Willett’s team falters when it jumps directly from hypothesis to dietary edict, Taubes says. “I would have no problem with Professor Willett making the statements he does if he attached caveats. If he said, ‘I think this is true, but we don’t know for sure.’ ”
Taubes points to a 2013 review of nearly 300 scientific papers. In more than half of them, the authors gave some kind of medical advice based on their observational studies. Yet only 14 percent of those papers called for randomized trials to validate the recommendations. That, Taubes argues, is a function of the blinders you don’t even realize you’re wearing when you’re so convinced that your study can give you the right answers.
The most dramatic example, he says, is hormone replacement therapy. The recommendation that post-menopausal women take HRT to guard against heart attacks, which was based on findings from the nurses’ study, was scrapped when trials concluded that the therapy actually increased the risk of heart disease, stroke, and breast cancer for some women.
To understand these two research models better, I get lunch with JoAnn Manson, Willett’s Harvard colleague and former student who is now the chief of preventive medicine at Brigham and Women’s Hospital. (In line at the expansive salad bar in the School of Public Health cafeteria, I ask her, “Is the food here good?” Yes, she says. “Walter makes sure of it.”)
Manson works both sides of the research fence, as an investigator on the nurses’ study and as someone also heavily involved in randomized trials. In between sips of her soup, she rattles off a half-dozen cases in which scientists reported promising findings based on observational studies (not necessarily Willett’s), but then the randomized trials found no benefit and even some elevated risks. Beta carotene was once thought to lower the risk of cancer and cardiovascular disease, but beta carotene supplements increased the risk of lung cancer in smokers in the trials. Vitamin E, a supplement many in the medical field were themselves taking 20 years ago, was linked to increased risk of hemorrhagic stroke in one trial and increased risk of prostate cancer in another.
Glancing down at my legal pad, where my notes of all her examples have formed sort of a wall of shame, I ask, “Are there any examples where the results of the randomized trials supported the initial hype?”
She pauses and then nods. “Yes, multivitamins.” The results of a trial published in 2012 showed a reduction in cancer. “But that was tested in men only,” she says. “We’re trying to do a follow-up study in women.” (She later adds several more apparent success stories to the list, including heart-healthy omega-3.)
Manson calls Willett “critically important — a pioneer in the field.” She says the fact that some of the trials she has been involved in have cast doubt on some of the findings from Willett’s studies doesn’t diminish the value of his contributions. “Our appreciation of the role of diet has really increased because of the research Walter has done,” she says.
For his part, Willet insists many of the findings from his studies have been supported in trials, such as the reduced risk of colorectal cancer with aspirin use and the lower risk of cardiovascular disease and Type 2 diabetes with — wait for it — nut consumption.
DURING THE MENUS OF CHANGE CONFERENCE in Harvard Square, Willett sits on a panel with Christopher Gardner, the director of nutrition studies at the Stanford Prevention Research Center and a specialist in randomized trials. While Willett is upbeat, extolling his wife’s nut loaf, Gardner confesses to suffering from a professional “midlife crisis.”
Afterward, I grab Gardner as he is heading for his vegan lunch. He explains that what we know about nutrition so far comes from big studies like Willett’s and small targeted trials like his. Willett watches thousands of people, year after year, to see who dies and who lives. But Willett can’t prove that it was, say, the whole grains in their diet that protected them — so Gardner runs randomized trials to isolate one element and try to determine cause and effect. Here’s what he finds instead: “One little thing at a time never makes a difference.”
A few years ago, Gardner ran a National Institutes of Health-funded study on garlic. Because he must test a specific hypothesis, lest he be accused of going on a fishing expedition, he asked whether one clove per day helped lower cholesterol in people with moderately elevated levels. Six months and $1.4 million later, he found no effect. “I couldn’t even answer: Is garlic good for you?” he says.Continued...