We're All Different Under the Skin



It swooped in below the public radar, passing the House and Senate under cover of the HIV immigration ban. Nobody talks about it, because hardly anyone knows it exists. Yet Section 131 of the National Institutes of Health Revitalization Amendments would fundamentally harm the process of clinical research in the United States.

The bill mandates that ''in conducting or supporting clinical research . . . the director of NIH shall . . . ensure that: (1) women are included as subjects in each project of such research; and (2) members of minority groups are included as subjects in such research.''

The amendments respond to the widespread view that medical research has systematically underrepresented women and minorities. Their goal is to ensure the fair distribution of benefits from federally funded medical research, an admirable and ostensibly reasonable aim. In the past, some medical studies arbitrarily excluded non-whites and women, and studies that did include women frequently excluded all who were pre-menopausal.

The trouble is that the bill fails to distinguish between these arbitrary exclusions and scientifically rational ones. By mandating that all medical research look for differences by race and gender, Congress has written into law a dubious biological assumption: that races and the sexes differ so fundamentally that data from one group cannot be applied to another. The result will be to raise that cost dramatically and decrease the number of clinical studies.

Because most diseases strike the races and sexes unevenly, it makes sense for many studies to focus on subgroups with high rates of the diseases under consideration. Since more men than women suffer heart attacks, for example, studies of heart-disease prevention that include only middle-aged men require many fewer persons to be tracked for a much shorter time than studies that include middle-aged women.

The scientific rationale for excluding women from these studies (and men from others) is that people of different races and sexes have more biological similarities than differences. If a daily dose of aspirin reduces men's chances of heart attacks, odds are it also reduces women's. The odds are good enough, in fact, that most researchers confidently project findings from men onto women, and vice versa. The assumption of biological homogeneity frees research dollars for other much-needed studies.

There are, of course, instances in which an assumption of different responses among subgroups makes sense. For example, the female hormone estrogen may naturally protect pre-menopausal women from heart disease. Thus, a study of estrogen therapy to prevent heart disease should include enough men and women to compare responses by gender, since female hormones have very different effects on men and women.

But in other cases, there is no scientific rationale for striking a balance between sexes and races. For example, if trials of osteoporosis prevention must include men, the incidence of the disease in the study population will be much lower than if only women are examined. Unless the sample size is greatly increased, studies will fail to observe enough cases of the disease to learn how to prevent it.

The most dangerous implication of the bill is not merely that every type of person must be included in all trials, but that all subgroups must be included in sufficient numbers to demonstrate possible treatment difference between them and ''other subjects,'' presumably white men.

To understand the practical consequence of such a requirement, consider the following hypothetical example: a study designed to test whether a treatment for hypertension reduces risk of stroke requires that roughly 5,000 persons be examined regularly for five years. Showing that such treatment works in women, men and minority groups requires an average of 5,000 in each subgroup.

But if, as the bill would require, the trial designers were compelled to quantify marginal treatment differences between each of these groups, the sample size would skyrocket. The trial would have to be large enough not only to assert reliably that the treatment lowers the overall stroke rate, but to demonstrate the differences between groups were due to more than chance. This would probably mean increasing the sample size by a factor of eight or more.

Such huge increases would carry enormous price tags. Last year the National Institutes of Health spent upward of $800 million on clinical trials; analysis of all studies by gender and race would require quadrupling this budget. Analyzing differences between both gender and race would require more than an eightfold increase. And this increase involves only the most basic definition of ''minority,'' one that only separates whites from non-whites. But why stop there? Conceivably, Mexican-Americans and Cuban-Americans could display subtle differences in treatment effect and disease progression, since all people of Hispanic origin are not ethnically homogeneous.

Baltimore Sun Articles
Please note the green-lined linked article text has been applied commercially without any involvement from our newsroom editors, reporters or any other editorial staff.