Partner Violence and Condom Use in HIV-Discordant Couples

Who Took Part in the California Partners Study?

The study drew on 145 heterosexual couples in the San Francisco Bay Area in which one partner had HIV and the other did not, 290 people in all (Eaton et al., 2014). They were recruited through a range of outreach methods between November 1996 and July 1999 for the California Partners Study II, a randomised trial of a behavioural programme to reduce sexual risk (Eaton et al., 2014). The 2014 paper is a cross-sectional analysis of the interviews both partners gave when they joined (Eaton et al., 2014).

The sample was lower-income and ethnically mixed, 39% African American and 36% white among those who gave their ethnicity, with a mean age of 40 and a median of 4.2 years together (Eaton et al., 2014).

Drug use was common: in the previous six months 42% had used crack cocaine and 29% had injected drugs, and 40% had at some point exchanged sex for money or drugs (Eaton et al., 2014).

How violence and condom use were measured

Interviewers asked each participant how often in the past six months the current partner had physically hurt or threatened to hurt them, and anyone answering once or more was counted as having experienced physical violence, so the measure merges attacks with threats (Eaton et al., 2014).

Condom use over the same six months was split into always and everything else, the outcome the authors called inconsistent condom use, and the model allowed for answers within a couple being related (Eaton et al., 2014).

What Did the Study Find?

People who reported violence or threats from their partner were more likely to report inconsistent condom use: after adjustment, their odds were 2.2 times higher (95% CI 1.1–4.1) (Eaton et al., 2014). Before adjustment the odds ratio was 2.4 (95% CI 1.4–4.1) (Eaton et al., 2014). The final model held gender, frequency of sex, injection drug use and a history of bisexual behaviour constant (Eaton et al., 2014).

Violence itself was common: 82 of the 290 participants, about 28%, said a partner had hurt or threatened them, with similar rates in women and men, 30% and 27% (Eaton et al., 2014). More than half did not always use condoms (Eaton et al., 2014).

An odds ratio of 2.2 does not mean the risk doubled, although the paper’s abstract says it did (Eaton et al., 2014). When an outcome is this common, odds ratios exaggerate relative risk: the unadjusted rates from the paper’s tables, about 70% against 49%, amount to a risk roughly 1.4 times higher (Eaton et al., 2014).

Violence and drug use were often shared

Violence tended to run in both directions: a person whose partner reported violence had 3.2 times the odds of reporting it too, and the tables show both partners reporting violence or threats in 19 of the 145 couples (Eaton et al., 2014). Drug use clustered even more strongly (Eaton et al., 2014).

Injection drug use was independently linked to inconsistent condom use (adjusted odds ratio 2.4, 95% CI 1.21–4.7), as were a history of bisexual behaviour and more frequent sex (Eaton et al., 2014). The authors suggest the link with frequency may simply reflect more occasions on which a condom could be skipped (Eaton et al., 2014).

Why Might Violence and Condom Use Be Linked?

The study cannot say why, and the authors describe explanations put forward by earlier researchers rather than testing any of them (Eaton et al., 2014). Because violence and condom use were recorded at the same interview, the survey cannot show which came first (Eaton et al., 2014).

One idea is that women in violent relationships feel less able to ask for condoms, or too “worn down” to resist unprotected sex with a partner who has HIV (Eaton et al., 2014). Another is that violence can follow disclosure of HIV status or a request to use a condom (Eaton et al., 2014). A third holds that men who abuse partners are more often HIV-positive and push risky sex on them, while also taking more risks themselves (Eaton et al., 2014).

These accounts mostly picture a male aggressor and a female victim, which this sample does not fully fit: men and women reported similar levels of violence, and the authors read it as, in some couples, a pattern with both people in both roles (Eaton et al., 2014).

Similar reporting does not mean similar harm. The study collected no measure of injury, and the authors cite other research in which men’s aggression injured women more often than women’s aggression injured men (Eaton et al., 2014).

Drug use is the other thread. Because the partners of people who injected drugs usually injected too, the authors describe injecting as mostly a shared problem that could cloud both partners’ judgement and reduce condom use (Eaton et al., 2014).

Are Condoms Still the Main Protection for Mixed-Status Couples?

No: condoms remain useful, but effective HIV treatment and pre-exposure prophylaxis (PrEP) now protect partners without relying on a condom every time (World Health Organization, 2023; World Health Organization, 2025). WHO’s 2023 policy brief states that people with HIV whose viral load is undetectable have zero risk of passing the virus to sexual partners (World Health Organization, 2023). A suppressed but detectable load, at or below 1,000 copies per millilitre, carries almost zero or negligible risk (World Health Organization, 2023).

The US PrEP guideline treats a viral load under 200 copies per millilitre for at least six months as undetectable and describes such people as posing effectively no risk to sexual partners (US Public Health Service & Centers for Disease Control and Prevention, 2021).

Some of the strongest evidence comes from mixed-status couples. In the PARTNER study of 888 serodifferent couples, no infection was traced to a partner whose viral load was below 200 copies per millilitre (Rodger et al., 2016), and in HPTN 052 early treatment cut linked infections by 93% (Cohen et al., 2016).

HIV-negative partners can also take PrEP. WHO now recommends daily oral PrEP, injectable cabotegravir, the dapivirine vaginal ring and, since July 2025, long-acting injectable lenacapavir (World Health Organization, 2025). Even so, WHO estimated 1.3 million new infections worldwide in 2024, when 31.6 million people were receiving antiretroviral treatment (World Health Organization, 2025).

These tools change the question the 2014 paper asked, because treatment and PrEP depend on keeping up pills or injections, and later studies suggest that violence can disrupt exactly that (Roberts et al., 2016; Hatcher et al., 2015).

How Does Partner Violence Affect HIV Prevention and Treatment Today?

Later studies suggest that partner violence is linked not only to condom use but also to HIV infection itself and to how well people keep up treatment and PrEP (Roberts et al., 2016; Hatcher et al., 2015; Li et al., 2014). A 2014 meta-analysis of 28 studies concluded that the available evidence suggests a moderate, statistically significant association between partner violence and HIV infection among women (Li et al., 2014). In the cohort studies it pooled, physical violence was linked to a risk about 22% higher (Li et al., 2014).

A 2015 meta-analysis of 13 studies, all cross-sectional, linked violence to lower use of antiretroviral treatment among women with HIV (Hatcher et al., 2015). Women reporting violence had about half the odds of good adherence (OR 0.48) and lower odds of viral suppression (OR 0.64) (Hatcher et al., 2015). Its authors cautioned that the lack of longitudinal data should temper how these results are read (Hatcher et al., 2015).

The clearest evidence on timing comes from the Partners PrEP study of African couples. Among 1,785 HIV-negative women whose partners had HIV, 16% reported verbal, economic or physical violence during follow-up (Roberts et al., 2016). Women who had reported violence in the previous three months were about 1.5 times as likely to have low PrEP adherence, measured both by pill counts and by tenofovir in the blood (Roberts et al., 2016). The effect was no longer significant three months after the violence was reported (Roberts et al., 2016).

How common partner violence is

Partner violence remains widespread. WHO estimates that 25.8% of ever-partnered women aged 15–49 have experienced physical or sexual violence, or both, from a husband or male partner (World Health Organization, 2026).

Men are affected too, as the California data suggested (Eaton et al., 2014). In the latest US national survey, 22.5% of women and 13.7% of men said a partner had been physically violent towards them at some point in their lives (Zhang Kudon et al., 2026). That survey changed its methods, so CDC cautions against comparing its figures with earlier rounds, and its response rate was only 16.5% (Zhang Kudon et al., 2026).

Prevention that depends on negotiating each act is weakest in exactly the relationships where negotiating is least safe.

How Strong Is the Evidence?

The evidence from the California study itself is limited. It is one cross-sectional analysis of 290 people interviewed in 1996–1999, so it shows an association at a single point in time, not that violence causes inconsistent condom use (Eaton et al., 2014). Both violence and condom use were self-reported, and the authors acknowledge that answers on both may be skewed by social desirability (Eaton et al., 2014). Threats and attacks were merged, injury was not measured and methamphetamine use was not recorded (Eaton et al., 2014). For these reasons the evidence on the article’s central question, whether partner violence undermines condom use in couples where one partner has HIV, is rated limited.

The reporting also needs care: some printed figures do not add up, and the manuscript was received in December 2012, when the interviews were already 13 to 16 years old (Eaton et al., 2014).

The wider literature is more consistent than any single study, but it is mostly observational (Hatcher et al., 2015; Li et al., 2014). The Partners PrEP analysis is stronger on timing (Roberts et al., 2016), and the one violence-prevention trial cited here reported no significant fall in physical violence (Abramsky et al., 2014).

Frequently asked questions

Does partner violence make people less likely to use condoms?

It is linked to it, but this study cannot show cause. In 145 California couples, people reporting violence or threats had 2.2 times the odds of not always using condoms after adjustment (Eaton et al., 2014). Before adjustment the rates were roughly 70% against 49%, and the survey could not show which came first (Eaton et al., 2014).

What is an HIV-discordant couple?

It is a couple in which one partner has HIV and the other does not, also called serodiscordant or serodifferent (Eaton et al., 2014; Rodger et al., 2016). Large studies of such couples shaped current treatment-as-prevention advice (Rodger et al., 2016; Cohen et al., 2016).

Can a partner with an undetectable viral load pass on HIV through sex?

No. WHO states that people with an undetectable viral load have zero risk of sexual transmission, and a suppressed but detectable load carries almost zero risk (World Health Organization, 2023). In the PARTNER study, which included 548 heterosexual couples, no infection was traced to a partner whose viral load was below 200 copies per millilitre (Rodger et al., 2016).

Do men also experience partner violence?

Yes. In the California study 27% of men and 30% of women reported violence or threats from their partner in six months (Eaton et al., 2014), and a recent US survey found lifetime physical partner violence among 13.7% of men and 22.5% of women (Zhang Kudon et al., 2026).

Why does partner violence matter for PrEP?

PrEP only protects people who can keep taking it. In the Partners PrEP study, African women who reported recent violence were about 1.5 times as likely to have low adherence (Roberts et al., 2016), and US guidance lists screening for partner violence among routine checks for PrEP users (US Public Health Service & Centers for Disease Control and Prevention, 2021).

This article is for general information and is not medical advice. It was reviewed for accuracy by a qualified clinician; decisions about your health should be made with your own doctor.