Limiting Eligibility for Medical Studies Can Omit Women, African-Americans, Stanford Study Finds
In reviewing data from a pool of 100,000 alcohol treatment patients, Keith Humphreys, PhD, professor of psychiatry and behavioral sciences, determined that women and African-Americans were substantially more likely to be excluded from treatment studies than men or non-African-American patients, because of eligibility requirements involving psychiatric problems, employment and housing problems, and drug use.
"Researchers' own study designs are thwarting their good-faith efforts to recruit representative patient samples," said Humphreys, whose paper will be published in the June issue of Alcoholism: Clinical and Experimental Research. "If we want health-care practice to be guided by research, we're going to have to do a better job at studying patients that clinicians actually see."
Medical studies typically exclude certain patients from participation. While some exclusions are often necessary--to protect patient safety, for example--Humphreys suspects researchers often use exclusions in their studies out of habit or tradition. In other cases, researchers may use eligibility criteria to enroll only "desirable" patients in their study, in an effort to make the trial run smoothly or to increase the chances that a favored treatment will show positive outcomes.
Humphreys said such exclusions could have dangerous implications. "If treatments are tested on, and developed for, only part of the population, that means everyone excluded is at greater risk when they use health care," said Humphreys. For this reason, the National Institutes of Health has instructed researchers to design their studies to ensure adequate enrollment of all populations, including women and racial minorities.
"The NIH, and we as a society, have decided that the burden and benefits of health research should be shared by everyone," said Humphreys, who is also on the staff of the Veterans Affairs Palo Alto Health Care System. "It's a moral and political stance and we're obligated to live up to it."
For the analysis, Humphreys and his colleagues looked at data sets from five previous alcohol treatment studies that enrolled more than 100,000 patients. They then explored the impact of six exclusion criteria commonly used in treatment outcome research: psychiatric problems, neurological problems, medical problems, drug use, non-compliance issues, and social and residential instability.
The researchers found that women were 62 percent more likely than men to be disqualified from studies that required social and residential stability. Women were 29 percent more likely to be excluded from studies that disqualified patients with psychiatric problems, and 20 percent more likely to be excluded for medical problems.
They also found that African-Americans were 46 percent more likely than non-African-Americans to be excluded from studies that required social and residential stability, and 23 percent more likely to be excluded based on drug-use criteria.
"The end result can be a sample that over-represents Caucasian males relative to the treatment population," the authors noted in their paper.
Humphreys stressed that he doesn't believe his fellow researchers are acting out of prejudice or purposefully excluding patients. "Most researchers want women in the studies, but they don't realize that the way they design the study undermines their intentions," he said. "A researcher may say, 'I don't want anyone who is depressed in my alcohol study,' and not stop to consider that the majority of depressed people are women."
Based on the paper's findings, Humphreys said he would encourage researchers to carefully examine the rationale of a study's eligibility criteria. "Researchers need to say to themselves, 'I'm only going to make the decision to exclude certain patients if there is a good reason for it,'" he said.
Humphreys and his colleagues are also urging designers of alcohol treatment studies to stop excluding patients with drug problems. "Eighty percent of patients with alcohol problems use some sort of drug, so there is a strong case for including these patients in studies," he said.
This study was funded by the National Institute on Alcohol Abuse and Alcoholism, and the Department of Veterans Affairs Health Services Research and Development Service. Humphreys' co-authors include the VA's Kenneth Weingardt, PhD, and Alex Harris, PhD.
Stanford University Medical Center integrates research, medical education and patient care at its three institutions -- Stanford University School of Medicine, Stanford Hospital & Clinics and Lucile Packard Children's Hospital at Stanford. For more information, please visit the Web site of the medical center's Office of Communication & Public Affairs at http://mednews.stanford.edu.
Posted: May 2007