Wednesday, May 16, 2007

FDA gives epidemiologists a black eye

I have grouched at some length about what epidemiologists customarily do -- take a large body of data and go fishing for correlations within it. Any correlations so found could well be random so are grounds for hypothesis testing only. They don't give us grounds to do anything else -- such as change our diets.

The proper scientific way to test something is to start out with a specific hypothesis and test it by an intervention. That is, take two groups that start out the same and give only one group some treatment or other that your hypothesis dictates. If the two groups end up differing in some predicted way, you might really be onto something.

It looks like an FDA consultant insisted on the proper method in the evaluation of a new prostate drug called Provenge:

"Hopes for Provenge were high because in late March a panel of FDA advisers voted 13-4 that there was substantial evidence that it was effective, potentially paving the way to approval. Shares tripled to more than $18 immediately, because the FDA usually follows the advice of such panels.

But from the start there were reasons Dendreon might be a special case. The question posed to the panel on efficacy had been changed, and some panelists seemed less than completely convinced of efficacy. Dendreon's clinical trials were small, containing only 225 patients, and had failed to meet their primary goals. It was only after taking a second look that Dendreon was able to argue that the trials seemed to show Provenge extended patients' lives.

Three top cancer researchers wrote to the FDA saying that Provenge should not be approved; all three letters were obtained by the Cancer Letter, an industry newsletter. Two of them had served on the panel and voiced their concerns then.

But the third letter, published last week, was the most compelling. It came from Thomas Fleming, a University of Washington statistician who is one of the world's experts on understanding clinical trials. Five years ago, he warned against approving AstraZeneca's lung cancer pill Iressa because clinical data were not convincing. He was right--Iressa failed in a big clinical trial, and its use was severely restricted.

Fleming wrote that he was invited to participate in the March 29 panel that voted to approve Provenge, but declined because he had done some consulting for Dendreon. He wrote that he was kept awake the night the panel voted. The survival data might look compelling, he wrote, but because the trial had not been designed to prove Provenge extended lives, there was a big risk that the results were a fluke.

So "fishing" for favourable results was ruled as telling us nothing. Bravo!

More on the "Mediterranean diet" religion

The journal article concerned is not yet online so I cannot critique it in detail but that tiny differences are being treated as meaning something goes without saying. Only a quarter of one percent of those studied got lung disease so how you hope to conclude anything from diffences among that group rather escapes me. I doubt that the differences were even statistically significant. And this is just more of the usual epidemiological data fishing.

The Mediterranean diet has been a fetish among epidemiologists ever since low levels of heart disease were observed in Greece and Italy. To acknowledge that Greeks and Italians might be GENETICALLY different from Northern Europeans would be politically incorrect, of course. Diet had to be the cause of why Greeks and Italians did better. Greeks and Italians don't live especially long however. For instance, Australians have a diet that is as far away from the Mediterranean in composition as Australia is geographically, yet they live longer than Greeks and Italians. And the long-lived Japanese have a quite different diet too. But we must not let facts upset a medical consensus, must we?

A traditional Mediterranean diet rich in fruit, vegetables, whole grains and fish can cut the risk of developing serious lung disease by half, researchers say. Chronic obstructive pulmonary disease (COPD) is an umbrella term for conditions such as emphysema and bronchitis. It leads to damaged air-ways, making it harder for people to breathe.

Smoking is the most important risk factor for getting COPD, which experts predict will become the third leading cause of death worldwide by 2020. Not all smokers develop the disease, which has led researchers to examine how genetic and environmental factors, such as diet, play a role. A study released before it is published today in the journal Thorax, found that eating a Mediterranean diet cut the risk of developing COPD in half. Meanwhile, people who ate a Western-style diet, with a high intake of refined grains, cured and red meats, desserts and French fries, were more than four times as likely to develop COPD.

The research, led by Dr Raphaelle Varraso, from the Harvard School of Public Health in Boston, followed the health of 42,917 men taking part in the US Health Professionals Follow-Up Study. All were asked about their medical history, diet and life-style, including how much they smoked and exercised. Between 1986 and 1998, 111 self-reported cases of newly diagnosed COPD were identified among the men. Even after adjusting for age, smoking and other risks, those men who ate a Mediterranean diet were found to have a far reduced risk of developing COPD. The more closely the diet was followed, the lower the risk of developing COPD over the 12-year study period.

The researchers concluded: “In men, a diet rich in fruits, vegetables and fish may reduce the risk of COPD, whereas a diet rich in refined grains, cured and red meats, desserts and French fries may increase the risk of COPD.”

It is not the first time that the Mediterranean diet has been hailed for helping to prevent medical conditions. Last month researchers from Britain’s National Heart and Lung Institute, the University of Crete, Venezelio General Hospital in Crete and the Centre for Research in Environmental Epidemiology in Barce-lona found that a Mediterranean diet helped to prevent children developing asthma and respiratory allergies. And last year US researchers found that eating a Mediterranean diet could reduce the risk of developing Alzheimer’s disease. Other researchers have shown that people live longer if they follow the diet.



Just some problems with the "Obesity" war:

1). It tries to impose behavior change on everybody -- when most of those targeted are not obese and hence have no reason to change their behaviour. It is a form of punishing the innocent and the guilty alike. (It is also typical of Leftist thinking: Scorning the individual and capable of dealing with large groups only).

2). The longevity research all leads to the conclusion that it is people of MIDDLING weight who live longest -- not slim people. So the "epidemic" of obesity is in fact largely an "epidemic" of living longer.

3). It is total calorie intake that makes you fat -- not where you get your calories. Policies that attack only the source of the calories (e.g. "junk food") without addressing total calorie intake are hence pissing into the wind. People involuntarily deprived of their preferred calorie intake from one source are highly likely to seek and find their calories elsewhere.

4). So-called junk food is perfectly nutritious. A big Mac meal comprises meat, bread, salad and potatoes -- which is a mainstream Western diet. If that is bad then we are all in big trouble.

5). Food warriors demonize salt and fat. But we need a daily salt intake to counter salt-loss through perspiration and the research shows that people on salt-restricted diets die SOONER. And Eskimos eat huge amounts of fat with no apparent ill-effects. And the average home-cooked roast dinner has LOTS of fat. Will we ban roast dinners?

6). The foods restricted are often no more calorific than those permitted -- such as milk and fruit-juice drinks.

7). Tendency to weight is mostly genetic and is therefore not readily susceptible to voluntary behaviour change.

8). And when are we going to ban cheese? Cheese is a concentrated calorie bomb and has lots of that wicked animal fat in it too. Wouldn't we all be better off without it? And what about butter and margarine? They are just about pure fat. Surely they should be treated as contraband in kids' lunchboxes! [/sarcasm].

Trans fats:

For one summary of the weak science behind the "trans-fat" hysteria, see here. Trans fats have only a temporary effect on blood chemistry and the evidence of lasting harm from them is dubious. By taking extreme groups in trans fats intake, some weak association with coronary heart disease has at times been shown in some sub-populations but extreme group studies are inherently at risk of confounding with other factors and are intrinsically of little interest to the average person.


No comments: