Two Fridays ago* I was interviewed on ABC radio about an article on the health effects of coffee that has been doing the rounds. The theme of the interview was how can ordinary people make sense of conflicting information about health research.
As this is also part of The Conversation’s mission, I thought that I would amplify some of the remarks I made on air here.
Health articles are often reports of a single study, although the articles (here and here) that prompted the interview started from a single study of whether that morning coffee was really picking you up, then tried to put this in a larger context of the health effects of coffee.
However, there is context and context, so to speak, we may read one article saying coffee is good for you, but in the back of your mind is something you read a while ago that said coffee was bad. You might be forgiven for thinking exasperatedly, “scientists keep changing their minds, one day coffee is bad the next day it is good!”
To make sense of these different reports you need to compare the studies. Are they looking at the same thing (coffee versus caffeine) are they looking at the same doses (high coffee intake versus moderate coffee intake) are they looking at the same thing (healthy people versus people with heart disease).
Obviously, the implications of modest consumption of coffee by healthy people and heavy coffee consumption by people with heart disease could be vastly different, and both statements (coffee is good for you coffee is bad for you) may be true in the appropriate context.
However, even in these days of Famous Internet Search Engines, finding that information may be a little tricky.
To illustrate this I’ll use just one paragraph from one of the articles on coffee.
And in a study published just a few weeks ago, Australian researchers suggested a chemical in coffee called chlorogenic acid may increase the risk of diabetes and even lead to the body storing excess fat.
After you have read this, your impression would be that consumption of coffee increases your risk of diabetes.
But in fact studies consistently show that coffee consumption actually reduces your risk of diabetes (see here, here and here for systematic reviews). In at least one study, the risk was reduced by around 35%, which is quite good. Now, before I go on to talk about the chlorogenic study in a little more depth, I’d like to talk about the kinds of evidence that are used in health research.
We use a range of observational and experimental studies when assessing health research. Epidemiological evidence, like the observation that coffee consumption is associated with a lower risk of diabetes, is important for drawing our attention to a possible cause and effect relationship but we cannot prove such a relationship by epidemiological evidence alone. Despite our best efforts to control as many variables as possible, some other factor may be responsible.
For example, the number of Nobel laureates in a country is correlated with chocolate consumption in that country, but that could simply be due to richer countries being able to support education and consumption of luxury items. The fact that there is a stronger association between the number of Nobel laureates and the number of IKEA stores in a country suggests the wealth explanation is more likely than some brain enhancing property of chocolate.
To investigate these epidemiological associations, we need other supporting evidence, from test tube studies (although no one uses test tubes these days, but “96 well plate studies” doesn’t have the same ring to it) to animal studies to intervention studies in humans.
Coffee is a complex brew of chemicals, it is chock full of antioxidants, many of these being the chemical class called polyphenols. Chlorogenic acid is one of them. In “test tube” studies it’s a good antioxidant, and produces a variety of biochemical responses in tissue culture that might be protective in an intact organism. When we give it to animals that develop diabetes it reduces their risk and severity of diabetes.
So how to make sense of the chlorogenic acid study that was reported in the newspaper article. What was missing from the article was the background that coffee is actually associated with protection from type II diabetes. In the actual research paper, this point is made clearly early on. The purpose of the paper was to find out which component of coffee was responsible for the protective effect. This is also omitted from the newspaper report.
The researchers used a high-fat diet in mice to induce pre-diabetic changes, like resistance to the hormone insulin. In this model, chlorogenic acid, despite all the other data suggesting it reduced the risk of diabetes, failed to alter the development of insulin resistance. Indeed it made it slightly worse. Again, in the research paper the authors carefully discussed their findings in the light of previous research showing protective effects.
Their conclusions, which you would not know from reading the newspaper article, was that either cholorogenic acid was not responsible for the protective effect of coffee but one or more of the other polyphenols was, or that there was a synergistic effect between cholorogenic acid and the other polyphenols, or that chlorogenic acid is protective at lower doses (they did use quite a lot of chlorogenic acid, equivalent to 5-6 cups of coffee, administered all in one go, which is not how humans consume it).
None of their conclusions could be construed as “coffee is bad for you”. Which is the implication from the way the study was reported.
Now, I’ve just spent over 800 words trying to explain one paragraph from a newspaper article. I haven’t even touched on the main areas of the article (but see here and my previous article on health and coffee). In terms of the stated goal of this article, to help people navigate reporting on health information, many readers will be a bit depressed by now. After all, I have a lot of background in this area, and have easy access to original research papers. How is the ordinary person supposed to navigate this mine field?
Look at the article again. Is it comparing like to like, or is the context unclear? As I emphasised at the beginning, context (who is at risk, under what circumstances) is very important. Use a Famous Internet Search engine to check for existing fact sheets from the NH&MRC, Heart Foundation and Cancer Foundations. These are far more helpful than many sites an unfocused search will bring up. Is there evidence from other studies? Look up the abstract (not the press release) associated with the study, it may be in technicalese, but you should be able to get a feel for whether the article reporting the study is going off the rails.
This may seem like a lot of work, but how much is your coffee worth to you?
*A bad bout of the flu and exam setting and marking has kept me away from this column, did you miss me?