Media Bias, Conflicts of Interest Distort Study Findings on SupplementsJune 2006
By Lyle MacWilliam, MSc, FP
Calcium-Vitamin D Study Flawed
The $18 million double-blind, placebo-controlled Women’s Health Initiative study was designed to test whether postmenopausal women who were given calcium and vitamin D would have a lower risk of hip fracture.9 In this study, 1000 mg of calcium in the form of calcium carbonate and 400 IU of vitamin D (generally the same dosages recommended by most doctors to their elderly patients) were provided each day to the intervention group, which was followed for eight years.
While the women receiving calcium and vitamin D showed a greater preservation of hip-bone density, over all there was a non-significant 12% lower risk of fracture. What most media reports failed to disclose, however, is that by the study’s end, compliance with the prescribed daily intake was only 59%. In other words, fully 41% of the study participants stopped taking the prescribed daily dosage of calcium and vitamin D. Moreover, 24% were no longer taking any of the supplements—a level of non-compliance that dramatically decreased the difference between the two groups.
This unexpectedly low compliance rate and a projected hip fracture rate that was more than twice the rate observed reduced the power of the study to a paltry 48%. Consequently, the trial, as designed, had insufficient power to detect anything but the largest of differences in fracture risk.
Despite this and other shortcomings, the data show that those women who mostly followed their prescribed regimem had a statistically significant 29% reduction in fractures, and women over the age of 60 experienced a statistically significant 21% reduction in the risk of fracture—a drop that the New York Times called a “hint” of gain.
Fatal Study Flaw: Excluding Magnesium and Other Minerals
In designing the Women’s Health Initiative study, the authors overlooked the fact that reducing fracture risk depends on factors other than calcium. Studies show that magnesium is equally important in treating and preventing osteoporosis, and its deficiency plays a central role in the development of the disease.10
Magnesium supplementation is believed to suppress excess bone turnover, which may help prevent age-related osteoporosis.11 It also contributes to the fortification of the bone mineral matrix and is critical for the proper function of vitamin D. It is well established that magnesium intake should be about half that of calcium intake.12
In particular, postmenopausal women and those with osteoporosis generally have low bone magnesium content and exhibit other indicators of magnesium deficiency not seen in non-osteoporotic women.13,14 Studies show that osteoporosis-related magnesium deficiency is associated with low blood levels of the most active form of vitamin D (1,25-dihydroxy-vitamin D), which in turn inhibits calcium uptake in the gut and its resorption in the bone.14
Consequently, postmenopausal women who increase their calcium intake without also increasing their magnesium intake—as was the case in the Women’s Health Initiative study—can reduce absorption of magnesium because calcium competes for absorption with magnesium.15 Within this context, the failure of the researchers to include magnesium supplementation, along with calcium and vitamin D, strongly biased the findings in favor of harm—a glaring and irresponsible oversight for an $18 million study.
Furthermore, emerging research suggests that additional minerals such as boron, zinc, and silicon may also be critical for maintaining healthy bones. (See “Osteoporosis: How Calcium Combines with Other Nutrients to Combat Bone Loss,” Life Extension, January 2005.) The failure of the Women’s Health Initiative study to examine the impact of these important mineral nutrients may have further compromised the study’s findings.
Other Crippling Design Failures
Several other factors conspire to erode the impact of the Women’s Health Initiative calcium trial. For one, the study’s prescribed dosage of 400 IU per day of vitamin D had already been shown to have a negligible effect on the risk of hip fracture.17,18 In fact, most of the studies supporting a benefit provide vitamin D at a dose of 600 IU per day or higher.19-23
In addition, more than half the women in both the intervention and comparison groups were already taking estrogen hormone therapy, known to increase bone mineral density, confounding interpretation of the study intervention. Moreover, all the study participants, including those in the comparison group, were allowed to continue their personal use of calcium and vitamin D. Therefore, it is quite conceivable that some of the women (those who normally take a calcium-vitamin D supplement) in the control group were actually taking more calcium and vitamin D than many of those women in the intervention group who did not take the prescribed daily amount. No wonder the results were confusing!
Despite the study’s glaring oversights and contrary to the media spin that the Women’s Health Initiative study dispels long-held beliefs about the benefits of calcium and vitamin D, the authors conclude that the results do provide evidence of a positive effect of calcium and vitamin D on the bone health of older, postmenopausal women.
One can only image what might have been the result had this $18 million boondoggle been properly designed in the first place.
Dietary Modification Trial Results Mixed
On February 8, 2006, the Journal of the American Medical Association (JAMA) published three studies based on data from the eight-year, $415 million Women’s Health Initiative Dietary Modification Trial, one of the largest long-term trials ever conducted, involving 48,835 postmenopausal women.
This dietary modification trial was designed to test whether behavioral intervention intended to produce a dietary pattern low in total fat—along with increased intake of vegetables, fruits, and grains—would decrease the risks of cardiovascular disease, breast cancer, and colorectal cancer in postmenopausal women. Women aged 50-59 were randomly assigned to a dietary intervention or comparison group, in an attempt to reduce total dietary fat intake in the intervention group to 20% of daily calorie intake. The primary outcomes were fatal and non-fatal cardiac events or stroke,26 invasive breast cancer,27 and invasive colorectal cancer.28 Each of these three primary outcomes was reported in a separate study.
The difference in fat intake between the intervention and comparison groups was expected to be 20%. However, the intervention group achieved only 70% of this design goal. This led to a substantial loss in the statistical power (ranging from 40% to 60%) of the studies to detect a reduction in each of the three outcomes—which is about as good as flipping a coin. In other words, despite its unusually large sample size, the Women’s Health Initiative study was dramatically underpowered and not particularly capable of detecting a difference, if one existed.
Studies like these require highly powered trials with large sample sizes and clear discrimination between the intervention and comparison groups in order to detect relatively rare events. (It is not every day, after all, that a person dies of a heart attack or cancer.) Consequently, in the Women’s Health Initiative trial, only a very large effect of decreased fat intake would have rendered the changes statistically significant. Put another way, the lack of evidence of a protective benefit does not mean there was none. Although these limitations were clearly reported in all three studies, the implications were either not understood or disregarded by the news media.
Finally, several of the participants in this three-part trial also participated in two other randomized studies. It is unclear whether the treatment regimens in these two other studies—the Women’s Health Initiative hormone replacement trial and Women’s Health Initiative calcium trial—confounded the effects of dietary intervention.
Fat Intake and Cardiovascular Disease Risk
The Women’s Health Initiative study investigating the effect of total dietary fat reduction on the risk of cardiovascular disease26 found that long-term reduction of total dietary fat did not affect the risk of coronary heart disease, stroke, or cardiovascular disease. It did, however, achieve a modest, yet significant, reduction in cardiovascular disease risk factors, including low-density lipoprotein (LDL) and diastolic blood pressure. There was also a trend toward reduction in cardiovascular disease risk among women who had the lowest intake of fat and the highest intake of fresh fruits and vegetables, as well as for women with no previous cardiovascular disease. Moreover, women in the intervention group who had the lowest fat intake had a lower risk of coronary heart disease than those in the control group.
The power of this trial, however, was crippled by its failure to lower the level of fat intake to that prescribed by the design criteria, and by an observed incidence of myocardial infarct (heart attack) and cardiovascular disease that was 30% lower than projected. Hobbled with a scant 40% chance of detecting a decrease in the cardiovascular disease rate, the chances of a positive finding were less than the flip of a coin. Consequently, it is not surprising that the trial failed to achieve a statistically valid reduction in cardiovascular disease risk. Nevertheless, what it does show is just how hard it is to achieve a dramatic reduction in total fat intake over the long haul through behavioral intervention.
The study was designed to reduce total fat intake without regard to the types of fat, a feature that has been harshly criticized by nutritional researchers and informed consumers alike. In a letter to the editor of the New York Times, one reader recalls her mother, who was involved in the Women’s Health Initiative studies, bemoaning the fact that the researchers “made no difference between lard and olive oil!” Given what we knew even back in 1991 (when the trial was designed) about the differences between “good” fats and “bad” fats, it appears an inexcusable oversight not to differentiate between healthy and unhealthy fats. However, within this context, the findings further our understanding that a nondiscretionary reduction of total fat is of limited value in reducing cardiovascular disease risk.
Far from being the “definitive answer” on the health effects of reduced dietary fat, as suggested by some “experts” and described by some reporters, the study only helps confirm what we already know—that simply eliminating all fats is not the answer to reducing cardiovascular risk.
Low Fat and Risk of Breast Cancer
Like the previous study, the Women’s Health Initiative investigation of fat reduction’s effect on invasive breast cancer27 did not find a significant decrease in cancer rates. After eight years of follow-up, the dietary intervention group had a relative decrease of 9% in the incidence of invasive breast cancer compared to the control group—a level of risk reduction that approached, but did not achieve, statistical significance. Cancer can take years, even decades, to develop. Considering the study’s relatively short time frame, it is very likely that, given more time for the trial to proceed, it would have revealed evidence of a preventive benefit.
As in the companion studies, the intervention group’s inability to reach the targeted level of fat reduction fatally compromised the trial’s power. The study design was simply not robust enough to detect anything but the most dramatic of changes between the intervention and comparison groups.
The investigators take care to point out certain trends in their findings. While it can be misleading to read too deeply into subgroup analyses, women who had the highest levels of fat intake at the start of the trial showed a stronger trend toward breast cancer reduction than did the intervention group as a whole. Such variation would not have been expected if the dietary intervention had no effect on breast cancer. The researchers also found that the low-fat diet was associated with a 15% reduction in circulating levels of estradiol, the form of estrogen that increases the risk of breast cancer. This finding is consistent with the results of other clinical trials demonstrating the protective effect of estradiol reduction in breast cancer treatment.29 Similarly, this would not have been expected if dietary intervention had no effect on cancer risk reduction.
All this leaves us with a study that lacked the diagnostic power to do the job it was designed to do, primary results that nudge the boundaries of statistical significance, and secondary findings that exhibit supportive trends for reduction in breast cancer risk. While it is disappointing that the results were not more definitive, it is hardly cause to “throw out the salad” and “give up health advice,” as one Canadian journalist suggested.