In this paper we have focused on two commonly used behavioural indicators—the number of lifetime sexual partners and the number of sexual partners in the last 12 months—and compared responses from four studies in Eastern and Southern Africa. Data have been provided from responses to questions asked at least once within four large population-based cohort studies that were set up to answer questions about HIV prevalence, incidence and longitudinal changes in sexual behaviour. Previous papers from these sites have described the studies and documented changes in sexual behaviour.6 7 17–19
The number of reported lifetime partners showed several similarities across age groups in the four sites. In all sites, at older ages (>35 years), men reported, on average, around 10 lifetime sexual partners, which is consistent with findings from DHS surveys in Uganda (10.6 lifetime sexual partners for men aged 40–49 years in 2004) and Zimbabwe (8.0 lifetime sexual partners for men aged 40–49 years in 2005).20 21
Women reported a much lower average of 2 or 3 lifetime partners in all sites, which again is similar to the findings reported by Wellings et al
Comparing birth cohorts of the same age, among men, those in later birth cohorts reported fewer lifetime sexual partners than those in earlier birth cohorts. However, among women there was less evidence of a decrease in the later birth cohorts compared with the earlier birth cohorts. Within some male birth cohorts, particularly those born in Masaka in the 1950s and 1960s, a decline in the mean reported number of lifetime sexual partners was observed with increasing age. This may be due to misreporting of lifetime partnerships as 25% of both men and women in Masaka reported fewer lifetime partners in a later survey compared with the earlier survey. However, selective mortality may also contribute to this decline, whereby those individuals with high numbers of partners are removed from the population via AIDS mortality. Furthermore, all sites used open cohort members in the analysis rather than restricting to a closed cohort, and thus changing participation may also contribute to this observation.
For both men and women there are some small differences in the proportion reporting more than one lifetime sexual partner between the sites. In Masaka, fewer men at all ages reported more than one lifetime sexual partner compared with other sites. In Rakai, more women across all age groups reported having more than one lifetime partner. In Umkhanyakude, a higher proportion of women aged >24 years reported more than one lifetime partner, which may be related to the low rates of marriage and later age at first marriage in this population.19 22
Despite these differences between the sites in reported lifetime partners, there is a consistency in the proportion of men who reported more than one sexual partner in the past 12 months. This suggests that reported partnerships in the past 12 months may be a more reliable measure than reported lifetime sexual partnerships, as these reports are less likely to be influenced by recall bias.
The median age at first sex for women is about 5 years younger than for men in Uganda and Umkhanyakude.23 24
However, by the age of 25 years, over 90% of both men and women reported having a sexual partner during the past 12 months and can be said to be sexually active. In older women (>40 years), a decline in the proportion that is sexually active was observed in all sites. However, in older men, such a decline in the proportion that is sexually active was only seen in the Masaka site in Uganda and was not seen in most countries included in a large comparative analysis by Wellings et al
In the Manicaland and Umkhanyakude cohorts there was some evidence of a reduction in the reported number of sexual partners in the past 12 months in later birth cohorts, especially among the younger age groups. Others have reported declines in the number of sexual partners in Uganda in the mid 1990s, which is before the period considered in this paper.17 18
In all four sites there was a striking gender differential with respect to reported number of partners and the proportion reporting more than one sexual partner, for both lifetime partners and partners in the past 12 months. Wellings et al
point out that this could naturally arise through the population age structure in Africa and the patterns of age mixing, whereby older men have sex with younger women.3
However, Boerma et al
note that community cohort surveys may miss highly sexually active women and sexually inactive men because of high levels of migration and mobility among these groups,25
or because of selection biases in sexual behaviour surveys. Many recent sexual behaviour surveys in Africa are conducted as part of HIV surveillance, and individuals with higher numbers of sexual partners may perceive themselves to be at higher risk of HIV and may be less likely to participate in the survey. In addition, women may under-report their number of sexual partners and men may tend to over-report.26
However, the proportion who are sexually active in the past year was similar for men and women in all four sites, and this did not change over time.
In all these sites there have been a number of HIV prevention interventions aimed at encouraging a reduction in the number of sexual partners. These interventions may have a different effect on sexual behaviour in men than in women. Results from Uganda showed that later birth cohorts of women (born in the 1980s) delayed sexual debut compared with earlier cohorts of women (born in the 1950s and 1960s), although this effect was not observed among men.23
In this paper we show that later cohorts of men have reported fewer sexual partners than earlier cohorts of men at the same age. Among men, the reduction in the number of reported sexual partners may have translated into lower demand for new partnerships with younger women, leading to less pressure on young women to start sex early and consequently later sexual debut for these women. More research on sexual mixing patterns in these sites is needed to explore this hypothesis.27 28
Analysis of longitudinal data in this way illustrates trends in behaviour change. Further analyses and modelling are required to relate these changes in sexual behaviour to changes in the HIV prevalence in these populations.6
- The patterns of reported sexual partners were consistent with each site, but differed by sex and by site.
- Across different surveys in all four sites, men reported 41–86% more sexual partners in the past year than women.
- For men, later birth cohorts reported fewer lifetime sexual partners than earlier birth cohorts at the same age in southern Africa, but not in Uganda.
- Taken by itself, the reported number of sexual partners is insufficient to explain the different levels of HIV infection.
There is much evidence linking reported number of sexual partners to the risk of HIV infection at an individual level.29–31
However, there is little evidence linking the number of sexual partners at a population level with the HIV prevalence in those populations.27
It has long been understood that African men do not have more sexual partners than men elsewhere.3
An ecological analysis of four cities with different levels of HIV prevalence showed that high-risk sexual behaviour, including a greater number of sexual partners, did not differ between the four cities.32
The limitation of the reported number of partners as a useful indicator of sexual behaviour has led researchers to analyse the types of partnership and levels of concurrency rather than the quantity of reported sexual partners.28 33
The collection of sexual behaviour and partnership data is subject to several possible biases.34
Social desirability may encourage men to exaggerate their number of sexual partners while married women may remain secretive about their own additional partners,26
although this may be less applicable to South Africa where a lower proportion of women are married.24
An individual’s sexual behaviour varies over time, and different individuals may have periods of greater sexual activity and periods of abstinence, including postpartum abstinence. Thus, averaging the reported number of partners within a population will not accurately capture the changes that occur in individual sexual behaviour. This bias may change over time if losses to follow-up are higher among those with high numbers of sexual partners due to death or migration.25
The observed patterns over time demonstrate similarities across sites, and the longitudinal cohort design enables trends to be observed over time within each of the sites. Other explanatory data, such as marital status and migration patterns, that are not available for this comparative paper, can be used to explore further the patterns in behaviour within each of the sites and to explain how the behaviour is associated with the prevalence of HIV. At the population level, the reported number of sexual partners is remarkably similar across these different sites despite different levels of HIV infection. This would indicate that analysing the number of sexual partners is insufficient to explain the differences in the HIV epidemic in these sites, but more detailed analysis of the type of partners and the duration of partnerships is needed. Qualitative research results may also be useful in the interpretation of these data.
HIV prevention messages encourage abstinence and faithfulness within marriage. These data show some evidence that men have reduced the number of sexual partners in recent years. The large difference between the number of sexual partners (both lifetime and in the past 12 months) reported by men and women needs further research to identify the reasons for this apparent discrepancy.