HRT Dangers Should Have Been Spotted Sooner
British researchers say many small studies pointed to problems before U.S. trial did
FRIDAY, Feb. 27, 2004 (HealthDayNews) -- In a sort of scientific slap on the hand, British researchers say experts should have realized the dangers of hormone replacement therapy for women much sooner than they did.
In fact, scientific research in general is fraught with problems, say the authors of a paper in the Feb. 28 issue of the British Medical Journal, including a lack of clear information on long-term side effects.
The saga of hormone replacement therapy (HRT) is just a case in point, they note.
Based primarily on observational studies alone, experts and consumers alike came to believe HRT not only eased the symptoms of menopause, but also lowered the risk of cardiovascular disease.
In 2002, however, the largest randomized trial on the subject, the U.S. government-sponsored Women's Health Initiative (WHI) was abruptly halted when it was found that the opposite was true: HRT increased the risk not only of cardiovascular disease but also of breast cancer and stroke.
Part of it is the sheer complexity represented by hormones, one expert notes.
"They affect so many different systems in the body that it's not as clear as it might be with another kind of drug that only has one purpose," says Dr. Margery L.S. Gass, a professor of clinical obstetrics and gynecology at the University of Cincinnati College of Medicine and a WHI investigator.
But part of the discrepancy can be explained by the differences between observational trials, which leave room for various types of bias, and randomized trials.
"It's an important difference," says Klim McPherson, co-author of the British paper and a visiting professor of public health epidemiology in the Nuffield Department of Obstetrics and Gynaecology at Churchill Hospital in Oxford, England. "If what you do is simply observe people who choose to take something as opposed to people who don't choose and see what happens, that is quite likely to be contaminated by biases. Namely, the people who take something are different from those who don't take it."
Randomized trials, on the other hand, start with a group of people randomly assigned to receive either the intervention or a placebo, thus eliminating this particular type of bias.
In the 1990s, McPherson and his colleagues examined existing trials on HRT. "We found 200 trials of efficacy, [but] most were useless because they didn't record any kind of side effects and many of them didn't have proper controls," McPherson says. "There were only 23 that were useful."
Their analysis of those 23 smaller, randomized trials foreshadowed what was to come. The conclusions, published in 1997, however, were not well-received.
They then located six more trials and had to obtain a court order to see them. "Six trials were unavailable because they weren't published and the manufacturers refused access to these data. We had to go to court to get access and, once again, it was confirmatory," McPherson says.
A synthesis of all this earlier data might have spared hundreds of thousands of women the risks involved with exposure to HRT. It might also have made the WHI trial unnecessary, the paper's authors state.
"What happens is that individual applications are assessed on their own merit," McPherson says. "Unless you pool these results, you don't get any answers."
Other experts don't agree, however. "It's very easy in retrospect, but doing prospective, randomized trials on important health issues is the absolute critical thing that we as a society need to do," says Dr. Jay Brooks, chief of hematology/oncology at the Ochsner Clinic Foundation in New Orleans. "One of the problems in American medicine today is that we try to rush things to judgment based on small, selective studies."
The study authors make a number of recommendations based on their experience, namely that regulators require that all studies on efficacy be made available and that adverse effects, even long-term and rare ones, should be recorded and made part of the public research record.
Also, McPherson adds, "studies ought to be big enough to detect rare events and most of these weren't big enough."
But even if a synthesis of smaller trials leads researchers to certain conclusions, a large randomized trial is still warranted, Brooks says. "There had been some previous studies that had hinted at [cardiovascular risk with HRT], but they hadn't really been well-designed ones that the WHI was. Everyone shot holes in the previous studies," he says. "We need to do the proper trials up front before we embark on these major scientific recommendations."
And even then, many mysteries often remain. "We have a lot of unanswered questions," says Gass. "What happens in science all the time is something we think is going to give us a simple answer often raises more questions."