Young Women's Stated Preferences for Biomedical HIV Prevention: Results of a Discrete Choice Experiment in Kenya and South Africa

Abstract

Background: Integrating end-user perspectives into the design of new biomedical HIV prevention products is recognized as vital to informing the product development pipeline.

Setting: Kisumu, Kenya; and Soshanguve, South Africa.

Methods: We conducted a discrete choice experiment survey with 536 women aged 18-30 years to assess preferences for hypothetical HIV prevention products characterized by the attributes of efficacy, pregnancy prevention, delivery form, dosing frequency, and side effects. Participants included product-experienced women from Tablet, Ring, Injection as Options TRIO, a cross-over clinical study evaluating 3 placebo delivery forms (oral tablets, vaginal rings, and injections), and a product-naive sample recruited from the same communities. Analyses used random parameters logit and latent class models.

Results: HIV prevention efficacy was a strong determinant of stated choice overall; however, in South Africa, delivery form was just as important, with an injection every 2-3 months most preferred and a daily oral tablet least preferred. In Kenya, product-experienced women preferred monthly injections and least preferred a monthly ring. Respondents indicated a preference for multipurpose prevention technologies that combine HIV and pregnancy protection. Latent class analyses confirmed these findings and delineated heterogeneity in preferences across subgroups defined by age, past experience with the delivery forms, and education.

Conclusions: Despite an overall preference for products with high efficacy, we identified attributes salient to future uptake and use of HIV prevention products. Preferences for injectable products underscored interest in this pre-exposure prophylaxis delivery form. Likewise, the multipurpose prevention technology feature was valued in both Kenya and South Africa and most influenced interest in vaginal rings.

FIGURE 1.

Discrete choice experiment design: Attributes and levels.

FIGURE 2.

Normalized preference weights from random parameters logit models for Kenyan and South African women. Kenyan product-experienced and product-naive estimated separately due to preference heterogeneity.

FIGURE 3.

Preference shares for TRIO products among product-experienced Kenyan women and South African women. When all product delivery forms offer both HIV and pregnancy protection (are MPTs), a monthly injection received the highest share. However, when the monthly injection only prevents HIV but the vaginal ring and tablet remain MPTs, the preference shifts away from injection. All products had no side effects and provided 90% HIV protection. Hashed fill indicates product is not an MPT.

FIGURE 4.

Normalized preference weights from latent class logit model of the South African sample (N = 268) and the Kenyan sample (N = 268). In Kenya (A, above), class membership was based on age and sample, with younger women (18–24 years) and those from the TRIO sample (product-experienced) more likely to be in class 2. In South Africa (B, above), class membership was differentiated based on education, with women who attended university more likely to be in class 1.