I’m starting to wonder if the editors of medical journals schedule a yearly Meat Causes Cancer! issue …something like their own equivalent of the Sports Illustrated Swimsuit issue.
Our cover model this year is pancreatic cancer, folks — and as you can see, she’s a hot little topic! We don’t want to start any rumors, but we have it on good authority she’s often seen in the company of some beefy hunks.
The hot little topic made a splash in the media last week, with headlines and opening paragraphs like these:
Eating two rashers of bacon or one sausage a day can increase the risk of a deadly form of cancer by almost a fifth, according to a new study. New research by the Karolinska Institute in Stockholm has found that eating 50g of processed meat a day can increase the risk of pancreatic cancer by 19%.
Some possible bad news for all the bacon lovers out there.
A new review in the British Journal of Cancer suggests a link between processed meats — like bacon and sausages — and an increased pancreatic cancer risk. In particular, eating an extra 50 grams a day of processed meat — or about a sausage — is enough to raise pancreatic cancer risk by 19 percent, BBC News reported, while an extra 100 grams of processed meat a day could raise the cancer risk by 38 percent.
“The authors of this study have suggested that one of the reasons could be that some of the chemicals that are used to preserve processed meat are turned in our bodies into some really harmful chemicals which can affect our DNA and increase the chance of cancer,” Jessica Harris, health information manager at Cancer Research UK, told Sky News.
Holy jumpin’ jiminy! A 19% increase in risk – that’s almost a fifth! Better drop that bacon right now, Mister. You don’t want to mutate your DNA and roll the dice with a 19% increase in the odds you’ll die of pancreatic cancer.
I tracked down the full study, and it was pretty much what I expected: a meta-analysis of several other studies, all of them based on food-recall surveys. So let’s put on our Science For Smart People hats (mine is cone-shaped; you can choose your own) and ask some critical-thinking questions:
Q: Was this an observational study or a clinical study?
A: It was a meta-analysis of 11 observational studies, the kind where the researchers pool the data and crunch the numbers.
Q: Did the researchers control the variables?
A: No, because they couldn’t. They were dealing with data published by other researchers who may or may not have done a good job controlling their variables. As the authors of the current study noted:
Our study has some limitations. First, as a meta-analysis of observational studies, we cannot rule out that individual studies may have failed to control for potential confounders, which may introduce bias in an unpredictable direction. All studies controlled for age and smoking, but only a few studies adjusted for other potential confounders such as body mass index and history of diabetes. Another limitation is that our findings were likely to be affected by imprecise measurement of red and processed meat consumption and potential confounders.
Let me put that into plain English: Our findings are meaningless. The studies we analyzed were based on food-recall surveys that are notoriously inaccurate, and most of them didn’t control for body mass index or diabetes, which essentially means they didn’t control for intake of sugars and refined carbohydrates.
Okay, folks, move along; nothing here to see.
What, you’re still here? Then we may as well continue.
Q: If A is linked to B, is it possible that they’re both caused by C?
A: Yes, of course it’s possible. As the researchers noted above, “All studies controlled for age and smoking, but only a few studies adjusted for other potential confounders.” Since processed meats are often served with a big wallop of refined carbohydrates – pizza, burritos, deli sandwiches, etc. – it’s entirely possible that people who consume more processed meats have higher rates of pancreatic cancer (if that’s even the case) because they also consume more white flour.
Q: If A is linked to B, do we see that connection consistently, or are there glaring exceptions?
A: We can answer that question by looking at the charts from the full study. This one shows the change in the relative risk of developing pancreatic cancer from consuming an additional 120 grams of red meat per day:
A relative risk of 1.0 is neutral – no change in risk. Below 1.0 means lower relative risk and above 1.0 means higher relative risk. The horizontal bars represent the range of values that fell within the “confidence interval,” the black squares represent the average relative risk for each study, and the white diamond in the last row represents the overall average obtained by pooling data from all the studies.
The first thing that jumped out at me is that in four of the 13 studies analyzed, the relative risk of developing pancreatic cancer was lower for the people who (supposedly) eat a lot of red meat. I wouldn’t call that a consistent result. If some studies show higher risk and some studies show lower risk, I’d conclude that we’re looking at the wrong variables.
But through the magic of statistical analysis, the researchers pooled the results (from studies that often failed to control the variables) and declared that consuming 120 grams of red meat per day raises your risk of developing pancreatic cancer by 13%.
Now here’s the change in relative risk from consuming an additional 50 grams of processed meat per day:
Nine studies, and in three of them the relative risk of developing pancreatic cancer was lower for people who consumed more processed meat. Once again, that’s hardly a consistent result, but the researchers pooled the data in order to declare that processed meat raises your risk of pancreatic cancer by 19% — which leads to our final question.
Q: What was the actual difference?
A: Almost nothing. That’s the short answer. Now for the longer answer:
Scientists like to cite relative risk instead of absolute risk because relative risk sounds far more impressive. Suppose that when I lived in sunny California, my odds of being struck by lightning were 1 in a million. But now that I live in Tennessee, suppose the odds are 1.5 in a million. That’s a 50% increase in relative risk … but a meaningless increase in absolute risk. The actual difference — the change in absolute risk — is 0.5 in a million.
According to the National Cancer Institute, the age-adjusted annual incidence rate of pancreatic cancer is 13.6 per 100,00 men and 10.3 per 100,00 women. We’ll split the difference and call it 12.15 per 100,000 people. Expressed as a percentage, here are the odds that you’ll be diagnosed with pancreatic cancer this year:
Just barely over one-hundredth of one percent. Now … let’s set aside the fact that this meta-analysis was 1) based on observational studies that 2) used unreliable food-recall surveys and 3) produced inconsistent results. Suppose we choose to believe that processed meat really and truly causes pancreatic cancer at the increased rate found by pooling all that data, but we keep on eating our bacon anyway. Here are the odds, expressed as a percentage, that we bacon-eaters will be diagnosed with pancreatic cancer this year:
And here’s the actual difference between those two numbers:
Well, maybe you’d prefer to deal with lifetime odds instead of annual odds. Okay, fine. According the National Cancer Institute, the lifetime odds of developing pancreatic cancer in the U.S. are 1.45%. If eating 50 grams per day of bacon or other processed meat really and truly (and all by itself) raised the rate by 19%, your lifetime odds would be 1.70%.
Here’s the actual difference between those two numbers:
Enjoy your bacon.