TCS Daily

It's the Fitness, Stupid

By Sandy Szwarc - September 16, 2004 12:00 AM

Consumers were left more confused than ever when the media reported on two obesity-related studies from the Journal of the American Medical Association last week. One seemed to find it was more important to be fit than thin for your heart health; the other that it was more important to be thin than fit to prevent diabetes. As Eric Berger of the Houston Chronicle wrote, the two studies disagreed and settled "nothing" in the debate over what matters most to a person's health: his overall fitness or his weight.

But in fact, the controversy has already been repeatedly answered in the scientific literature. The trouble is, it's not what a lot of people want to hear...and others without science backgrounds don't realize.

These side-by-side JAMA studies provided an invaluable opportunity for the media to help consumers sort through medical information and come away with a very important message: not all studies are created equal.

Most anything can be and is called a "study" nowadays and many aren't actual clinical studies or even examining real people. That surprises many consumers. Medical news is all too often taken at face value. But before believing headlines or making any changes or decisions about one's health, it's important to know that some studies are sounder and have more validity than others. So those "should have more influence," said Dr. Peter Jones, an associate professor of medicine at Baylor. A study also needs to be considered in light of the entire body of scientific evidence.

When we look at these two JAMA studies, their differences quickly become apparent.

When a Study Isn't a Study

The first study led by Amy Weinstein, MD claimed to find obesity was more important in predicting type 2 diabetes than physical activity. This was a type of study often called a "data dredge" because it mines a large database of characteristics from a population, looking for correlations. This study used numbers from the Women's Health Study -- a database of questionnaires collected from nearly 40,000 women since 1992, based at Brigham and Women's Hospital in Boston. I call these databanks "Rorschach tests" of epidemiology, because researchers can pull out numbers in almost unlimited combinations to find all sorts of correlations and conclude just about anything they set out to find. Just like the Rorschach test, seeing patterns where none exists, finding connections that are there but not as strongly as believed, and seeing what one expects to see, are common.

The biggest and longest-running Rorschach test is the Nurses Health Study -- a huge quarry of questionnaires (essentially unverified anecdotes) gathered since 1976 from over 120,000 nurses, headed by JoAnne Manson, MD, DrPH of Brigham and Women's Hospital in Boston. Over 500 "studies" have been released all using the Nurses Health Study numbers and, interestingly, they frequently disagree. You may remember they reported just a week ago that sugary drinks were associated with obesity and type 2 diabetes. That contradicted their own earlier 2002 "study" which found sugar and glycemic load was unrelated to diabetes or weight.

No wonder consumers are confused -- hearing "science" has found one thing one day, only to find something entirely different the next. My advice is to ignore these data dredge studies. Wait until a real study comes out.

These types of epidemiological studies can only manipulate numbers from groups of people to find possible relationships. They were never meant to apply to individuals, nor can they ever show a cause for anything because they don't actually test anything. Their purpose is to identify possible hypotheses for later clinical tests, which can then try to identify if and explain why something happens. But correlations themselves can never prove causation. For example, ice cream sales correlate with drownings, as the winner of the Purdue University 1998 Spurious Correlations Contest noted, but it is not sound to conclude that ice cream causes drownings!

Yet even to claim a correlation in these types of studies, it must also be statistically significant enough to be over what one would find by chance -- more than the toss of the dice -- or due to error, missing relevant factors, or statistical bias. So scientists do not take as credible anything which doesn't appear two to three times over baseline, or a 100- to 200-percent difference. The National Cancer Institute found anything less than 100 percent wasn't tenable. But Ernst Wynder, MD, founder and director of the American Health Foundation and editor of Preventative Medicine prior to his death, said anything less than three is suspect. And Marcia Angell, MD, former editor-in-chief of the New England Journal of Medicine said they looked for three or more before accepting a study for publication. Yet, we frequently see epidemiological findings of 5-, 30- or 50-percent reported as if that means anything.

Type 2 diabetes appears to be correlated to obesity but it and the "diabetes epidemic" are just as spurious as ice cream. Rising diabetes rates can also be partially explained by an aging population, increased awareness and testing for the condition, and because of changes in the definition itself.

Just like "overweight," diabetes has been redefined to include increasing numbers of the population. It used to be defined as fasting blood sugars >140, in 1998 that was changed to >126, and last November they deemed 100 abnormal, creating a new malady "prediabetes." That latest change alone added another 40 percent of the adult population to the roster in need of treatment, according to Dr. Frank Vinicor, diabetes chief for the Centers for Disease Control and Prevention. Vinicor told CNN on April 29, lowering the cut-off was "based on emerging science from the last two to three years." But others question if prediabetes even exists. According to Paul Zimmet, MD, PhD, world-renowned diabetes researcher and founding director of the International Diabetes Institute, it may be "inappropriate to use the term 'prediabetes' when there is only a 50% chance of developing diabetes in the next ten years."

Sadly data dredge studies are increasingly being misused and misinterpreted. In the Weinstein study, you'll see several classic strategies identified by Dr. John Brignell, PhD in The Epidemiologists: Have they got scares for you! (Brignell, 2004). For example:

· The trojan horse number to increase the impressiveness of the study. They claimed the study was of nearly 38,000 women, yet the researchers actually only clinically confirmed type 2 diabetes in 89 women of the (1361) self-reported cases.

· The authors selected the characteristics they wanted to look at, but disregarded ones they didn't, and didn't even consider factors that might have been important, such as race and ethnicity. The groups with different body mass indexes were similar in many lifestyle factors, except the slender group drank three times more alcohol than the fat group. But you sure didn't hear them conclude that drinking alcohol, especially more than one drink a day, prevents diabetes!

· Another notable drawback to this study is that it relied on self-reported information of recreational physical activity. That's a more nebulous way to arrive at determinations of metabolic fitness than studying fitness on a treadmill, for example.

Most noteworthy is that this study's findings contradict many stronger clinical and epidemiological studies that have found that exercise reduces type 2 diabetes and improves insulin resistance, unrelated to weight. For example, researchers at the Cooper Institute in Dallas, Texas led by Timothy S. Church, MD, PhD, followed over 2,000 diabetics for 25 years, using a range of health assessments, including treadmill tests to gauge their fitness levels. They found that premature deaths from all causes were significantly lower among the fit. Weight was irrelevant. Researchers at the Veterans Affairs, Palo Alto Health Care System, Stanford University studied over 6,000 men for six years and found exercise capacity was more important in risks of dying than "known" risk factors including obesity, cholesterol, hypertension, smoking and even diabetes. Even a small clinical study at Queen's University, Kingston, Ontario, Canada following 54 obese women found daily exercise, without dieting or weight loss, substantially reduced insulin resistance in just 14 weeks.

Studies on Real People

Which brings us to that second study led by Timothy R. Wessel, MD. It concluded that being fit significantly reduces heart attacks, strokes and heart problems among women. Weight was again irrelevant. It was a cohort study -- a type of study where researchers gather clinical information on a group of people and then follow them, observing their health outcomes. The soundness of the data they used is notably superior to the Weinstein study. For four years, the researchers followed more than 900 women who had been examined at four major medical centers using multiple clinical diagnostic tests of heart disease and correlated it with treadmill capacity assessments.

Most significant, this study is just one of dozens of clinical studies over decades which have found the exact same thing in men and women: when fitness is considered, weight is irrelevant to long-term health, heart disease, diabetes or premature death from all causes.

The list is too extensive to cite here, but clinical studies concluding 'fitness not weight is what counts' include the Harvard Alumni Health Study of 12,516 men followed for 16 years; the St. James Women Take Heart Project of 5,721 women studied for 8 years; and the Aerobics Center Longitudinal Study, an ongoing study that includes 25,389 patients examined at the Cooper Clinic in Dallas from 1970 to 1989. Even the Women's Health Study published findings in 2001 that found merely light to moderate activity was dramatically associated with lowered heart disease in women, including those who were overweight, had high cholesterol or smoked.

Ending the Debate

What the public didn't read last week was a poignant editorial in this same issue of JAMA by Dr. Steven Blair, PED and Dr. Church, of the Cooper Institute for Aerobic Research in Dallas, Texas, where much of the country's landmark research on fitness and preventative health has been done. They chastised today's obesity researchers, saying that "failure to adequately quantify physical activity when examining the risks of obesity is similar to exploring risk factors for cancer and misclassifying tobacco use."

Drs. Blair and Church emphasized that death rates and heart disease among obese people, with just moderate fitness, are half that of "normal" weight people who aren't fit. The amount of exercise to attain this health-giving level of moderate fitness isn't much, either, and has been proven in 24 clinical studies: it's merely 150 minutes of moderate-intensity activity a week. They say that's equivalent to 30 minutes, 5 times a week of: walking, gardening, housework, bicycling, swimming or other activities enjoyed in daily life.

Now that's news that people can use.


TCS Daily Archives