Risks V: Alcohol, Tobacco, and Other Substances

23 minutes ago

13 min read

Part V of the Risks series. Parts I through IV covered cardiovascular disease, cancer, cognitive decline, and bones and joints.

The short version

This is the first part of the series that isn't a disease. It's an input. Alcohol, tobacco, and the rest have shown up as a line item in every article so far. This one puts the line items on one ledger.
The story on moderate drinking changed, and the guidelines got vaguer instead of clearer. The old idea that a drink a day protects you rested largely on a comparison-group error. What's left is a dose-dependent risk, and the right question is what it's costing you, not whether it's healthy.
The treatments work and almost nobody uses them. Fewer than 2 in 100 people with an alcohol problem ever get medication for it. If you've tried to cut back and couldn't, that's not a character verdict. It's a condition with treatment options.

Substance use is still a major problem in this country. That sentence would have been true in any year I've practiced, but the shape of the problem keeps changing, and the current version has some genuinely good news in it.

Overdose deaths are falling. The CDC's provisional count for 2025 was about 70,000, down roughly 14% from the year before and more than a third below the peak in early 2023. Cigarette smoking among adults dropped below 10% for the first time in 2024. And the share of Americans who drink at all fell to 54% last year, the lowest Gallup has recorded since it started asking in 1939. For the first time, a majority of Americans told Gallup they think even moderate drinking is bad for your health.

Now the other side of the ledger. Excessive drinking still kills about 178,000 Americans a year, and that number rose 29% over five years. Fentanyl still accounts for more than half of overdose deaths. And the number of Americans who use cannabis daily or nearly daily now exceeds the number who drink that often. That crossover happened in 2022, and it went largely unremarked.

The first four parts of this series were about diseases: the heart, cancer, the brain, the frame. This one is different. Alcohol, tobacco, and cannabis don't have their own organ. They show up as a contributing factor in every disease we've already covered. So the useful way to think about this article is as an audit. We're going to find the line items that have been hiding in the other four and add them up.

The frame is the same one as the rest of the series. Front end, middle, back end. The difference is that for substances, the front end is nearly the whole game.

Alcohol: the story changed

For most of my training, the conventional wisdom was that a drink a day was good for you. Moderate drinkers lived longer than nondrinkers in study after study. Red wine got its own mythology. The curve had a dip in it, and the dip was where everybody wanted to live.

The problem was the comparison group. In a lot of those studies, the "nondrinkers" included people who had stopped drinking because they were already sick. Former heavy drinkers, people with liver disease, people on medications that don't mix with alcohol. Of course they died sooner. Comparing moderate drinkers to that group made moderate drinking look protective when it was really just making the sick people look like abstainers.

In 2023, a group of researchers pooled 107 studies covering nearly 5 million people and corrected for that error. Once former drinkers were separated out from lifetime nondrinkers, the protective dip disappeared. Low-volume drinking was not associated with a meaningful reduction in all-cause mortality. Risk started climbing at higher intake, and it started climbing sooner for women than for men.

I want to be fair to the other side, the same way I tried to be fair to the oncologists and the surgeons. This is still observational data. There is no randomized trial of a drink a day for thirty years, and there won't be. And when the federal government commissioned two separate reviews to inform the newest dietary guidelines, they didn't agree. One found that risk begins at low levels of drinking. The other, commissioned by Congress, found moderate drinking associated with a lower risk of death from any cause. Reasonable people are still arguing about the bottom of the curve.

Nobody is arguing about the slope. More drinking means more risk, and the cancer evidence is the clearest piece of it. In January 2025 the Surgeon General called alcohol the third leading preventable cause of cancer in this country, behind tobacco and obesity. It's tied to at least seven cancers, including breast, colorectal, esophageal, liver, mouth, and throat. The estimate is about 100,000 cancer cases and 20,000 cancer deaths a year in the U.S. For breast cancer specifically, roughly 16% of cases are attributable to alcohol. That one gets the least attention and probably deserves the most, because I see a lot of women in their forties and fifties who were never told.

Then, this January, the new Dietary Guidelines for Americans came out and dropped the specific limits. For more than forty years the federal guidance had been up to one drink a day for women and two for men. The 2025–2030 version replaced the numbers with a general instruction to drink less. Depending on who you ask, that's either a reasonable acknowledgment that no number is safe or a step back from telling people what the number is. I'll leave that argument to others. The practical effect for you is that the guideline no longer gives you a number to hide behind, so you're going to need your own.

The ledger

Here's what I mean by an audit. Go back through the first four parts of this series and pull out every place alcohol showed up.

In the cardiovascular article, it's blood pressure. Regular drinking raises it, and cutting back lowers it. It's also atrial fibrillation, and this is where we have unusually good evidence. In a randomized trial published in 2020, regular drinkers with AF who were assigned to stop drinking had fewer recurrences over six months, 53% versus 73%, and spent less time in the arrhythmia. These were people averaging around 17 drinks a week, which is not unusual in my panel.

In the cancer article, it's the seven cancers above. In the cognitive decline article, excessive alcohol is one of the fourteen Lancet risk factors. In the bones and joints article, it's falls, and it's bone density, and it's the 6 a.m. trip to the bathroom in the dark after three glasses of wine at dinner.

And then there's the line item I see most: sleep. Alcohol helps you fall asleep and then wrecks the second half of the night. People who track their sleep can often see it in their resting heart rate and heart rate variability the next morning. If you wear a ring or a watch, look at your data after a night with two drinks and a night with none. Most people don't need me to interpret it for them.

The analogy I use is an expense ratio. Most people with a retirement account have no idea what they're paying in fees. One percent a year doesn't feel like anything. But over thirty years, compounded, it can cost you a quarter of the account. Alcohol works the same way. Two drinks on a Tuesday doesn't feel like a decision. Two drinks most nights for twenty years shows up in your blood pressure, your sleep, your waistline, your liver enzymes, and your cancer risk, and you never see a single bill. Charlie Munger liked to say smart people go broke three ways, and liquor was the first one on his list.

I am not a prohibitionist, and I'm not going to tell you that a glass of wine at your daughter's wedding is a health decision. It isn't. The question isn't whether alcohol is healthy. As a substance, it isn't at any dose. Arthur Brooks has provided the counter argument that it improves socialization as it reduces anxiety and this can contribute to socialization.

The question is what it's costing you and whether you're getting your money's worth.

Tobacco and nicotine

This is the success story, and it deserves to be told as one. In the 1960s more than 40% of American adults smoked. In 2024 it fell below 10%. That is one of the largest behavior changes in the history of public health, and it happened through taxes, smoking bans, advertising restrictions, and a slow change in what was socially acceptable. Not through a breakthrough drug.

If you still smoke, the most important number in this article is this one. Smokers die about ten years earlier than people who never smoked. People who quit before 40 avoid about 90% of that excess risk. People who quit in their mid-forties to mid-fifties still gain back about six years. There is no age at which quitting stops paying.

Vaping is the harder conversation. Adult e-cigarette use has roughly doubled since 2020, to about 7%, and among people 18 to 24, vaping is now far more common than smoking. Most young adults who vape never smoked.

Nicotine pouches are the newest version, and they get more attention than their numbers do. Zyn and its competitors are small tobacco-free pouches that sit between the lip and gum. In January 2025, Zyn became the first pouch the FDA authorized for sale, after the agency found it carried far lower levels of harmful compounds than cigarettes or most chewing tobacco. The FDA was explicit that authorization doesn't mean safe, and it doesn't mean approved. Among adults, the best national data we have, from 2022 and 2023, found about 2.7% had ever tried a pouch, and fewer than half a percent were current users. Use among people who had never touched tobacco was close to zero. It was highest among people who had recently quit something else, which suggests a lot of adults are using pouches as an off-ramp. The teenage numbers are less reassuring. Pouch use among middle and high school students roughly doubled between 2021 and 2025, to 1.7%. While cigarettes and vapes both fell, pouches became the second most common nicotine product kids use.

Here's the honest version. What kills smokers is mostly the smoke: the combustion products that drive lung cancer, COPD, and vascular disease. Nicotine is what keeps them coming back. So for a long-term smoker who can't quit any other way, switching completely to a non-combustible nicotine product is very likely a reduction in risk. That's a real argument, and I make it to some patients. But it's an argument for switching, not for starting. Nicotine raises heart rate and blood pressure, it's highly addictive, and we don't have thirty-year data on daily vaping because nobody has been doing it for thirty years. A product that's safer than cigarettes is a low bar.

Cannabis

If you'd told me early in my career that daily cannabis use would outnumber daily drinking, I wouldn't have believed you. In 2022, an estimated 17.7 million Americans reported using cannabis daily or nearly daily, compared to 14.7 million daily or near-daily drinkers. Around 40% of current cannabis users now use it that often. That's a pattern that looks more like tobacco than like how most people drink.

The version I see in my practice is not the stereotype. It's a 50-year-old professional who takes an edible every night to sleep. They don't think of it as drug use. They think of it as a sleep aid that happens to be legal in a lot of states. Some of them have been doing it for years.

Two things worth knowing. First, the cardiovascular data. A 2024 analysis of about 430,000 adults found that daily cannabis users had 25% higher odds of heart attack and 42% higher odds of stroke compared to non-users, and the association held up in people who had never used tobacco. It's survey data and it can't prove cause. But it's consistent with other studies, and it applies whether the cannabis is smoked, vaped, or eaten.

Second, sleep. Cannabis can shorten the time it takes to fall asleep, the same way alcohol does. Tolerance builds, and when people stop, sleep often gets worse for a while before it gets better. That rebound is a big part of why the nightly edible becomes the permanent nightly edible. Daily use also carries a real risk of dependence, and in younger users, heavy use is linked to psychosis. That last one matters if you have teenagers or college-aged kids.

The legal picture is moving. In April, the Department of Justice moved FDA-approved and state-licensed medical marijuana to Schedule III, with a hearing on broader rescheduling that began this summer. Separately, Congress closed the loophole that allowed delta-8 and other hemp-derived THC products to be sold at gas stations and smoke shops, and those restrictions are scheduled to take effect by the end of this year. None of that changes biology. Rescheduling is about research access and regulation, not a statement that daily use is harmless.

The other substances

The overdose numbers are better, and I don't want to undersell that. But the drug supply is more dangerous than it was a generation ago. Fentanyl is still involved in more than half of overdose deaths, and it turns up in counterfeit pills that look exactly like the real thing. A pill that came from a friend, a coworker, or a website and not from a pharmacy should be treated as unknown.

That's why I think every household where anyone takes opioids, or where a teenager or young adult lives, should have naloxone in the cabinet. It's sold over the counter now. It reverses an opioid overdose, it doesn't hurt someone who isn't overdosing, and the people who use it are almost always family members, not paramedics. It's the fire extinguisher of this article. You hope you never need it, and it's cheap enough that there's no reason not to own one.

I covered the medicine cabinet in the cognitive decline article, and I won't repeat it here. The short version is that over-the-counter sleep aids, benzodiazepines, and other sedating medications deserve a review with your doctor, especially combined with alcohol.

What's worth measuring

The first measurement is the one almost nobody takes: an honest count.

A standard drink in this country is about 14 grams of alcohol. That's 12 ounces of regular beer, 5 ounces of wine, or a shot and a half of spirits. A restaurant pour of wine is often closer to 6 or 8 ounces, and a craft IPA can be nearly two standard drinks in one can. Most people who tell me they have "a glass or two" are underestimating, and not because they're lying. Nobody measures.

So count for two weeks. Write it down when you pour it, not at the end of the day. Do the same for cannabis if you use it. Most people find the real number is higher than the one they would have reported, and that gap is the most useful piece of information in this section.

The screening question I find most useful is a simple one. How many times in the past year have you had four or more drinks in a day if you're a woman, or five or more if you're a man? If the answer is anything other than zero, that's worth a conversation. Not a lecture. A conversation.

Labs can corroborate. Liver enzymes, especially GGT, rise with regular drinking. So does the average size of red blood cells, called MCV. None of these should be used as a gotcha. They're useful because they give us a number we can watch when we change something. That's the same logic as the Health Score Card. Your blood pressure, waist, and sleep data are all partly alcohol numbers. We just don't label them that way.

Treatment, and where the money goes

This is the part that frustrates me most. Alcohol use disorder is one of the most common conditions in medicine, and we have effective medications for it. In one national estimate, among people with an alcohol use disorder, only 7.6% got any treatment at all, and 1.6% got medication.

Naltrexone is the one I use most. It blunts the reward from drinking and reduces cravings. It doesn't require that you've already stopped, and some people take it only on days they expect to drink. It's inexpensive and generic. Acamprosate is another option, mainly for maintaining abstinence once someone has stopped. Neither one is new, and neither one is complicated to prescribe.

The newer development is the GLP-1 drugs. In the first randomized trial designed to test it, published last year, semaglutide reduced cravings and cut the amount people drank on drinking days by nearly 30%, compared to about 2% on placebo. It was a small trial, 48 people over nine weeks, and I'd hold it loosely until the larger trials read out. But it fits what I hear from patients on these drugs, many of whom say they just stopped wanting the second drink. I wrote about the trade-offs of these medications last month. This is one more thing to weigh.

For smoking, the medications that work best are varenicline and combination nicotine replacement, a patch plus a short-acting form like gum or a lozenge. Most people who quit successfully needed more than one attempt. That is the expected course, not a failure.

The broader point is that treatment for substance use isn't a choice between doing nothing and a 28-day residential program. Most of it can happen in a primary care office, with a prescription, a plan, and follow-up. The largest gap here isn't science. It's that people don't ask, and doctors don't offer.

What's in your control

Not in your control. Your genetics and your family history. Addiction runs in families, partly through genes and partly through what you grew up around. If you have a parent or sibling who struggled, your risk is higher, and it's worth being more deliberate about your own patterns than the people around you need to be.

In your control. Almost everything else. This is the first article in the series where that's true. You can't choose your LDL entirely, or your APOE status, or the knee you injured at 19. You can choose what's in your glass tonight.

That doesn't make it easy. Alcohol and nicotine are built to be hard to put down, and the social pressure in a lot of professional circles runs in one direction. I wrote in an earlier piece about the drunk driver who makes it home. Getting away with something for years is not the same as it being a good decision. And in December, I'll come back to the question underneath all of this: when a risk is one you choose with your eyes open, how much of it is worth taking?

For now, the question is simpler. Do you know what you're actually consuming, and do you know what it's costing you?

Three things to do

1. Count for two weeks before you change anything. Standard drinks, written down when you pour them. Do the same for cannabis or nicotine if you use them. The goal isn't to judge the number. It's to replace your estimate with a measurement.

2. Take the substance out of your sleep routine for two weeks. If the nightcap or the edible is how you fall asleep, stop it for fourteen days and watch your sleep data and how you feel. Expect the first few nights to be worse. Judge it at the end of the second week, not the first.

3. If you've tried to cut back and couldn't, ask about medication. Naltrexone is cheap, generic, and doesn't require that you've already quit. Needing help with this is common and treatable. Bring it to your next visit, or send us a message before then.

Next in this series: Risks VI — Infection and Immunity.