Thursday August 20 2015
"Just one drink a day raises breast cancer risk," is the front page headline in the Daily Mail following the results of a new study. While the health risks of heavy drinking are well established, the effects of light drinking are less clear.
The study, which involved almost 136,000 people, found women who drank the equivalent of a glass of wine a day over a 30-year period were 13% more likely to develop one of the alcohol-related cancers (breast cancer being the most common) than women who didn't drink at all.
The study found low to moderate drinking increased the risk of certain types of cancers already thought to be linked to alcohol, but only among women or people who smoked. Men who didn't smoke and drank moderately had no increased risk of any type of cancer. But these were relatively small risk increases for low to moderate drinking – those for heavier drinking were much greater.
In an accompanying editorial, the lead researcher recommends "women with a family history of breast cancer should consider reducing their alcohol intake to below recommended limits [no more than 2-3 units a day], or even abstaining altogether". And all people, whatever their medical history, are recommended to take a break from drinking a few days a week.
Where did the story come from?
The study was done by researchers at Harvard TH Chan School of Public Health and Harvard Medical School in the US.
It was funded by grants from the US National Institutes of Health, and was published in the peer-reviewed BMJ. It is available to read free online on an open-access basis.
The research was reported widely in the UK media, but with some confusing contradictions. In part this may be because the study itself made lots of different comparisons, resulting in many different statistics.
The Daily Mail's coverage, which was front page news, was generally accurate but failed to make clear that while the increase in cancer risk for women who drank moderately was statistically significant, it was also relatively small.
The Independent's claim that men who drank moderately had an increased cancer risk did not make it clear this only applied to men who had smoked.
The Daily Telegraph's headline figure that one drink a day increases breast cancer risk by 15% does not appear anywhere in the study. This may be a rounding up of the 13% increased risk of any alcohol-related cancer for women. Specific risk figures for breast cancer are not given.
The BBC News story is balanced and accurate, and points out that the overall increase in cancer risk is likely to be small for those who stick to the recommended alcohol intake.
What kind of research was this?
The research was based on two prospective cohort studies that followed large groups of people over time, recording detailed information about their lifestyles and health.
Cohort studies are good at showing whether there are links between lifestyle factors such as how much alcohol you drink and outcomes such as getting cancer. But they cannot prove that one thing causes another, as other factors may be involved.
What did the research involve?
The researchers used the records of more than 88,000 women and 47,000 men in two long-running US studies of health professionals, which started in 1980 and 1986.
They asked the people involved in the studies questions about their lives, including family history of cancer, their diet, whether they smoked and how much they exercised, and recorded their height and weight.
They followed people until 2010, and looked at how many got cancer during that time. They wanted to know whether people who drank alcohol were more likely to get cancer than people who did not.
The researchers were particularly interested in whether people who drank a small amount of alcohol had a higher chance of cancer. They looked at cancer overall, and then at specific cancers we know are linked to alcohol: bowel cancer, breast cancer, mouth or throat cancer, and liver cancer. Although we know drinking a lot of alcohol is linked to these types of cancer, we don't know whether light to moderate drinking affects the risk.
When analysing the results, the researchers adjusted their figures to take account of the effect of other things that may raise the chance of getting cancer, including smoking, obesity and a family history of cancer. Smoking in particular can affect the results, as people who smoke are more likely to drink alcohol – and drink more of it.
The researchers used US measures of alcohol, which do not convert directly into UK units. They defined light to moderate drinking as less than 15g of alcohol for women and less than 30g of alcohol for men.
One UK unit includes 8g of alcohol – for women, 15g of alcohol is just over 2 units. For men, 30g of alcohol is just under 4 units. One unit of alcohol is about equal to half a pint of ordinary-strength beer, or a small pub measure of spirits. A small glass of ordinary-strength wine is about 1.5 units. UK recommended limits are no more than 2-3 units a day for women and 3-4 units for men.
What were the basic results?
The study found people who drank alcohol were more likely to get cancer than people who did not drink alcohol. The more alcohol people drank, the more likely they were to get cancer.
For people who drank light to moderate amounts of alcohol, the chances of getting any type of cancer overall were "minimally increased". This means that though there was a suggestive trend for an increased risk, the risk differences between light to moderate drinkers and teetotallers were non-significant, or only on the border of being significant.
However, for cancers known to be linked to alcohol, the picture was more complicated. Women in the study who didn't smoke and drank moderate amounts of alcohol were 13% more likely than teetotal women to get an alcohol-related cancer (relative risk 1.13, 95% confidence interval 1.06 to 1.2).
This was mainly accounted for by breast cancer. Though no risk figures are provided specifically for breast cancer, removing breast cancers from their risk calculations gave a non-significant result. Men who didn't smoke and drank moderately were not more likely to get alcohol-related cancer.
The researchers also looked at whether how often people drank alcohol, or whether they drank heavily on individual days, had an effect. They found the total amount of alcohol people drank was more important than patterns of drinking.
How did the researchers interpret the results?
The researchers said their results showed that, in women who have never smoked, the risk of breast cancer "increases even within the range of one alcoholic drink a day".
They said people's decisions about whether to drink alcohol, and how much, should take account of factors such as whether they have ever smoked and if they have a family history of alcohol-related cancers.
The impact of drinking alcohol on health has been debated for years. While some research suggests alcohol can benefit heart health, other findings suggest the negative effects outweigh any positives.
It is clear that heavy drinking is bad for health in many ways, including being linked to a higher chance of getting certain cancers, such as those of the breast, liver and bowel. The big question is the effect of drinking low or moderate amounts of alcohol.
This large study found non-smoking women who drank the equivalent of one small glass of wine a day had a small increased chance of getting one of these alcohol-related cancers – mainly attributed to increased breast cancer risk. But it did not find that light or moderate drinking increased the risk of non-smoking men or women getting any type of cancer in general.
The study benefited from having detailed health information about a large group of men and women, followed for a maximum of 30 years. People were asked about their alcohol consumption at the start of the study and again about every four years, as people's drinking habits may change over time. This makes the results stronger. But there is always a possibility that people will not accurately report how much alcohol they drink.
The researchers adjusted their figures to take account of some of the other things that can affect the risk of cancer. For example, they found that if people smoked, smoking was likely to have a bigger effect on their risk of cancer than drinking.
However, it is possible that the influence of other health and lifestyle factors may not have been fully taken into account. These may be things influencing cancer risk in general, or related to specific cancers.
Some factors known to affect the risk of breast cancer, for example, such as the number of children a woman had, whether she breastfed, and use of the contraceptive pill, were not included. We don't know whether these could have affected the results.
Even if we had all this information, we could not say for sure that there is a set threshold level of alcohol below which it is safe to drink, and above which alcohol will cause cancer. We can only say that, in this large group of people, those who drank increasing amounts of alcohol were more likely to get cancer than those who did not.
The researchers speculated that substances that form when alcohol is broken down by the body could have a toxic effect that may increase the risk of cells turning cancerous. They suggest breast tissue may be particularly susceptible to the potentially harmful effects of alcohol.
In an editorial in the BMJ accompanying the piece, one doctor wrote this study means women with a family history of breast cancer should consider abstaining from alcohol altogether. This advice was widely reported, although the research did not find a stronger link between alcohol and cancer risk for women with a family history of breast cancer. Only 3% of breast cancers are caused by inherited faulty genes, according to Cancer Research UK.
The main finding of this study is that low to moderate drinking may increase a woman's chances of getting alcohol-related cancer by 13%, relative to the chances of teetotal women.
This is a relatively small risk increase, and may be much less than the influence of other factors – some you can change (weight, diet and physical activity), and some you can't (age, gender and hereditary factors). Whether this small increased chance of cancer merits cutting down to less than the current recommended limits, or stopping altogether, depends on your attitude to risk.