Stanford University Home

Stanford News Archive

Stanford Report, April 5, 2000

Research criteria unintentionally exclude minorities  


African-Americans, the poor and people with severe drug, alcohol or behavior problems are being excluded from psychiatric studies intended to help them, researchers from Stanford and the University of California, San Francisco (UCSF), have discovered. The fault lies with the rules for deciding who qualifies for a study -- guidelines designed with the best motives but that often weed out the people who would benefit most from the research.

The researchers say that the findings also explain why many treatments that perform well in controlled studies often fail in real-world situations: they are tailored for patients who tend to be wealthier and more stable than those showing up at psychiatric clinics and hospitals.

Almost every medical study applies some sort of guidelines, or exclusion criteria, to winnow the pool of potential subjects, said Keith Humphreys, PhD, assistant professor of psychiatry at Stanford University Medical Center and associate director of the Veterans Affairs Program Evaluation and Resource Center. Would-be participants can be turned away because of age, health, where they live or how supportive their family is, to give just a few examples. Although selection rules make the researcher's job easier, they can also skew the composition of the study group so that it no longer reflects the real patient population, Humphreys said. "The kinds of people who end up in research studies aren't the ones who are seeking care," he said.

Humphreys and colleague Constance Weisner, DPH, professor of psychiatry at UCSF, examined how exclusion criteria operate to favor one kind of patient over another by studying 593 patients in eight alcohol treatment clinics in Northern California. They selected eight commonly used exclusion criteria, then asked which of the real patients would be eligible to participate in a hypothetical study if one of these rules was in force. Grounds for exclusion from the study included non-
compliance, living too far from treatment, drug dependence, social instability, psychiatric or emotional problems, medical problems, having no fixed address and failure in prior treatment.

Overall, between eight percent and 75 percent of patients would be disqualified, depending on which rule was applied. The criterion with the lowest disqualification rate was being non-compliant or unmotivated, while failure at previous treatment ruled out the largest number of patients.

The researchers then analyzed the results according to whether the clinics were private or public, since the two types of clinics draw different kinds of patients. As they report in the April issue of American Journal of Psychiatry, Humphreys and Weisner found significant differences between excluded and included groups at both kinds of facilities. For instance, though blacks made up only 22 percent of the patients at private clinics, they constituted 31 percent of the excluded patients under the drug dependence criterion. Using the same rule at public clinics, where 46 percent of patients were African-American, 62 percent of the excluded patients were black.

When Humphreys and Weisner considered income, they discovered that, under several criteria, patients faced higher odds of exclusion at both types of clinics if they earned under $10,000. The same was true for patients who were dependent on drugs or had severe alcohol or psychiatric problems.

Humphreys sees no evil intent behind the exclusion criteria that produced these findings. Instead, he thinks that researchers are motivated by legitimate concerns such as convenience and cost as well as by the desire for a smooth-running, successful study. While their intentions are good, researchers are essentially "creaming" the patients most likely to succeed, which tend to be well-off whites with less severe psychiatric and alcohol problems, Humphreys said. Thus "the research gives an unrealistically rosy picture of how well the treatment works."

Scientists should change their ways for ethical and practical reasons, Humphreys said. "By law and by tradition, government-funded research is supposed to be useful to all groups in society." To that end, National Institutes of Health guidelines require studies to enroll representative numbers of minorities and women. However, because of the extensive use of exclusion criteria, many researchers may be unintentionally violating these rules, Humphreys said.

On the practical side, the homogeneity of study groups means, many people are not getting appropriate treatment. Scientists already know that the effectiveness and toxicity of drugs vary among races. But to develop treatment programs that work for a larger fraction of the population, researchers will have to take into account the cultural differences of the participants, Humphreys said. One corrective measure, he said, may be to reduce the use of exclusion criteria in medicine. As a result, scientists may have to work harder to locate and track patients who are homeless, poor or emotionally unstable, and they may have to plan larger studies, knowing that a certain fraction of the participants will drop out.

Those who fund research should be prepared for bigger bills, Humphreys said. With scientists running larger studies or pursuing subjects more diligently, costs will rise.

Humphreys and Weisner's study was funded by the National Institute on Alcohol Abuse and Alcoholism, one of the National Institutes of Health. Humphreys works at Stanford and at the Palo Alto Veterans Affairs Health Care System. SR