the way STDs interact with HIV infection (Pepin et al., 1989; Piot and Tezzo, 1990; Laga et al., 1991, 1993; Wasserheit, 1992; Wald et al., 1993; Laga, Diallo, et al., 1994; Grosskurth et al., 1995).
The interrelationship between HIV infection and chlamydial infection is defined strictly as an increase in HIV transmission probability (the relative risk) due to increased HIV susceptibility and infectivity in presence of the cofactor chlamydial (May and Anderson, 1989; Boily and Anderson, 1996). To account for the variability in the estimates from different studies (Plummer et al., 1991; Wasserheit, 1992; Laga et al., 1993; Wald et al., 1993), it was assumed that chlamydial increases HIV transmission probability by 3.6- and 5-fold. In this model, the epidemic is seeded by introducing one HIV infected person in the female activity class 6 in 1980. The annual rates of new partner acquisition of the six sexual activity classes for the model (Table C-1) were derived from different national sex surveys of the general population (Anderson and Dahlberg, 1992; Leigh et al., 1993; Laumann et al., 1994; Seidman and Rieder, 1994) that report approximately 10 percent of the general population had more than two partners in the previous year (Table C-2). The problems with such data, as with most data on sexual behavior in the general population (ACSF Investigators, 1992; Anderson and Dahlberg, 1992; Johnson et al., 1992; Leigh et al., 1993; Seidman and Reider, 1994; Laumann et al., 1994; Turner et al., 1995), are that for a variety of reasons (Morris, 1993; Wadsworth et al., 1996), men usually report more female partners than females do male partners. This is inconsistent with the fact that men and women are having sex with each other (Blower and McLean, 1991; Boily and Anderson, 1991; Morris, 1993; Wadsworth et al., 1996). Thus, for simplicity and to ensure that the mean number of sex partners between the male and female population is balanced (Blower and McLean, 1991; Boily and Anderson, 1991; Lepont and Blower, 1991), we assumed that males and females have a similar distribution in sexual activity. The simulations have been performed under an assortative mixing scenario (Garnett and Anderson, 1993b; Garnett et al., 1996) or, in other words, one where the individual prefers to choose his or her partners within the same activity class (a minimum of 44 percent of partner formation occurs within members of the same activity class). The fact that, under proportionate mixing (individuals choose their sexual partners at random, depending on availability only) (Haralosottir et al., 1992; Boily and Brunham, 1993), chlamydial and HIV infections cannot establish in the population even with a relative risk of 5 further supports this hypothesis.
The predicted HIV and AIDS trends and estimates of the fraction of cases attributable to cofactor chlamydial were produced using different sets of realistic parameter assumptions in which HIV transmission probabilities were varied depending on the magnitude of association used. The various biological and demographic parameter values used for chlamydial and HIV infection are summarized in Table C-3, and those on sexual behavior and initial chlamydial prevalence are