According To Study: Does Bacon Actually Cause Cancer?
Who's saying what
The man pictured above is Ron Swanson, a character in the US TV show Parks and Recreation. As you may be able to gather from his statement, he enjoys bacon. But does Swanson have an increased risk of cancer because of it? If you believed the UK tabloid The Daily Mail (who, Ben Goldacre points out, have "an ongoing project to divide all the inanimate objects in the world into ones that either cause or prevent cancer"), then you would say yes, that Ron will eventually pay a heavy price for his bacon eating. After all, the Daily Mail has published several articles saying so (see here, here, here, and here)! Of course, last year, the Daily Mail also published an article saying that, actually, bacon is good for you. So maybe Swanson will be okay?
By now, some of you are probably sitting there thinking, "who cares about the Daily Mail? This is Straya, mate". And fair enough! Oh, except for that Rupert Murdoch's stable of tabloids here in Australia very commonly use content from the Daily Mail, which our beloved Rupester also owns. Your mum and your greengrocer read articles about cancer largely sourced from the Daily Mail if they read the Herald Sun and Courier Mail (see here, here, or here). After reading these articles, you probably sit there thinking, "why can't scientists make up their minds about this stuff? Is bacon good or bad? MAKE UP YOUR MIND, BOFFINS!"
Somewhat surprisingly, in some ways, it's not that the Daily Mail are making things up - they do base their articles on actual proper research done by serious scientists. Instead, the reason why scientists seemingly can't make their minds up is basically just a matter of statistics. Cancer is a statistical disease. We (should) talk about cancer risks. Some people get cancer, others don't. But there are certain things which definitely increase your risk of getting cancer that most people know about - smoking all your life, for example, or being exposed to asbestos or radiation, or sun damage. The older men on the Channel 9 cricket commentary team recently spent about 10 minutes of downtime discussing the large amount of skin cancer they've had removed over the years. Ian Chappell said that he now regrets wearing shirts open to show off his chest hair in the 1970s before they knew about sunscreen, because that's where he's had cancers removed.
But remember that even smoking doesn't automatically lead to cancer - it just substantially increases the risk. Risks aren't uniform. If someone is twice as likely to get cancer because they smoke, it doesn't necessarily mean that there will be exactly twice as many cases of cancer in those who smoke. Mathematically, the likely amount of cases can vary a surprising amount - if you toss a coin 100 times, you probably would expect to get about 50 heads and 50 tails. But it probably wouldn't bug you if you got 40 heads and 60 tails - it happens sometimes. And it'd be really weird, but it's not entirely impossible that someone tossing a coin 100 times one day might get all heads and no tails. And cancer is like this. Think about it: cancer is basically a small bunch of cells in your body that have rebelled. They're just in it for themselves, so they grow and grow where the rest of your body stays still. The reason why they keep growing is because the genes in those cells have mutated. What the nicotine in cigarettes does - or sun damage, or, maybe, bacon - is make cells more likely to mutate. Every cigarette is like tossing a coin. You might be lucky to toss more heads than tails, and through sheer luck avoid having nicotine mutate your genes in the kind of specific ways that cause cancer (of course, you very likely won't be this lucky if you chain-smoke for fifty years. I'm not endorsing your filthy habit.)
So how do researchers figure out whether, say, bacon actually has an increased risk of causing cancer? Generally they either do cohort studies or case-control studies. A cohort study is where they interview thousands of people about their diet, and then monitor them to see who gets cancer and who doesn't. A case-control study is where they interview the dozens or hundreds of people who, say, come to a hospital with a particular cancer over a particular period, and then they find and interview people of equivalent age, gender, income and so on. The researchers in a case-control study then look at the data to see if there are any differences between the people with cancer and the people who don't have cancer.
For example, one cohort study interviewed almost eight thousand 46-65 year old Hawaiian men of Japanese ancestry in 1965-1968, and then followed up on them in 1986. In that time period, they wanted to know, how many of them had been diagnosed with stomach cancer, one of the harder-to-treat and more-common cancers? I wouldn't wish stomach cancer on my worst enemy, but a fair few of the people in the study were diagnosed with it between 1965 and 1986: 1.2% of non-smokers got stomach cancer, while 2.9% of smokers got stomach cancer (this is a big difference in the world of cancer studies, and not really surprising, considering). The same study also looked at their diet. 1.9% of those who never ate bacon (or sausages) got stomach cancer. 2.2% of those who ate bacon or sausages at least 5 times a week got stomach cancer. So it's an open and shut case, right? An extra 0.3% of old Hawaiian men who eat Ron Swanson-worthy levels of bacon get stomach cancer. BAN BACON, THE NEWSPAPERS SAY!
Well...not so fast. The thing is that this 0.3% is within the bounds of random chance. The researchers gave a mathematical index of the range of possibilities of what that 0.3% means, called a 95% confidence interval. The range of possibilities inherent in that 0.3% basically goes from eating lots of bacon is very slightly better for you than going without all the way to eating lots of bacon doubles your risk of cancer. This is because there's just not that many people who get stomach cancer, and extrapolating from them is bound to be imprecise; maybe it's just pure chance that more of them happen to eat a lot of bacon. Maybe not. But believe you me, the Daily Mail would feel free to report this finding with the headline 'BAN BACON, IT CAUSES CANCER'.
To get a better idea of whether something is a cancer risk, people studying cancer reduce the range of possibilities by combining the results from lots and lots of studies - this is called a meta-analysis. A Swedish team recently did a meta-analysis on studies that looked at bacon's effect on stomach cancer, where they combined the data of 15 separate studies (using some statistical wizardry). Some of the studies the Swedes found suggested that it was likely eating lots of bacon halved your chance of stomach cancer, some of them found that it was likely that eating lots of bacon doubled your chance of stomach cancer. This wide variation isn't necessarily because one of the studies was done by the Ron Swanson Foundation and the other was done by PETA's Commercial Pig Farmers Are Scum Foundation; it's all to do with the difficulty of discerning risk from a small amount of people. If you add up all the studies together, you're discerning risk from a larger amount of people, and so your estimates get more accurate. When the Swedish team added up all the studies (and thus looked at data from lots and lots of people in order to reduce the range of possibilities), it turned out that bacon had a 'relative risk' of 1.37, meaning that eating loads of bacon made you slightly more likely to get stomach cancer. This basically means, say, that if 100 in 10000 people who don't eat bacon get stomach cancer, then 137 in 10000 who do eat loads of bacon do get stomach cancer.
So what does this mean? Should you take this study seriously? Is eating loads of bacon bad for you? Seemingly. Jonathan Schoenfeld and John Ioannides (in a recent meta-analysis of meta-analyses, which is basically Inception-level medical stats) argued convincingly that you should basically only take seriously studies on various foods causing cancer if there's less than a 1 in 1000 chance that the data is wrong because of random luck. The first study I mentioned, on Hawaiian men, might sound convincing on the surface, but it didn't quite have this level of statistical solidity. But the Swedish meta-analysis from the last paragraph was solid like this. It checks out. There was less than a 1 in 1000 chance of their data being wrong because of random luck. People who eat loads of bacon are more likely to have cancer.
The question, however, that the Swedes don't really answer is why people who eat loads of bacon are more likely to have cancer. There's an association, sure. But why? Is it cos of carcinogens in the bacon? Or is it because of something else that goes along with the bacon? Another meta-analysis looking at the effect of eating loads of bacon on gliomas (a kind of cancer found in the brain/spine) found a similar increased risk of cancer for the Ron Swansons of the world. However, these researchers cautioned against worrying too much about it. They pointed out that it's reasonably likely that the bacon isn't what's causing the cancer. Instead, people who eat lots of bacon very often eat too much in general, and it's the eating too much in general that's actually the problem. All in all, more research needs to be done, as scientists truthfully but possibly boringly often say.
What the studies do also seem to show is that bacon is fine in moderation - if most of the bacon you have in a given week is a couple of rashers on a Sunday morning as part of a big breakfast, it probably won't hurt you too much. Instead, it's the This Is Why You're Fat levels of bacon that make you slightly more likely to get stomach cancer (or brain cancer). Of course, there might be other reasons why you'd want to avoid even small amounts of bacon - pigs are quite smart, and commercial pig farming can definitely be very cruel to them. Still, being Ron Swanson does probably give you an increased risk of cancer, whether Swanson eats too much bacon, or just eats too much. Mind you, Swanson would probably argue strenuously against the idea that the words "too much" are applicable to any category in which bacon fits.
You shouldn't trust those articles in the paper when it comes to making changes to your diet, unless there are very large numbers of people included in the study, and thus an extremely good chance that the data is right. The average half-arsed journalist isn't going to bother looking to see whether there's less than a 1 in 1000 chance of the study being wrong because of random chance (I mean, I had to work out the numbers for the Swedish meta-analysis using equations in Excel!). Instead of reading articles in the newspaper, you should be getting your dietary advice from places like the Cancer Council website if you actually want to reduce your risk of getting cancer. If you look at their page on nutrition, they ignore most of the wild and wacky things that the Daily Mail reckons causes cancer (which includes cakes, cayenne pepper and Worcestershire sauce, apparently) based on what is thus far flimsy evidence. Instead, they say that you should eat more vegies and wholegrain cereals, and that you shouldn't overdo the red meat and processed meat (i.e., bacon). This is responsible evidence-based advice.