Is It Safe to Quit Drinking Cold Turkey? What the NHS and NICE Say Before You Cut Down

Is cutting down a real treatment goal, or only a compromise

Here is the part most English-language coverage skips. In 2013, nalmefene was approved and marketed in Europe with an indication for reducing alcohol consumption. It was the first medicine for alcohol dependence licensed around a reduction goal rather than abstinence.

Japan followed with both a drug and a clinic. On January 8, 2019 the same compound was approved there for reducing the amount of drinking in patients with alcohol dependence, and it launched that March as Japan's first medicine of that kind. The approval rested on a multicenter, randomized, double-blind phase 3 trial in Japanese patients with alcohol dependence at a high drinking risk level: alongside psychosocial treatment, it significantly reduced heavy drinking days and total alcohol intake over 12 weeks compared with placebo.

Germany put the same idea in a guideline instead. The 2021 update of the S3 guideline on alcohol-related disorders, led by the German psychiatric and addiction medicine societies, says that where the initial treatment goal is a reduction in drinking, this medicine can be offered as part of an overall treatment plan to alcohol-dependent people outside inpatient detoxification, after the risks have been considered and explained. The guideline puts its evidence at four placebo-controlled trials covering roughly 2,500 people, and describes the effect on heavy drinking days and drinks per drinking day as modest.

The United Kingdom took the appraisal route. NICE published its technology appraisal in November 2014 and, after a review in May 2018, kept the recommendation in place. The conditions are specific: the person has no physical withdrawal symptoms and does not need immediate detoxification, is at a high drinking risk level, receives continuous psychosocial support alongside, and starts only if they are still at a high drinking risk level two weeks after the initial assessment.

In the United States the word "reduction" sits somewhere else entirely. It is not attached to a medicine but to a stage of care. The US Preventive Services Task Force recommended in November 2018 that clinicians screen all adults 18 and over, including pregnant people, for unhealthy alcohol use and give brief behavioral counseling to those who screen positive, at grade B. Meanwhile the FDA's approved list for alcohol use disorder holds three medicines: naltrexone, acamprosate and disulfiram. Nalmefene is not on it. We found no source describing why, so this article says only that it is absent from the list.

That difference is the most useful thing in this whole comparison. Two systems use the same word, and one files it under treatment after a diagnosis while the other files it under prevention before one.

France shows how contested this territory can get. Its medicines agency granted a temporary recommendation for use of baclofen in alcohol use disorder in March 2014, then set a daily ceiling in a 2017 revision to reduce serious adverse events at high doses. In June 2018 a temporary specialist scientific committee concluded that the benefit-harm balance in alcohol use disorder was negative. Weeks later the agency said it opposed marketing authorization in the form applied for while supporting use in alcohol-dependent patients, and in October 2018 use was authorized, as a second-line option added to psychosocial support and with a daily ceiling. A cohort followed under the temporary scheme reported large drops in average daily consumption and craving scores among the small group tracked for a year or more, but it had no control group, so it cannot be read as an effect size. Both halves of that story belong in the same paragraph.

One more thing before the practical part. The yes-or-no question in most people's heads, am I an alcoholic or not, has already left the diagnostic manual. According to the US National Institute on Alcohol Abuse and Alcoholism, the current diagnosis is alcohol use disorder, graded mild, moderate or severe by how many criteria a person has met in the past 12 months. The older split between abuse and dependence was merged, craving was added as a criterion, and the legal-problems criterion was dropped. The international classification still lists alcohol dependence as a diagnosis, and the old dependence category maps onto the moderate-to-severe range of the current one.

This article does not reproduce those criteria, and it does not carry a self-test. Screening tools exist and health services use them to assess people. That is as far as we go, on purpose. Counting is something to do with a clinician, and the next sections cover where to find one.

None of the above makes reduction a better goal than stopping. The two facts we can support are that several health systems now accept reduction as a legitimate goal, and that where the problem is severe, abstinence is what gets recommended. The Japanese clinic that pioneered reduction-focused care wrote that rule into its own intake policy.

Is it safe to stop drinking cold turkey

This section comes before the practical tips. In this particular subject, reversing that order would be a mistake.

The NHS is blunt about it. "It can be very dangerous to stop drinking suddenly if you're dependent on alcohol." On its alcohol support page it goes further: "If you have become physically dependent and need to stop drinking completely, stopping overnight could be harmful." The advice is to get help with how to do it and with the medicine needed to do it safely. So the standard motivational line, just stop today, is the wrong instruction for some readers of this article.

The withdrawal symptoms the NHS lists are anxiety, trouble sleeping, nausea and vomiting, a fast heartbeat, sweating and shaky hands. Beyond those come seeing, hearing or feeling things that are not there, confusion and seizures. Symptoms typically begin 6 to 12 hours after the last drink and usually last 3 to 7 days.

Signs to get help with immediately

• Hallucinations, severe shaking, or a seizure. The NHS treats all three as emergencies and directs people to call the UK emergency number.
• Signs of acute alcohol overdose: mental confusion, difficulty staying conscious, vomiting, seizure, trouble breathing, a slow heart rate, cold and clammy skin, dulled responses, and extremely low body temperature. This can lead to permanent brain damage or death. Someone who is vomiting should be leaned forward, and someone lying down or unconscious should be turned onto their side with the ear toward the ground so they do not choke.
• Chronic liver disease such as hepatitis B or C. Korea's National Cancer Center guidance notes that liver cancer risk from drinking rises sharply in this situation, which is why it recommends cutting down.

When withdrawal needs to happen in a hospital is not left to guesswork. The NICE guideline on alcohol-use disorders recommends admission for medically managed withdrawal when someone has, or is assessed as being at high risk of, withdrawal seizures or delirium tremens. It also recommends inpatient or residential assisted withdrawal for people with a history of seizures or delirium, a history of epilepsy, a need to withdraw from benzodiazepines at the same time, or regular heavy drinking combined with a significant psychiatric or physical comorbidity such as severe depression, psychosis, heart failure or chronic liver disease. For people who are frail, cognitively impaired, or without social support, the threshold is set lower. The guideline states its drinking figures in UK units, and we have left them out here rather than hand anyone a number to measure themselves against.

The opposite side deserves equal space. The American Society of Addiction Medicine's withdrawal management guideline says the level of care should be decided from current signs and symptoms, the risk of severe or complicated withdrawal, and the person's recovery resources and environment. Patients whose risk factors are limited can be managed safely in outpatient settings. Where psychosocial support is low or the living environment is not safe, a more intensive level may be needed. The same guideline notes that a patient with active suicide risk should be treated in a setting equipped to respond to it. If that describes you or someone you live with, the national lines further down this page are entry points for both mental health and substance use, and emergency services are the route when someone is in immediate danger.

The range is also wider than most people picture. A review in an NIAAA journal describes withdrawal running from mild insomnia to life-threatening complications, with severe features such as hallucinations, seizures and delirium appearing between one and four days after withdrawal begins. Delirium can persist for three to four days and comes with dangerous signs such as a racing heart and fever. Alcohol withdrawal is one of the most common causes of status epilepticus in the United States, and about 5 percent of patients who experience delirium die from metabolic or cardiovascular complications, trauma or infection. That review dates from 1998, which we note because the field has kept moving.

Why no two countries agree on how much is too much

Put the official numbers side by side and the first thing you notice is that they are not measured with the same ruler.

Table: low-risk drinking guidance and standard drink definitions by country. Drink counts are not comparable across these rows.

The standard drink is where the confusion starts. The WHO defines one standard drink as about 10 g of pure alcohol, Korea's health ministry defines it as 7 g, and one UK unit is 8 g. So 14 units a week in Britain, 2 drinks a week in Canada and 2 drinks a day in Korea are quantities measured on three different scales. We could not confirm the gram conversion behind the American and Canadian figures from primary sources, so those cells are blank rather than filled in with a number we could not stand behind.

Korea publishes the arithmetic, which makes it a useful reference point. A 360 ml bottle of soju at 20 percent alcohol contains about 58 g of pure alcohol, which is 8.3 standard drinks by the Korean definition. A 225 ml can of beer at 4.5 percent works out at about 8 g, or 1.2 drinks. The formula is volume in millilitres times the percentage times 0.7947.

Two official Korean documents also point in different directions, and both are current. The low-risk drinking guideline's figures date from 2014. The national cancer prevention code, revised for the sixth time in 2023, advises avoiding even one or two drinks a day, and the accompanying guidance answers plainly that from a cancer standpoint there is no minimum level of drinking that prevents cancer. The documents have different jobs, one managing alcohol-related harm and the other preventing cancer.

Behind the WHO position is a classification most people have not connected to alcohol. Ethanol was classified decades ago by the International Agency for Research on Cancer as a Group 1 carcinogen, the same category as asbestos, radiation and tobacco, and current evidence cannot establish a threshold at which the carcinogenic effect switches on. Alcohol is linked to at least seven types of cancer. The WHO's January 2023 statement in The Lancet Public Health follows from that: no amount of drinking is safe for health.

And in January 2026 the United States removed its number, which is where the country sections start.

Where you actually go, and what you get

The practical differences are not in the numbers. They are in who you have to go through to get seen.

Table: national routes into alcohol treatment and reduction support. The reduction column describes institutional handling, not efficacy.

Having medicines available is not the same as having results guaranteed. A Cochrane review of opioid antagonists pooled 50 randomized double-blind trials with 7,793 patients and found that oral naltrexone lowered the risk of returning to heavy drinking to 83 percent of the placebo group. Twelve people needed treating to prevent one return to heavy drinking, 20 to prevent a return to any drinking, and drinking days fell by an average of four a month.

United States: the year the number came out of the guidance

US

On January 7, 2026 the Department of Health and Human Services and the Department of Agriculture published the 2025-2030 Dietary Guidelines for Americans without the daily limit that earlier editions carried. The line about no more than one drink a day for women and two for men is gone, replaced by a sentence telling people to consume less alcohol for better health. No numeric cap, and no distinction by sex.

Two days later the American Association for the Study of Liver Diseases issued a statement of concern. Its objections were that the new guidance sets no daily limit, does not reflect differences in how men and women metabolize alcohol, and makes no mention of the link between alcohol and cancer. The association set that against the 2025 Surgeon General's advisory, which it says identified alcohol as a leading preventable cause of cancer and described roughly 100,000 cancer cases and about 20,000 cancer deaths a year. We were not able to read that advisory directly, so those figures appear here only as what the liver association cited.

Which side has it right is not something this article decides. What is checkable is that two federal documents pointing in different directions came out of the same country inside twelve months.

The part of the American system that actually addresses drinking less is upstream of all that. The Preventive Services Task Force recommendation from November 2018 covers screening every adult 18 and over, pregnant people included, for unhealthy alcohol use, and offering brief behavioral counseling to those who screen positive. The grade is B and the net benefit for adults was judged moderate. The tools named are the WHO's AUDIT, its shorter consumption-focused version, and a single-question instrument. We are not reprinting any of their items here.

The front door is a single number. SAMHSA's National Helpline is 1-800-662-HELP (4357), with TTY at 1-800-487-4889. It is free, confidential and open 24 hours a day, every day of the year, in English and Spanish. It is a referral and information service rather than a counseling line, connecting people with mental health and substance use conditions, and their families, to local treatment facilities, support groups and community organizations. You can also text a ZIP code to 435748, or search FindTreatment.gov.

Then comes the number that undercuts all of it. In the 2024 national survey, 7.6 percent of people with alcohol use disorder had received treatment of any kind in the previous year, and 2.5 percent had received any of the three approved medications. One in 40. The barriers named are stigma, lack of insurance coverage, cost, access to specialty care, and a shortage of awareness that alcohol use disorder is a treatable medical condition. Three medicines, a free national helpline, a B-graded preventive recommendation, and one person in 40 getting the medication. The gap here is not written into the rules.

United Kingdom: a door that opens without a referral

GB

The British answer to the access problem was to remove the gatekeeper. The NHS tells people worried about their drinking to see a GP or an alcohol support service, and adds that if talking to a GP feels uncomfortable, they can contact their local drug and alcohol treatment service directly, with no referral. Registration happens online, by phone or through an online form. Whether self-referral is available, and how long the wait is, varies by area. Treatment is provided in confidence through the NHS and a range of other organisations.

The sentence in the NHS treatment pages that readers of this article are most likely to need is about goals rather than doors. The NHS writes that the aim of treatment is usually to help someone stop drinking completely, particularly where there is dependence, and then says that if you are not ready to stop, or only want to cut down, you will be supported to reduce your drinking to safer levels. A national health service put a patient's choice of goal in writing.

On medicines, the NHS overview page names three. Acamprosate and naltrexone are described as reducing cravings, and disulfiram as producing an unpleasant reaction if you drink. Nalmefene, the reduction-indicated one, does not appear on that patient-facing page, which is not the same as the country not using it: the NICE appraisal from 2014 survived its 2018 review with the recommendation intact.

The national line is Drinkline on 0300 123 1110. The NHS also points to Alcoholics Anonymous, Al-Anon Family Groups and Alateen for families, We Are With You, Adfam, Nacoa on 0800 358 3456, and SMART Recovery.

Japan: reduction written into a law, a plan and a prescription

JP

Japan legislated. The Basic Act on Alcohol Health Damage passed in December 2013 and took effect in June 2014, and the second basic plan under it was adopted by cabinet decision in March 2021. That plan requires liaison meetings among the parties involved to be held more than once in designated cities as well as at prefectural level.

The plan also carries an acronym worth borrowing: SBIRTS, a route running from early detection through brief intervention and specialist care to self-help groups. Alongside it came guidelines for regional and medical coordination, treatment guidelines and treatment manuals. Handing someone off to a self-help group is written into the national plan as a named step. In Britain that job falls to local services, and in the United States to a helpline.

Reduction-focused care existed as an actual clinic. In May 2017 Kurihama Medical Center, part of the national hospital organization, opened a reduction outpatient service. It did not require abstinence as a premise, and supported goals set patient by patient, such as drinking less or building a drinking pattern that no longer causes problems. It was the first of its kind in Japan and saw 389 new patients between opening and August 2020. People with a diagnosis of alcohol dependence could be seen, but where the problem was severe enough that abstinence was indicated, abstinence was what the clinic recommended, and patients needing inpatient care were sent to the existing alcohol department. New patient intake for that service ended at the end of August 2025. No reason appears in the published material, so we are not supplying one.

Self-help is where Japan diverges most visibly. Danshukai was formed in 1958 with Alcoholics Anonymous as a reference point, and a national federation oversees the local groups. Three differences stand out. AA keeps no organizational structure while danshukai is run as an organization. AA members participate anonymously while danshukai members give their real names. AA has no dues or joining fee and passes a collection, while danshukai charges membership and joining fees. Its distinctive feature is that family members speak about their own experience in the same room, and there are meetings held for families alone.

A room you can enter anonymously and a room where you give your name and bring your family sit side by side as two answers to one problem. Nothing in the material we have says which works better. Kurihama Medical Center reports that maintained abstinence is higher among people who take part in self-help groups, while noting from its own data that about one in four of its discharged patients participates, compared with roughly four in five of treatment-seeking patients in the United States joining AA. People who join and people who do not differ in motivation and severity to begin with, so that comparison cannot be read as an effect size.

South Korea: an entry point that is not a hospital

KR

The Korean route starts before medicine. The order set out in the National Cancer Center's guidance is a district public health center first, then an addiction management center, and only then a mental health welfare center or a search for a clinic.

There are 67 addiction management centers nationwide, 17 metropolitan and 50 district-level. They serve people in the community with alcohol, drug, gambling or internet-related addiction problems along with their families and neighbours, and are used by calling or by walking in. The mandatory services at district-level centers include early detection and brief intervention counseling, registration and case management of people with addictive disorders, rehabilitation programs and family support. The health ministry material we worked from does not state a fee, so this article does not call the service free. Ask the center. Separately, mental health welfare centers operate at municipal level, reachable on a single national number, 1577-0199, with the welfare counseling line on 129. For specialist inpatient care, the health ministry designates seven alcohol specialty hospitals, and designations can change from year to year.

That is the supply side. The demand side is the reason this section exists.

Korea's 2021 national mental health survey interviewed 5,511 people aged 18 to 79 in their homes. The 12-month prevalence of alcohol use disorder came out at 2.6 percent, 3.4 percent among men and 1.8 percent among women. That figure counts people who met the diagnostic criteria in the previous year.

The same survey produced a second 2.6 percent, and it counts something else. Among people diagnosed with alcohol use disorder, 2.6 percent had used a mental health service, the lowest figure among the conditions surveyed. That one is a reach rate, not a prevalence. Across all diagnosed mental disorders, 7.2 percent had used a service within the year and 12.1 percent had ever used one.

So the country with 67 centers, a national number and designated hospitals has the lowest service reach among the disorders it surveyed. The structure is there and the connection to it is not. One thing this article cannot tell you is how alcohol use disorder medication is licensed or covered by insurance in Korea, because we could not verify it. That cell in the table above is blank for the same reason, and nothing in the other country sections should be read across to Korea. Looking forward, the sixth National Health Promotion Plan for 2026 to 2030 lists a new health promotion levy on alcoholic drinks among the items under review, which is not the same as a decision.

Stack the four together and the failure points are different in each. America keeps a single phone number and three medicines while one person in 40 receives one of them. Britain took out the referral form and wrote the choice of goal into its patient guidance. Japan built the route into legislation, then the clinic that pioneered reduction closed its intake. Korea has the densest list of places to go and the lowest reach among the disorders it counted.

What you can start doing this week

Line up the official documents from different countries and the practical advice converges almost to the word. This is the least glamorous finding in this article and probably the most useful one.

Table: alcohol reduction practices published by British and Korean health authorities. Both lists come from official guidance, and they overlap almost entirely.

A few items did not fit the grid. The Korean list also suggests freezing leftover wine for cooking, not topping up a glass that still has drink in it, socializing with family and friends rather than around alcohol, and building up hobbies as a way of handling stress. The British list includes cutting down a little each day, with a note attached that on this approach every day you keep to counts as a success.

On record-keeping the Korean guidance is the more specific of the two. It asks for a week of logging while you work toward a target, then three to four weeks of continuous logging so you can see when and how much you actually drink, and a comparison of your original target against your real consumption. The template has columns for the day, the place, the type of drink, the number of drinks you intended, the number you had, and glass size.

Canada runs the free online version of all this. The Canadian Centre on Substance Use and Addiction, whose 2023 guidance was written by a panel of 23 experts from 16 organizations, operates a public information site built around its continuum-of-risk framing, with the Canadian standard drink definitions, reduction tips such as drinking slowly and drinking plenty of water, and separate guidance for pregnancy, young people and differences by sex. Its central line is the one this article would pick as a closing thought: any reduction in alcohol use has benefits.

There is also a myth to retire about counseling length. A Cochrane review of brief interventions in primary care found, across 34 trials and 15,197 people in the main analysis, that people who received a brief intervention were drinking 20 g less pure alcohol a week than controls a year later, on moderate-quality evidence. Twenty grams a week is about a third of a bottle of soju, and there is no point dressing that up as more than it is. Still, a conversation of 5 to 15 minutes leaving a measurable difference a year later is a real finding. The same review concluded that longer counseling probably adds little on top.

Where to make the first call

• United States: call 1-800-662-HELP (4357) or search FindTreatment.gov. You can also text your ZIP code to 435748.
• United Kingdom: see a GP, or contact your local drug and alcohol treatment service directly with no referral. The national line is Drinkline, 0300 123 1110.
• South Korea: call or visit your district public health center or an addiction management center. Mental health counseling and clinic referrals go through 1577-0199, and general welfare counseling through 129.
• Canada: the substance use and addiction centre's public information site carries the reduction tips and standard drink definitions.

Numbers and addresses reflect the date of the sources behind this article. If a line does not connect, check the current one on the relevant health authority's page.

One last passage, from the Korean guidance, because of what it says about tone. Its section on relapse states that not giving up is what matters most, that a failed attempt is material for working out the cause rather than grounds for discouragement, and that you should retrace when, with whom and in what setting the drinking ran over, then rebuild the plan with a solution to that cause in it. A government document with a chapter that assumes failure is a government document that does not treat this as a question of willpower.

What did not hold up

The line about a drink or two being good for your heart has been re-examined. The cardiovascular benefit that observational studies kept producing did not reproduce in genetic analysis. In a nonlinear Mendelian randomization study, light drinking was associated with a small increase in cardiovascular risk, and the protective effect the J-curve implied did not appear. Light to moderate drinking tends to travel with other, healthier lifestyle factors, and adjusting for those weakens the apparent heart protection.

Korea's official documents point the same way. The National Cancer Center guidance answers that a non-drinker starting to drink for the purpose of preventing cardiovascular disease is not recommended, and states that no minimum level of drinking prevents cancer.

The type of drink does not sort the risk either. Some argue that wine helps prevent certain cancers, but the conclusion so far is that the amount is what decides, and that differences in cancer risk between types of alcohol are close to nil.

Totals alone also miss something. For the same weekly total, drinking one or two glasses a day with food carries a lower risk of certain diseases than putting away seven or more in a single sitting, even once a week.

On drinking to cure a hangover, this article makes no efficacy judgment, because the only thing we can confirm about it here is telling. Morning drinking appears as one of the items in an international standard assessment instrument for drinking problems. It is treated as a risk signal, not as a remedy. On the remedies themselves, 21 placebo-controlled trials produced nothing that has been independently replicated. On why some people flush, a single ALDH2 variant leaves acetaldehyde stalling, and tolerating the flush is what raises the cancer risk. And on the table where the pressure comes from, what actually changed at East Asian company dinners, and what did not.

Frequently asked questions

Frequently asked

Do I need a referral, or can I contact a service myself?

It depends on the country, and the answer is more open than most people expect. In England you can contact your local drug and alcohol treatment service directly without going through a GP, though availability and waiting times vary by area. In the United States the national helpline and FindTreatment.gov both work as direct entry points. Korea's addiction management centers are used by phone or by visiting. For Japan we could not confirm whether self-referral is available, so that cell in our table is blank.

What happened to the American one-or-two-drinks-a-day limit?

It was removed. The 2025-2030 Dietary Guidelines for Americans, published on January 7, 2026, dropped the numeric daily limit and the distinction between men and women, and replaced them with advice to consume less alcohol for better health. The liver disease association raised concerns two days later, including the absence of any mention of alcohol and cancer. The WHO's position from 2023 is that no level is safe for health, and Korea's cancer prevention code advises avoiding even one or two drinks a day. The guidance you find depends on which body issued it.

Is there a medicine for drinking less rather than quitting, and can I get it in the US?

A medicine licensed specifically for reducing alcohol consumption exists in Europe, where it was approved in 2013, and in Japan, approved in 2019. The FDA's approved list for alcohol use disorder holds naltrexone, acamprosate and disulfiram, and does not include it. We have no source on the review history behind that, so we will not speculate. What the American system does offer for cutting down is brief counseling in primary care, recommended at grade B. Which medication, if any, fits your situation is a question for a clinician, and every guideline we read attaches psychosocial support as a condition rather than treating medication as a standalone.

Can I compare drink counts between countries?

Not directly, and this trips up more people than any other detail in this subject. One standard drink is about 10 g of pure alcohol by the WHO definition and by Australia's, 8 g in a UK unit, and 7 g under Korea's health ministry definition. So 14 units a week and 10 drinks a week and 2 drinks a day are quantities on different scales. We could not verify the gram figures behind the American and Canadian numbers, so we left them out rather than guess.

Does a five-minute conversation with a doctor actually change anything?

By a measurable amount, yes, and by a modest one. In the Cochrane review of brief interventions in primary care, people who got one were drinking 20 g of pure alcohol less per week than controls a year later, on moderate-quality evidence, while effects on binge frequency and drinking days were small. The same review found that longer counseling probably adds little. So the value is in having the conversation at all rather than in its length.