Exploratory data analysis of three sexual behaviour surveys: implications for HIV-1 transmission in the U.K.

Abstract

Recently, the number of European and North American heterosexually transmitted AIDS cases has begun to increase. It appears that injecting drug users may be acting as an important infection source for heterosexual transmission, and that the HIV-1 epidemic may be slowly diffusing out into the general population. Therefore, the question arises, as to whether there will be an extensive heterosexually transmitted epidemic in Europe or North America. The possibility of such an epidemic occurring depends upon three factors: sexual mixing patterns, the prevalence and the distribution of specific risk factors, and changes in risk behaviours. To assess the potential for an extensive sustained heterosexually transmitted epidemic it is necessary to estimate these three factors by conducting large-scale linked behavioural and seroprevalence surveys. Unfortunately, such surveys have yet to be conducted in any country. However, a limited amount of sexual behaviour data are available from three small studies that have been conducted in the U.K. Although each of the studies suffers from methodological imperfections, together they may be used as the basis for a preliminary assessment of the potential effects of heterosexual transmission in the U.K. In this paper I discuss the results from an exploratory analysis of these data; the results are organized into five sections: (i) assessment of the prevalence and distribution of specific risk behaviours; (ii) determination of covariates of risk behaviours (and hence characterization of the subgroups that may be at risk); (iii) assessment of age-dependent sexual mixing patterns; (iv) assessment of sexual behaviour changes; and (v) calculation of epidemiological parameters. The validity and the reliability of the data from the three studies are assessed. The analysis suggests that it is possible that a heterosexually transmitted HIV-1 epidemic could occur in the U.K. Hence, it is essential to conduct longitudinal surveys that collect linked seroprevalence and behavioural data.