factors that may influence either the infectiousness of or susceptibility to HIV are listed in Table 3-1.

Areas with the highest HIV prevalence correspond roughly to geographic areas where most men are not circumcised (Bongaarts et al., 1989; Caldwell and Caldwell, 1993; Caldwell, 1995a, 1995b; Piot et al., 1994). Lack of circumcision in men has been associated with an increased prevalence and incidence of HIV in several studies in Nairobi (Cameron et al., 1989), although not in studies in Rwanda or Tanzania (Borgdorff et al., 1991, cited in Piot et al., 1994; see also Chapter 4). Vaginal trauma or abrasions caused by traumatic sex or vaginal application of plant extracts or desiccating products have been associated with an increased risk of HIV acquisition among women in several countries (see Chapter 4).

Since β is defined as the probability of transmission per partnership, it depends in part on sexual behaviors, such as coital frequency, anal intercourse, and dry sex. The most important measure of sexual behavior is c, the rate of sexual partner change within a population (which through its effect on STDs also affects β). Note that c is not the average annual number of new sexual partners per person, µ, but instead is σ+ σ²/µ where σ² is the variance of the number of new sexual partners per year. It is the number of partners associated with the average new partnership, which exceeds the average number of new partners per person, just as the size of the city in which the average person resides exceeds the average size of cities. The lesson is that both mean and variance matter. Those few individuals with many new partners ensure that infections will spread more rapidly.