Introduction

Models of the transmission dynamics of sexually acquired infections have recently turned greater attention to the importance of sexual mixing patterns and sexual networks in the spread of such diseases (Hethcote and Yorke 1984, Ramstedt et al. 1992, Haraldsdottir et al. 1992). Apart from the theoretical problems of constructing models which can adequately characterise the complexity of human sexual partnership formation, there remain practical problems in obtaining sufficiently robust empirical data to measure parameters of interest for use by modellers. This paper addresses some aspects of empirical measurement and discusses the validity of some basic parameter estimates commonly used in models of sexually transmitted disease (STD) transmission, particularly where they are used to demonstrate quantitative rather than qualitative considerations.

Key parameters of interest in deterministic models include the population ‘rate of partner change’ and its variance, the probability of transmission per sexual partnership (or per act of intercourse) and the duration of infectiousness of the organism under consideration (Hethcote and Yorke 1984, Anderson and May 1988).

Recent studies of sexual behaviour (Johnson et al. 1992, ACSF investigators 1992, Catania et al 1992, MMWR. 1988) have emphasised the marked heterogeneity in sexual behaviour in human communities as measured by numbers of sexual partners in different time intervals. These distributions suggest that models need to take account not only of simple population means, but also to consider stratification of the population into high- and low-activity classes (Hethcote and Yorke 1984). This can, in part, be achieved by considering the demographic correlates of variability in sexual behaviour. These include, for example, varying effects in different cultures of age cohort, lifecourse, gender, marital and socio-economic status (Johnson et al. 1992).