Estimation of gender bias in clinical trials

The perception is that the clinical trials enterprise has been biased in fa vour of males by devoting a disproportionate effort to males and to the dis eases and conditions afflicting them - a perception reinforced by a few hig h profile male-only heart trials undertaken in the 1970s and 1980s. The per ception was sufficient to cause the U.S.A. Congress to enact legislation to require that a clinical trial Lis designed and carried out in a manner suf ficient to provide for a valid analysis of whether the variables being stud ied in the trial affect women... differently, than other subjects in the tr ial'. Observed effort differentials are based on counts of single-gender tr ials indexed in MEDLINE and published in U.S. journals. Differentials are c ompared to those expected using male-female differentials in mortality and years of potential life loss due to mortality before age 65 to estimate eff ort bias. The ratios of female-only to male-only published trials were 0.53 , 0.89 and 0.95 for the decades of 1966-1975, 1976-1985 and 1986-1995, resp ectively. The expected ratios, if single-gender trials were done in proport ion to female -male mortality differentials, would be 0.57, 0.56 and 0.57, respectively. The differences in observed versus expected female to male ra tios correspond to a slight excess of male-only trials in the decade of 196 6-1975 and to sizeable excesses in female-only trials in the decades of 197 6-1985 and 1986 -1995. The results do not support the perception that women have been understudied relative to males in clinical trials. Most differen tials favour females, whether based on mortality or years of potential life loss due to mortality before age 65 years. Copyright (C) 2001 John Wiley & Sons, Ltd.