Review of evidence on the linkages between GBV and HIV

1. Review of evidence on the linkages between GBV and HIV Naeemah Abrahams Ph.D. Gender & Health Research Unit: South African Medical Research Council On behalf of: Global Burden of Disease Injuries and Risk Factor Study2010 study expert group on Interpersonal Violence: chaired by Charlotte Watts1 and Claudia Garcia-Moreno 2 and co-ordinated by Karen DeVries1 1London School of Hygiene and Tropical Medicine 2World Health Organization

2. Introduction • Recognition from quite early in AIDS epidemic that gender inequality and violence placed women at risk of HIV • High level endorsement: e.g. Ending VAW is among 9 priority areas in the UNAIDS Outcome Framework 2009-11 • PEPFAR funding in Sub-Saharan Africa • UN Stop Rape Campaign • Many NSPs include interventions to address gender issues and reduce violence • Key question: • What is the evidence?

3. Many pathways through which IPV experiences increase risk of HIV • Forced sexual intercourse • Women in abusive relationships less likely to refuse and negotiate sex and condom use • Abusive men more risky behaviours (more likely to have multiple partners/ use alcohol/ visit sex workers / have STIs/ • Child sexual abuse • Reverse causality – abuse/ abandoned because of HIV status

4. This presentation • Results from a systematic review and meta analysis on associations between IPV and HIV/STIs • Non Partner sexual violence • Other evidence • Perpetration studies • Child abuse analysis

5. Aims & methods of IPV and HIV/STI review Aims • Compile existing epidemiological evidence on the association between exposure to intimate partner violence (IPV) and HIV/STI infection- focus on physical and sexual violence Methods • Systematic review • Searches of Pubmed, Embase, Cinahl, other databases until Dec 1 2010- 20 electronic data bases screened as well as unpublished studies • > 3,000 abstracts screened • Identified all papers reporting associations with HIV/STI • Inclusion: any population, any definition of IPV, HIV/STI • Analysis stratified by study quality: • Prospective studies • High quality cross-sectional studies (biological outcome data, unexposed reference group) • Meta-analysis done to produce a pooled estimate

6. Results • 35 papers, describing 41 datasets with 121,479 participants, reporting 115 estimates included • 5 prospective datasets • 3 large studies with biological outcomes • 2 incident HIV, 1 incident STIs • 3 case-control datasets • 35 cross-sectional datasets • With biological outcome data AND unexposed reference groups • HIV: 12 datasets, 25 estimates • STI: 6 datasets, 6 estimates

7. Quality considerations to assess confounding and bias • Best quality – prospective studies (exposure of violence precede the outcome – HIV) • Biological outcomes vs self reports • Misclassification of exposure to violence • Type of violence measured i.e. physical / sexual alone vs physical and or sexual • Current & past partner (DHS )vs any partner ever • Control of confounding • Male partner variables (alcohol use, concurrent partners) • Women’s concurrent partners

8. Prospective studies find associations

9. Different analyses of same cross-sectional data have different findings depending on restricting of analysis to current partner women only /controlling of variables/ us of weighted data

10. Cross sectional studies and different forms of IPV considered: Biological HIV outcome PHYSICAL SEXUAL PHYSICAL AND/OR SEXUAL PHYSICAL AND SEXUAL Figure:. Cross-sectional studies. Pooled OR, biological data only, HIV outcome, where reference group is no physical or sexual violence

11. Non Partner Sexual Violence & HIV/STI • Associations from 3 studies included – (US & SA) • 3 studies reported on association between HIV and NPSV • 1 longitudinal study (SA); 1 case control; 1 cross sectional (US veteran studies)

12. Girl children abused in childhood have a higher HIV incidence: (Jewkes et al Child Abuse & Neglect, 2010)

13. Evidence from research with men: • Survey of adult men in Eastern Cape and KZN: perpetration of physical IPV was associated with elevated HIV prevalence in young men (aged <25): (Jewkes et al 2009) • > 1 episode of physical IPV aOR2.08(1.07, 4.06) • Similar findings from India: Men who have perpetrated IPV have elevated HIV prevalence • aOR for 1.91 (95%CI 1.11, 3.27) (Decker et al 2009)

14. Incidence and relative incidence of HIV infection in women exposed to forms of violence and inequity: (analysis provided by R Jewkes- based on SA study )

15. Growing evidence of a clustering of risk behaviours • Men who are abusive to their partners are also more likely to have: • Concurrent sexual partners • A sexually transmitted infection • Problematic use of alcohol • Refuse to use a condom • Clustering of risk linked to common underlying risk factors

16. Gender inequality & social norms condoning some use of violence Poverty & economic stresses • Social constructions • of masculinity • Problematic • alcohol use Reduced access to info & HIV services Low or inconsistent condom use Increased probability partner has HIV and/or STI Increased likelihood that woman is HIV infected Potential pathways of association between IPV & women’s risk of HIV • Early experiences or witnessing of violence RISK FACTORS FOR PERPETRATION OF INTIMATE PARTNER VIOLENCE Partner physically and/or sexually violent Child abuse sexual /physical Physical Sexual Partner has concurrent sexual partners Woman has concurrent sexual partners DETERMINANTS OF HIV RISK FROM PARTNER Genital trauma

17. Conclusions • Prospective studies show an association between physical and/or sexual IPV and incident HIV in South Africa • Prospective data also find association between sexual IPV & HIV in Uganda and sexual violence & STI in India • Cross-sectional data analysis find less consistent findings- but better quality studies show significant associations • Despite growing studies – quality of studies lead to mix results • Need Longitudinal studies – to answer causal questions • Measurement of exposure to violence • Control of confounding variables – male partner variables • Unclear how generalizable findings are across different epidemic settings • Pathways between IPV & HIV complex – need to be better understood to inform effective programmes

18. Priorities to improve evidence base… Identify opportunities to collect additional evidence from longitudinal studies • Take advantage of opportunities within ongoing intervention trials with HIV outcomes Make best use of DHS & other population data collection • Address methodological issues in DHS such as collecting data on violence exposure from all partners Integrate questions on violence in HIV intervention research • Provide deeper understanding of how violence and the fear of violence may undermine effectiveness of proven HIV interventions

19. Thank you