BETHESDA, MD. — The biggest problem with the diet modification study of the Women's Health Initiative was the diet. Not enough women stuck with it, and it had an outdated design.
Despite these flaws, the trial's results, first reported in February, came close to proving that a reduced-fat diet—followed by thousands of postmenopausal women for an average of 8.1 years—could significantly reduce the incidence of invasive breast cancer.
“It's not an optimal diet; it's not feasible” for many women, commented Dr. JoAnn E. Manson at a conference on the Women's Health Initiative sponsored by the Department of Health and Human Services. The average level of fat reduction that most women achieved in their diet was not as substantial as had been planned, blunting the diet's effects.
Despite this shortcoming, the results showed “signals” that a low-fat diet produced some benefits, including a trend toward a reduced number of invasive breast cancers, a bigger reduction in breast cancers among women who had the highest level of fat in their diet at baseline, and a reduction in the incidence of colonic polyps or adenomas, said Dr. Manson, chief of the division of preventive medicine at Brigham and Women's Hospital, Boston, and a principal investigator for the WHI.
The study failed to show that the reduced-fat diet with increased consumption of fruits, vegetables, and grains could lower the incidence of cardiovascular disease.
“You need a 20%–30% decrease in low-density lipoprotein cholesterol to get an effect” on cardiovascular disease. “In this study, the effect was way too low,” commented Dr. Lewis H. Kuller, professor of epidemiology at the University of Pittsburgh and a principal investigator for the WHI. The diet was also “too low in polyunsaturated fats. You blunt the LDL effect by reducing polyunsaturated fats,” he said in an interview.
In this respect, the diet reflected what was known when it was designed in the early 1990s. Today, researchers have a better understanding of the benefit of polyunsaturated fat for cardiovascular disease and of the danger from saturated fat.
The diet modification study of the WHI was a companion to the hormone therapy study, as well as a third study that assessed the effect of calcium and vitamin D supplements.
All three of the studies enrolled women who were postmenopausal and aged 50–79 years at entry.
The diet modification study involved more than 48,000 women who, on their usual diet, got 32% or more of their calories from fat, based on a food frequency questionnaire, and who had no history of breast or colorectal cancer. More than 19,000 women were randomized to an intensive behavior-modification program designed to change their eating habits so that their fat intake would be no more than 20% of calories, their consumption of fruits and vegetables would be at least five servings a day, and their consumption of grains would be at least six servings daily. This diet did not involve a reduction in total calorie intake and had no weight loss goals. The more than 29,000 women randomized to the control arm were not asked to make any changes in their diet.
The diet intervention was modestly successful. One year into the study, average fat intake by women in the intervention group was 24% of daily calories. After 1 year of the study, 31% of women in the intervention group had met the goal of consuming 20% or less of their daily calories as fat. In the sixth year of the study, 14% of women in the intervention group were at this goal level. In addition, differences were modest between the intervention and control groups in the consumption of fruits, vegetables, and grains. One year into the study, women in the intervention group ate an average of 5.1 servings of fruits and vegetables daily and an average of 5.1 grain servings a day. The daily averages for women in the control group were 3.9 servings and 4.2 servings, respectively.
The study's primary outcome was the incidence of invasive breast cancer, which was reduced by a relative 9% in the intervention group, compared with the control arm, a nonsignificant difference with a P value of .07 (JAMA 2006;295:629–42). Two secondary outcomes of the study were the rates of invasive colorectal cancer and cardiovascular disease. These rates did not significantly differ between the two study groups.
Researchers who ran the trial noted that the average follow-up was 8.1 years instead of the planned 9 years, a reduction caused by slower than expected recruitment of women into the study. The shortened follow-up limited the study's power to show a statistically significant difference in breast cancer rates.