To not understand why health research is so contradictory?(30 Posts)
I can't link but the main headline today was that a large scale study has concluded that the majority of cancer is caused by lifestyle factors like smoking and diet.
This is the complete opposite to the findings of a similar study only a few months ago which concluded that the vast majority of cancer is due to 'bad luck' and not to anything that the person did.
Advice about diet seems to change every 5 minutes. For years and years, we've been told that fat is bad but that seems to be changing and it's now sugar that we need to avoid. I had Tesco sushi for lunch yesterday and the labelling said that it constituted 30% of my daily sugar intake! How can fish, vegetables and rice be full of sugar?
Fruit juice and smoothies are no longer seen as healthy because they're full of sugar.
I know there are some research findings that seem generally well established eg I can't imagine a study concluding that smoking loads of fags is good for you but can any science/research people explain why advice keeps changing and how 2 studies about cancer carried out at around the same time can produce directly opposing conclusions? No wonder nobody knows what to believe.
Ok, well I'm no expert, but there could never be such a thing as one definitive study. They're all different, carried out on different groups of people and in differing ways. Some studies are weak because they only look at relatively small groups of people.
The best types of evidence are those which collate the results of lots studies, to look for any findings which they have in common. Only then can you start saying that X treatment is the best intervention for Y disease.
But when you're looking at different cancers and their varying causes, it's so incredibly complex, I don't think we'll ever know whether genetics or environmental factors play the biggest role. And there are so many different types of cancer anyway, I don't think it's possible to conclude anything definitive about so many different types of disease.
The problem, actually, is that research papers are really, REALLY badly reported in the press.
Lots of reasons
A lot of research isn't primary research but a rehash of other studies - i.e. they take other primary research and look for correlations. However, the primary purpose of the study being used may not have been the same, so you get a dilution of controls.
For example, research designed to look at impact of childhood immunisations may have collected data points (age of child, IQ, mortality data, feeding practice) which can be used by someone looking at impact of breastfeeding. However, the first study may have been controlled in a different way or the data set may lack data points that in an ideal world would be available (e.g. educational background of mother).
With long term studies, like cancer, it's also very difficult to control as it relies heavily on self-reporting, and people lie (some of the most interesting research is how much people lie to give "good" answers, even when surveys are anonymous).Also, there is often a predisposition and a lifestyle impact- not everyone who smokes dies of it, but there is a clear correlation.
Some studies are way too small to be statistically significant and the way in which the results are reported are often sensationalist- e.g. doing x doubles your risk of Y. Well, if your baseline risk of x is one in a million, there's no need to stress.
Also, there is a lot of "sponsored" research where researchers are under pressure to reach a certain conclusion or to cherry pick data. I'm no fan of woo, but I'll admit that Big Pharma don't exactly help themselves in this regard (e.g The Great Cholesterol Con etc).
There has been numerous pieces of research that suggests that some Cancers are largely lifestyle related, dating back over at least a decade.
It does depend on the Cancer.
""How can fish, vegetables and rice be full of sugar?""
They aren't, it's been added in to keep the costs down.
People won't accept that lifestyle matters and we know start filling our children with high levels of Sugar/Unhealthy fat and Chemicals, from Birth.
I had Tesco sushi for lunch yesterday and the labelling said that it constituted 30% of my daily sugar intake! How can fish, vegetables and rice be full of sugar?
Fish, vegetables and rice are not full of sugar. Sugar is added to make it taste better. If you made your own without adding sugar it most certainly wouldn't be 30% of your daily allowance. That's the problem with buying processed foods.
Mainly I'd say
1) Study design - very complex but what are you measuring and what else is at play (most of the cholesterol studies were confounded by people also eating high levels of sugar, which went unmeasured and ignored)
2) Reporting - bias, poor understanding of statistics, need for headlines, etc
Look back at the vaccination/ autism debacle. It displays all of these.
Also, people (and the press) not understanding the difference between correlation and "cause and effect".
e.g. if you plot autism diagnoses against Ribera consumption, there's a positive correlation. Does that mean that drinking Ribena causes autism? Probably not- they are 2 independent factors and the increase in both is caused by a third factor- increase in age.
Also the difference between what is actually in the research papers, and how the mainstream media reports it.
Which is all too often terribly badly. Because they are looking for a definitive, newsworthy breakthrough. Not a proper consideration of the patient accumulation of knowledge (which is incremental, and often quite dull).
Because if you look at, for example, evidence based dietary advice, it's been remarkably consistent for ages. Eat plenty of fruit and vegetables, avoid highly processed foods, eat a wide variety of foods, consume in amounts that keep you at a steady weight that is appropriate for your height. That's not headline stuff.
Population studies are notoriously difficult to get decent data from. If everyone in the study is a smoker then who gets long cancer would be mostly down to genetics. If most are not smokers then who gets lung cancer will be mostly down to smoking. Then there are confounding factors which need to be controlled for e.g. is red meat bad for your health or do meat eaters have worse health because they generally ignore healthy living advice?
Lots of reasons. Confounding variables. Poor study design. Studies that ask and answer different questions. Studies that are reported badly in the mainstream press (most journalists are humanities graduates who squint at you suspiciously if you use terms like "end point" and "statistical significance").
Whether or not you get a particular cancer is very largely down to a combination of genes and luck, but environmental factors can massively increase the risk of some types of cancer. It's not either/ or, and depending on the type of cancer, the environmental influence may be smaller or larger. For eg it's well known that environmental factors (smoking, industrial toxins) play a large role in many lung cancers, but environmental factors probably play only a very small or even a non-existent role in childhood leukaemia.
Problem is, the gold standard trials that really give you an answer about cause and effect, are randomised control trials. This would involve taking randomised groups of babies and giving each a different possible cancer trigger for their whole lives while simultaneously keeping everything else about their lives controlled so it was the same for each group. Can you imagine an ethics committee ever giving ethical approval to that? No. Which is why we have to make the best of the info obtained from less robust types of trials, which don't always answer the question adequately.
Try the systematic reviews in the Cochrane Library for rigorous research.
This is fascinating and goes to show science still doesn't have all the answers, it's a wonder it's relied on to the extent it is, when it is so contradictory and easily manipulated.
YY to Freda I'm an author for Cochrane. In one our latest reviews we looked at something like 60,000 references (other pieces of work). These got reduced to about 50 included studies (those studies where the treatment, problem and population matched the question we were trying to answer). So it gives an over view of the research. Problem is that often small, poorly designed studies give the most controversial results & these are the ones picked up by the press.
Really sorry to highjack, but could you explain what you mean by 'The Great Cholesterol Con' KeyserSophie?
Cholesterol is an issue with me at the moment and I'm puzzled with my GP's less than helpful suggestion that I cut out fatty foods and lose weight - my weight is under the ideal for my age and height and I don't eat fatty foods.
There was never anything wrong with butter and scientists that stated such at the time were laughed out of the building in favour of processed candles spreads.
I see the same thing happening with sugar and honey, people are turning to processed sweeteners.
For the most part it isn't actually the fault of the researchers (I admit there are some really crap studies out there - but a lot are excellent). Also research evidence tends to be cumulative ie group A could find out that a certain gene is found in some breast cancer patients (I'm making this up btw), another group (B) could then independently find out that the gene Group A is also looking at stops cells from dying when they are damaged. Group C could then come along and link group A's work to group B's work and find that this is the gene that is causing breast cancer. All of that takes years, lots of publications involving speculation by the researchers about the possible functions of the gene - which is allowed and actually a good thing to do.
What gets reported in the press is, for the most part, the speculation (remember deodorants causing breast cancer? Pure speculation that was written in a paper and then not really ever followed up or proved definitively).
Also, as others have said, journalists for the most part are not scientifically trained so don't understand the nuances of scientific speak and think that researchers are talking absolutes rather than possibles.
I believe very strongly that now we've got more open access articles, members of the public are intelligent enough to go and find the papers themselves (look on Google scholar) and make up their own minds.
Lies, damn lies and statistics. It is possible to design a study to prove what you want it to prove and in this way many small studies are biased particular by the group they use for the study.
Large, multi- centre doubled blinded trials give the least biased results, however, these are the most expensive type of trials to run often running into millions of pounds.
You need to read the full paper of the study to determine how valuable the evidence and conclusion in that study is.
Of course the media will publish the study with the result that suits today's purpose regardless of quality. It's just before Christmas we need to remind people to not smoke and drink and eat too much so we publish a study saying lifestyle choices are the main cause of cancer.
Vegetables and rice are full of carbohydrates, fruits are indeed full of sugar - and carbohydrates (well the ones we can digest, anyway) are broken down into simple sugars exactly the same as table sugar. Table sugar (sucrose) is a 2-molecule compound made up from glucose and fructose - fruit sugar is mainly fructose (as is honey), and starches and complex carbohydrates break down to glucose.
The more carbs you eat, then, the more "sugar" you are putting into your body, even though it doesn't start off as being actual sugar.
Research is also usually funded. Very little of it is done without funding, and therefore the funders often have an agenda for the outcome that they expect from the research. Not saying that they're exactly falsifying data, just that they do the stats, and design the protocols, to skew the results in the direction they would like. Not ALL research is done this way, but a fair amount is.
Re. the fats thing - I watched a documentary on the TV (and OH how I wish I'd video'd it!!) back in the 1980s that suggested there was a report done into the causes of heart disease in the early 1970s, and that initially 4 dietary factors were identified - high salt, saturated fat, cholesterol and sugar. But, according to this documentary, lobbying from the Tate family manage to get the sugar part of this report removed (big supporters of the Govt, lots of clout) which left us with salt, sat. fat and cholesterol.
The dietary cholesterol thing was the first to be debunked, when it was realised that we make 80%+ of our own cholesterol, and conserve it quite highly (i.e. reabsorb it) - dietary cholesterol has very little effect on overall cholesterol levels, it's an internal biochemistry issue that causes increases in cholesterol. Then came the subdivision of cholesterol into "good" and "bad" - neither being true, both are essential - and the ratio being more important, then add in the triglycerides (fatty acids) and it became clear that it was a much more complex picture than just "stop eating eggs and prawns".
It has now been realised that all this high-sugar, high-carb, low fat food we've been led towards has in fact increased the obesity rate; excess sugar/carb in your diet leading to it being laid down as fat once the body has reached its limit of carb storage, and leading to insulin resistance and then type II diabetes.
You could say that vested interests are entirely to blame; or you could say that it's new learning that comes out that changes our beliefs, or changes the "facts" as we know them; or a combination of the two.
I used to get information from the coffee council quite regularly, on how caffeine/coffee was shown to be good for you in various conditions - beneficial for diabetes, cancer, you name it. And then at the same time there would be conflicting research coming out explaining how caffeine exacerbated diabetes and could be damaging in other health conditions - leaving you with a face like and wondering whether the good outweighed the bad in a cup of coffee!
But I have to say that I tend to "follow the money" when I see conflicting research - look at who has done it, who has funded it, conflicts of interest etc.
Something else entertaining - several years ago, the New England Journal of Medicine (NEJM) decided that they wouldn't take any papers where money had exchanged hands between the primary researchers and the people who wrote up the paper (a common practice) - they apparently didn't get enough papers to publish! So they had to revise their restriction to "only taking papers where there was an upper limit on how much the authors had been paid" (not sure how much that was).
I've linked to this before, but it is very good (and very funny, especially as someone who has published clinical research) about the level of science 'journalism' in the media:
Oh and one other thing I noticed over the years - when doing paper research, it's really necessary to go back to the original paper. Sometimes people only read abstracts, and can draw the wrong conclusions from them, and then if they cite those abstracts in their own work, the wrong conclusion is perpetuated - so it's important to go back to the original to make sure you're getting the correct results from what they did.
Biology is squishy and difficult
<struggling biology researcher>
Dino that is brilliant!
My MIL doesn't believe in science because "the scientists are always changing their minds". She means those studies that are reported in tabloids every couple of weeks that say things like "wine is good for you" "wine will kill you" "chocolate will make you live longer" "chocolate shortens your life".
I've tried explaining to her that that's not actually what the studies say, it's how they're regurgitated for the great unwashed. But it falls on deaf ears!
smoothies never were good for you unless you had no teeth or couldn't digest food that wasn't liquidised; there are some people to whom that applies. Otherwise, mushing up a huge amount of fruit, far more than you'd eat by the normal chewing and tasting technique, is obviously going to give you a greatly increased calorie intake.
fish and rice aren't full of sugar. Heavily processed food is.
as for cancer and luck; simple probability theory. If you smoke like a fashion model you increase your chances of getting cancer, as stink-sticks are a known carcinogen. But it is only CHANCE; it doesn't mean you will get cancer, or perhaps you'll die of something else before you do. This is why we all know of an auntie gertie who smoked 80 a day and lived to 90.
You can get cancer even if you've never smoked or done any of the other known triggers, but you have a lower chance of it.
the Ribena and autism study is a very good example too.
Join the discussion
Please login first.