were smokers. Only about 10 to 15 percent of smokers will develop lung cancer. So it's not a guarantee. It increases the risk. Just like alcohol (https://www.cancer.gov/about-cancer/causes-prevention/risk/alcohol/alcohol-fact-sheet https://www.cancer.gov/about-cancer/causes-prevention/risk/a...) The problem with doing RCT with alcohol and with smoking is that we're talking about lifestyle studies. You can't really do a study with the goal being "see if we can give this dude cancer during his lifetime". And let's also remember that the study that people who drink moderately had longer lifespans just really notes correlation. It does not prove a causation. Drinking is ubiquitous. Most people drink. Drinking is also a luxury. So if you're poor, you either drink a lot or very little. Because you either have a problem or you're too broke to afford it. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3185179/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3185179/) Guess which income bracket engages in moderate drinking the most? That's right, the income bracket that also allows you to afford better health outcomes overall. So while it may be true that moderate drinking correlates with longer lifespans, it's a knock-on effect to the fact that being poor sucks.
Scoundreller 5y ago
The issue is, the difficulty of doing an randomized study doesn’t nullify the benefits of randomization (a big one being eliminating confounding variables, for which there can be a lot for who decides to take up an addiction and why). The control part would be particularly difficult (unless maybe you gave everyone a real or sham nicotine patch? But then you can’t exclude some other benefit from straight nicotine itself). One could control for all the confounders they want, but you still risk missing some that are unknown or undervalued.
beebmam 5y ago
So your argument here is that it is ridiculous to suggest that an RCT is possible in this circumstance, so we should therefore accept observational studies which derive correlations as truth. Not only does that conclusion not follow from your premise (a logical fallacy), but it's an absurd suggestion as a process for deriving truth. Observational studies, at best, offer an insight into possible hypotheses, and should by no means be considered persuasive unless all interacting systems have been controlled for. In the case of ethanol, there are an enormous amount of interacting systems that need to be controlled for. Your premise is also strictly false, as it's clear that not only is an RCT for ethanol consumption possible, but it was planned and partially in progress before it was terminated. It's truly incredible to see the kind of language that you use here, how confident you are about your fallacious argument. I kindly request that you turn down your confident language. Science is a long and careful process of pushing back the fog of ignorance, and if you are serious about the search for truth you shouldn't use such confident language.

abra0 5y ago

Because only people with a high enough conviction of benefits of moderate drinking are commenting here and arguing for. However I have no idea how they do become so strongly convinced in that, given the low quality of the data that exists, the general reproducibility crisis in bio-sciences, precedent of complete compromise of a large study, things like https://slatestarcodex.com/2014/12/12/beware-the-man-of-one-study/ https://slatestarcodex.com/2014/12/12/beware-the-man-of-one-..., the argument that if moderate amounts of ethanol were beneficial we would have just evolved to synthesize it, etc. I think the most likely outcome is boring: moderate amounts of alcohol are moderately harmful, but not harmful enough to be immensely obvious and not harmful enough to outweigh the non-biological benefits some people extract from it.