Kouyoumdjian, Fiona G.a; Calzavara, Liviana M.a; Bondy, Susan J.a; O’Campo, Patriciaa; Serwadda, Davidb; Nalugoda, Fredc; Kagaayi, Josephc; Kigozi, Godfreyc; Wawer, Mariad; Gray, Rond
Violence against women is a serious and common human rights and public health problem, which causes significant morbidity and mortality worldwide . Violence against women is a public health concern per se and as a risk factor for other adverse health outcomes, including unintended pregnancy, nonuse of contraception, poor pregnancy outcomes, gynecological morbidity, and sexually transmitted infections including HIV .
Intimate partner violence (IPV) is one form of violence against women, and is defined as ‘behaviour within an intimate relationship that causes physical, sexual, or psychological harm, including acts of physical aggression, sexual coercion, psychological abuse and controlling behaviours’ . IPV is prevalent internationally; the WHO Multi-Country Study on Women's Health and Domestic Violence identified lifetime prevalence of physical and/or sexual partner violence ranging from 15 to 71% and a past year prevalence between 4 and 54% , and these are likely underestimates of IPV prevalence as they do not include verbal violence or controlling behaviours.
There is evidence of an association between IPV and HIV from cross-sectional [4,5–14] and prospective studies [15,16]. However, several studies have also failed to find an association between IPV and HIV [17–20]. Interpretation of these findings is limited by use of cross-sectional data [4–8,10,11,13,14,17–20], measurement of only limited forms of IPV [4,5,7,8,13,15–20], the lack of population-based data [4–7,14,16–18], and inconsistent collection of data on and inappropriate analysis of potentially confounding and mediating variables [7,8,11–13,15] (e.g. treating a potential mediator as a confounder, which would lead to an underestimate of a true association).
There is also a lack of understanding of the direct or indirect mechanism for the association between IPV and HIV. Putative mechanisms for a direct effect include forced or coercive sex and associated physical trauma , which could increase susceptibility to HIV; women who experience violence may have limited ability to negotiate safer sex practices [4,21,22], which may lead to low rates of condom use; women who experience violence may subsequently engage in early sexual initiation, anal sex, commercial sex work, and unprotected sex with unfamiliar partners [21,23,24]; and the psychological stress of violence may affect susceptibility to HIV through immune suppression . Alternatively, IPV could be a consequence of infection with HIV if, for example, a woman's disclosure of HIV infection to her partner precipitated violence . IPV may also be an indirect marker of risk for HIV, as men who perpetrate violence may have higher rates of sexual risk behaviours, which increase the risk of exposure to their female partners .
Longitudinal data from the Rakai Community Cohort Study (RCCS) were used to determine the risk of incident HIV infection for women who had experienced IPV compared with women who had not experienced IPV, and to examine potential mediators of the association between IPV and HIV in Rakai, Uganda.
Since 1994, the Rakai Health Sciences Program has followed an open cohort of about 12 000 participants aged 15–49 years in 50 communities in the Rakai district of southwestern Uganda. The cohort has been described in detail elsewhere [26,27]. In brief, participants provide consent and are interviewed privately by interviewers of the same sex every 10–18 months, using a standardized questionnaire. Venous blood is collected for HIV-1 testing. More than 90% of eligible individuals participate in any given survey round.
Women who participated in the study between 2000 and 2009 were included in analyses if they were seronegative in 2000 or at baseline if they entered the cohort subsequent to 2000; participated in at least two surveys during the period under study; and were in a sexual relationship for all or part of the period under study. Once a woman became HIV-positive, further rounds of study participation were excluded from the analysis.
IPV was defined as any physical, sexual, or verbal violence by a partner in an intimate relationship. Questions on IPV were modified from the Revised Conflict Tactics Scales (CTS2), and, in some analyses, type of IPV was classified as minor or severe as per the CTS2 (as shown in Appendix 1, see online supplement, http://links.lww.com/QAD/A323), based on the putative harm resulting from IPV [28,29]. Data were collected in all survey rounds on experiences of IPV in the prior year, and in one round on experiences of all three forms of IPV ever. Data were collected on frequency of IPV in the past year for physical and verbal IPV in four rounds and for sexual IPV in three rounds. All data on IPV, as well as other survey data, were reported by women; no data from men were included in this study.
HIV infection was detected by two enzyme immunoassays (Vironostika HIV; Organon Teknika, Charlotte, North Carolina, USA, and Cambridge Biotech, Worcester, Massachusetts, USA), and confirmed by western blot (bioMérieux VITEK, St. Louis, Missouri, USA) or RT-PCR (Roche, Molecular Systems, Branchburg, New Jersey, USA). The time of HIV infection was assumed to be the round in which the participant first tested positive for HIV, unless data on HIV test results were missing from one or more immediately preceding rounds; for these participants, the time of infection was assumed to be the round at the midpoint between the last negative and first positive test.
Potential confounders included in analyses were age, marital status, education, religion, occupation, partner's occupation, type of relationship, length of relationship, and difference in age between the participant and her partner. Potential effect modifiers assessed were sexual abuse in childhood or adolescence and pregnancy intent. Childhood or adolescent sexual abuse was hypothesized to affect subsequent sexual risk behaviours. Pregnancy intent was hypothesized to modify the association between IPV and HIV as women who were trying to conceive would putatively be less likely to use condoms, or the dynamics relevant to the IPV–HIV association might differ in relationships in which women intend to conceive.
Potential mediators assessed were condom use and number of partners in the past year. Condom use was hypothesized to be a mediator if women who experience IPV may be unable to negotiate safer sex. We hypothesized that women with a history of IPV may be more likely to have more sexual partners, which could increase the risk of exposure to HIV.
We conducted longitudinal data analysis using a Poisson model with person-time as the offset and an exchangeable correlation matrix to assess the association between IPV and HIV, with each participant modeled as a random effect to account for correlation in repeated measures and given the unbalanced nature of the data. IPV was modeled by timing of exposure, by type and severity, and by frequency; no data on IPV were imputed for women who were missing data on IPV in certain rounds, and women who were missing specific data were excluded from relevant analyses. Covariates were considered as potential confounders by assessing associations with IPV and HIV infection, by strata of the covariate. Covariates which were associated with the exposure and outcome and which did not function as effect modifiers were included in a multivariable model, and considered for removal based on whether their removal changed the coefficient for the exposure-outcome association by more than 10–15% . Putative effect modifiers, sexual abuse in childhood or adolescence and pregnancy intent, were assessed in stratified analyses for multiplicative interactions .
To assess whether number of partners and condom use in the past year functioned as mediators of the association between IPV and HIV, we conducted longitudinal analyses to estimate the associations between the predictor and outcome, the predictor and mediator, the mediator and outcome while controlling for the predictor, and the predictor and outcome while controlling for the mediator, as per Baron and Kenny , using data from sequential rounds for the exposure, mediator, and outcome, respectively. Sobel–Goodman mediation tests were used to calculate the indirect effect and its standard error . Bootstrapping of estimates of the indirect effect was performed with 500 repetitions .
To assess for differential loss to follow up, rates of participation were calculated by exposure to IPV. Plausible estimates for HIV incidence in those lost to follow-up were used to assess the robustness of calculated estimates of the association between IPV and HIV, that is, if the incidence rates in the participants lost to follow-up were the same as the calculated rates in the exposed and unexposed. These values were selected based on the incidence rates of HIV in people who were not lost to follow-up (by history of exposure to IPV) and in consideration of the estimated incidence rates in the overall population in Uganda over this period . The assumptions were: 1.5 times the rate in the exposed and 1.25 times the rate in the unexposed, 1.25 times the rate in the exposed and 1.5 times the rate in the unexposed, 1.5 times the rate in the exposed and unexposed, and 0.75 times the rate in the exposed and unexposed.
The adjusted population attributable fraction was calculated , assuming that the association between IPV and HIV was causal. The adjusted incidence rate ratio (IRR) was obtained from a Poisson mixed methods model with individual treated as a random effect and controlling for relevant confounding variables. Confidence intervals (CI) were calculated using the 95% confidence intervals for the adjusted IRR. Analyses were done using Stata 11 (Stata Corp., College Station, Texas, USA).
Of 20 584 women in the study during this period, 10 252 (49.8%) met criteria for inclusion in these analyses. Compared with the total population of women, the sample women were older (54.4% were younger than 25 compared with 57.5%, P = 0.0001) and more likely to be married (68.7 compared with 55.1%, P < 0.0001), but not different in terms of education level (25.7% with less than 5 years of education compared with 25.3%, P = 0.68). The average number of years of follow-up was 5.7 for women who reported having experienced any abuse ever and 5.2 years for women who did not report any abuse. The baseline characteristics of women included in these analyses are shown in Table 1, by their experience of IPV ever.
IPV was prevalent, as shown in Table 2, with more than a quarter of participants reporting any IPV in the past year and more than half reporting any IPV ever. Verbal IPV was most common, followed by physical IPV and then sexual IPV. Most women experienced more than one form of IPV concurrently. In the last year of participation for the 2905 women who experienced IPV, 761 (26.2%) experienced only verbal violence, 57 (2.0%) only physical violence, and 266 (9.2%) only sexual violence, whereas 456 (15.7%) experienced both verbal and physical violence, 298 (10.3%) physical and sexual violence, and 165 (5.7%) verbal and sexual violence. Eight hundred and ninety-eight women (30.9%) experienced all three forms of violence.
As shown in Table 3, the HIV infection rate was higher for women who experienced IPV ever than for women who did not experience IPV, and these differences were statistically significant for any IPV, physical IPV, and verbal IPV.
Longitudinal data analyses adjusted for confounders revealed that experiencing IPV ever was consistently associated with an increased risk of incident HIV infection, as shown in Table 4. IPV in the past year was associated with HIV infection for all types of IPV except sexual IPV. The IRRs tended to be greater for models of longer periods of exposure, that is, for IPV ever as compared with IPV in the past year. IRRs for HIV infection were similar for sexual, physical, and verbal IPV, especially in models of IPV ever. There was a tendency for severe IPV to be associated with higher IRRs for HIV infection, though the differences between severe and minor IPV, as compared to no IPV, were not statistically significant. Finally, women who experienced more frequent IPV over the course of the study tended to have higher risk of HIV infection than women who experienced IPV less frequently, with an IRR of 0.84 (95% CI 0.47–1.51) for having experienced any IPV once, 1.49 (95% CI 0.87–2.55) for IPV twice, 1.48 (95% CI 1.00–2.20) for IPV between three and five times, 1.82 (95% CI 1.06–3.10) for IPV between six and 10 times, 1.38 (95% CI 0.57–3.34) for IPV between 11 and 20 times, and 3.03 (95% CI 1.83–5.01) for IPV more than 20 times in at least 1 year. A similar trend existed for IPV exposure in the past year, and no data were collected on the frequency of IPV ever.
Exposure to IPV ever prior to the round of infection was 62.7% in cases, and the adjusted IRR from a Poisson model for HIV infection of women who experienced IPV ever compared with women who never experienced IPV was 1.55 (95% CI 1.25–1.94). The adjusted population attributable fraction of HIV infection associated with any IPV ever was therefore 22.2% (95% CI 12.5–30.4).
There was evidence of modification of the effect of IPV on HIV by the experience of sexual abuse in childhood or adolescence. For those who had experienced childhood sexual abuse, the IRR of HIV infection was 2.50 (95% CI 0.90–6.97) for women who had ever experienced IPV compared with women who had never experienced IPV, whereas the IRR of infection for women who had not experienced childhood sexual abuse was 1.22 (95% CI 0.68–2.18). There was no evidence of effect modification by pregnancy intent.
Retention rates were 62.1% of follow-up rounds for which participants were eligible, with women who ever experienced IPV participating in 62.2% of rounds and women who never experienced IPV participating in 62.0% of rounds. Sensitivity analyses of plausible values for HIV incidence rates in participants who were lost to follow-up indicated that under most circumstances, the IRR for the complete study population would be similar to the calculated IRR for those in the study sample who were not lost to follow-up, ranging between 1.19 and 1.32 for the complete study population, as compared with 1.21 for those who were not lost to follow-up. The exception was for the scenario in which the rate in those exposed to IPV was 1.25 times the calculated rate and the rate in those not exposed to IPV was 1.5 times the calculated rate for those lost to follow up, that is, in which the increase in HIV incidence was relatively lower in those with IPV who were lost to follow-up compared with those with no IPV who were lost to follow-up; for this scenario, the IRR was 1.08.
Neither condom use nor number of partners in the past year functioned as a mediator. Using Baron and Kenny's criteria, the association between IPV and HIV did not change after controlling for either condom use or number of partners. Consistent with this, the Sobel–Goodman test revealed an IRR of 1.0 for an indirect pathway from IPV to HIV mediated through either condom use or number of partners.
This study found a significant association between IPV and subsequent incident HIV infection. Women who had ever experienced IPV had an adjusted IRR of 1.55 (95% CI 1.25–1.94) of acquiring HIV compared with women who had never experienced IPV. These results are consistent with two other prospective studies, which found adjusted IRRs of 1.51 (95% CI 1.04–2.21) in women who had experienced physical or sexual IPV  and of 1.62 (P = 0.35) in women and 1.69 (P = 0.31) in men who had experienced verbal or physical IPV , respectively. The adjusted population fraction of HIV attributable to IPV in this study, 22.2% (95% CI 12.5–30.4) is also similar to that identified in another prospective study, 11.9% (95% CI 1.4–19.3) , and indicates that IPV is responsible for a modest proportion of HIV infection.
These findings from the RCCS are noteworthy both because of their consistency with the other two prospective studies [15,16] and because this study addresses several limitations of these other studies. In the study by Jewkes et al. of women in South Africa, only physical and sexual IPV were measured, and in the study by Were et al. of HIV-serodiscordant couples in East and Southern African countries, only physical and verbal abuse were measured. In contrast, in the present analyses, physical, sexual, and verbal IPV were included. In the Jewkes et al. study, estimates of the risk of HIV infection were calculated for women who experienced IPV more than one time compared with those who experienced IPV one time or not at all. In light of the findings in this study of an increase in the risk of HIV infection with increasingly frequent exposure to IPV during the study, this categorization may underestimate the association between IPV and HIV. In the Were et al. study, there were very few reports of IPV in women who seroconverted (6.0% of 146 women), which may explain why the positive association between IPV and HIV infection was not statistically significant. Another issue in the Were et al. study was the relatively short period of follow-up (with a median of 18 months), especially as they did not gather data on the experience of violence ever.
In this study, the association between IPV ever and HIV infection was present and similar for different types of IPV, which suggests that it may be the experience of violence per se rather than a particular type of violence which increases risk of HIV infection, although there is overlap in the forms of violence experienced by most women in this study. The association between IPV and HIV tended to be greater for severe forms of violence compared with minor forms of violence. The association tended to be stronger for more frequent exposure to violence, as noted in another study . There was no evidence that condom use or number of partners mediated the association between IPV and HIV, which is consistent with the finding of another study, which showed no change in the magnitude of the association between IPV and HIV after controlling for each of these variables .
Strengths of this study are the longitudinal nature and size of the RCCS, and the inclusion of data on three types of violence (physical, verbal, and sexual) and of relevant confounders, mediators, and effect modifiers. Importantly, as the study is population-based, the findings are likely externally generalizable to other rural Ugandan populations and potentially elsewhere.
There are several limitations to this study. The lack of data on partner behaviors precludes determination of whether the increase in risk of HIV is due to increased risk of HIV infection in men who perpetrate IPV. Also, questions on ever having experienced IPV and on frequency of IPV were included only in specific surveys, which may limit power. As with much research on violence, there may have been underreporting due to stigma associated with IPV, that is, social desirability bias, however, various measures taken in the study procedures such as ensuring privacy and confidentiality of responses should have mitigated this. Further, since women are unlikely to know their HIV status until after they have completed the survey interview, any bias due to underreporting of IPV would likely be nondifferential and would therefore lead to an underestimate of the association between IPV and HIV. Another important limitation is the loss to follow-up, however, sensitivity analyses suggested that the association between IPV and HIV would not be significantly changed under most plausible scenarios for the HIV infection rates of those lost to follow-up.
Although it remains unclear whether IPV is a risk marker or causal antecedent of HIV infection, these prospective data suggest that women who have experienced IPV are at greater risk of HIV infection. There are two priorities that follow from this. First, HIV prevention programs should include discussion about IPV in HIV voluntary counseling and testing programs. This could lead to referring women who are experiencing IPV or who have experienced past IPV to counseling, legal assistance, and other community services, as well as consideration of female-controlled HIV prevention interventions, such as preexposure prophylaxis. Second, there is a strong imperative to prevent violence in childhood and IPV in adulthood, as a means of both stopping the psychological and physical consequences of IPV and potentially of preventing HIV.
F.K. developed the research question and protocol, conducted analyses, and drafted the manuscript. L.C., S.B., and P.O. contributed to the analysis and interpretation of data. R.G. was involved in conception and design of the study, acquisition of data, and analysis and interpretation of data. D.S., F.N., J.K., G.K., and M.W. were involved in conception and design of the study, and acquisition of data. All authors revised the article and approved the final draft.
Funding: The Rakai Community Cohort Study was supported in part by research grants through National Institute of Child Health and Human Development (grant F31HD063345) and National Institute of Allergy and Infectious Diseases (grants 5P30HD06826, U01AI075115) of the National Institutes of Health, Henry M. Jackson Foundation, a grant (5D43TW00010) from the Fogarty Foundation, and the Bill and Melinda Gates Institute for Population and Reproductive Health. FK received funding in the form of a Fellowship from the Public Health Agency of Canada and a Student Award from the CIHR Social Research Centre in HIV Prevention.
The views expressed herein do not necessarily represent the views of the funding sources.
Conflicts of interest
There are no conflicts of interest.
1. Ellsberg M, Jansen HA, Heise L, Watts CH, Garcia-Moreno C. Intimate partner violence and women's physical and mental health in the WHO multicountry study on women's health and domestic violence: an observational study
2. World Health Organization and London School of Hygiene and Tropical Medicine.Preventing intimate partner and sexual violence against women: Taking action and generating evidence.
Geneva World Health Organization; 2010.
3. Garcia-Moreno C. WHO Multi-Country study on women's health and domestic violence against women
. Geneva:World Health Organization; 2005.
4. van der Straten A, King R, Grinstead O, Serufilira A, Allen S. Couple communication, sexual coercion and HIV risk reduction in Kigali, Rwanda
5. Fonck K, Leye E, Kidula N, Ndinya-Achola J, Temmerman M. Increased risk of HIV in women experiencing physical partner violence in Nairobi, Kenya.[Erratum appears in AIDS Behav. 2007 Mar;11(2):337 Note: Els, Leye [corrected to Leye, Els]]
. AIDS & Behavior
6. Maman S, Mbwambo JK, Hogan NM, Kilonzo GP, Campbell JC, Weiss E, Sweat MD. HIV-positive women report more lifetime partner violence: findings from a voluntary counseling and testing clinic in Dar es Salaam, Tanzania
. Am J Public Health
7. Dunkle KL, Jewkes RK, Brown HC, Gray GE, McIntryre JA, Harlow SD. Gender-based violence, relationship power, and risk of HIV infection in women attending antenatal clinics in South Africa
8. Jewkes R, Dunkle K, Nduna M, Levin J, Jama N, Khuzwayo N, et al. Factors associated with HIV sero-status in young rural South African women: connections between intimate partner violence and HIV
. Int J Epidemiol
9. Chandrasekaran V, Krupp K, George R, Madhivanan P. Determinants of domestic violence among women attending an human immunodeficiency virus voluntary counseling and testing center in Bangalore, India
. Indian J Med Sci
10. Dude AM. Spousal intimate partner violence is associated with HIV and Other STIs among married Rwandan women
. AIDS Behav
11. Kayibanda JF, Bitera R, Alary M. Violence towards women, men's sexual risk factors and HIV infection among women: findings from a national household survey in Rwanda
. J Acquir Immune Defic Syndr
12. Silverman JG, Decker MR, Saggurti N, Balaiah D, Raj A. Intimate partner violence and HIV infection among married Indian women
13. Decker MR, Seage GR 3rd, Hemenway D, Raj A, Saggurti N, Balaiah D, Silverman JG. Intimate partner violence functions as both a risk marker and risk factor for women's HIV infection: findings from Indian husband-wife dyads
. J Acquir Immune Defic Syndr
14. Ntaganira J, Muula AS, Masaisa F, Dusabeyezu F, Siziya S, Rudatsikira E. Intimate partner violence among pregnant women in Rwanda
. BMC Womens Health
2008; 8:1–7.[Aailable at http://www.biomedcentral.com/1472-6874/8/17
15. Jewkes RK, Dunkle K, Nduna M, Shai N. Intimate partner violence, relationship power inequity, and incidence of HIV infection in young women in South Africa: a cohort study
16. Were E, Curran K, Delany-Moretlwe S, Nakku-Joloba E, Mugo NR, Kiarie J, et al. A prospective study of frequency and correlates of intimate partner violence among African heterosexual HIV serodiscordant couples
17. Prabhu M, McHome B, Ostermann J, Itemba D, Njau B, Thielman N. Prevalence and correlates of intimate partner violence among women attending HIV voluntary counseling and testing in northern Tanzania, 2005–2008
. Int J Gynaecol Obstet
18. Shamu S, Abrahams N, Temmerman M, Musekiwa A, Zarowsky C. A systematic review of African studies on intimate partner violence against pregnant women: prevalence and risk factors
. PLoS One
19. Harling G, Msisha W, Subramanian SV. No association between HIV and intimate partner violence among women in 10 developing countries
. PLoS One
20. Jewkes R, Sikweyiya Y, Morrell R, Dunkle K. The relationship between intimate partner violence, rape and HIV amongst South African men: a cross-sectional study
. PLoS One
21. Maman S, Campbell J, Sweat MD, Gielen AC. The intersections of HIV and violence: directions for future research and interventions
. Soc Sci Med
22. Wingood GM, DiClemente RJ. The effects of an abusive primary partner on the condom use and sexual negotiation practices of African-American women
. Am J Public Health
23. Zierler S, Feingold L, Laufer D, Velentgas P, Kantrowitz-Gordon I, Mayer K. Adult survivors of childhood sexual abuse and subsequent risk of HIV infection
. Am J Public Health
24. Zierler S, Witbeck B, Mayer K. Sexual violence against women living with or at risk for HIV infection
. Am J Prev Med
25. Balbin EG, Ironson GH, Solomon GF. Stress and coping: the psychoneuroimmunology of HIV/AIDS
. Baillieres Best Pract Res Clin Endocrinol Metab
26. Wawer MJ, Gray RH, Sewankambo NK, Serwadda D, Paxton L, Berkley S, et al. A randomized, community trial of intensive sexually transmitted disease control for AIDS prevention, Rakai, Uganda
27. Wawer MJ, Sewankambo NK, Serwadda D, Quinn TC, Paxton LA, Kiwanuka N, et al. Control of sexually transmitted diseases for AIDS prevention in Uganda: a randomised community trial. Rakai Project Study Group
28. Straus MA. Measuring intrafamily conflict and violence. The conflict tactics (CT) scales
. J Marriage Fam
30. Vittinghoff E, Glidden DV, Shiboski SC, McCulloch CE. Regression methods in biostatistics: linear, logistic, survival, and repeated measures models
. New York:Springer Science and Business Media Inc; 2005.
31. Szklo M, Nieto FJ. Epidemiology: beyond the basics
. Second ed.Sudbury, Massachusetts, USA:Jones and Bartlett Publishers, Inc; 2007.
32. Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations
. J Pers Soc Psychol
34. Shrout PE, Bolger N. Mediation in experimental and nonexperimental studies: new procedures and recommendations
. Psychol Methods
36. Rothman KJ, Greenland S, Lash TL. Modern epidemiology
. 3rd ed.Philadelphia:Lippincott Williams & WIlkins; 2008.
© 2013 Lippincott Williams & Wilkins, Inc.