Johns Hopkins Magazine -- February 2001
Johns Hopkins 
     Magazine Home


Curt Meinert's new study debunks a widely held belief regarding bias in medical research. Will anyone listen?
P U B L I C    P O L I C Y    &    I N T E R N A T I O N A L

Subject to Dispute
By Sue De Pasquale
Illustration by Kim Barnes

Have American women traditionally been understudied in clinical trials? Whether you live in Duluth, Minnesota, or New York City; whether you're a middle-aged male or a college-age female, Curtis Meinert is betting you'd respond to that question without hesitation. Yes, of course they have.

After all, that's the message we've been hearing since the 1980s, from community officials, the National Institutes of Health, and even Congress, which in 1993 issued legislation explicitly requiring women--and certain ethnic groups--to be included in clinical trials.

The only problem with this prevailing perception, contends Meinert, is that it has no basis in fact. Zilch. Zero. Nada. Meinert, the epidemiologist who heads the Center for Clinical Trials at Hopkins's School of Public Health, has been conducting clinical trials for three decades, and has sat on the influential Institute of Medicine's Committee on the Ethical and Legal Issues Relating to the Inclusion of Women in Clinical Studies.

It was while serving on this committee in the early 1990s that he began to wonder: How do we actually know this widely held assumption to be true? Meinert's curiosity eventually led him and his Hopkins colleagues to undertake an exhaustive study, published last fall, that examined more than 100,000 clinical trials reported in leading U.S. journals between 1966 and 1998. The findings: 55 percent of the published trials involved both men and women, 12 percent involved men only, and 11 percent involved women only; 21 percent did not specify gender. With cancer trials, there were 2.5 times as many female-only trials as male-only trials. Even with heart disease, traditionally assumed to affect men disproportionately, 64 percent involved both genders (though male-only heart studies comprised 14 percent, and female-only heart studies, just .1 percent).

Perhaps the legislative changes of the late '80s and early '90s skewed the overall results? Not so, says Meinert, who points out that the biggest change in the gender mix actually occurred years earlier. And what about that 21 percent in the "no-gender specified" category--couldn't that have encompassed many male-only studies? Not really, explains Meinert, since the "no-gender" percentage dropped over time, but there was no corresponding increase in male-only studies.

"Overall," concluded Meinert, "we find little to support the perception that women have been underrepresented or understudied in trials, or that there is an effort bias in favor of men, even in the decade previous to the one in which Congress acted." The report appeared in the October 2000 issue of Controlled Clinical Trials. A press release trumpeting the results went out and then...nothing. While other findings emanating out of Public Health routinely find their way onto the pages of The New York Times or The Wall Street Journal, this one generated barely a ripple of interest.

Meinert, a friendly man who doesn't mince words, wasn't surprised. "Being the cynic I am," he says, "I didn't expect it to get much play in the press. If you come up with a message the world doesn't want to hear, they don't pay much attention."

Why does it matter whether this message does get heard? First off, says Meinert, there's the issue of public trust. Researchers get their funding from taxpayer dollars. If suspicions arise as to how those dollars get apportioned, he says, "it erodes the base for research."

Second, Meinert knows firsthand how difficult it can be to round up enough participants who fit the necessary criteria to conduct an effective study. "Ninety percent of the battle is trying to recruit and enroll," he says. "You're never able to recruit at the speed you would like." The fewer the restrictions, he says, the easier it is to recruit.

Many diseases don't fall neatly into 50/50 gender mixes; they may affect one gender, or other subgroup, disproportionately, he says. The recruitment process only becomes more difficult, time-consuming, and expensive, he argues, when researchers must jump through demographic hoops. And while current NIH guidelines require female study participants (unless their exclusion can be justified), there is no additional money to fund any increased costs for recruitment, he says. The danger, according to Meinert: Important research that could benefit those of both genders could go unexplored.

"There is no point in worrying about whether a treatment works the same or differently in men and women, until it has been shown to work in someone," he contends. Meinert gives the example of a Veterans Administration hospital, which has a predominantly male population. Should a researcher forgo doing a trial--or live with this less-than-ideal study population? "The answer is obvious," believes Meinert. "Information is information, and some, even if imperfect, is better than none at all."

The epidemiologist argues, too, that there are serious ethical issues raised when researchers turn away some patients who are in need of treatment (even experimental treatment) while continuing to recruit others simply to achieve "quotas."

The bottom line, according to this veteran trialist, is that focusing on gender differences just doesn't make scientific sense. He contends that despite obvious anatomical and reproductive differences, males and females are in most ways the same. Example: Clinical trials since the 1960s have led to breakthroughs in curbing heart disease. Meinert notes that between 1960 and 1990, mortality from heart disease has dropped in half--a drop that has been uniform in both men and women, despite the fact that clinical trials often enrolled more men. He says the point is clear: Both genders will benefit from clinical trials that are well-designed and carried out, even if those trials don't draw on a representational population.

To find out how the perception that women have been overlooked in medical research got its genesis, it is necessary to look back to the 1950s and '60s, when revelations came to light regarding abuses in the system involving human research subjects. Around the same time, the public learned that babies born to mothers who took particular prescribed drugs (such as DES) could suffer birth defects. Against this backdrop, the Food and Drug Administration issued guidelines intended to protect human test subjects and unborn fetuses--specifically, a 1977 guideline that discouraged any women of childbearing potential from participating in phase 1 or phase 2 drug trials.

But the tide began to shift in the mid-1980s, with the advent of AIDS, and the concern that women and ethnic groups were being unfairly prevented from taking part in experimental treatments that could help them. The shift from "protectionism" to "access" in medical research gained further momentum in the late 1980s when activists rightfully pointed out that several large-scale heart trials of the '70s and '80s--among them the Multiple Risk Factor Intervention Trial (MRFIT), and the Physicians' Health Study--were male only (despite the fact that 10 percent of physicians were female).

These studies became viewed as "smoking guns," says Meinert. Elected leaders like U.S. Senator Barbara Mikulski and U.S. Representative Patricia Shroeder in the early '90s mobilized the Women's Congressional Caucus to push for greater inclusion of women in medical research. In the 1991 confirmation hearings for NIH director Bernadine Healy (MD fellow, '76), she made it clear that women's health issues would be a centerpiece of her administration.

What may have added to the spur to action for Congress, according to Meinert, was a report prepared by the NIH Advisory Committee on Women's Health Issues. The committee was asked to provide data on what percentage of NIH funding was dedicated to female-specific illness and disease. The totals: 12.2 percent in FY 86 and 13.5 percent in FY 87. Because there were no corresponding figures presented for male-specific illness, it was assumed that the rest of the funding pie was reserved for male-only illnesses. (Meinert points out that in fact the corresponding percentage for male-specific diseases was about 6 percent.) Nevertheless, he says, "the [report] created a great indignation."

So, in 1993, Congress forced the NIH to take action; under legislation known as the Revitalization Act, the funding body established a "valid analysis" mandate. It states, in effect, that in designing studies of illnesses common to both genders, researchers must provide a "valid analysis" of whether variables in the study affect women or minority group members differently than any other subjects in the trial. To do this, there must be sufficient numbers of women and ethnic group members enrolled in the trial. And therein, for Meinert, lies the rub.

He has written papers and commentaries in Science, Statistics in Medicine, and other journals, criticizing the "valid analysis" requirement. Notes Meinert, "The way to better, more robust trials is not by legislation and recruitment quotas, but rather by making them larger and more inclusive. Let the demographics of the disease dictate." He believes the issue has become politicized, to the detriment of good science, asserting, "The mandate springs from parochial interests in who is studied and in the politics of votes."

To see just what the economic impact might be, Meinert looked back in time and asked: Suppose MRFIT had been changed to include a representative number of women and ethnic minorities? Would differing recruitment restrictions have driven up the cost of the study?

Scientists spent years lobbying Congress for funds to do MRFIT, which cost $115 million and took eight years. Its goal: to see whether reducing such risk factors as cigarette smoking, high cholesterol, and high blood pressure in otherwise healthy men ages 35 to 57 could decrease their risk of coronary heart disease. The concentration of risk factors made recruitment challenging: Researchers had to screen nearly 362,000 men in order to enroll 12,688 for the trial.

When Meinert reran the numbers, assuming a gender-race mix of 45 white and 5 black females, and 45 white and 5 black males, per 100 people, he concluded that screening would have taken an additional six years and nearly doubled the cost of the trial--to $209.8 million. That's primarily because finding women with the same constellation of risk factors as men is much more difficult. His results appeared in the 1999 Statistics in Medicine.

"The way to better, more robust trials is not by legislation and recruitment quotas, but rather by making them larger and more inclusive," says Meinert.
But Meinert drew a different conclusion when he re-examined the Physicians' Health Study, which looked at whether low-dose aspirin and beta-carotene could reduce the risk of heart disease and cancer among an all-physician population. The rationale for not including women in that study was "inadequate numbers"--the 10 percent female physician population would not be enough to reliably test whether the interventions worked, or whether the effect differed between men and women. Not a compelling argument, Meinert concluded. Even in smaller numbers, the female data could have yielded important clues for further study. And the cost for recruitment would have been minimal, since it was only necessary to find female physicians willing to take the aspirin and beta-carotene.

Not surprisingly, Meinert has his critics. Defenders of the "valid analysis" mandate, like biostatistician Nancy Geller, of the U.S. Office of Biostatistics Research, take issue with some of his conclusions. Geller notes in a 1999 commentary in Statistics in Medicine, for instance, that the MRFIT study could have been redesigned at a much lower cost by using a smaller proportion of women--enough to suggest hypotheses worthy of further study. She points to a slew of ongoing multi-center trials aimed at women and minorities (such as WAVE, which is examining the use of hormones and antioxidants in preventing heart disease in post-menopausal women), and concludes, "The law has had the effect of making trialists consider differences and similarities of various subpopulations when designing clinical trials. The result is that far more is known specifically about women's and minorities' health than previously."

Ruth Faden, who directs Hopkins's Bioethics Institute and who co-chaired the Institute of Medicine committee on which Meinert served, calls his recent work "a very important study that will help to get rid of some of the ill-informed rhetoric around the claim that women" have been understudied compared to men. "But," Faden adds, "it doesn't end the story." The fact that the numbers show women are included in studies doesn't necessarily mean their interests are being advanced. Early clinical studies involving AIDS, for example, looked primarily at women as either "vessels" for babies, or as "vectors of transmission," Faden says. "Gynecological manifestations of HIV were not even on the radar screen." Currently, she points to "a huge hole in our understanding" of the impact various drugs have on women during pregnancy; as well, she says, there's been little work that looks at the impact of menstrual cycling on how therapeutic interventions work.

While Meinert is clearly no fan of the "valid analysis" requirement, he says, "If this [law] causes us to do bigger and better trials, then I take my hat off to everybody. That's what benefits us as people: bigger, better trials--not more female-only trials or male-only trials."

Then, almost compelled to end on a cautionary note, he adds, "If you try to slice this pie 100 different ways, there's no piece left to slice."