The supply and demand dynamics of sexual behavior: implications for heterosexual HIV epidemics.

This article investigates the supply and demand dynamics of sexual behavior in a simple epidemiological model of heterosexual transmission of human immunodeficiency virus (HIV). The supply and demand dynamics of sexual behavior are modeled by specifying implicit sexual behavior change (ISBC) mechanisms. These mechanisms specify how males and females modify their rates of sex partner change in response to changes in the availability of the opposite sex. During an epidemic, the availability of both sexes will change owing to recruitment and mortality; consequently, implicit sexual behavior changes will be induced. These behavior changes will be independent of any explicit sexual behavioral changes that may occur as a result of intervention strategies. We investigate four different ISBC mechanisms: the two extremes of the continuum of possible mechanisms (where only one sex changes behavior), as well as two intermediate possibilities (where both sexes change behavior). The results show that the epidemiological effects of these ISBC mechanisms depend upon the transmission speed of the virus. If HIV transmission is slow, the epidemiological effects of the four ISBC mechanisms cannot be differentiated. If HIV transmission is fast, the four ISBC mechanisms differ considerably in the degree to which they modify the gender-specific rates of sex partner change, but the sexual behavioral changes occur too late to significantly decrease the cumulative number of infected persons. However, if HIV transmission is moderately fast, the four different ISBC mechanisms produce significantly different epidemics. The differences between the epidemics, produced by the four ISBC mechanisms, are magnified as the degree of asymmetry between the heterosexual transmission efficiencies increases. We discuss the implications of our results for both the future number of AIDS cases that will be observed in the "real world" and the number of AIDS cases that will be predicted from mathematical models. We also discuss the implications for studies that evaluate the causation of sexual behavior changes.