Validating HEART: a HEAlthy relationships assessment tool for PrEP use


BACKGROUND: HEART is a counselor-administered tool, first developed through formative research and then piloted within a clinic-based intervention (CHARISMA) to reduce social harms and increase healthy relationship dynamics for PrEP use. The tool assesses participants' levels of partner support, control or abuse, suggesting module-based content for counselor delivery. This analysis assessed the validity and reliability of the HEART scales across three CHARISMA research phases (formative, pilot, and RCT), all conducted at Wits RHI in Johannesburg, South Africa.
METHODS: We compared the reliabilities of baseline scores on the five HEART scales: traditional values (TV), partner support (PS), partner abuse & control (PAC), partner resistance to HIV (PR) & HIV prevention readiness (HPR) and conducted a confirmatory factor analysis (CFA) for each sample. To assess construct validity, we examined correlations among HEART scale scores and between scales and independent measures of partner communication, disclosure and intimate partner violence (IPV) within the RCT sample. Using RCT scale scores, we also identified latent classes of participants and assessed whether classes exhibited different demographic, risk or behavioral characteristics.
RESULTS: Internal reliability in four RCT scales was strong (a=.85-.86), similar or higher than in formative and pilot samples. CFA statistics were acceptable. As in formative and pilot samples, the PAC scale in the RCT had a strong inverse relationship with PS (-0.61- -.71) and strong positive correlation with PR (0.48-0.68). PAC, PS and PR were strongly correlated with independent measures of IPV. The latent class analysis produced two distinct groups with significant differences on partner communication, decision-making and IPV behaviors. Further, 96% of participants had a high probability (.8 or higher) of belonging to a single class.

Expansion Study Treatment ArmClass A "Supportive"Class B "Conflictual"p-value

Scale mean scores (range 1=disagree a lot; 6=agree a lot)
Traditional Values (TV) 13 items, a=0.851.32.0<0.001
Partner Support (PS) 10 items, a=0.865.74.7<0.001
Partner Abuse & Control (PAC) 9 items, a=0.861.11.8<0.001
Partner Resistance to HIV (PR) 5 items, a=0.851.02.1<0.001
Age (years)28.628.30.76
Education/some college or more37.7%27.3%0.38
Decision-making (8 items, scale 0-8)
WHO "Violence Against Women" measures
Partner communication (4 items, scale 0-4)3.83.450.001

IPV controlling behaviors (7 items, scale 0-7)0.81.7<0.001

IPV abusive behaviors (4 items, scale 0-4)0.30.8<0.001

IPV physical violence (6 items, scale 0-6)

IPV sexual violence (3 items, scale 0-3)<

HIV Risk Perception

Not at all worried34.4%29.1%0.36

Somewhat worried50.5%46.4%

Very worried15.1%24.5%

Independent PrEP Disclosure, Support and Opposition mean scores

Disclosed PrEP use to partner (yes)69.9%54.5%0.08

PrEP Support (5 items, scale 0-5)1.140.750.02

PrEP Opposition (7 items, scale 0-7)

Class probability (mean)0.970.98

CONCLUSIONS: This multidimensional tool offers an efficient, integrated measurement of social support and IPV. These data offer further evidence for HEART reliability and validity and support its utility in IPV and PrEP-related services.