Antiretroviral treatment (ART) offers the possibility of significant improvements in length and quality of life for many HIV-infected individuals. However, it remains to be seen whether wide-scale ART in a high prevalence resource-poor setting will lead to a change in HIV incidence in the general population through its influence on sexual behaviour and attitudes. More information about partnership patterns in South Africa and their contribution to the spread of HIV in the population is thus needed to inform rational behavioural interventions that might reduce HIV incidence.
Using ACDIS data a number of projects are underway or have recently been completed:
A. Validation of Multiple Imputation methods for ACDIS sexual behaviour surveys
Research question
Can we validate the use of multiple imputation (MI) for drawing statistical inferences about sexual behaviour that are representative for the entire ACDIS population?
Rationale
The routine womens’ and mens’ general health surveys in the ACDIS have been subject to non-response rates of between 40 and 60 percent in each round of surveillance. With large amounts of missing data, it may be that inferences based on only the complete data are not representative of the entire population.
Data source/methods
This study will consider two rounds of the ACDIS general health survey, collected in 2003-4 and in 2007, and will focus on modelling two outcomes, the number of partners in the past 12 months, and whether the respondent has more than one current partnership. The approach will be to create a “complete” dataset based on the ACDIS data, then omitting a portion of the data, imputing a dataset from this, and the testing imputation by comparing the imputed results with the results from the complete data.
Implications
This work will result in an imputation model suitable for implementation in substantive work that uses the AC sexual behaviour data.
B. Measurement of the prevalence of concurrent relationships, 2003-2007, using different definitions, in rural South Africa
Research question
To contrast different approaches to measuring concurrency, and to describe the prevalence of concurrency, 2003-2007, in a rural South African population.
Rationale
Concurrent sexual partnerships have been identified by modelers as a significant driver of the spread of HIV. It is argued by some that concurrency should be a key message focus in HIV prevention strategies. Studies measure concurrency in different ways, in some cases not distinguishing concurrency from serial monogamy. Africa Centre has used two different approaches to measuring concurrency which provides an opportunity for comparison.
Data source / Methods
The following definitions are considered:
A. A response of 2+ partners to any of the 2003 questions
B. Using the 2005-2007 surveys:
Ø Currently in 2+ relationships
Ø 2+ partners in the last year
Ø Sex with 2+ partners in last month using time since last sex data
Ø Overlapping relationships
Preliminary findings
The prevalence of concurrency was very low among women compared to men, and lower among residents than non-residents for both sexes. Even if social desirability results in under-reporting by women, the data consistently indicate concurrency is rare among women.
Implications
The findings suggest that in rural South Africa prevention messages focusing on reducing concurrency would be best targeted at migrants, particularly male migrants; and that efforts should be made to raise women’s awareness to the prevalence of concurrency (and possible HIV acquisition risk) among their potential male partners.
C. Older people’s sexual behaviour and HIV risk in four high prevalence sub-Saharan African populations
Research question
To compare sexual behaviour indicators of HIV risk in men and women aged 40-60 years by marital status, including monogamous first marriage and monogamous re-marriage as separate categories, in four African cohorts, and with younger age groups.
Rationale
HIV prevalence in older people (40 years and older) is expected to increase rapidly in sub-Saharan Africa due to the increased survival of HIV infected adults on treatment and new HIV acquisition by older people. In contrast to sexual risk behaviours in young people in sub-Saharan Africa, HIV/STD research and intervention efforts have seldom focused on high risk sexual activity behaviour in older people.
Data source
This paper uses data from four cohort studies: the Manicaland cohort in eastern Zimbabwe; the Kisesa Demographic Surveillance System (DSS) in the Mwanza region of Tanzania; the Kyambuliwa cohort in south-western Uganda, and the Africa Centre DSS in KwaZulu-Natal, South Africa. We not only use a novel approach to classify sexual behaviours in the past year as “safe” or relatively “unsafe” in terms of individual HIV risk, but also classify the risk of individuals’ sexual behaviour for their partners as a way to examine the extent to which older people are exposed to HIV.
Findings
We find that many people over aged 40+ years have sexual behaviours that put them and their partners at risk of HIV.
Implications
There is a need for HIV prevention programmes to creatively find ways to promote safer sexual behaviours, particularly condom use, in older people aged 40+ years. HIV incidence at older ages may be relatively small compared to younger age groups, nonetheless older people’s sexual behaviours place them at an unacceptable high risk of HIV.
D. Age at second sexual partnership: data from sub-Saharan Africa on a poorly described determinant of HIV transmission
Research question
To describe the timing of transition from zero or one lifetime partner to at least two lifetime partners.
Rationale
Age at first sex marks the start of an individual’s exposure to HIV infection: the acquisition of a second partner marks the point at which that individual could transmit infection. Unlike age at first sex, progression to a second sexual partner is not well documented.
Date source
Data from the Africa Centre Demographic Information System (South Africa), the Masaka District Cohort Study (Uganda) and Demographic and Health Surveys (Namibia, Tanzania, Uganda, Zambia and Zimbabwe).
Findings
In all countries there are substantial differences between men and women with respect to acquiring a second partner. Fewer women than men acquire a second partner and, on average, women are older when they do make the transition. Men in all countries seem to acquire second partners at a similar rate, although the age at which these partners are acquired varies between countries. By their late twenties around 80% of men have had a second partner and this is remarkably similar across the different sources of data. There is more difference between countries in the pattern of second partner acquisition for women and Zimbabwean women are markedly different from those in the other countries. The distribution of current marital status is different for respondents with only one lifetime partner and those with more than one lifetime partner.
Implications
The results suggest some important areas for further research. It is likely that marriage patterns have an important effect on the acquisition of a second partner.
E. Marriage and Partnership patterns in the context of low rates of marriage & continued high HIV incidence in rural South Africa
Research question
To explore determinants of first marriage among 18-35 year olds in a rural South African population with high HIV prevalence.
Rationale
Marriage rates have continued to fall among African South Africans, and now are the lowest in sub-Saharan Africa. Never married individuals have been shown to be at significantly higher risk of being HIV positive than those currently married. Are certain characteristics associated with the incidence of first marriage in this age group?
Data source / methods
An analytical sample of never married individuals aged 18-35 years who were members (resident or non-resident) of at least one household in the ACDIS surveillance area on June 1, 2001 (9,265 women and 10,988 men), were considered at risk of first marriage, and followed for 7 years (through June 30, 2008) using survival analysis methods (Cox proportional hazard models) to accommodate censoring of those who had not yet married by the time of their last report during the period.
Preliminary Findings
In this population, marriage is uncommon in young adults (18-35 years). This contrasts with other sub-Saharan countries where while age at first marriage has increased, marriage remains almost universal. There is strong selection into marriage, with marriage concentrated in higher SES groups and/or particular religious groups. Employment appears to promote marriage among men, while employed women are more likely to remain unmarried.
Implications
Further analyses are needed to explore the pathways by which marriage, as a distal and proximate determinant, reduces individual HIV risk
1. Silungisa Impilo: This study aims to investigate the impact of antiretroviral treatment (ART) on family and partner relationships, and sexual behaviour of HIV positive individuals in the Hlabisa sub-district of Umkhanyakude.
A cohort of HIV positive men and women, who i) are initiating ART (‘ART initiators’) or ii) are being monitored for ART (‘Monitoring’), are recruited at three primary health care (PHC) clinics. Participants are interviewed every six months by a member of the study team. The study will enrol 600 individuals; 300 ART initiators and 300 Monitoring with a follow-up of 3 years. The study started in January 2009 and, as of January, 521 have been enrolled; 374 ART initiators and 147 Monitoring. We have increased the accrual of ART initiators on the basis of observed lost to follow-up rate by 6 months. Recruitment is still ongoing. Six month visits started in July 2009, and 12 month started in January 2010.
Analyses of baseline data are planned and a baseline cohort paper will be written in mid-2010 to describe baseline characteristics, including stigma and gender norm scores, and how they are associated with disclosure of HIV positive status and with sexual behaviours reported at baseline in the two recruited groups.
F. Non-residents living arrangements study (Ikhaya ngekhaya or ‘Home from Home’).
Data collection for this cross-sectional study, added as a module to the non-resident HIV surveillance visits in 2009, was completed in mid-December 2009. Data entry and cleaning are ongoing. The study aimed to investigate the cohabitation and partnering arrangements of non-resident members of households in the demographic surveillance area (DSA), to determine the migration factors associated with risky sexual behaviour, and to investigate the heterogeneity of HIV risk among migrants with respect to their living arrangements while away from home, in particular cohabitation with sexual partners.