KEY POINTS:
It's been a torrid two weeks for knowing what to eat. At the beginning on the month the World Cancer Research Fund report linking cancer and obesity told us we should limit our intake of red meat and alcohol, and avoid processed meat, salt-preserved, and salty foods. A few days later dieticians told us how not all vegetables are good and that we should limit our intake of the starchy ones - potato, corn and sweet potato.
The next day, we were again warned about drinking alcohol - "a more dangerous drug than the party pill Fantasy". And, to top it off, some scientists told us eating certain icecreams - the ones containing beta-casomorphin 7 from A1 milk - may cause serious health problems.
It seems nothing is safe. The confusing aspect of these studies doesn't help either. The cancer study, for example, says: "the evidence on cancer justifies a recommendation not to drink alcoholic drinks". But then it adds "modest amounts of alcoholic drinks are likely to reduce the risk of coronary heart disease". In other words, drink a little to avoid getting a heart attack - but it might give you cancer. Then there is the stream of studies about the health effects of coffee or chocolate - one week it's good for you and the next week it's bad. Not to mention the just-published study by the Centre for Disease Control and Prevention in Atlanta, Georgia, which finds that carrying a little extra flab - though not too much - might help people to live longer.
Even worse are the studies that get it entirely wrong - the classic being the much prescribed hormone-replacement therapy which was promoted as a result of epidemiological studies but after more studies turned out to be a hazard to women's health rather than a benefit.
Small wonder many - in the face of conflicting messages and information overload - take health studies with a pinch of health-endangering salt.
"Epidemiologists have got something of a reputation for conflicting studies," says Anthony Rodgers, professor of epidemiology at Auckland University's Clinical Trials Research Unit. A big issue is newsworthiness. He says if a study finds the same as the last 10 studies it won't get any media attention, whereas a study that conflicts with previous ones is far more likely to hit the headlines.
"Also, results we want to believe get a lot of press. Chocolate-heart disease studies seem to be in the paper every few months."
Rodgers says that's why studies shouldn't be taken in isolation. What's needed to avoid chance results or selective reporting are reviews of all the evidence or "meta-analyses". So if you do meta-analyse multiple studies and review all the evidence, what does it say about coffee? "Coffee is completely safe." And chocolate? "There may be minor benefits, but they are really tiny. It's been played out of all proportion - people want the story to be true."
Red wine? "It looks like the protective thing is alcohol itself, not the antioxidants in red wine. People who drink a little bit of spirits have the same protection against heart disease." But again, says Rodgers, this is often taken out of context, for many people the adverse health effects of alcohol are much more important.
And how should people process the protective effects of alcohol in relation to heart disease against the World Cancer Research Fund study which finds any level of alcohol is convincing as a cause of some cancers? In such cases Rodgers says it's necessary to take all health risk factors into account and balance the risk - which to be fair, the Cancer Research Study did, changing its no alcohol finding to a "limit alcohol" recommendation.
Dr Chris Atkinson, director of Oncology Service at Christchurch Hospital, agrees: "If you had a glass of wine every day, I think it would be hard to prove dogmatically that had a caused a much greater risk of the various cancers linked in this study." But he points out also that oncologists have known for some time that too much alcohol is clearly associated with head and neck cancer and oesophagus cancer.
Atkinson also takes a moderate line on the report's other startling recommendation - that you should limit red meat intake, and avoid processed meats. How these foods are seen as a cause of colorectal cancer involves N-nitroso compounds - suspected mutagens and carcinogens - produced in the stomach and colon of people who eat large amounts of red meat, and also resulting from high salt intake. Many processed meats also contain high levels of salt and nitrite. Atkinson's view of the N-nitroso cancer causing mechanism is that it's a hypothesis rather than an absolute known association.
"I'll probably be shot for saying that. But we don't know that 1 plus 1 equals 2 in many of these things. I think they are just observations that people have looked at."
Which is an apt description of the studies themselves - observations and monitoring of disease rates and lifestyle factors in populations. That means looking at things like what large groups of people eat, how they exercise, what prescription drugs or supplements people take, or whether they are exposed to pollutants. The investigators then hypothesise about what caused the disease variations they've observed.
So it's not surprising the studies generate speculations - leapt on by the media - about the causes or prevention of chronic diseases. The potential benefits of not eating red or processed meats and the dangers of getting fat and sedentary living are always going to make good, anxiety producing copy. But it's worth noting that in the cancer study - a meta analysis of thousands of studies and including assessments of other papers examining biological evidence - none of the papers involve a randomised controlled clinical trial. The latter is an experiment, not an observational study, and is the surest way we have of establishing cause and effect. It's a point often forgotten in newspaper reports - that an association between two events is not the same as a cause and effect.
That's not to say what epidemiological studies observe is rubbish, it's just that it's not proven beyond doubt, like lowering cholesterol and preventing heart disease, where the evidence is overwhelming. The general public isn't helped in making such fine distinctions when reading reports such as the cancer study, which is full of causality statements, using a scale from "limited", through "probable" to "convincing" evidence.
"It's very hard to be completely certain about some of these things because definitive randomised trials would need to be absolutely enormous," says Rodgers. "It's pretty consistent evidence but you can't be certain."
How certain? "There is a spectrum. At one end, grey hair is associated with hip fracture but we are certain doesn't cause it, whereas smoking is associated with lung cancer and we are certain causes it. I'd say we are about 70-90 per cent along this spectrum with body fatness and types of cancer". He points out that many of the findings in the cancer report have been known about for some time.
"These studies are done looking at whole populations and saying the people who have lower BMIs (body mass index) for example, tend to get less bowel cancer. It looks like it's consistent across studies and doesn't look to be caused by smoking or anything else."
But Rodgers points out that for anyone who gets bowel cancer, it will be due to a number of different things and BMI is only one of them. Which is why epidemiologists talk about reducing your risk. "Making these changes would reduce your risk, but not remove it completely."
How much benefit might one get following the obesity recommendations of the cancer study? Rodgers says the health benefits would be huge, but most would come from avoiding heart disease, stroke and diabetes. Of the lives saved by losing weight, less than 10 per cent are from avoiding cancer. "If an obese individual lost weight to the ideal weight recommended in the study, he or she might have a five per cent lower chance of developing all types of cancer and perhaps a 15-30 per cent lower risk for specific cancers like bowel cancer."
What's worth keeping in mind is that aside from a few cases, such as smoking and lung cancer, and sexual activity to spread the papilloma virus and cervical cancer, we can't explain what causes most cases of cancer.
Rodgers describes the links in the report between things like body fatness and types of cancer as "modest" associations.
"What epidemiologists worry about when you've got those smaller associations is that they could, in fact, be due to something else." So while a study may show, for example, that people who are obese are more likely to get colorectal cancer, there is a possibility that something else such as not eating fruit and vegetables could be causing that association. "It's what we call confounding," says Rodgers. Which also, almost certainly confounds the general public.
But it helps to remember that epidemiological studies are focused on populations, not individuals. And when it comes to specific recommendations, the net gain in terms of life expectancy for an individual is usually minimal. Rodgers agrees that even though the link between cholesterol and heart disease is well established, diet has little effect. "Unfortunately, diet can't change cholesterol levels quickly. In countries like New Zealand where high saturated fat diets are the norm, your cholesterol levels build up steadily from when you start on solids. For a 40-year-old, an intensive one year diet still only covers a 40th of the time you've been accumulating cholesterol. Even if you work really hard at it, you might only get 10-15 per cent cholesterol reduction in that year."
The impotence of the individual is vividly demonstrated in the work of Dr Cliona Ni Mhurchu, director of the Nutrition & Physical Activity programme at the Clinical Trials Research Unit. Some of the unit's research involves the association between sodium (salt) content in the diet and population blood pressure levels.. It seems a simple enough equation. But as Ni Mhurchu points out the vast majority of sodium - at least 75 per cent - is non-discretionary and comes from salt added to foods during manufacturing and processing.
Why epidemiologists are interested in small effects is that the impact on public health can be huge if the findings effect on an entire country. If millions of women decrease their breast cancer risk by 20-30 per cent, thousands of such cancers will be prevented each year. The downside is that public-health logic tends to err on the side of prudence - even if it means persuading us all to engage in an activity such as keeping out of the sun, eating certain foods, or getting inoculated with something that may do nothing for us and sometimes could have unforeseen harmful consequences.
Atkinson acknowledges it's a debate which the Cancer Society has had to wrestle with. For years it has had a programme telling us to put on sun block and avoid sunburn because of the risk of melanoma. But there is a more recent body of literature which says that if you don't have any sunlight and don't have enough vitamin D, you're likely get other illnesses - in some cases, cancers.
Today's public health message is a little different. "The sensible advice is don't never have any exposure to the sun because that could be bad for you, but don't have over exposure because that is definitely bad for you."
Apply that sort of censor to all pieces of health information and, trite as it sounds, what you get, says Atkinson is: "Moderation in capital letters and neon lights". He advocates a common sense approach - "that a little bit is better than too much and is probably more sensible than none at all."
