Cup of coffee

A new study doesn't actually show that coffee can prevent cancer, despite the headlines.

Coffee is one of those things that feels like it should be bad for you, probably because it tastes nice and lots of people drink it. So it might surprise some people to see headlines this morning claiming that coffee could protect against breast cancer.

Here’s what the press release said:

Coffee reduces breast cancer risk

Recently published research shows that coffee drinkers enjoy not only the taste of their coffee but also a reduced risk of cancer with their cuppa. More detailed research published today in BioMed Central’s open-access journal Breast Cancer Research shows that drinking coffee specifically reduces the risk of antiestrogen-resistant estrogen-receptor (ER)-negative breast cancer.

Pretty strong words. The message is clear: drink coffee, and your risk of breast cancer (or at least one type of it) will go down. Is that what the study showed?

Not really.

The press release leaves out important details in the study, which either contradict these conclusions, or severely weaken them.

Let’s take a closer look.

About the study

The study itself was a “case-control study” that compared around 2,800 Swedish people with breast cancer (cases) and 3,100 people without it (controls). The researchers asked these people to remember their coffee-drinking habits and other aspects of their lives, many years ago.

Although they can be useful, case-control studies have several inherent problems. For example, some people might remember things inaccurately, or there might be important but undetected differences between your cases and controls.

The study found that people who said they drank more than 5 cups of coffee a day were 20 per cent less likely to have breast cancer, compared to those who didn’t drink any.

But this apparent difference between coffee-drinkers and coffee-abstainers could be explained in other ways, such as how much exercise they get, how much alcohol they drink, how old they are, and so on. After adjusting their stats for these other factors, the team found that the protective effect of coffee had weakened to the point where it was no longer statistically significant.

The only interesting result from the study came when the team focused on women with a specific type of breast cancer, called ER-negative breast cancer. Here, even after adjusting for all the other factors, people who drank more than 5 cups a day were 57 per cent less likely to have breast cancer than those who didn’t drink any.

This effect was statistically significant, as was the trend (which means that the more people drank, the lower their risk).

A storm in a teacup?

There are two possible ways to interpret this.

First, you could argue that there is indeed a link between coffee and one specific type of breast cancer.

This doesn’t mean that coffee protects against breast cancer, as the press release suggests. It is simply an association, rather than proven cause and effect. But it’s not an unreasonable one.

We know that coffee contains thousands of chemicals, some of which have proven effective against cancer in lab experiments (such as these two recent examples). We also know that breast cancer is not just one uniform disease. There are different types, and they behave in subtly different ways. So, it’s possible that some chemical in coffee prevents the development of some type of breast cancer.

Alternatively, you could argue that coffee doesn’t have anything to do with breast cancer at all, and the single positive result from this study is a fluke. This is a recurring problem, especially when studies like this divide their initial set of people into smaller and smaller groups, each with fewer and fewer people.

It means you end up doing more calculations, and you increase the odds that any one of them will throw up a positive result, just through chance. These “subgroup analyses” can end up as fishing expeditions, where you throw several fishing lines into the water and see if anything catches the statistical bait. (More on this here)

History repeating

Normally, you might look at a result like this, file it away, and wait to see if other studies find the same thing. This is critically important in science, and something that many news reports fail to mention. Positive results need to be checked and repeated to ensure that they’re not just one-offs. Science survives on replication.

The thing is, these researchers effectively did another study to check their results.

They repeated their analysis in a different case-control study from Germany, including 3,464 women with breast cancer and 6,657 women without it. Did they find the same result?

In a word, no.

In this validation study, there was not a single statistically significant result, even when looking only at ER-negative breast cancers.

But the researchers are undeterred. They write:

“Though not reaching statistical significance, the strongest protective effect from heavy coffee consumption was similarly observed for the ER-negative subtype in the validation study. We believe that, collectively, the results from the two studies in this paper support a protective effect of high intakes of coffee against ER-negative breast cancer. The weaker associations found within the [German] study may perhaps be attributed to other factors related to coffee drinking, such as brewing method, bean type, and caffeine content.”

This seems to be over-reaching. Another possible explanation is that the result in the first Swedish study was a fluke, which is why analysing the larger study failed to validate it. They may well be right about the difference between methods of preparing coffee – the point is that their data don’t lend a lot of confidence to that interpretation.

So what have other studies found?

In 2009, a Chinese group combined the results of 15 different studies and found that every 2 cups of coffee a day were associated with a 2% lower risk of breast cancer – a tiny effect, and one that was only just statistically significant. A similar study reported a 6% lower risk – slightly larger, but still pretty small – and much smaller than the effects of well-known things like exercise and cutting down on alcohol consumption.

What about specific types of breast cancer?

An American study of 52,000 women found no significant link between coffee and any type of breast cancer. A Swedish study of 61,000 women similarly found no significant links. Another American study of 85,000 women found a weak association between coffee and ER-positive breast cancers but not ER-negative ones. And finally, a fourth study of 38,000 women found that women who had the most caffeine in their diets had higher risks of ER-negative breast cancer.

To summarise in technical terms, the data are “all over the place”.

There are many reasons to drink coffee. You might like the taste. You might be very tired. But don’t count on it to reduce your risk of breast cancer.


Breast Cancer Research 2011, 13:R49 doi:10.1186/bcr2879