Hormone replacement therapy (HRT) for post-menopausal women was thought to decrease the chance of cardiovascular problems like heart attack and stroke. This ‘made sense’ because the risk of cardiovascular disease in women rose sharply after menopause, indicating that female hormones had a protective effect. Many large observational studies supported this belief, and HRT was widely prescribed in the 1980s and 1990s. Later evidence from large, placebo controlled, randomised trials failed to show any cardiovascular benefit. Again, observational evidence was shown to overestimate the effectiveness of a common medical treatment and again, practice became established before the definitive trials were done.
How it started
In the 1960s and 70s, there was considerable support for HRT based on theoretical advantages and some observations. Books were written about it and just the titles of some of the articles from that time give you a flavour of the message, like “The fate of the nontreated postmenopausal woman: a plea for the maintenance of adequate estrogen from puberty to the grave.”
There were many reasons why HRT was so commonly prescribed, including minimising the risk of fracture from osteoporosis and treating the uncomfortable symptoms of menopause, but it was also thought to protect women from heart disease. In fact, in a summary of the observational (non-randomised, non-placebo) studies from 1991 (here) it was estimated that HRT halved the risk of cardiovascular disease.
Because the evidence came purely from observational studies, consideration was given to the possibility of confounding (error) from several possible sources, but it was thought that the cardio-protective effect was too strong to be explained by any form of bias.
The placebo studies
Large-scale placebo controlled studies (mainly the Women’s Health Initiative, two large US trials) showed no decrease in heart attacks, but instead an increase in strokes, deep venous thrombosis and breast cancer with HRT. The release of this evidence in the early 2000’s led to a dramatic decrease in the use of HRT.
It has been suggested that the observational studies may have been confounded (biased) because healthier women tended to be prescribed HRT (so they made the HRT patients look better) and possibly because physicians were less likely to prescribe HRT to women at high risk of cardiovascular disease (again making the HRT group look better).
There is ongoing debate about the role of dosages, types of HRT, and how soon after menopause the treatment is started, in explaining the mismatch between the placebo studies and the observational studies, but the bottom line is that the beneficial effects of HRT were considerably overestimated by the observational studies.
This is also another example of the medicalisation of a natural phenomenon, this time menopause. In our (very human) search for cause-and-effect for every phenomenon, menopause had previously been attributed to sin and to neurosis (here). The addition of the scientific understanding of hormones and the biochemical features of menopause meant that doctors could take over and label menopause as a disease: a deficiency of hormones. Once doctors (with the help of the pharmaceutical industry) had a treatment (HRT, regardless of how simplistic, naïve, ineffective and harmful it was) they needed to medicalise the process in order to control it. The pharmaceutical companies contributed to this problem by ghostwriting articles that overplayed the benefits and underplayed the harms of HRT (here).
The bottom line
The role of HRT in ‘treating’ menopause continues to be debated, but the point that I would like to make is that our reliance on observational studies led to an overestimation of the benefit of HRT and an underestimation of the harms. And once again, treatment became widespread before the randomised placebo trials were done.
Good coverage of the history of this treatment and the reasons for the differences between the observational studies and the placebo controlled randomised trials is available here.