A population's higher-risk sexual behaviour is associated with its average sexual behaviour—An ecological analysis of subpopulations in Ethiopia, Kenya, South Africa, Uganda and the United States

• Previous research has found mean population behaviour (such as salt and alcohol intake) to be closely associated with the prevalence of those with high risk behaviour.
• We investigated if this was true for variations in sexual behaviour by ethnic group within Ethiopia, Kenya, South Africa, Uganda and the United States of America.
• We found that the mean number of lifetime sex partners in a population is strongly correlated with the prevalence of those classified as being in the top decile of number of lifetime partners.
• Our results suggest that sexual behaviours vary coherently between different populations and differences in mean may act as important determinants of prevalence of those with an excessive number of partners.
• As such, sexually transmitted infection control efforts would benefit from including population-strategies that seek to address factors responsible for high mean risk behaviour.

BackgroundGiven the large variation in the prevalence of sexually transmitted infections between populations it is important to characterize how sexual vary between populations. In this paper we assess how the distribution of lifetime sexual partners varies between populations. We ask: do populations with higher mean lifetime sexual partners merely differ by virtue of the presence of a core-group with increased number of partners or do the population distributions move as more coherent wholes?MethodsWe defined those in the top decile of lifetime sex partners in each country as constituting the higher-risk behaviour group (HRB). Spearman's correlation was used to evaluate the relationship between mean lifetime partners and prevalence of those in the HRB by ethnic group within Ethiopia, Kenya, South Africa, Uganda and the United States. In each case, data from nationally representative surveys were used. Two-sample Wilcoxon rank-sum were applied to test if there was a difference in the number of lifetime partners between the highest and lowest-risk subpopulations, separately for men and women. To account for autocorrelation, all analyses were conducted using means/medians excluding those in the HRB. A P-value of <0.05 was considered statistically significant.ResultsIn each country, a positive correlation existed between subpopulations’ mean lifetime partners and the prevalence of the HRB. Spearman's correlation varied from 0.20 to 0.99 for men and 0.74 to 1.0 for women. This association was statistically significant in 6 out of 10 comparisons. There was a statistically significant difference in the number of lifetime partners between the highest and lowest-risk ethnic groups in all countries except South Africa, where the difference was only significant for women.ConclusionsOur results suggest that sexual behaviours vary coherently between different populations. As such, sexually transmitted infection control efforts would benefit from including both targeted campaigns focusing on the HRB and population-strategies that seek to address factors responsible for high mean risk behaviour.