Based on a new analysis, the health benefits of having a drink a day may be exaggerated, UK researchers say.

Only women over age 65, if anyone, might get a protective effect from light drinking, compared to people who never drank, the study found.

Much past research has shown that people who drink a low to moderate amount of alcohol – about one drink per day – fare better than both heavy drinkers and those who abstain completely.

But that non-drinker comparison group may be skewing the results, according the new study’s lead author Craig S. Knott of the department of Epidemiology and Public Health at University College London.

People who used to drink but don’t anymore, for whatever reason, tend to be less healthy and more likely to die than others, he told Reuters Health by email.

“With existing research having largely grouped former and never drinkers together, there was the possibility that reductions in risk among lighter drinkers may be partly due to their comparison against people who are simply less healthy,” he said.

And people who have never had alcohol may have other health conditions that make drinking unsafe and make them less healthy than others, he noted.

When he and his coauthors left out former drinkers and compared light, moderate and heavy drinkers to those who only occasionally drank, the apparent benefits of drinking disappeared almost entirely.

They used annual, nationally-representative surveys of English adults age 50 or older collected from 1994 to 2008 and including a total of more than 45,000 people. At each survey, the participants reported how often they consumed alcohol during an average week and how much they drank on the heaviest drinking day of a typical week.

As other studies have found, currently drinking alcohol was associated with a lower risk of death in almost all age groups and levels of drinking compared to people who were not current drinkers, based on a linked database of deaths in England.

But when the researchers excluded those who used to indulge but had given it up, the apparent protective effect remained only for men between ages 50 and 65 and women over age 65. Protection was minimal for the men and only applied to those who drank at least seven pints of beer in an average week. It was broader, but still low, for the women, the authors write in BMJ.

Some other studies have in fact found the opposite, that alcohol in moderation is more effective for men than women, according to Dr. Giovanni de Gaetano of the University of Milan in Italy, who was not involved in the new study.

“Relative to occasional drinkers, there was little to no indication that regular light consumption may actually be of benefit in any age-sex group, at least in terms of mortality risk,” Knott said.

Any reductions in risk that remained after the study team adjusted for personal, socioeconomic and lifestyle factors, and left out the former drinkers, could still be the result of other biases in group selection, inaccurate alcohol reporting or residual factors that haven’t been accounted for, Knott said.

“On this basis, it seems sensible for current drinkers to consider moderating the amount of alcohol they consume, and for non-drinkers to remain abstinent,” he said.

Assigning a group to drink alcohol and another to abstain for a study would be unethical, so researchers have to try more sophisticated ways of comparing people in the real world and trying to account for other differences between them, he noted.

There is no good biological explanation for why light drinking might benefit older women, according to Jürgen Rehm, director of the Social and Epidemiological Research (SER) Department of the Center for Addiction and Mental Health in Toronto, Canada.

“Let us face it: alcohol is not consumed for health reasons,” Rehm told Reuters Health by email. “To avoid too much risk, people should drink lightly, best one drink at most per day. Most European drinkers exceed this limit and put themselves into risk.”

As age increases, it gets harder and takes longer for the body to eliminate ethanol from the blood, the authors note. Alcohol-related hospitalizations and deaths are most common for older age groups, they write.

“Nobody should drink alcohol for health reasons,” Rehm said.