Healthy Skepticism

Fat or Fiction?

(Julia Ewan - Twp)
By Lisa M. Schwartz, Steven Woloshin and H. Gilbert Welch
Special to The Washington Post
Tuesday, March 14, 2006

The public is bombarded with messages about diet and cancer prevention. Unfortunately, the advice is pretty inconsistent. One day a diet prevents cancer, the next day it doesn't. In the early '90s, beta carotene (a vitamin A precursor present in fruits and vegetables) was said to prevent lung cancer. But several years later, headlines read, "Beta carotene pills yield no benefit" (The Post, 1996). And while people have been told for years to eat a high-fiber diet to reduce the risk of colon cancer, recently we were told "High-fiber diets are not anti-cancer miracle" (Montreal Gazette, 2005).

The latest apparent flip-flop is about low-fat diet and breast cancer. Last month, a front-page headline in The Post read: "Low-Fat Diet's Benefit Rejected: Study Finds No Drop in Risk for Disease." But less than a year ago, a prominent headline sent a different message: "Study of Breast Cancer Patients Finds Benefit in Low-Fat Diet" (New York Times, May 17, 2005). We believe that understanding why these studies seem to have come to opposite conclusions will help shed light on why medical research on diet seems to change so often.

Despite longstanding interest in the subject, there have been surprisingly few randomized trials -- gold-standard scientific experiments -- testing whether a low-fat diet really decreases cancer risk. That is why these two recent studies -- both large, randomized trials -- received wide media attention.

The newer study, part of the federally funded Women's Health Initiative (WHI), involved about 50,000 women in an eight-year experiment to see if fat reduction in diet might reduce cancer risk. Findings were published last month in the Journal of the American Medical Association.

The earlier trial, called the Women's Intervention Nutrition Study (WINS), involved about 2,500 women in a similar dietary experiment. Results were presented in May 2005 at the American Society of Clinical Oncology.

The two studies (summarized in "Two Different Findings on Low-Fat Diet and Breast Cancer," at right) were alike in important ways. The goal of each was to see whether the low-fat diet reduced the occurrence of new breast cancers. In each, post-menopausal women were randomly assigned either to a low-fat diet group that received intensive dietary counseling or to a comparison group whose members were given standard advice about a "healthy diet" but were not asked to change what they ate.

How the Studies Differed

The studies differed in two major ways. First, they studied different groups of women: one with prior breast cancer (WINS) and one without (WHI). Women with breast cancer are substantially more likely to develop a new breast cancer than women without breast cancer. In general, people at high risk stand to gain more from interventions than people at low risk. In other words, if dietary fat were going to have an effect, it would be more obvious in women with breast cancer.

And this is exactly what happened. In both studies, the low-fat diet group developed fewer new breast cancers. But the effect of diet was larger in the women who already had the disease: For every 1,000 women in the study with breast cancer, 25 fewer new breast cancers were detected. In comparison, for every 1,000 women without breast cancer, two fewer breast cancers were detected.

The second major difference between the studies was in how statisticians judged the findings. This difference was profound. Based on conventional statistical practices, the effect of the low-fat diet in women with breast cancer was deemed to be real -- the result of more than chance. But in women without breast cancer, the effect of diet was judged to be statistical noise.

Reality vs. Chance

The judgments about what is real and what is statistical noise are based on what researchers call p values. [See "Research Basics: Accounting for Chance," below.]

Based on the size of the study groups and the number of cancers in each, the p value communicates how often you would expect to see an effect this big simply as a result of chance. By convention, scientists say p values below 5 percent are "statistically significant"-- meaning not likely attributable to chance. And p values of 5 percent and higher are considered statistical noise (that is, likely due to chance).

The p values for the effect of low-fat diet on breast cancer in the two studies were quite similar. For women with breast cancer, the p value was 3 percent. For women without breast cancer, the p value was 7 percent.

So even though, by convention, one finding is called "statistically significant" and the other "not-significant," we would say that the statistics of the two studies are not that different: Both are close to the conventional cutoff point of 5 percent. Since the p values are actually quite close, we would argue that the role of chance was about the same. That is, if you believe one is real, you should probably believe the other is real.

What Should Women Do?

To decide whether to act on the results of a study, you need to understand what it will take to make the change -- here, how much dietary sacrifice you must make to reduce risk of breast cancer -- and whether you think the sacrifice is worth the benefit.

In these studies, women assigned to the test groups had to limit themselves to 20 to 40 grams of fat a day (roughly the equivalent of a cup of Ben and Jerry's ice cream). To achieve this goal, women in both studies had to eliminate things like butter on bread, cream cheese on bagels, oil in salad dressing, regular ice cream, most cakes, cheeses and so on. In the WHI study, they also were encouraged to eat at least five servings a day of fruit and vegetables and at least six servings of grains.

For women with breast cancer, the diet change may well be worth the effort: Their risk of another breast cancer is high and the diet may lower it enough to be worth the sacrifice.

For women without breast cancer, whose breast cancer risk is low to begin with, the small possible benefit of the diet may not be worth the sacrifice. (Study participants' poor compliance with diet suggests they did indeed view it as a sacrifice. See "Why It's Hard to Study the Role of Diet in Cancer," at right.) The point is that individuals need to weigh the pros and cons themselves.

Tough Challenge

While it is easy to get frustrated with the apparent flip-flops of medical research, it is important to emphasize just how hard it is to study the role of diet in cancer (or, for that matter, in any disease).

Most of the evidence used to support the idea that diet affects cancer risk comes from observational studies -- studies where scientists simply observe what happens to people who eat different diets. Unfortunately, observational research can be misleading because people who eat differently also differ in countless other ways that affect their cancer risk.

People who choose to follow different diets may well differ in their cancer risk because of their family history (genetic makeup) or habits (for example, smoking). In general, observational studies tend to overstate the effect of diet because a "healthy diet" is often a marker for a healthy lifestyle. People who watch what they eat probably exercise more, avoid smoking and follow other behaviors linked to good health. Any of those behaviors might also help explain why these people have lower rates of cancer.

In fact, most of the diet flip-flops come from observational studies that suggest a benefit (or harm) from diet that is not subsequently confirmed in randomized trials. Since people in randomized trials are assigned a diet essentially by the flip of a coin, we can be confident that any difference found is the result of the diet. Even in randomized trials, however, diet is hard to measure and hard to change, and it is hard to figure out what part of a diet change makes the difference (See "Why It's Hard to Study the Role of Diet in Cancer.")

Many have criticized the WHI study (the study of women without breast cancer), saying the low-fat diet didn't lower fat intake enough, focused on total fat (rather than changing from animal to plant fat), or wasn't a Mediterranean diet--based on grains, olive oil, vegetables, fruits and fish. While intriguing, these criticisms are themselves based on observational research -- no more credible than the low-fat findings that stimulated this study to begin with.

Bottom Line

The best data to date suggesting the potential for diet -- or any lifestyle alterations -- to affect cancer risk is limited. The single notable exception is smoking: There is no doubt that not smoking dramatically lowers cancer risk.

The effect of diet on cancer is likely to be small for most people because diet is so heterogeneous and the effect of any given food may depend on its interaction with other foods. And the smaller the effect, the harder it is to demonstrate statistically. So it is really not surprising that results of research about diet and cancer flip-flop. Low-fat diets probably do lower the risk of breast cancer -- but the effect on risk is small -- particularly for women with no prior history of the disease. Changing diet to reduce breast cancer -- or any other cancer -- is a personal decision, not an imperative. ยท

Lisa M. Schwartz, Steven Woloshin and H. Gilbert Welch are physician researchers in the VA Outcomes Group in White River Junction, Vt., and faculty members at the Dartmouth Medical School. They conduct regular seminars on how to interpret medical studies. (See

The views expressed do not necessarily represent the views of the Department of Veterans Affairs or the United States Government. E-mail:

© 2006 The Washington Post Company