Scholarly article on topic 'HIV Incidence and Predictors of HIV Acquisition From an Outside Partner in Serodiscordant Couples in Lusaka, Zambia'

HIV Incidence and Predictors of HIV Acquisition From an Outside Partner in Serodiscordant Couples in Lusaka, Zambia Academic research paper on "Health sciences"

Share paper
OECD Field of science

Academic research paper on topic "HIV Incidence and Predictors of HIV Acquisition From an Outside Partner in Serodiscordant Couples in Lusaka, Zambia"


JAIDS Journal of Acquired Immune Deficiency Syndromes Publish Ahead of Print DOI: 10.1097/QAI.0000000000001494

HIV incidence and predictors of HIV acquisition from an outside partner in serodiscordant couples in Lusaka, Zambia, 1995-2012

Dvora L JOSEPH DAVEY3, g, Kristin M WALL a,b, William KILEMBEa, Htee Khu NAWa, Ilene BRILLad, Bellington VWALIKAa, e, Elwyn CHOMBA af, Joseph MULENGAa, Amanda TICHACEKa, Marjan Javanbaktg, Pamina GORBACHg, Susan ALLENa

Running head: Unlinked HIV infection in couples

a. Rwanda Zambia HIV Research Group, Department of Pathology & Laboratory Medicine, School of Medicine and Hubert Department of Global Health, Rollins School of Public Health, Emory University, Atlanta, GA, USA

b. Department of Epidemiology, Rollins School of Public Health, Laney Graduate School, Emory University, Atlanta, Georgia, USA

c. Department of Gynecology and Obstetrics, Emory University, School of Medicine, Atlanta, Georgia, USA

d. Department of Epidemiology, Ryals School of Public Health, University of Alabama at Birmingham, Birmingham, Alabama, USA

e. Department of Gynecology and Obstetrics, School of Medicine, University of Zambia, Lusaka, Zambia

f. Ministry of Community Development, Mother and Child Health, Lusaka, Zambia

g. Department of Epidemiology, Fielding School of Public Health, UCLA, Los Angeles, California, USA

Corresponding author and author

Dvora Joseph Davey, PhD, MPH 650 Charles E. Young Drive South Los Angeles, CA 90095 Email: Telephone: +001 310-701-1526

Funding: This study was supported by the National Institutes of Child Health and Development (NICHD R01 HD40125); National Institute of Mental Health (NIMH R01 66767); National Institute of Allergy and Infectious Diseases (NIAID R01 AI51231; NIAID R01 AI040951; NIAID R01 AI023980; NIAID R01 AI64060; NIAID R37 AI51231); the AIDS International Training and Research Program Fogarty International Center (D43 TW001042); the Emory Center for AIDS Research (P30 AI050409); the US Centers for Disease Control and Prevention (5U2GPS000758); and the International AIDS Vaccine Initiative. This study was made possible by the generous support of the American people through the United States Agency for International Development (USAID). The contents are the responsibility of the International AIDS Vaccine Initiative and do not necessarily reflect the views of USAID or the United States Government. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

The authors report no conflicts of interest related to this work.

This is an open-access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.


Objective: Evaluate the incidence and predictors of HIV acquisition from outside partners in serodiscordant couples.

Methods: Demographic, behavioral, and clinical exposures were measured quarterly in a cohort of serodiscordant cohabiting couples in Zambia from 1995-2012 (n=3049). Genetic analysis classified incident infections as those acquired from the study partner (linked) or acquired from an outside partner (unlinked). Factors associated with time to unlinked HIV infection were evaluated using multivariable Cox proportional hazards regression stratified by gender.

Results: There were 100 unlinked infections in couples followed for a median of 806 days. Forty-five infections occurred in women (1.85/100 couple-years [CY]; 95%CI:1.35-2.47). Risk of female unlinked infection (vs. non-seroconverting females) was associated with reporting being drunk weekly/daily vs. moderate/non-drinkers at baseline (aHR=5.44;95%CI:1.03-28.73), genital ulcers (aHR=6.09;95%CI:2.72-13.64) or genital inflammation (aHR=11.92;95%CI:5.60-25.37) during follow-up adjusting for age, years cohabiting, income, contraceptive use, previous pregnancies, history of sexually transmitted infections (STI), and condomless sex with study partner. Fifty-five infections occurred in men (1.82/100 CY;95%CI:1.37-2.37). Risk of male unlinked infection was associated with genital inflammation (aHR=8.52;95%CI:3.82-19.03) or genital ulceration (aHR=2.31;95% CI:2.05-8.89), reporting =>1 outside sexual partner (aHR=3.86;95%CI:0.98-15.17) during follow-up, and reporting being drunk weekly/daily vs. moderate/non-drinkers at baseline (aHR=3.84;95%CI=1.28-11.55), controlling for age, income, circumcision status, and history of STI.

Conclusions: Predictors of unlinked infection in serodiscordant relationships were alcohol use, genital inflammation, and ulceration. Causes of genital inflammation and ulceration should be screened for and treated in HIV-negative individuals. Counseling on risk of alcohol use and sex with outside partners should be discussed with couples where one or both are

HIV-negative, including in counseling on use of pre-exposure prophylaxis to prevent HIV acquisition in the HIV-negative partner.

Keywords: serodiscordant; unlinked infection; couples; Zambia; HIV transmission; HIV acquisition Background

HIV incidence in studies of HIV-serodiscordant couples ranges from 2.0 to 11.8 per 100 person-years depending on the whether they know their joint serostatus, the type of study, the study location, and the accompanying services made available to the couples [1, 2, 3]. Among couples with HIV infections acquired from a partner outside the primary serodiscordant partnership, as indicated by viral sequencing, incidence rates range from 0.51 per 100 person years among women reporting no outside sex partners, to 3.46 per 100 person years among women reporting outside sex partners [4]. Though uptake of antiretroviral treatment (ART) among HIV-serodiscordant couples has been recommended where feasible, it does not remove the need for behavioral interventions to reduce HIV acquisition from outside partners [4-6]. Several African countries are adopting couples-based HIV counseling and testing as an HIV prevention and care intervention, thus more couples are becoming aware of each other's HIV status, and many find out that they are HIV-serodiscordant [7]. Despite this increased knowledge of HIV-serodiscordance among couples, there has been less focus on effective messages and interventions to prevent HIV acquisition from outside partners. Appropriate and effective interventions for HIV prevention in serodiscordant couples depend on accurate assessments of the sexual risk behaviors among the HIV-positive and HIV-negative partners, as well as understanding their own sexual behaviors with other men and women.

Improved evidence around factors that increase the risk of HIV infection in cohabiting relationships is needed in order to develop and improve evidence-based interventions. Specifically, more research is needed on factors with a disputed and unclear influence on HIV acquisition and transmission at the individual- and couple-level, including hormonal contraception [8-10, 30], concurrent sexual partnerships [11-16], alcohol [16-19, 43], and age discordancy [2022]. Studies among African serodiscordant couples have demonstrated that HIV acquisition is associated with younger age, alcohol use, genital inflammation and discharge, and non-ART use by HIV-positive partner [19, 22, 23]. However, there is limited research on factors associated with acquiring HIV from an outside partner, or an 'unlinked infection'. Learning that one is in an HIV-serodiscordant relationship may cause the HIV-negative partner to have sex with outside

partners, often of unknown HIV serostatus, who might be perceived to pose lower risk of HIV infection than the primary HIV-positive partner [1, 3, 4]. Greater understanding of HIV-negative partners' behaviors with outside or concurrent sex partners, and these behaviors' associations with HIV acquisition, will inform HIV risk-reduction messages for uninfected partners.

A recent study of serodiscordant couples found that the proportion of subjects reporting sex with an outside partner increased from 3.1% to 13.9% (p<0.001) after learning of their serodiscordant results [3]. Condomless sex was more common with outside partners than with the primary HIV-positive partners, and a small portion (<5%) reported concurrent sexual partnership with their infected partner and an outside partner within the same month [3]. Another study showed that sex with infected partners decreases during follow-up and increases with outside partners over time, possibly reflecting relationship dissolution [24]. A recent Kenyan study found 24% of serodiscordant couples separated during 2-years of follow-up [24]. However, in our Zambian cohort, only 3.7% separated shortly after CVCT [25] and the annual rate of separation thereafter was <5%. This study examined the incidence of unlinked infections, and risk factors

prospective cohort which enrolled 3,049 adult heterosexual serodiscordant couples recruited from a couples' counseling and testing site in Lusaka, Zambia. Data collection began in January 1995 and ended in December 2012. Couples were enrolled following couples' HIV counseling and testing services and their HIV serodiscordant diagnosis.

Ethics. Couples provided joint written informed consent to participate in this study. This study was approved by the Office for Human Research Protections-registered Institutional Review Boards (IRB) at Emory and the University of Zambia.

Study participants. Married or cohabiting couples in Lusaka, Zambia, attended voluntary couples' HIV counseling and testing (CVCT) services, either spontaneously or after receiving an invitation from a community promoter [26, 27]. CVCT

services include group counseling, rapid HIV testing, and post-test couples' counseling [27, 28]. HIV-serodiscordant couples were invited to enroll in a longitudinal open cohort follow-up study between 1995 and 2012. Routine outpatient care and family planning were provided at the research clinic. Beginning in 2007 when ART became available in government health clinics, HIV-positive individuals were referred to the facility nearest to their residence for ART initiation, which was prescribed based on national guidelines in force at the time.

Study eligibility. The cohort's inclusion criteria were: (1) confirmed HIV-1 serodiscordance after attending CVCT services, (2) the participants had been married or cohabitating for at least three-months, and (3) the participants planned on staying in the Lusaka region for the next year. Couples were ineligible if either partner was on ART. Couples were censored if either partner died, the couple separated, the HIV-positive partner initiated ART, or either partner was lost to follow-up. Couples were considered lost to follow up if they did not return for two consecutive quarterly study visits

Data collection. Study questionnaires were administered by study nurses or counselors at baseline and every subsequent visit (monthly or quarterly). Study participants completed behavioral and medical history questionnaires in the local language and underwent a full physical examination, including pelvic/genital exams, as well as HIV counseling and testing, and screening for sexually transmitted infections (STI) at baseline. Follow-up visits included physical exam, vaginal swab wet preparation, repeat Rapid Plasma Reagin (RPR) screening, repeat HIV testing of the HIV-negative partner, and completion of study questionnaires that collected demographic, psychosocial, behavioral, medical history, and health-services data. The data were collected by interviewer-administered questionnaires in face-to-face interviews. Couples were given coital diaries to keep and record sexual exposures with and without condoms. Coital diaries were used to validate responses about sex frequency and condom use and non-use

Data collection varied over the 18 years of follow-up. In 1999, the study began plasma banking for viral load testing, and in 2003 p24 antigen ELISA screening began. From 1995-2003 both partners were seen quarterly, had routine physical and genital exams, and received laboratory screening for HIV (rapid antibody test) trichomonas (vaginal wet prep) and syphilis (RPR: Positive RPR were batched for later TPHA, confirming a low rate of false positive RPR) [23, 29]. In 2003, the study procedures were streamlined and physical exams and STI screenings were performed at baseline, annually, and when participants reported symptoms. Vaginal wet mounts were prepared using self-administered swabs at every visit that did not include a gynecologic exam (validated at the time of use in the study). From 2002-2011, fertility intentions were

recorded. From 2007-2008, HIV-partners came in monthly for the first three-months, and quarterly thereafter at which time a sexual exposure risk assessment including self-reported unprotected sex or incident STI (in men and women); or sperm (a biomarker of condomless sex in women) or trichomonas on a wet mount and incident pregnancy (for women) was completed. HIV-negative partners in couples with at least one exposure received monthly HIV testing until the next quarterly visit, when the risk assessment was repeated. From 2009-2012, HIV-negative partners were tested monthly for HIV. We did not collect data on HIV status or gender of outside partner and assumed that outside partners were heterosexual.

Table 1: Data collection and laboratory test by year of collection

Year of data collection Data that was collected Laboratory tests used

1995-2003 Both partners seen quarterly with physical and genital exams with lab screening for trichomonas and syphilis RPR, TPHA for syphilis Vaginal wet prep for trichomonas

1999 Viral load testing began Viral load

2003 p24 antigen ELISA screening began STI screenings done at baseline, annually and if reported symptoms. Vaginal wet mounts at every visit that did not include a gynecological exam.

2007-8 HIV- partners came in monthly for first 3 months then quarterly thereafter

2009-2012 HIV- partners were tested monthly for HIV

ed that outside par >

Genital ulceration and inflammation. Measures of chronic/recurrent or acute genital or perianal ulcers were assessed via self-report, physical exam (including erosion or friability of the cervix or vagina in women), chart review of outpatient treatment visits and laboratory testing. A time-varying composite genital ulcer variable was developed and considered positive if the patient reported ulcers, had ulcers noted on physical exam, had been treated for ulcers, or had a newly positive RPR [23].

Genital inflammation measures included in the time-varying composite included physical exam findings including cervical, vaginal or external genital inflammation (redness, swelling, exudate, irritation, or tenderness) or inguinal adenopathy (80% of which was bilateral adenopathy); candida or bacterial vaginosis (BV) on vaginal wet mount including KOH prep, sniff test; and reported vaginal discharge, odor, or discomfort; self-report or treatment for

gonorrhea/chlamydia (diagnosed syndromically based on endocervical pus in women or urethral discharge in men) or trichomonas (vaginal wet mount) in the last 3-months. [23, 30, 31];

HIV testing. To define the time of incident infection, when available, plasma from the last antibody-negative sample was tested with RNA polymerase chain reaction (PCR, Roche). The date of HIV infection was defined based on available laboratory results as the minimum of: the midpoint between the last negative and first positive antibody date; three weeks prior to the first p24 antigen positive test date; or two weeks prior to the first positive HIV viral RNA/rapid test negative test date.

The molecular epidemiology of the HIV incident transmission events that occurred during study follow-up was reported by examining the genetic characterization of HIV viral strains. Comparison of PCR-amplified HIV nucleotide sequences in peripheral blood mononuclear cell DNA samples collected from both partners at the time of seroconversion classified incident infections as linked or unlinked to the study partner.

To establish linkage for HIV-1 viral strains in transmission pairs, amplified viral sequences encoding the extracellular domain of gp41 were first aligned and then subjected to pairwise sequence comparisons. Uncorrected nucleotide sequence distances were then calculated for each transmission pair and compared to the mean sequence distance calculated for a reference set of gp41 sequences from 81 unrelated Zambian HIV-1 infected individuals from the same Zambian cohort of serodiscordant couples [32]. The latter value (9.3%) minus two standard deviations (5.5%) was arbitrarily assigned as the maximum diversity cutoff value for epidemiologically linked sequence pairs. Transmission pairs were tentatively classified as having epidemiologically linked viruses when their pairwise sequence distances fell below this limit, and unlinked viruses when their pairwise distances exceeded this limit. Transmission linkage was confirmed by phylogenetic analysis using a neighbor-joining tree of the same gp41 sequences.

Analyses. HIV incidence rates were calculated as the number of unlinked incident infections per 100 couple-year (CY) of follow-up, excluding those couples with a linked infection in the numerator but not the denominator, as they were at risk of HIV acquisition from a study partner throughout the exposure period. We used all follow-up intervals rather than follow-up intervals with reported outside exposure to report a cumulative measure of exposure to outside partners. CYs of follow-up were calculated from enrollment until either an outcome of interest occurred or the couple was censored. Rates were also calculated by calendar time, duration of follow-up time and by gender of seroconverted. Baseline exposures

were stratified by gender of the HIV-negative partner. Baseline and time-varying exposures were described using counts and percentages (for categorical variables) or means and standard deviations (for continuous variables). For descriptive analyses, we tested all differences using chi-square tests for bivariate variables and t-tests for continuous variables. We included other variables with p<0.10 because they were a priori exposures of interest (e.g. alcohol). We conducted sensitivity analyses around heavy alcohol use and grouped weekly and daily drinkers in the final model and moderate and non-drinkers in the reference group to provide enough power to evaluate the effect of heavy alcohol use. We then included variables with p<0.10. Bivariate hazard ratios, p-values, and 95% confidence intervals between covariates and unlinked infections are based on bivariate Cox proportional hazards regression [44].

Four multivariable Cox proportional hazards regression models were constructed to evaluate predictors of time to acquisition of genetically unlinked HIV infection among (1) female and (2) male partners, compared to no acquisition of infection. We included all variables that were significant in bivariate analyses (p<0.05), and were also a priori risk factors for HIV acquisition [2, 3, 10, 11, 18, 20, 22]. We included the following interaction terms in our analyses:, age and literacy, HIV status and condom use, none of which were significant (p<0.05) in the models. The proportional hazards assumption was confirmed for time-independent covariates. Because we identified effect-measure modification for gender of seroconverter, stratified analysis was used for all subsequent models. Multi-collinearity was assessed by analyzing the variance inflation factor, and tolerance to check the degree of collinearity between variables. For collinear variables such as number of children and number of previous pregnancies, we selected one variable (e.g., number of previous pregnancies in this example), which was more strongly associated with the outcome of the model. All analyses were conducted with SAS v9.4 (Cary, NC).


The primary analyses comprised a total of 3,049 couples. In these couples, there were 478 seroconversions, 100 of which were determined to be acquired from an outside partner (i.e. their virus was not genetically linked to the partner's virus: 45 women, 55 men). The remaining 354 individuals had a linked seroconversion (i.e. their virus was genetically linked to the study partner's virus), or were unknown linkage (n=24), and were excluded from the analyses here. There were no seroconversions among the remaining 2,571 couples. Couples with an unlinked infection were followed for a median of 806 days. The unlinked HIV incidence rate among women was 1.32 per 100 couple years (CY) from 1995 to 2002, and it

rose to 2.21 per 100 CY from 2003 to 2012 though this difference was not significant. Similarly, the male unlinked incidence was 1.42 per 100 CY from 1995 to 2002 with a non-significant increase to 2.01 per 100 CY from 2003 to 2012.The incidence rate was similar for patients followed for the first 12 months was similar to the rate for participants followed for more than 12 months and beyond for men and women (Figure 1) (p>0.05), confirming no cohort effect.

Unlinked infections in women. Forty-five genetically unlinked seroconversions were observed in women over 2432.79 CY of M+F- couple follow-up, for a seroconversion rate of 1.85 seroconversions/100 CY (95% CI: 1.35, 2.47). Unless specified, the following differences are statistically significant at p<0.05. HIV-negative women who did not seroconvert were older (28.8 vs, 25.5 compared to women who acquired HIV from an outside partner), had lived with partners more years (mean of 8.3 years vs. 5.6 years respectively) and had more children (mean of 3.7 previous pregnancies vs. 2.7 respectively) than those who did not seroconvert. Monthly mean family income was lower in couples in which the woman had an unlinked seroconversion ($50.9 vs. $94.0/month, respectively). The proportion of women who read Nyanja (a measure of education) was lower among those with an unlinked infection compared with those who did not seroconvert (11% vs. 24%, p=0.06). Women with an unlinked infection reported more heavy alcohol drinking (drunk weekly or daily/almost daily) at baseline (24% vs. 17%, p=0.052). In 1% (n=157) of intervals non-seroconvertors reported having an outside sex partner during the study compared to 4% (n=2) of women who reported having an outside partner the interval before they acquired HIV from an outside partner (p=0.054). Women with an unlinked infection were more likely to report a history of STI in the year prior to enrollment (42% vs. 26%). In addition, women with genital inflammation in the interval before seroconversion had increased risk of having an unlinked infection compared to non-seroconverting women (HR=15.46, 95% CI=7.91, 30.20). Similarly, women with genital ulceration in the interval before seroconversion had increased risk of unlinked infection compared to non-seroconverting women (HR=8.05, 95% CI=3.95, 16.40). Women who used oral contraceptive pills had an increased risk of compared to non-seroconverting women (HR=3.07, 95% CI=1.36, 6.95) (Table 1).

Controlling for woman's age, years cohabiting, couple income, contraceptive method use at baseline, number of previous pregnancies, self-reported STI in the year prior to baseline, and condomless sex with study partner, women's risk of acquiring HIV from an outside partner, when compared with non-seroconverters, was associated with the woman reporting being drunk weekly or daily/almost daily in the past year at baseline (vs. moderate/non-drinkers) (adjusted

hazard ratio [aHR]=5.44; 95%CI:L03, 28.73), time-varying genital ulcers (aHR=6.09; 95% CI=2.72, 13.64), and genital inflammation (aHR=11.92; 95% CI=5.60, 25.37). However self-reporting =>1 outside partner and hormonal contraceptive use were not independently associated with incident unlinked infection among women (aHR=2.07, CI=0.42, 10.28) (Table 3).

Unlinked infections in men. Fifty-five genetically unlinked incident seroconversions were observed in men over 3025.40 CY of follow-up, for a seroconversion rate of 1.82 seroconversions/100 CY (95% CI: 1.37, 2.37). Unless specified, the following differences are statistically significant at p<0.05. In descriptive analyses, men who had an unlinked seroconversion tended to be younger than men who did not seroconvert (mean age of 32.8 vs. 35.4) and had a lower monthly mean income ($68.3 vs. $95.6). Men with an unlinked infection were more likely to report heavy alcohol use in the past year at baseline (drunk daily or almost daily) (73% vs. 54%) and to self-report having a STI in the year prior to enrollment (42% vs. 25% of men who did not seroconvert). In addition, men who had genital inflammation in the interval before seroconversion had increased risk of an unlinked infection compared to non-seroconverting men (HR=13.75, 95% CI=6.76, 27.96). Similarly, men with genital ulceration in the interval before seroconversion had increased risk of an unlinked infection compared to non-seroconverting men (HR=6.47, 95% CI=3.42, 12.25). Approximately 21% of men with an unlinked infection reported having outside partner in the interval before they seroconverted, compared with 8% of intervals among non-seroconverting men (HR=3.75, 95% CI=1.86, 7.54) (Table 2)

Controlling for man's age, couple income, male circumcision (before or during the study), and history of STI in past year at baseline, the risk of unlinked incident HIV infection among men was associated with male alcohol use (reporting being drunk weekly or daily vs. moderate and non-drinkers in the year before enrollment) (aHR=3.84, 95% CI=1.28, 11.55); and with genital ulcers (aHR=2.31, 95% CI=2.05, 8.89), genital inflammation (aHR=8.52, 95% CI=3.82, 19.03), and reporting =>1 outside sex partner in the past 3-months (aHR=3.86, 95% CI=0.98, 15.17) (Table 3).


This study contributes to the understanding of acquisition of HIV from concurrent sex partners among uninfected partners in cohabiting discordant couples in Southern Africa [3, 33]. We observed a genetically unlinked HIV incidence rate of 1.85/100 CY among women, and of 1.82/100 CY among men in this large 18-year prospective study. Overall, 21% of seroconversions were genetically unlinked during the 18-year study follow up, with a higher proportion among men

(24%) compared with women (18%). Our unlinked infection incidence rate is similar to other studies in the region [3], demonstrating the importance of outside partners as a risk for HIV acquisition in generalized HIV epidemics. Our study also demonstrated that factors associated with unlinked infection are similar to those reported for HIV acquisition more broadly: heavy alcohol use among both men and women [18, 43], and genital discharge, ulceration and inflammation [1, 19, 23] were important predictors of unlinked seroconversion. Studies of the newly infected partner in this cohort also confirm that genital inflammation is associated with transmission of multiple variants from outside partners [34, 35]. Similar to linked infections in this cohort [30, 36], after controlling for potential confounders, we did not find that pregnancy (prevalent or incident) or contraceptive use were risk factors for unlinked seroconversion.

A greater proportion of new infections were acquired outside the marriage in men compared with women (24% vs. 18% respectively). This may reflect an increased social or cultural acceptance of men having outside sexual partnerships, particularly once they know their main partner is HIV-positive. This may also reflect travel for work, which was not measured but is more likely among men compared with women [12-16]. Self-reporting of outside partnerships was low and not associated with unlinked seroconversion among female partners, likely the result of underreporting. This highlights the challenges of self-reports collected by face-to-face interviews, especially among women, when aiming to identify individuals at risk from outside partnerships.

Approximately 55-60% of Zambian adults requiring treatment were on ART in 2013 [37]. Zambia has adopted the WHO-recommended approach to couples in whom one person is HIV-positive by offering treatment regardless of their CD4 cell count [38]. However, the impact of ART as prevention in real world settings is mitigated by delays in initiation, poor retention and adherence [39]. Therefore, primary prevention strategies must begin with widespread implementation of CVCT and be followed by ART adherence counseling for the HIV-positive partner. Prevention of HIV exposure through concurrent partnerships in the HIV-negative partner should include emphasis on the fact that partners with an unknown serostatus still present a risk of acquiring HIV.

Recent research on how to deliver pre-exposure prophylaxis (PrEP) to HIV-negative partners in discordant couples has focused on time-limited PrEP acting as a "bridge" for approximately 6-months until the HIV-positive partner is on ART and virally suppressed [45]. However, in the context of a large proportion of HIV-negative partners having outside sex partners, PrEP use should continue due to continued high risk exposure. PrEP counseling should be adapted in

this case for HIV-negative partners who may have other sex partners of unknown serostatus. In addition, health care providers should focus on counseling and services for alcohol abuse for HIV-positive and -negative individuals, as our study confirmed what other studies have shown about the risk of HIV acquisition among men and women who reported heavy alcohol use [40, 43].

The unlinked HIV incidence rate was higher from 2003 to 2012 compared with 1995-2002 among both men and women though this difference was not significant. In contrast, during this time frame linked seroincidence rates decreased [23]. It is not clear what may have contributed to these findings, which reinforce the need to emphasize protective behaviors with concurrent as well as primary partner during counseling [28, 31].

Limitations. Though couples were demographically similar to the general population using DHS data [41], they were different from the general population and other serodiscordant couples in that they self-selected into both couples' counseling/testing and the cohort study which may reflect higher health-motivation and a greater commitment to safer sexual practices [25]. Use of self-reported information on sensitive exposure variables such as heavy alcohol use or outside sex partners may have low sensitivity, and this information bias may be differential by the outcome of interest, biasing our results in an unknown direction. The use of audio computer assisted self-interviews (ACASI) or other types of self-data collection could have improved the collection of sensitive questions such as alcohol and condom use and outside sex partners. Couples were censored if the HIV+ participant started ART, which may have over-estimated the proportion of participants who acquire HIV from their study partner (as they are not on ART, and most likely will not be virally suppressed), and under-estimate the proportion of participants who acquire HIV from an outside partner. Couples were followed-up every 3-months; however, that amount of time could lead to recall bias on certain measures that may not be salient to the participants. In addition, we did not collect data on condom use with outside partners. As a result, there may be residual confounding in our analysis. Since sex with outside partners was under-reported, we used all follow-up intervals rather than follow-up intervals with reported outside exposure in our cumulative measure of exposure to outside partners.. As a result, our analysis misclassified the exposures as we included person-time among couples who had not reported any partners outside the serodiscordant relationship, likely underestimating the incidence rates for those exposed to outside partners.


Our study found that HIV acquisition from outside partners of serodiscordant couples is high enough to warrant further attention in research and prevention efforts. Factors associated with unlinked infection among men and women included heavy alcohol use, genital inflammation and ulceration, and reporting having outside sex partners among men. Greater understanding of associations between HIV-negative partners' behaviors with outside partners and HIV acquisition will be important for the design of HIV prevention strategies for uninfected partners. Considering the continued HIV exposure from concurrent and outside sex partners, counseling on time-limited PrEP as a "bridge" until the HIV-positive partner is virally suppressed may need to be adapted to allow for longer PrEP use for uninfected partners who may have outside sex partners. In addition, our findings provide further evidence that HIV-negative partners in serodiscordant couples should be regularly screened for symptomatic and, when possible, asymptomatic STIs and other non-sexually transmitted causes of genital inflammation in women. Screening and prophylactic treatment will help reduce genital inflammation and ulceration, which will reduce the risk of HIV acquisition. We also bring attention to the role of heavy alcohol use by males and females in acquisition of HIV from outside partners.

Int AIDS Soc 2014; 17: 18765.

2. Chemaitelly H, Awad SF, Abu-Raddad LJ. The risk of HIV transmission within HIV sero-discordant couples appears to vary across sub-Saharan Africa. Epidemics 2014; 6: 1-9.

3. Kamali A, Price MA, Lakhi S, et al. Creating an African HIV Clinical Research and Prevention Trials Network: HIV Prevalence, Incidence and Transmission. Barnabas RV, ed. PLoS ONE. 2015;10(1):e0116100. doi:10.1371/journal.pone.0116100.

4. Ndase P, Celum C, Thomas K, et al. Partners in Prevention HSV/HIV Transmission Study Team. Outside sexual partnerships and risk of HIV acquisition for HIV uninfected partners in African HIV serodiscordant partnerships. J Acquir Immune Defic Syndr 2012; 59(1): 65-71.

5. Jia Z, Mao Y, Zhang F, et al. Antiretroviral therapy to prevent HIV transmission in serodiscordant couples in China (2003-11): a national observational cohort study. Lancet 2013; 382(9899): 1195-1203.

6. Ondoa P, Gautam R, Rusine J, et al. Twelve-Month Antiretroviral Therapy Suppresses Plasma and Genital Viral Loads but Fails to Alter Genital Levels of Cytokines, in a Cohort of HIV-Infected Rwandan Women. PLoS One. 2015 May 26;10(5):e0127201.

7. Karita E, Nsanzimana S, Ndagije F, Wall KM, Mukamuyango J, Mugwaneza P, Remera E, Raghunathan PL, Bayingana R, Kayitenkore K, Bekan-Homawoo B, Tichacek A, Allen S. Implementation and Operational Research: Evolution of Couples' Voluntary Counseling and Testing for HIV in Rwanda: From Research to Public Health Practice. J Acquir Immune Defic Syndr. 2016 Nov 1;73(3):e51-e58. PubMed PMID: 27741033; PubMed Central PMCID: PMC5367509.

8. Polis CB, Curtis KM. Use of hormonal contraceptives and HIV acquisition in women: a systematic review of the epidemiological evidence. Lancet Infect Dis 2013; 13(9): 797-808.

9. Polis CB, Phillips SJ, Curtis KM. Hormonal contraceptive use and female-to-male HIV transmission: a systematic review of the epidemiologic evidence. Aids 2013; 27(4): 493-505.

10. Haddad LB, Polis CB, Sheth AN, et al. Contraceptive methods and risk of HIV acquisition or female-to-male transmission. Curr HIV/AIDS Rep 2014.

11. Tanser F, Barnighausen T, Hund L, et al. Effect of concurrent sexual partnerships on rate of new HIV infections in a high-prevalence, rural South African population: a cohort study. Lancet 2011; 378(9787): 247-55.

12. Shelton JD. A tale of two-component generalised HIV epidemics. Lancet 2010; 375(9719): 964-966.

13. Mah TL, Shelton JD. Concurrency revisited: increasing and compelling epidemiological evidence. J Int AIDS Soc 2011; 14: 33.

14. Mah TL, Halperi DT. Concurrent sexual partnerships and the HIV epidemics in Africa: evidence to move forward. AIDS Behav 2010; 14(1): p. 11-6; dicussion 34-7.

15. Halperin DT, Epstein H. Concurrent sexual partnerships help to explain Africa's high HIV prevalence: implications for prevention. Lancet 2004; 364(9428): 4-6.

16. Seeley J, Nakiyingi-Miiro J, Kamali A, et al.; CHIVTUM Study Team. High HIV incidence and socio-behavioral risk patterns in fishing communities on the shores of Lake Victoria, Uganda. Sex Transm Dis 2012; 39(6): 433-439.

17. Hahn JA, Woolf-King SE, Muyindike W. Adding fuel to the fire: alcohol's effect on the HIV epidemic in Sub-Saharan Africa. Curr HIV/AIDS Rep 2011; 8(3): 172-180.

18. Coldiron ME, Stephenson R, Chomba E, et al. The relationship between alcohol consumption and unprotected sex among known HIV-discordant couples in Rwanda and Zambia. AIDS Behav 2008; 12(4): 594-603.

19. Ruzagira E, Wandiembe S, Abaasa A, et al. HIV incidence and risk factors for acquisition in HIV discordant couples in Masaka, Uganda: an HIV vaccine preparedness study. PLoS One 2011; 6(8): e24037.

20. Harling G, Newell ML, Tanser F,et al. Do age-disparate relationships drive HIV incidence in young women? Evidence from a population cohort in rural KwaZulu-Natal, South Africa. J Acquir Immune Defic Syndr 2014; 66(4): 443-451.

21. WHO, National AIDS Programmes: A Guide to Indicators for Monitoring and Evaluating National HIV/AIDS Prevention Programmes for Young People. 2004, World Health Organization: Geneva, Switzerland.

22. Biraro S, Ruzagira E, Kamali A, et al. HIV transmission within marriage in rural Uganda: a longitudinal study. PLoS One 2013; 8(2):5506

23. Wall KM, Kilembe W, Vwalika B, et al. Risk of heterosexual HIV transmission attributable to sexually transmitted infections and non-specific genital inflammation in Zambian discordant couples, 1994-2012. Int J Epidemiol. 2017 Apr 11. doi: 10.1093/ije/dyx045. [Epub ahead of print] PubMed PMID: 28402442.

24. Mackelprang RD, Bosire R, Guthrie BL, et al. High rates of relationship dissolution among heterosexual HIV-serodiscordant couples in Kenya. AIDS Behav. 2014 Jan;18(1):189-93.

25. Kempf MC, Allen S, Zulu I, et al. Enrollment and retention of HIV discordant couples in Lusaka, Zambia. J Acquir Immune Defic Syndr 2008;47(1):116-25.

26. Wall KM, Kilembe W, Nizam A, et al. Promotion of couples' voluntary HIV counselling and testing in Lusaka, Zambia by influence network leaders and agents. BMJ Open. 2012 Sep 6;2(5). pii: e001171. doi: 10.1136/bmjopen-2012-001171. Print 2012. PubMed PMID: 22956641; PubMed Central PMCID: PMC3467632

27. Chomba, E., Allen S, Kanweka W, et al., Evolution of couples' voluntary counseling and testing for HIV in Lusaka, Zambia. J Acquir Immune Defic Syndr, 2008. 47(1): p. 108-15

28. McKenna SL, Muyinda GK, Roth D, et al. Rapid HIV testing and counseling for voluntary testing centers in Africa. AIDS 1997; 11 Suppl 1:S103-10.

29. Dionne-Odom, J., Karita E, Kilembe W, et al. (2013). Syphilis treatment response among HIV-discordant couples in Zambia and Rwanda. Clin Infect Dis 56(12): 1829-1837.

30. Wall KM, Kilembe W, Vwalika B, et al. Hormonal contraception does not increase women's HIV acquisition risk in Zambian discordant couples, 1994-2012. Contraception. 2015 Jun; 91(6): 480-487.

31. Allen S, Meinzen-Derr J, Kautzman M, et al. Sexual behavior of HIV discordant couples after HIV counseling and testing. AIDS 2003;17(5):733-40.

32. Trask SA, Derdeyn CA, Fideli U, et al. Molecular epidemiology of human immunodeficiency virus type 1 transmission in a heterosexual cohort of discordant couples in Zambia. J Virol. 2002 Jan;76(1):397-405.

33. Campbell MS, Mullins JI, Hughes JP, et al. Viral linkage in HIV-1 seroconverters and their partners in an HIV-1 prevention clinical trial. PLoS One. 2011 Mar 2;6(3):e16986.

34. Haaland RE, Hawkins PA, Salazar-Gonzalez J, et al. Inflammatory genital infections mitigate a severe genetic bottleneck in heterosexual transmission of subtype A and C HIV-1. PLoS Pathog. 2009 Jan;5(1).

35. Boeras DI, Hraber PT, Hurlston M, et al. Role of donor genital tract HIV-1 diversity in the transmission bottleneck. Proc Natl Acad Sci U S A. 2011 Nov 15;108(46):E 1156-63.

36. Stephenson R, Vwalika B, Greenberg L, et al. A randomized-control trial to promote long-term contraceptive use among HIV sero-discordant and concordant positive couples in Zambia. J Womens Health 2011; 20(3).

37., Accessed June 2, 2016

38. Zambian Guidelines on Antiertroviral Therapy, Ministry of Health, Zambia, 2013., Accessed June 2, 2016

39. Mujugira A, Celum C, Thomas KK, et al. Delay of antiretroviral therapy initiation is common in East African HIV-infected individuals in serodiscordant partnerships. J Acquir Immune Defic Syndr. 2014 Aug 1;66(4):436-42.

40. Kalichman SC, Simbayi LC, Kaufman M, et al. Alcohol use and sexual risks for HIV/AIDS in sub-Saharan Africa: systematic review of empirical findings. Prev Sci 2007; 8: 141-151.

41. Dunkle KL, Stephenson R, Karita E, et al. New heterosexually transmitted HIV infections in married or cohabiting couples in urban Zambia and Rwanda: an analysis of survey and clinical data. Lancet. 2008 Jun 28;371(9631):2183-91. doi: 10.1016/S0140-6736(08)60953-8. PubMed PMID: 18586173.

42. Kelley AL, Karita E, Sullivan PS, Katangulia F, Chomba E, Carael M, Telfair J, Dunham SM, Vwalika CM, Kautzman MG, Wall KM, Allen SA. Knowledge and perceptions of couples' voluntary counseling and testing in urban Rwanda and Zambia: a cross-sectional household survey. PLoS One. 2011 May 9;6(5):e19573. doi: 10.1371/journal.pone.0019573. PubMed PMID: 21573068; PubMed Central PMCID: PMC3090401.

43. Joseph Davey D, Kilembe W, Wall KM, et al. Risky Sex and HIV Acquisition Among HIV Serodiscordant Couples in Zambia, 2002-2012: What Does Alcohol Have To Do With It? AIDS Behav. 2017 Feb 27. doi: 10.1007/s10461-017-1733-6. PubMed PMID: 28243934.

44. Young, J. G., Hernan, M. A., Picciotto, S., & Robins, J. M. (2010). Relation between three classes of structural models for the effect of a time-varying exposure on survival. Lifetime Data Analysis, 16(1), 71-84.

45. Morton JF, Celum C, Njoroge J, et al. Counseling Framework for HIV-serodiscordant Couples on the Integrated Use of Antiretroviral Therapy and Pre-Exposure Prophylaxis for HIV Prevention. Journal of acquired immune deficiency syndromes (1999). 2017;74(Suppl 1):S15-S22. doi:10.1097/QAI.0000000000001210.

Table 1. Demographic and behavioral characteristics of HIV-negative women in serodiscordant couples by seroconversion outcome N =

1186 Male HIV-positive/ Female HIV-negative couples)

Non-seroconverting Unlinked infections in HR 95% CI

women women

Total 1141 96% 45 4%

N/mean %/SD N/mean %/SD

Follow-up time in days (mean, SD) 750 862 731 715


Woman age (mean, SD) * * 28.8 7.0 25.5 6.3 0.93 0.88, 0.97

Age disparity (mean, SD) 6.9 4.4 6.8 3.9 0.99 0.93, 1.06

Years cohabiting (mean, SD) * * 8.3 6.8 5.6 4.9 0.92 0.86, 0.98

Family income/month (mean, SD) ** $94.0 118.8 $50.9 41.6 0.99 0.99, 1.00

Woman does not read Nyanja * 869 76% 40 89% 2.28 0.89, 5.81

Woman reporting being drunk monthly or more frequently 192 17% 11 24% 2.82 1.16, 3.23

at baseline** (n=1038 couples)

Family planning and sexual history

# previous pregnancies (mean, SD) ** 3.7 2.5 k2.7 1.7 0.80 0.69, 0.93

Pregnant at baseline 155 14% 6 13% 0.92 0.39, 2.17

Fertility intentions of woman (n=472) §

Yes, next year 87 19% 4 21% 1.83 0.57, 5.88

Yes, but not next year 88 19% 5 26% 1.63 0.56, 4.79

No/Don't know 278 61% 10 53% Ref

Woman lifetime sex partners (mean, SD) 2.8 3.4 3.0 2.5 1.02 0.97, 1.08

Woman last year sex partners (mean, SD)** 1.1 0.36 1.3 0.9 1.52 1.17, 1.97

Woman history of STI last year (self-reported) * 292 26% 19 42% 1.83 1.00, 3.35

Time-varying variables (n=interval, % of intervals during the study)

Clinical characteristics

Genital inflammation in past 3-months ** 1799 15% 24 53% 15.46 7.91, 30.20

Genital ulcer in past 3-months ** 724 5% 12 27% 8.05 3.95, 16.40

Sexual behaviors

# of times sex with study partner with a condom in the last 18.6 18.9 21.3 24.1 1.00 0.98, 1.01

3 -months (mean, SD) *

# of times sex with study partner without a condom in the 2.3 8.2 5.0 13.9 1.01 0.98, 1.03

last 3 -months (mean, SD) *

Sex with study partner with a condom in past 3-months 10,036 85% 39 87% 1.19 0.50, 2.84

Sex with study partner without a condom in past 3-months 3299 28% 12 27% 0.90 0.46, 2.84

Sperm present on wet prep during study* 4028 37% 14 32% 0.49 0.07, 3.68

Reported sex with =>1 outside partner during study * 157 1% 2 4% 3.43 0.98, 5.75

Family planning characteristics

Non-hormonal only * * 9060 65% 20 44% Ref

Implant 1000 7% 3 7% 1.12 0.20, 6.14

Injectables 1899 14% 8 18% 1.92 0.79, 4.68

Oral contraceptive pill 1921 14% 14 31% 3.07 1.36, 6.95

Pregnant during interval 1130 9% 2 5% 0.15 0.02, 1.12

*=p<0.10; ** p<0.05; SD= standard deviation; STI= sexually transmitted infection; §= fertility intentions were only collected from 2005

Table 2. Demographic and behavioral characteristics of HIV-negative men in serodiscordant couples by seroconversion outcome (N = 1485 Male HIV-negative/ Female HIV-positive)

Non-seroconverting men Unlinked infections in men

Total 1430 97% 55 3% HR 95% CI

N/mean %/SD N/mean %/SD

Follow up time in days (mean, SD)** 750 834 992 1047


Man age (mean, SD) * * 35.4 8.5 32.8 É6.8 0.96 0.92, 0.99

Age disparity (mean, SD)* 6.8 5.1 5.8 4.2 J0.95 0.89, 1.01

Years cohabiting (mean, SD) 6.0 6.0 5.5 4.8 0.97 0.92, 1.02

Monthly family income (USD) (mean, SD) $95.6 117.5 $68.3 50.9 1.00 1.00, 1.00

Man does not read Nyanja 800 56% 35 64% 1.16 0.94, 1.34

Man heavy alcohol use last year (drunk daily or almost daily)** (n=1014 non-seroconverting couples; n=35 unlinked couples) 544 54% 24 69% 1.26 1.04, 1.58

Family planning and sexual history

# of previous pregnancies (mean, SD) 3.2 2.2 3.0 2.2 0.86 0.73, 1.03

Pregnant at baseline (N, %) 155 11% 6 11% 1.56 0.81, 3.03

Fertility intentions of man (n=542 couples)§

Yes, next year 92 19% 3 5% 0.90 0.24, 3.34

Yes, but not next year * 169 ^35% 11 19% 2.22 0.96, 5.14

No/Don't Know 226 46% 44 76% Ref

Man lifetime sex partners (mean, SD) 11.2 17.8 10.1 11.5 0.99 0.98, 1.01

Man last year sex partners (mean, SD) 1.6 1.8 2.1 1.9 0.59 0.25, 1.38

Man history of STI last year (self-reported)** 358 25% 23 42% 2.02 1.18, 3.48

Circumcised male partner 267 19% 5 9% 0.47 0.19, 1.18

Time-varying variables (n=interval, % of intervals during the study)

Clinical characteristics

Genital inflammation in past 3 months (man)** 1589 12% 19 35% 13.75 6.76,

Genital ulcer in past 3 months (man)** 851 6% 14 25% 6.47 3.42,

Sexual behaviors

# of times sex with study partner with a condom in the last 3-months (mean, SD) 21.7 22.9 20.4 20.1 1.01 1.00, 1.02

# of times sex with study partner in without a condom in the last 3-months (mean, SD) 3.4 11.8 3.9 11.0 1.00 0.98, 1.02

Sex with study partner with a condom in past 3-months 10,037 85% 49 89% 1.42 0.61, 3.34

Sex with study partner without a condom in past 3-months 4056 34% 19 35% 1.06 0.60, 1.87

Reported sex with =>1 partner during study (n=11,584 3.75 1.86, 7.54

non-seroconverting intervals and n=53 unlinked 866 8% 11 21%

seroconverting intervals) * *

*=p<0.10; ** p<0.05; SD= standard deviation; STI= sexually transmitted infection; §= fertility intentions were only collected from 2005

Table 3. Multivariable Cox proportional hazards regression of predictors of time to female acquiring unlinked infection (Model 1) and men acquiring unlinked infection (Model 2) compared with non-seroconverting couples

Adjusted Hazard Ratio (95% CI)

MODEL 1: Female acquiring unlinked infection * (Male HIV-positive/Female HIV- ^tive c >u _ les, n= 186 couples)


Woman reported being drunk weekly or daily/almost daily (vs. moderate and non-drinkers) 5.44 (1.03-28.73)


Genital inflammation in past 3-months 11.92 (5.60, 25.37)

Genital ulcer in past 3-months 6.09 (2.72, 13.64)

>1 outside partner during study 2.07 (0.42, 10.28)

MODEL 2: Male acquiring unlinked infection * ale HIV -negative/- Female HIV-positive couples, n=1485 couples)


Man reported being drunk weekly or daily (vs. moderate and non-drinkers) 3.84 (1.28, 11.55)

Man wants a child 2.81 (0.94, 8.44)


Genital inflammation in past 3-months 8.52 (3.82, 19.03)

Genital ulcer in past 3-months 2.31 (2.05, 8.89)

>1 outside sex partner during study 3.86 (0.98, 15.17)

* Adjusting for woman's age, years cohabiting, couple income (in USD), contraceptive method used at baseline, # of previous pregnancies, history of STI in past year at baseline, and frequency of recent condomless sex with study partner.

** Adjusting for man's age, couple income (in USD), male circumcision (before or during study), and history of STI in past year at baseline

Figure 1. Genetically unlinked HIV incidence rates (and 95% confidence intervals) among serodiscordant couples in Zambia stratified by year of enrollment, follow-up time and gender (1995-2012)

c J ¿j

a. 1.50

u 0.50

Female Female Male Male Female Female Male Male

seroconversion seroconversion seroconversion seroconversion seroconversion seroconversion seroconversion seroconversion (1994-2002) (2003-2012) (1994-2002) (2003-2012) (0-12 months) (>12 months) ¡0-12 months) (>12 months)