Drink for health? You and that scientist were cheated.

Author | Li Chenyang Wang Dongli

"Drink some red wine, beauty beauty, to soften blood vessels." "It doesn't matter if you drink less, it can also prevent coronary heart disease." Have you ever heard such a toast on the wine table?

Although the harm of alcoholism is certain, many people still think that drinking a small amount of alcohol is not only harmless to health, but also beneficial.

However, a study published in Nature-Communication in June 65438+1October 65438+March by Yang Jian, a professor at West Lake University, poured a pot of cold water on the chicken soup for wine lovers.

Contradiction: Is it negative correlation or positive correlation?

In various articles circulating on the Internet, the benefits of drinking a small amount of alcohol include softening blood vessels, lowering cholesterol, controlling weight and even preventing tumors.

To the delight of wine lovers, some studies published in regular academic journals also support "moderate drinking".

20 19 A study published in Nature-Genetics pointed out that weekly alcohol consumption was negatively correlated with diseases such as type 2 diabetes, chronic kidney disease, Parkinson's disease, major depression and autism spectrum disorder.

However, a paper published in Molecular Psychiatry on 20 17 claimed that the increase of alcohol intake was negatively correlated with the degree of obesity.

In some studies, the relationship between alcohol consumption and disease risk presents a "J" curve, which seems to indicate that there is an "optimal" alcohol consumption. At this point, the risk of various diseases is minimized, even lower than that of teetotalers.

This curve is like putting a wonderful "tick" on the behavior of drinkers.

Xue, a doctoral student at the University of Queensland, Australia, studies the causal relationship between drinking, smoking, exercise and other lifestyle habits and diseases.

Soon, he found many contradictions.

"In genome-wide association studies's paper, there are indeed many conclusions that drinking is negatively correlated with diabetes and cardiovascular diseases. But if we use Mendel's randomization method for causal analysis, we will find that drinking has a risk effect on those diseases. " He said, "This makes me feel very strange."

When looking at the literature further, the researchers noticed the crux like detectives.

Mystery: Who lied in the questionnaire?

"You can only investigate people, but you can't do experiments with people-this is the biggest difference between human epidemiological research and animal experiments."

Yang Jian, correspondent of the paper, told China Science Journal, "In this case, questionnaire survey has become the most commonly used research method, although it is not ideal."

The loophole is hidden in the questionnaire.

The British Biobank is one of the largest human information resource banks built so far.

Researchers around the world can use their data to explore the relationship between genes, lifestyles and human diseases.

The researchers analyzed the data of more than 450,000 people in the British Biobank, among whom about 1.4 million people claimed that they never drank alcohol.

Interestingly, the database also provides medical records of these people in British hospitals.

The researchers found that at least 10% of the14,000 self-proclaimed "teetotalers" suffered from diseases closely related to drinking, such as alcoholic hepatitis.

Others told the truth in front of the doctor and admitted drinking.

There are also some questions about the type, frequency and quantity of drinking, such as "how often to drink wine" and "how much beer and red wine to drink". Results Nearly 654.38+00000 people just said that they had drunk alcohol in the past year, and then filled in "0" in the column of various alcohol intake.

"False positives are actually a very common phenomenon." Xue Anli said, "In the questionnaire we analyzed, 10%~20% people have this problem."

But this may be just the tip of the iceberg, because some false positives are undetectable by researchers. If someone drinks two glasses of wine a day and only reports one and a half glasses in the questionnaire, his lies may never be discovered.

Strange talk: abstainers get sick instead?

In addition to lying intentionally or unintentionally, another more subtle factor is also affecting scientists' judgment, which researchers call "the vertical change of time." How to understand this concept?

The way to ask questions in a questionnaire is usually this: How much wine have you drunk (please answer the situation in the past year)? In other words, the data collected by the questionnaire can only reflect the situation within one year.

However, the questionnaire of British Biobank only added another question: Compared with 10 years ago, did you drink too much, drink less or remain the same? The data showed that people who drank less than 10 years ago were much more likely to get sick than the other two groups. Did abstinence make their health worse?

"You have to think about it: some people who have drinking habits have suddenly stopped drinking or even drinking in recent years. What may be the reason? " Jackyang laughed.

The questionnaire further questioned the reasons for these people to reduce drinking, the most important of which were "illness" and "doctor's advice".

"This means that a large number of people with diseases have been enriched in this part of the population who have reduced drinking."

Yang Jian explained, "It can be seen that individual behavior changes with time, which often leads to very large systematic errors in the data."

After correcting the above-mentioned factors leading to errors, the researchers found that the "negative correlation" between drinking and many diseases disappeared, while the "positive correlation" between drinking and diseases such as hypertension, diabetes and iron deficiency anemia began to appear.

As for the "hook"-"J" curve that relaxes wine lovers, it has also become a boring straight line after removing inconsistent data.

In other words, there is a simple linear relationship between disease risk and alcohol consumption-drinking less and getting sick, drinking more and getting sick, which is very simple and rude.

"We can basically determine now that drinking is basically bad for our health. Those studies that come to the conclusion that a small amount of drinking is good for health are probably because someone lied about drinking behavior or changed their living habits because of illness. " Yang Jian concluded.