Estimation of coital frequency and condom use from cross-sectional survey data

Abstract

Coital frequency is an important theoretical determinant of the rate at which an STI can spread through a population. Differences in frequency of sex may bias measures of condom use based on survey data because survey respondents who have infrequent sex and those who have frequent sex contribute equally to commonly used measures of condom use. Data on sexual behaviour are widely available from surveys but detailed information on coital frequency is seldom collected. This thesis examines the available information, examines the utility of a method to make the most use of existing data and investigates whether condom use measures are biased by differences in coital frequency using, for the most part, data from Australia and Tanzania. The existing data and literature show a lack of information on coital frequency for men and for unmarried people. Certain factors are correlated with coital frequency but there are no stable patterns of variation between different populations. Data from Australia and Tanzania show that condom use and coital frequency both vary according to the types of partnership for which they are reported. Further analysis of commonly used measures of condom use shows that these measure are influenced by differences in coital frequency and demonstrates that additional measures, describing the proportion of sex acts protected by condoms, provide complementary information. The scarcity of data on coital frequency is unlikely to be resolved by using the more widely available information on time since most recent sex. Although it is theoretically possible to work backwards from this to the number of sex acts in a given time period, this does not work in practice. Recommendations are made for improved methods to collect coital frequency information in large-scale surveys of the general population.