Drinking seven cans of beer a week can contribute to cognitive decline, according to a new study.

It’s long been understood that drinking alcohol can affect the brain negatively, which is why everyone is encouraged to consume it moderately.

A large observational study recently found that moderate drinking – say, seven cans of beer – can still increase iron buildup in the brain, contributing to cognitive decline.

Published in PLOS Medicine, the researchers behind the study analyzed existing U.K. Biobank data for about 20,729 people. As with all observational studies, researchers can draw associations that point to further avenues of exploration.

About 48.6% of the individuals analyzed were female, with a mean age of 55 years and an average intake of around 17.5 units of alcohol per week. Meanwhile, 2.7% of the study population were never-drinkers.

The participants self-reported their alcohol intake through questionnaires, while the researchers assessed cognitive function using a series of executive function tests. They include trail-making, fluid intelligence with puzzle tasks, and reaction time based on a car game.

The researchers used MRI scans to assess the presence of iron in the brain, which revealed that individuals drinking seven or more alcoholic units per week showed markers of increased accumulations of iron in their brain.

According to the study author Dr. Anya Topiwala, of the Department of Psychiatry at the University of Oxford in the U.K., while the study showed higher iron at seven units and above, a separate study they published in medRxiv found evidence that there’s no safe level over zero alcohol units.

Topiwala said elevated iron is a possible mechanism that could contribute to brain damage and memory problems.

“Understanding the mechanism of damage is, of course, important if you want to be able to intervene in patients,” she said.

The current analysis also revealed that those with higher levels of iron in the brain were more likely to develop cognitive issues down the line, if they don’t have it yet.

However, the authors indicated that the causal association for the observational trial was not well-established.