WHI and the fallout for women's health
- ruchitandon
- 4 days ago
- 8 min read
How flawed data and headlines mislead a generation of women on HRT
Writing about the Women’s Health Initiative (WHI) feels impossible. The study is vast, the data complex, and the fallout unparalleled in women’s health.
But here’s why it matters: for more than two decades, its misinterpretation has shaped the way doctors, policymakers and women themselves think about HRT.
When I first learned about the WHI, its data on cardiovascular disease and breast cancer risk was presented to me like any other research.
It wasn’t until I read Oestrogen Matters that I understood the full story: how results were delivered, how they were spun, and how a single study cast a shadow that stopped millions of women from accessing treatment.
It’s not just the data that counts; it’s how that data is presented and received. And in the case of the WHI, presentation became the story.
So, this first Substack explains context.
Why the WHI was designed, who it studied, and why you should question whether it even answered the question it set out to ask.
But most of all, we women, everyone, should be appalled that a group of researchers plied a hungry media with statistical ‘smoke’ to fuel a lie that would stop women from using HRT, and for no known reason.
The theory behind the Women’s Health Initiative
In the 1990s, HRT was common. Between 30–40% of postmenopausal women used it, not just for symptoms, but to prevent osteoporosis, heart disease and other age-related conditions (1). Many doctors prescribed it almost routinely.
But the FDA refused to approve HRT for heart disease prevention without harder evidence. Observational studies suggested oestrogen lowered LDL (“bad cholesterol”), raised HDL (“good cholesterol”), and reduced heart disease risk.
But observational data is biased: women who used HRT tended to be wealthier, healthier and better educated. That left one big question: was it the hormones or the lifestyle?
One of the key investigators of the WHI trial, Jacques Rossouw, wrote an article in Circulation in 1996, lamenting the widespread use of HRT and arguing that it was time to put “the brakes on that bandwagon” (2).
And that’s what he did when he and his colleagues designed the WHI, the largest randomised controlled trial on HRT.
Concerns about oestrogen use
There was good reason to be concerned about unrestricted, non-validated use of hormones.
Oestrogen use began in the 1960s under the promise of eternal youth, but it was quickly discovered that oestrogen was increasing the risk of endometrial cancer.
By the 1980s, it was recognised that adding progesterone counteracted this stimulation and reversed the risk of endometrial cancer (4).
Once again, oestrogen use began in earnest, this time with the concomitant use of progesterone, but without any long-term studies to support the safety of this new combined regime.
The aims of the WHI study and the population recruited
The WHI recruited 27,500 women, aged 50–79, and split them into two groups:
Women without a uterus: conjugated equine oestrogen (CEE) vs placebo
Women with a uterus: CEE plus medroxyprogesterone acetate (MPA) vs placebo
Yes, “equine” means horse. The main form of oestrogen at the time came from pregnant mare’s urine. It contained estrone (human) but also equilin and equilenin (non-human). Oral, not transdermal. Today, we mainly use plant-derived, body-identical oestrogen instead.
The progesterone wasn’t bio-identical either, but a synthetic, soluble derivative: MPA. Not widely used today. Modern HRT uses micronised (powdered) bio-identical progesterone instead.
In short, the WHI didn’t test the hormones we prescribe now.
The WHI groups
Here’s where things go awry. The WHI set out to test whether HRT prevented cardiovascular disease.
Let’s remind ourselves that most women start HRT at the age of menopause, at 51. Oestrogen is protective, particularly for the heart and vessels, and cardiovascular disease in women before this age is uncommon (5).
A data analyst at CVS in the US studying trends in cardiovascular drug use explained to me that “men take drugs for cardiovascular disease from very early on, women don’t start taking the same drugs until after 50, their use rapidly increases, and eventually they overtake the men”.
The group of women recruited into the WHI looked like they were already far along the slippery slope of post menopausal cardiovascular disease.
Let’s unpick this.
1. Age
Why allow women from such a large age range (50-79) to enter the study?
Initiation of HRT over a decade post menopause would have been unusual, yet these results were reported as applicable to all women and all ages.
Importantly, women who had already used HRT in their 50s were neither excluded nor randomised separately. That decision introduced bias and muddied the outcomes.
2. Risks
This is a group of high-risk women with comorbidities that would have predisposed them to multiple diseases
Obesity, hypertension, and smoking are themselves risk factors for cancer, cardiovascular disease, osteoporosis, venous thrombosis, and stroke.
In effect, they swapped the “wealthy, healthy, well-informed” women of earlier observational studies for a cohort that was older, less fit, and often already carrying chronic disease.
While obesity and comorbidities do reflect much of the midlife population in developed countries, it is inappropriate to generalise these findings to younger, healthier women initiating HRT before disease develops.
3. Exclusions
To preserve the trial’s “blind” design, women with moderate or severe menopausal symptoms (such as hot flushes) were excluded, perhaps explaining why so many participants were long past menopause.
Despite this, the WHI still measured HRT’s effect on symptom relief and, predictably, reported no benefit in a group that had few symptoms to begin with.
4. Randomisation
Randomisation was performed for the risk of cardiovascular disease only; this meant that each arm had similar numbers of patients with risk factors such as obesity, smoking history and hypertension, etc.
They did not randomise for the other outcomes they measured
That meant the results were skewed and should have been presented after the proper adjustments were made
This was most obvious in the case of previous HRT use
Women in the placebo group who had already taken HRT had unexpectedly ‘favourable’ results, making the treatment group’s appear dramatically worse
When they were removed from the analysis, the supposed risks disappeared (8)
Delivery to the media
Typically, medical research is tested for validity by a group of editors at a journal. In a trial of this size, all of the principal investigators, responsible for the different study arms, would have had to ‘sign off’ on the findings before submission to a journal.
And, typically, the ‘statistical significance’ of a finding needs to meet a medically recognised probability that it has occurred by ‘chance’ of less than 1 in 20. If it doesn’t, we can infer that the findings could have happened by luck, irrespective of the intervention applied or not.
Instead, the trial was halted early after 5.2 years in 2002, announced first by a press release. Ten days later, before full group consensus, a JAMA publication followed (9), citing an almost statistically significant increase in breast cancer risk.
Why?
Jacques Rossouw told Tara Parker-Pope (of the Wall Street Journal) that the WHI “was intentionally going for ‘high impact’ when it called the press conference”, because their goal “was to shake up the medical establishment and change the thinking about hormones” (10).
Reporting directly to a press conference was highly unusual. The risks were not only presented in a sensational and inflammatory way, but were later shown to have been factually exaggerated; recalculations in 2003 reduced the apparent risks (11).
Even more unusual, as reported in 2017 by epidemiologist Robert Langer (one of the principal investigators), “the investigators most capable of correcting the critical misinterpretations of the data were actively excluded from the writing and dissemination activities” (12).
At a meeting between all the investigators and the National Institute for Health (the US government agency funding the trial), the paper was presented, already written, for discussion. “Some of the investigators were aghast” at how conclusions had been drawn, ignoring their own published methods protocol.
The identified breast cancer risk that triggered the trial’s early stop came from the skewed, unadjusted pool of data; numbers that still carried bias from unequal randomisation. When the data were properly adjusted for confounders (like prior HRT use), the apparent increase in breast cancer risk disappeared. Had the adjusted figures been used, the trial would almost certainly not have been halted (13).
When concerns were raised about the propriety of producing a paper on behalf of the entire study group, the other investigators were allowed to amend calculations and alter the tone of the paper.
This was couriered to the journal, only to be told that the issue had already been printed and was waiting in warehouses ready to be shipped out.
The press release
The NIH didn’t just issue a press release; they dispatched emissaries to the press and TV to amplify the message: women should stop HRT immediately.
On NBC, Chief Medical Editor Dr Nancy Snyderman told viewers there was conclusive “proof that breast cancer rates go up,” and added, with finality, that “[hormones] don’t make your bones stronger, they don’t make your heart better, they don’t make you smarter or your skin better — but they do cause
disease.”
She was wrong on all counts, but the damage was done. Headlines around the world ran with the same inflammatory story, and HRT prescriptions plummeted.
The BBC, bless them, did reproduce the headlines but also sought out British gynaecologists, cancer specialists and menopause experts, all of whom urged restraint: “no British woman should stop taking HRT on the basis of these results” (14).
The media frenzy was so out of hand it forced researchers to publish the obvious: that doctors actually have a duty to tell the public the truth when they talk to the press (15).
The fallout
It is challenging to quantify the downstream effects of the trial, the inaccurate statistics and the media frenzy.
Some effects are measurable, such as the drop in global HRT prescriptions by 40-80% (16) and the dramatic jump in hip fractures in those who stopped (17).
Others will take decades to untangle, masked by lifestyle shifts, such as dementia and cardiovascular disease.
Then there are the unknown unknowns: the stagnation of hormone research and women’s health when funding evaporated after WHI.
And there are the cultural scars. In my own medical training, I saw how women’s menopausal symptoms were dismissed, their concerns trivialised, and they were casually branded as “moaners.”
But, I hear you ask, what about breast cancer?
After the headlines, after HRT prescriptions plummeted, there was a brief window where breast cancer incidence dipped, particularly in women over 50 who had been most likely to use hormones. But the reprieve was short-lived.
By the late 2000s, breast cancer rates were rising again, despite HRT use remaining far below pre-2002 levels.
The long-term reality is that breast cancer has continued to climb, driven by age, obesity, and lifestyle factors, while millions of women were left untreated for menopausal symptoms for no gain.
Finally, I leave you with the following:
In a masterclass of irony, the NIH’s 30-year review in 2024 came with yet another press release. This time, they stressed that HRT “should not be used to prevent chronic disease” (despite never actually testing that theory) while conceding it could be an option for “managing moderate to severe menopausal symptoms,” the very group they had systematically excluded from the trial (17).
Comments