IJE Advance Access originally published online on July 17, 2006
International Journal of Epidemiology 2006 35(4):814-816; doi:10.1093/ije/dyl133
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Editorial |
The women's health initiative—curse or blessing?
1 Obstetrics and Gynecology Epidemiology Center, Department of Obstetrics, Gynecology and Reproductive Medicine, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA
2 Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
* Correspondence to: Obstetrics and Gynecology Epidemiology Center, Brigham and Women's Hospital, Harvard Medical School, 221 Longwood Avenue, Boston, MA 02115, USA. E-mail: kmichels{at}rics.bwh.harvard.edu
Some 15 years ago, 3 weeks into her tenure as the first female director of the National Institutes of Health, Bernadine Healy, MD, announced to Congress plans for a new and ‘holistic’ study: the Women's Health Initiative (WHI). Hopes were high for the ambitious project. The largest randomized controlled trial (RCT) in history, costing the government an estimated $700 million, was expected to provide definitive answers to three key questions in women's health: (i) whether hormone replacement therapy (HRT) reduces the risk of heart disease, (ii) whether a high fat consumption increases the incidence of breast cancer, and (iii) whether calcium and vitamin D supplements prevent fractures.
The main results for all three trial arms have now been released.1–4 The outcome of the mega-trial including more than 50 000 postmenopausal women randomized to one, two, or three of the factorial trials has left consumers confused and timid and doctors puzzled. What, if anything, can we make of these data?
When the first trial of the WHI was stopped short in June 2002, an unexpected 24% higher incidence of coronary heart disease (CHD) had been observed among women randomized to HRT including oestrogen plus progestin than among women randomized to placebo. At the time the trial was prematurely halted, participants randomized to oestrogen plus progestin were advised to discontinue their medication. This news awakened alarm among postmenopausal women around the globe, many of whom discontinued their HRT. Physicians found themselves in a difficult situation, unsure what to advise. A closer look at the WHI, however, may help to disentangle some of the mysteries.
The WHI was building on the premise that oestrogen may aid the heart drawing largely on the results from observational studies. After the WHI news broke, observational studies were criticized as biased and unreliable. However, many important differences between the observational studies and the WHI may help explain the apparent discrepancy in findings: while observational studies follow women who choose together with their physician to take HRT and initiate it usually because they experience menopausal symptoms as they transition into menopause, WHI participants were randomized to start HRT, mostly for the first time, at ages 50–79 years when they entered the trial and they had no or only modest menopausal symptoms.5 While the observational studies reflected clinical practice, the treatment regimen laid out in the WHI differed from clinical practice in randomizing women to first use of HRT years or decades into menopause. In fact, the observational studies and the WHI addressed different questions. The observational studies asked whether HRT use among women with menopausal symptoms initiated at the onset of menopause affected the risk of heart disease; the WHI explored whether HRT use among women largely free of menopausal symptoms and initiated at any time between the ages of 50 and 79 may have an effect on the incidence of CHD. Subsequent investigations into both types of studies have consistently revealed that the timing of HRT initiation may indeed be critical in its impact on the heart.6,7 Time since the onset of menopause at initiation of HRT and severity of menopausal symptoms may be important modifiers of the effect of HRT on CHD. Women a decade or more into their menopause may be unlikely to derive cardiovascular benefit from HRT and may even have an increased risk of a thromboembolic event precipitated by treatment.
In February 2006, the long-awaited results from the dietary component of the WHI were released.3,8,9 Unlike the HRT component of the trial,1,2 the diet part was not prematurely terminated but ended after 8.1 years as initially planned. The intervention, aiming to reduce fat intake from 
38% of calories to 20% of calories did not produce a significant change in the incidence of the primary endpoints, breast cancer and colorectal cancer, or in the secondary endpoint, cardiovascular disease (CVD).
How should these results be interpreted? Or can they be interpreted at all? As we suggested when the WHI was in its planning stages, randomizing diet is a problematic endeavour since healthy individuals are unlikely to maintain a drastically altered dietary regimen over many years.10,11 We were concerned that it would be impossible for individuals accustomed for many years to a diet with 38% of calories from fat to suddenly change to a diet with 20% of calories from fat—even if they agreed to do so at the time of randomization. Conversely, it seemed likely that women in the control group while not advised to change their diet would do so on their own merely by being part of the study and being fully informed of its interventions and goals. We expected that over time, no meaningful difference in fat consumption between intervention and control group would be maintained. Now the results of the WHI diet trial are indeed non-significant and are difficult to interpret. Compliance with the dietary regimen assigned was monitored with repeated food frequency questionnaires (FFQs), which seemed to indicate a 10.7% difference in fat intake as percentage of energy in year 1 of follow-up, which diminished thereafter. The FFQs are probably afflicted with considerable differential and non-differential misclassification: participants of both groups trying to be or appear compliant will report a diet closer to the one they were assigned to than the one they truly consume. Hence, the contrast between intervention and control group during follow-up is probably even smaller than reflected by the diet questionnaires. This concern is supported by the low total energy intake self-reported during follow-up, which fell from 1790 to 1500 calories in the intervention group and from 1789 to 1594 in the comparison group in the first year. While there was a slight weight loss in the intervention group in the first year, there was none in the comparison group, and only a difference of 0.8 kg between intervention and comparison group after 6 years of follow-up. Furthermore, differences in serum antioxidants, glucose, insulin, cholesterol, and triglycerides in intervention and control group at year 3 were minor, indicating a lack of substantial differences in dietary intake. Although most of these blood analytes are of limited value as biomarkers of dietary intake, the DASH (Dietary Approaches to Stop Hypertension) diet prescribing 27% of calories from fat to participants who consumed 38% of calories from fat at baseline lowered their total cholesterol levels by a significant 0.35 mmol/l [95% confidence interval (95% CI) –0.49 to –0.22 mmol/l] and low density lipoprotein levels by 0.28 mmol/l (95% CI –0.40 to –0.16 mmol/l).12 In the DASH trial all meals were provided to the participants, which may be the only successful approach to randomizing diet.
Moreover, changing a postmenopausal woman's diet at ages 50–79 years may not include the most favourable window of opportunity for diet to influence the development of breast cancer. Diet and dietary change in childhood, adolescence, and prior to menopause are more likely to affect the risk of mammary carcinoma.13 Even if WHI participants would be able to adhere to a low-fat diet, they would probably have been exposed to a high-fat diet during ages when their breast tissue was particularly susceptible to environmental influences. A change in diet after menopause lasting a few years as aimed for by the WHI investigators, is probably not sufficient to have any significant impact on breast cancer risk. The intervention period of on average 8.1 years was probably too short to impact on the cancer endpoints.
Another fundamental problem with this study is that fat and grain subtypes were not considered. Participants in the intervention group were advised to lower total fat intake and increase consumption of grains. It is apparent from prospective cohort studies that frequent consumption of vegetable oils rich in monounsaturated and polyunsaturated fats such as olive and canola oils and high consumption of whole grains significantly reduce the risk of CHD, while intake of trans and saturated fats, and refined carbohydrates promote heart disease.14 Participants of the WHI who increased their intake of refined carbohydrates and reduced their consumption of vegetable oils may have increased their risk of CHD.
Observational studies of diet are difficult to interpret if they use a case–control design, which is almost inevitably affected by recall and selection bias and because diet is difficult to assess in human populations. Randomizing diet in a healthy population is similarly problematic because of difficulties with compliance. In the diet trial of the WHI no significant difference in the incidence of cancer and CVD was found in the two groups. It remains unresolved whether this lack of difference should be interpreted to mean that a low-fat, high-carbohydrate diet does not affect cancer and CVD incidence, that participants did not follow their assigned diets sufficiently to achieve a meaningful contrast, that the age of participants at intervention was too old, or that the duration of intervention was too short. But the results should not be interpreted to mean that diet does not matter for health. The types of fat and carbohydrates can have a major impact on the risk of CHD and staying lean can substantially reduce the risk of post-menopausal breast cancer and of colorectal cancer. The hazard ratio for invasive breast cancer of 0.9 among the women in the intervention group compared with the women in the control group is compatible with the modest weight loss among the women randomized to low fat intake.
The final instalment of the WHI trials arrived in the form of another surprising null result, a lack of association between calcium and vitamin D supplementation and the risk of fractures.4 While an RCT of supplements should be straightforward, even this part of the WHI leaves room for error. A large proportion of women included in this trial were already well-nourished with calcium and vitamin D from diet and non-study supplements; 69% of WHI participants were taking non-study calcium supplements and ~30% used multivitamin supplements with vitamin D. Moreover, only 59% of participants were still taking 80% or more or the assigned study medication at the end of the study. Given these caveats, it is possible that the modestly dosed intervention of 1000 mg calcium carbonate plus 400 IU vitamin D did not achieve a sufficient contrast between intervention and control group to impact on fracture risk.
What do we take away from this mega-trial? Randomizing pills and supplements was compromised by lack of compliance and randomizing dietary recommendations seemed next to impossible. We have learned from the WHI that women well into their menopause who have no or modest menopausal symptoms do not benefit from starting HRT and long-term use of HRT for chronic disease prevention should generally not be recommended. Whether calcium and vitamin D supplements are beneficial may depend on the baseline intake of these nutrients and on the bone density of the individual. The WHI has not substantially advanced our current understanding of the role of diet in maintaining health and preventing disease, and does not affect dietary recommendations. For more answers we may have to rely on observational studies or attempt controlled feeding trials when feasible.
References
1 Writing Group for the Women's Health Initiative Investigators. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. JAMA 2002;288:321–33.
2 Women's Health Initiative Steering Committee. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. JAMA 2004;291:1701–12.
3 Prentice RL, Caan B, Chlebowski RT et al. Low-fat dietary pattern and risk of invasive breast cancer. JAMA 2006;295:629–42.
4 Jackson RD, LaCroix AZ, Gass M et al. Calcium plus vitamin D supplementation and the risk of fractures. N Engl J Med 2006;354:669–83.
5 Michels KB, Manson JE. Postmenopausal hormone therapy—a reversal of fortune. Circulation 2003;107:1830–33.
6 Grodstein F, Manson JE, Stampfer MJ. Hormone therapy and coronary heart disease: the role of time since menopause and age at hormone initiation. J Womens Health 2006;15:35–44.[CrossRef][ISI]
7 Hsia J, Langer RD, Manson JE et al. Conjugated equine estrogens and coronary heart disease: The Women's Health Initiative. Arch Intern Med 2006;166:357–65.
8 Beresford SAA, Johnson KC, Ritenbaugh C et al. Low-fat dietary pattern and risk of colorectal cancer. JAMA 2006;295:643–54.
9 Howard BV, Van Horn L, Hsia J et al. Low-fat dietary pattern and risk of cardiovascular disease. JAMA 2006;295:655–66.
10 Michels KB, Willett WC. The Women's Health Initiative: daughter of politics or science? In: DeVita VT, Hellman S, Rosenberg SA (eds). Cancer Prevention. Philadelphia: Lippincott, 1991.
11 Michels KB, Willett WC. The Women's Health Initiative: will it resolve the issues? Recent Results Cancer Res 1996;140:295–305.[Medline]
12 Obarzanek E, Sacks FM, Vollmer WM et al. Effects on blood lipids of a blood pressure-lowering diet: the Dietary Approaches to Stop Hypertension (DASH) Trial. Am J Clin Nutr 2001;74:80–89.
13 Michels KB, Willett WC. Breast cancer—early life matters. N Engl J Med 2004;351:1679–81.
14 Stampfer MJ, Hu FB, Manson JE, Rimm EB, Willett WC. Primary prevention of coronary heart disease in women through diet and lifestyle. N Engl J Med 2000;343:16–22.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  D. A Lawlor Quality in epidemiological research: should we be submitting papers before we have the results and submitting more hypothesis-generating research? Int. J. Epidemiol., October 1, 2007; 36(5): 940 - 943. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||