Sorry, There Is No "Healthy" Drinking

Read the article on Substack

For decades, the message seemed clear: a little alcohol might protect your brain. The famous U-shaped curve suggested light drinkers were safer than both heavy drinkers and non-drinkers when it came to dementia risk.

But what if this reassuring narrative was built on a fundamental misreading of the science?

In this paradigm-shifting episode, we explore a massive study using genetic data from 2.4 million people that completely dismantles the protective drinking myth. Through Mendelian randomization—a powerful technique that uses genetics to establish causation—researchers discovered the U-curve was an illusion created by reverse causation and confounding.

You'll discover:

• Why observational studies consistently got it wrong
• How people in early dementia often stop drinking BEFORE diagnosis, creating misleading patterns
• What genetic analysis actually reveals: a straight-line increase in risk with any alcohol consumption
• The staggering finding that reducing alcohol use disorder could prevent 16% of dementia cases
• Why there appears to be no safe threshold for brain health

References:  Alcohol use and risk of dementia in diverse populations: evidence from cohort, case-control and Mendelian randomisation approaches

This is Heliox: Where Evidence Meets Empathy

Independent, moderated, timely, deep, gentle, clinical, global, and community conversations about things that matter.  Breathe Easy, we go deep and lightly surface the big ideas.

Thanks for listening today!

Four recurring narratives underlie every episode: boundary dissolution, adaptive complexity, embodied knowledge, and quantum-like uncertainty. These aren’t just philosophical musings but frameworks for understanding our modern world. 

We hope you continue exploring our other podcasts, responding to the content, and checking out our related articles on the Heliox Podcast on Substack. 

Transcript

Speaker 1: 0:25

This is Heliox, where evidence meets empathy. Independent, moderated, timely, deep, gentle, clinical, global, and community conversations about things that matter. Breathe easy. We go deep and lightly surface the big ideas.

Speaker 2: 0:50

Welcome to the Deep Dive. Today we are taking aim at one of those really widely accepted pieces of health advice. You know, the kind of thing you hear all the time, the idea that maybe light drinking, just a little bit, might actually be good for your brain health, particularly when it comes to dementia risk.

Speaker 1: 1:08

Right. The famous U-shaped curve.

Speaker 2: 1:10

Exactly. If you follow any health news, you've definitely seen it. It suggests non-drinkers and heavy drinkers are at higher risk, but the light to moderate drinkers, they're somehow protected down there at the bottom of the U.

Speaker 1: 1:22

Yeah, and that's become almost like standard advice, but it's a classic case of maybe mistaking correlation for causation. It's mostly built on years and years of observational studies just watching groups of people over time. So our mission today is to really dig into a huge recent study that challenges this whole idea. It uses a powerful genetic technique called Mendelian randomization alongside that observational data.

Speaker 2: 1:46

Okay.

Speaker 1: 1:47

And spoiler alert, it pretty much dismantles the light drinking as protective theory.

Speaker 2: 1:52

And this isn't some small study you're talking about.

Speaker 1: 1:54

Oh, no, not at all. The sources we looked at are massive. They combined data from huge population groups, the U.S. Million Veteran Program, the U.K. Biobank. The observational part alone tracked almost 560,000 adults. But the really powerful part, the genetic analysis, pulled together genome-wide association study GWSS data from about 2.4 million people.

Speaker 2: 2:21

2.4 million, okay.

Speaker 1: 2:22

Yeah. So we're moving beyond just watching habits to actually looking for genetic evidence of cause and effect.

Speaker 2: 2:28

All right. Let's unpack this properly then. We should start where the original research did the observational findings, because that's where the whole U-curve myth came from in the first place.

Speaker 1: 2:35

Exactly. You have to see what they saw initially.

Speaker 2: 2:37

So they watched these, what, nearly 560,000 adults for several years looking for who developed dementia.

Speaker 1: 2:45

That's right. I mean, a follow up was between four and 12 years across the different cohorts. And when they just crunched the numbers based on reported drinking habits.

Speaker 2: 2:52

They found the U-curve.

Speaker 1: 2:53

They found exactly the U-curve, just like dozens of studies before them. The risk pattern looked, well, U-shaped.

Speaker 2: 2:59

Okay. So who was at the highest risk in that data?

Speaker 1: 3:02

Unsurprisingly, the heavy drinkers. They define that as over 40 drinks a week. Compared to light drinkers, their risk was about 41% higher. Hazard ratio of 1.41.

Speaker 2: 3:14

Okay, significant.

Speaker 1: 3:14

And also individuals with a diagnosis of alcohol use disorder, AUD, they had an even higher risk, about 51 percent higher compared to light drinkers. So, yeah, heavy drinking, definitely bad news for dementia risk.

Speaker 2: 3:27

OK, no huge surprises there. Heavy drinking is harmful. We know that. Yeah. But the U-curve really hinges on the other side of the U, doesn't it? The non-drinkers.

Speaker 1: 3:36

That's the twist. Exactly. And crucially, in this observational data, too, the non-drinkers appeared to have a higher risk than the light drinkers. So based purely on looking at these correlations, the people having just a few drinks a week, they look like the safest group. They seem to protect it.

Speaker 2: 3:52

And that fuels this incredibly persistent idea, doesn't it? That maybe avoiding alcohol completely isn't the best strategy or maybe non-drinkers are just less healthy overall for other reasons.

Speaker 1: 4:01

Precisely. If we stopped right here, the takeaway would be, see, a little bit is better than none.

Speaker 2: 4:07

But, and this is the big but, that observational U-shape might be an illusion, right? It's riddled with potential problems.

Speaker 1: 4:14

Absolutely riddled. Correlation isn't causation. Even with half a million people, just finding an association doesn't prove the alcohol itself is protective. The huge issue is confounding.

Speaker 2: 4:25

Confounding, meaning other factors that get mixed up with drinking habits.

Speaker 1: 4:29

Exactly. Think about it. On average, who are the light drinkers in these studies? They often tend to be a bit wealthier, maybe better educated, more socially connected. Perhaps they exercise more, smoke less.

Speaker 2: 4:40

Healthier lifestyles generally.

Speaker 1: 4:42

Right. Compared sometimes to groups of non-drinkers, which can include people who stop drinking due to poor health. So is it the light drinking that's protecting them? Or is it their overall healthier life and socioeconomic advantage? Observational studies find it really hard to untangle all that.

Speaker 2: 4:58

So you need a method that can somehow isolate the effect of the alcohol itself, stripping away all that other lifestyle noise. How do you even do that?

Speaker 1: 5:07

That's where the study shifts gears dramatically. They brought in Mendelian randomization, M. Mosho. It's a technique that uses genetics to get much closer to understanding actual causality.

Speaker 2: 5:20

Okay, genetics. How does that work in this context?

Speaker 1: 5:22

Well, the basic idea is pretty neat. We all inherit slightly different versions of genes. And some of these genetic variations influence things related to alcohol, like how quickly our body processes it or even our predisposition towards drinking more or less or developing AUD.

Speaker 2: 5:37

Right. Some people flush red. Others can drink more.

Speaker 1: 5:40

Exactly. And the key is these genetic variants are distributed randomly throughout the population when we're conceived. It's like nature's own randomized trial. Your genes are assigned before your lifestyle choices happen.

Speaker 2: 5:51

Ah, okay. So instead of relying on people telling you how much they drank, which can be inaccurate, you look at their genetic tendency towards certain drinking patterns.

Speaker 1: 6:00

Precisely. You use these randomly assigned genetic variants as a sort of proxy or an instrument for their lifelong exposure to alcohol. And because these genes are random, they generally aren't linked to those confounding factors like wealth or diet or education.

Speaker 2: 6:15

That makes sense. It bypasses a lot of the social stuff.

Speaker 1: 6:18

It gets much closer to isolating the biological effect of alcohol exposure itself.

Speaker 2: 6:22

And this MR analysis wasn't just on the 560,000 cohort people. You said it used data from 2.4 million.

Speaker 1: 6:29

Correct. It used summary statistics from multiple large GW West consortia. Just a massive pool of genetic information to get reliable estimates.

Speaker 2: 6:38

Okay, so the moment of truth. What happened when they applied this powerful genetic method? Did the U-curve hold up?

Speaker 1: 6:45

Not at all. It vanished completely. The genetic analysis directly contradicted the observational findings about light drinking being protective.

Speaker 2: 6:51

Wow. So what did it show instead?

Speaker 1: 6:53

It showed a monotonic increase in dementia risk with genetically predicted alcohol consumption.

Speaker 2: 6:58

Monotonic. Meaning?

Speaker 1: 6:59

Meaning it just goes up. The more the genetic prediction indicated higher alcohol consumption, the higher the dementia risk. No dip in the middle. No protective effect at low levels. Just a straight upward trend.

Speaker 2: 7:12

So the genetics basically say more alcohol equals more risk.

Speaker 1: 7:16

period. No safe haven for like drinkers. That's what the MR results strongly suggest. Let's get

Speaker 2: 7:22

specific. What were the numbers from the genetic analysis? They found that a genetically predicted

Speaker 1: 7:28

increase of one standard deviation in log transformed drinks per week basically, a standard jump in regular consumption was associated with about a 15% increase in dementia odds. 15%, okay. And similarly, a genetically predicted two-fold increase in the likelihood of having alcohol use disorder was linked to about a 16% increase in dementia odds. So consistent evidence from

Speaker 2: 7:50

different genetic angles pointing towards a linear dose-response relationship. More alcohol,

Speaker 1: 7:54

more risk. Exactly. Which brings us back to the big question. If the genetics, which should be less confounded, show this straight line up, why did the observational studies always show that dip? Why did light drinkers look protected compared to non-drinkers? Yeah, what explains that illusion

Speaker 2: 8:11

in the observational data. There must be something systematically misleading going on there.

Speaker 1: 8:16

And the study provides a really compelling explanation. Reverse causation.

Speaker 2: 8:20

Reverse causation. Okay, what does that mean here?

Speaker 1: 8:23

It means the supposed cause, not drinking, might actually be an early effect of the disease itself. Dementia.

Speaker 2: 8:31

Hmm, okay, explain that.

Speaker 1: 8:33

Think about dementia. It doesn't just appear overnight. It develops slowly, often over many years. Long before someone gets a formal diagnosis, they can start experiencing Very subtle cognitive changes, memory slips, maybe changes in behavior or planning ability.

Speaker 2: 8:48

Right, the preclinical stage.

Speaker 1: 8:49

Exactly. And what might someone do if they're starting to feel not quite right? Maybe they're finding alcohol affects them more strongly, or they're just simplifying their lives.

Speaker 2: 8:59

They might cut back on drinking or stop altogether.

Speaker 1: 9:02

Precisely. People in the very early stages of cognitive decline often reduce or cease their alcohol consumption because of those emerging symptoms.

Speaker 2: 9:12

Ah, I think I see where this is going.

Speaker 1: 9:13

So by the time researchers come along years later for their study and ask, do you drink? So these individuals say no. They get categorized as non-drinkers.

Speaker 2: 9:23

But they didn't stop drinking randomly. They stopped because the dementia process had already subtly begun.

Speaker 1: 9:30

Exactly. Their abstinence isn't the cause of their later dementia diagnosis. It's an early consequence of the underlying disease process.

Speaker 2: 9:37

So that group of non-drinkers in observational studies is effectively contaminated with people who are already on the path to dementia.

Speaker 1: 9:44

Correct. It artificially inflates the apparent risk in the non-drinker category. And when you compare the other groups to this artificially risky non-drunker group, the light drinkers suddenly look protected by comparison.

Speaker 2: 9:56

It's like sickness was masquerading as abstinence. Wow.

Speaker 1: 9:59

That's a good way to put it. The reverse causation creates the illusion of the protective U-curve dip.

Speaker 2: 10:04

Okay, that is a really crucial insight. It fundamentally reframes how we should interpret all that observational data we've relied on for decades.

Speaker 1: 10:13

It really does. The synthesis here seems pretty clear. The supposed protective effect of light drinking seen in observational studies is likely an artifact, probably due to this reverse causation. The stronger causal evidence from genetics doesn't support any protection. If anything, it points the other way.

Speaker 2: 10:30

So connecting this to the bigger picture, what does this mean for our understanding of alcohol and brain health?

Speaker 1: 10:37

Well, it strongly suggests that basically all levels of alcohol consumption likely contribute incrementally to dementia risk. There doesn't seem to be a safe threshold below which there's no risk or even a benefit, neurologically speaking.

Speaker 2: 10:50

Which forces a major rethink of moderate drinking guidelines when it comes to brain health, doesn't it?

Speaker 1: 10:55

It certainly challenges the idea that moderate drinking is actively neuroprotective. The message shifts more towards. Less is likely always better for reducing dementia risk over the long term.

Speaker 2: 11:06

And the study highlighted a specific public health angle too, right? Something about alcohol use disorder.

Speaker 1: 11:11

Yes, very clearly. Since the genetic analysis supported a causal link between AUD and dementia risk, the researchers estimated the potential impact of tackling AUD.

Speaker 2: 11:22

And what was that estimate?

Speaker 1: 11:24

Their calculations suggested that public health strategies effective enough to substantially reduce the prevalence of AUD could potentially lower the overall incidence of dementia across the population by as much as 16 percent.

Speaker 2: 11:36

16 percent.

Speaker 1: 11:38

That's huge.

Speaker 2: 11:39

That's a massive potential public health gain from focusing on reducing harmful alcohol use patterns.

Speaker 1: 11:44

It really underscores that AUD isn't just a social or acute health issue. It has profound long-term consequences for brain health, potentially driving a significant fraction of dementia cases.

Speaker 2: 11:55

Okay. So we started with the comforting U-curve and ended with a much starker picture from the genetics, a steady increase in risk.

Speaker 1: 12:02

Which leaves us and you, the listener, with a really important thought to consider. If the best causal evidence we have now points to a monotonic risk, basically, the less alcohol the better, with no safe minimum for the brain, what does that truly imply for global health advice?

Speaker 2: 12:17

Yeah.

Speaker 1: 12:18

How should health bodies and maybe even individuals adjust their thinking about moderate versus heavy drinking if the underlying biological risk seems to just climb steadily upwards with any level of consumption? It really forces a reevaluation.

Speaker 2: 12:33

Thanks for listening today. Four recurring narratives underlie every episode. Boundary dissolution, adaptive complexity, embodied knowledge, and quantum-like uncertainty. These aren't just philosophical musings, but frameworks for understanding our modern world. We hope you continue exploring our other podcasts, responding to the content, and checking out our related articles at heliocspodcast.substack.com.