Background
Household couples represent a population wherein there is a clear unit to intervene for HIV testing and treatment, and numerous couples-driven interventions exist (e.g. couples HIV testing and counseling, partner-provided HIV self-tests). However, it is unknown the extent to which there is homophily (like partnered with like) among household couples by HIV testing and treatment engagement.
Methods
We analyzed data from 18 Population-based HIV Impact Assessment surveys (PHIAs) conducted in Cameroon, Cote D’Ivoire, Ethiopia, Mozambique, Kenya, Lesotho, Malawi, Namibia, Rwanda, eSwatini, Tanzania, Uganda, Zambia, and Zimbabwe between 2015-2021. Heterosexual household couples with self-reported HIV testing history and biological data on HIV status, presence of ART, and viral load suppression (VS) were analyzed. Each survey was given equal weight and survey design was accounted for in analysis. Among serostatus groups, homophily was assessed by comparing couples’ observed mixing patterns to that expected under random mixing, with bootstrapped confidence intervals.
Results
There were 62,639 couples, representing 123,856 individuals, with 86.4% seroconcordant negative, 7.0% seroconcordant positive, 3.5% serodiscordant with a female partner living with HIV (LHIV), and 3.1% serodiscordant with a male partner LHIV. Among seroconcordant couples LHIV, more couples than expected (Figure) were both positive but unaware of their status (4.30 times greater than under random mixing, 95% CI=4.01-4.63), aware of status but not on ART (3.26, 95% CI=2.66-3.93), or on ART but not virally suppressed (VS) (2.34, 95% CI=1.55-3.10). Among seroconcordant HIV-negative couples, never tested individuals were over twice as likely to be coupled together than expected by random chance (2.54, 95% CI=2.52-2.64). Similarly, among serodiscordant couples, more couples between individuals LHIV but unaware of status with HIV-negative and never tested were observed than expected by random chance (1.76, 95% CI=1.55-1.96 when female partner LHIV; 1.95, 95% CI=1.67-2.33 when male partner LHIV).
Conclusions
We find evidence of homophily in HIV testing behaviors and treatment engagement in household couples in sub-Saharan Africa, supporting couples-based interventions when reaching individuals that have been previously unengaged. Furthermore, HIV testing and treatment interventions that target household couples as a unit may be efficient in increasing reach.