The pendulum set in motion during the summer of 2002 continues its return swing toward a balanced perspective on hormone therapy.1,2 Now we have the opportunity to help our menopausal patients make decisions in a less emotional environment. Nonetheless, explaining the findings of the initial Women’s Health Initiative report—added to the findings of many subsequent WHI reports—is tricky even for the statistics experts.
How can we confidently counsel our patients?
American women fear breast cancer more than any other disease.3 That is understandable. Recent large studies, including WHI and the British Million Women Study, have been remarkably consistent in finding that combination hormone therapy (HT) modestly elevates the risk of being diagnosed with invasive breast cancer (relative risks or hazard ratios >1.0 and <2.0).2,4,5
Our challenge is to understand what the relative risk (RR) and hazard ratios mean as we guide our patients in making decisions about HT. To place these risks in a meaningful context, let’s compare them with risk factors for other cancers, other risk factors for breast cancer, and translate these relative risks into absolute risks, which patients understand much better.
We might start with lung cancer as a comparison. While the relative risk for breast cancer is less than 2 for menopausal women using combination HT, the relative risk of lung cancer is 15 to 30 for cigarette smokers.6
Other breast cancer risk factors of magnitudes similar to that of combination HT (RR <2.0) include menarche prior to age 12, high socioeconomic status, nulli-parity, never having nursed an infant, first full-term pregnancy after age 30, and alcohol consumption.7
Relative risk vs attributable risk
If our patients understand how the relative risks used in clinical trials translate into absolute or attributable risks, they will be better prepared to make sound choices regarding HT. Too often, however, relative risks are confused with attributable risk, which in this context is the incidence of an outcome (breast cancer) in women exposed to HT, minus the incidence in those not exposed.
The WHI trial of combination HT found an RR of 1.26 for breast cancer, meaning that HT users were 26% more likely to be diagnosed with this disease than were participants randomized to the placebo arm. Applying this RR to the absolute incidence of breast cancer observed in participants, the study’s authors noted that the attributable risk associated with use of combination HT was “low”: 8 additional breast tumors were diagnosed annually per 10,000 women (0.08% or ~0.1%) in the combination HT arm compared with the placebo arm.2 The annual breast cancer incidence in the HT arm (38 of 10,000 participants) is indeed some 26% higher than in the placebo arm (30 of 10,000 participants).
Keep in mind that WHI participants’ mean age at screening was 63 years and mean duration of HT use was 5.2 years. Because the incidence of breast cancer rises with age, and risk at baseline relates to attributable risk, the attributable risk associated with use of HT by younger menopausal women (those most likely to be seeking treatment for bothersome vasomotor symptoms) would be substantially lower than the 0.08% additional risk noted by the WHI investigators.
- The findings of the Women’s Health Initiative are misunderstood by most primary care specialists, although ObGyns have a better understanding of the risks and benefits compared to other specialties.8
- We hypothesize that physicians who overestimated the increase or decrease in risk were making the error of confusing relative risk with absolute risk difference. There is a great need for physician education about the attributable risks and benefits of HRT.8
A survey of physicians underscored the difficulty of trying to translate relative risks into attributable risks, and thereby helps us understand how readily our patients may overestimate their own risk. In Williams and colleagues’ survey8 of Florida physicians, conducted in 2004, prior to publication of the results of the WHI estrogen-only trial, all respondents correctly indicated that HT was associated with an elevated risk of breast cancer.
When asked to characterize the attributable risk of breast cancer associated with HT (the choices were 0.1%, 3%, 10%, and 30%), fewer than half of physicians answered correctly that the attributable risk is 0.1%. More than half of physicians picked one of the wrong choices—all of which were higher than the correct attributable risk of breast cancer associated with HT.
Breast cancer risk: Estrogen-only vs combined HT