According to the new recommendations from the National Board of Health and Welfare, health services should offer support to people who drink ten standard glasses or more per week, or four standard glasses or more per occasion once a month or more often. Ephi spoke to epidemiologist Måns Rosén, who questions the National Board of Health and Welfare's assessment after reviewing the research behind the advice.
- When the National Board of Health and Welfare says that the risk of disease or premature death is lower the less alcohol you drink, it is simply not true.
The words are Måns Rosén's. Rosén is an epidemiologist and former head of the Epidemiological Centre at the National Board of Health and Welfare, former director general of the Swedish Agency for Health and Social Evaluation (SBU) and previously active at Karolinska Institutet and Umeå University.
He says this in the context of the much-publicised new limits for high-risk alcohol use in the health sector, which the National Board of Health and Welfare updated and published in September 2023.
The threshold for risky drinking is now the same for women as for men: a maximum of ten standard glasses of alcohol per week, or a maximum of four standard glasses on a single occasion once a month. The new guidelines are not aimed directly at the general public, but at healthcare professionals to assess when healthcare support should be offered to help a patient reduce their alcohol consumption.
The National Board of Health and Welfare's communication on the new guidelines is clear: the more alcohol a person drinks, the higher the risk of ill health or even premature death. It is not possible to set a limit for when alcohol consumption is risk-free.
But is that really true?
To understand how the National Board of Health and Welfare came to the conclusion that less alcohol is always better, some background is needed. Almost everyone knows that people who drink alcohol in moderation have better health than those who drink a lot of alcohol. It is undisputed that excessive alcohol consumption can have devastating consequences, both for the drinker and for those around them.
But a lot of research has also long shown that people who drink moderately also seem to have slightly better health compared to those who don't drink at all. There may be several reasons for this: it could be due to some effect of the alcohol itself, but it could also be that people with moderate alcohol consumption have more social contacts than others, which we know is good for health. The poorer health of teetotalers could be due to the fact that they have fewer social contacts than others, but also to the fact that some people who do not drink at all have had problems with alcohol in the past. But even if you adjust for these factors, in many cases there is still a tendency for people with moderate alcohol consumption to have slightly better health than others.
I the knowledge base which forms the basis of the National Board of Health and Welfare's new guidelines, refers to new research. This research is intended to show that the previous findings that moderate drinking could be healthy are not correct. Måns Rosén has read the evidence base and reacts to the fact that it is not particularly comprehensive:
- 'I am surprised that the evidence base is so thin, and focuses a lot on the recommendations of other countries. There are many systematic reviews in the field of alcohol consumption and health, but the one referred to by the National Board of Health and Welfare is questionable," says Rosén.
The overview study that Rosén refers to, which according to Rosén has had a major impact on the National Board of Health and Welfare's new assessment, is a article from January 2023 which was published in the journal JAMA Network Open, and is a reference in the National Board of Health and Welfare's knowledge base. There are reasons to delve into that article.
The title is 'Association between Daily Alcohol Intake and Risk of All-Cause Mortality', and it reviews 107 other studies covering 4.8 million people whose alcohol consumption and health were studied. It is thus a so-called meta-study, which uses and compiles the results of other studies and analyses them. If you read the summary in layman's terms, the results of the meta-study are clear:
"Low-volume alcohol drinking was not associated with protection against death from all causes."
Hard to misunderstand, one might think. According to the article's summary, low alcohol consumption has no protective effect, contrary to what many previously believed.
But the study has has been criticised for how it handles statistics. A criticism shared by Måns Rosén.
The study distinguishes between 'non-drinkers' (which we can call teetotalers) who do not drink alcohol at all, and 'low-volume drinkers' (which we can call moderate drinkers) who drink alcohol, but no more than 24 grams of pure alcohol per day. This corresponds to two standard Swedish glasses: about two small glasses of wine or two 33cl cans of strong beer per day.
The study analyses the relative risk of death for moderate drinkers compared to teetotalers. If the result is above 1, it means that the relative risk of death is higher among moderate drinkers than among non-drinkers. It also indicates a confidence interval, which is simply an upper and lower limit of the result that shows how certain the estimate is. A 95 per cent confidence level is often used in this type of research.
The relative risk for moderate consumers is, perhaps somewhat surprisingly given the conclusion, 0.86. Moderate alcohol consumption thus appears to lead to a 14 per cent lower risk of dying compared to the death risk of sober people. The confidence interval is between 0.83 and 0.88.
- But then the study takes the next step and tries to adjust for confounders, i.e. factors that may co-vary with mortality and disturb the results. Some of this makes sense, such as adjusting for socioeconomics and smoking. Other parameters included are more questionable. It is difficult to know how these parameters interact with each other,' says Måns Rosén.
When these parameters are taken into account, the result is instead 0.93. So there still appears to be a seven per cent lower relative mortality risk for moderate drinkers. But the result also becomes more uncertain, as shown by the increase in the confidence interval: it becomes 0.85-1.01 instead. It therefore extends on both sides of 1, albeit by the smallest possible margin. From this, the study concludes that there is no significant result. Formally, this is true.
- Now the authors have used the 95% confidence level. But if they had used the 92% confidence level instead, the confidence interval would have been slightly smaller, and the results would still have shown that moderate consumers had a lower relative risk of mortality compared to those who do not drink alcohol. In other words, we can simply say that with a 92 per cent probability, the study's conclusion is not correct," concludes Måns Rosén.