might also serve as a useful model for menopause, to look at the effect of vaginal transmission of viral infection in an absence of hormones.

Willard Cates, Jr., M.D., M.P.H.

Family Health International

In a recent evaluation, Family Health International (FHI) reviewed the role contraception plays in a range of factors affecting sexual transmission of human immunodeficiency virus (HIV). Our basic knowledge includes a general understanding that the most powerful predictor of HIV transmission is the stage of infection during which sexual contact occurs, with probability of transmission highest during late-stage infection, during very early-stage infection when viral load is highest, and/or concurrent with the presence of other sexually transmitted diseases (STDs) of the type that are ulcerative or productive of discharge.

Important among the elements of the review was an examination of relationships, either protective or facilitative, between hormonal contraceptives and HIV transmission. Because no data are available from Level 1 studies—that is, randomized controlled trials-the review depended on epidemiologically based data from Level 2-3 studies-that is, well-controlled cohort studies. Of approximately 25 observational studies which collected data on HIV transmission and oral contraception, the majority were cross-sectional, which meant that no conclusions could be drawn about direction of causality. Only 9 of the studies were of Level 2 quality.

The range of association, in a variety of populations for combined oral contraceptives containing both estrogens and progestins, extended from a protective effect of 0.6 to a harmful effect of 4.5; no conclusions can be drawn from such a range. Since a measure of relative risk equal to 1 means no effect, any measure less than 1 is considered protective, and any measure greater than 1 indicates a harmful association. In studies where the quality of evidence is weaker, relative risks below 2 may be confounded by many biases, including risk of sexual exposure, contextual and biological factors affecting transmission, and behavioral variables that may mask biological impact.

As for the relationship between implanted hormones and HIV transmission, no studies with large enough populations to permit solid, directly attributable conclusions have been conducted. However, the injectable hormone Depo-Provera