A total of 450 PBFW participated in the study (52% pregnant and 47% breastfeeding). The median age of participants overall was 26 years (IQR 22–31). Participant characteristics for each site are shown in Table 2. In Gaborone, participants were more likely to be unmarried/not living with their partner (62%) than married/cohabiting in East London and Cape Town (51–53%). More participants in East London reported having no partner (n = 21/150,14%). Of all participants, 20% (n = 88) did not know the HIV status of their partner, and more participants reported a partner living with HIV in Cape Town (9%).
Table 2
Characteristics of participants in the discrete choice PrEP experiment stratified by site (n = 450)
Characteristics | | Cape Town (n = 150) | East London (n = 150) | Gaborone (n = 150) | Total (n = 450) |
Age in years (median, IQR) | | 26 (22—32) | 27 (22—32) | 25 (22—29) | 26 (22—31) |
Pregnant or Postpartum/breastfeeding | Pregnant | 76 (51%) | 75 (50%) | 85 (56%) | 236 (52%) |
Postpartum | 74 (49%) | 75 (50%) | 66 (44%) | 215 (47%) |
Relationship status | Married/living with partner | 80 (53%) | 77 (51%) | 54 (36%) | 211 (47%) |
Unmarried/not living with partner | 67 (45%) | 52 (35%) | 93 (62%) | 212 (47%) |
No partner | 3 (2%) | 21 (14%) | 4 (3%) | 28 (6%) |
Partner HIV status | Don’t know | 43 (29%) | 25 (17%) | 20 (13%) | 88 (20%) |
HIV Negative | 91 (61%) | 100 (67%) | 124 (82%) | 315 (70%) |
Living with HIV | 13 (9%) | 4 (3%) | 1 (1%) | 18 (4%) |
Refused | 0 (0%) | 0 (0%) | 2 (1%) | 2 (0%) |
Among 150 participants in Cape Town, 76 pregnant women and 74 postpartum women participated in a behavioural survey to determine their oral PrEP experiences and preferences (Supplemental Table 2). They had been using PrEP for a median of 84 days (40–152). The majority (> 90%) reported previous use of injectable contraceptives and/or condoms. All participants emphasized HIV prevention as the most liked characteristic of PrEP over other features. Compared with 20% of breastfeeding women, 32% pregnant women disliked side effects of PrEP. Similarly, more pregnant women (16%) disliked daily dosing than breastfeeding women (4%). One-quarter of pregnant women expressed fears of side effects. Most participants (> 95%) reported no shame about PrEP or concerns regarding their partner finding out.
Main effects across all sites
Almost all attribute levels had significant coefficients, demonstrating where preferences for certain attribute characteristics most diverged (Supplemental Table 3). Figure 2 depicts PBFW’s PrEP attribute preferences across all three settings (Cape Town, East London, and Gaborone). Results demonstrate that participants preferred not to receive vaginally inserted (coefficient − 1.57, 95% CI=-1.84, -1.29) or implanted PrEP (-0.79, 95% CI=-1.00, -0.59) versus oral PrEP. Similarly, participants strongly favoured combination prevention, including HIV, STIs, and pregnancy (1.02, 95% CI = 0.80, 1.24), with a notable preference for combinations beyond HIV prevention alone. Notably, community delivery was less preferred (-0.31, 95% CI=-0.46, -0.17), while private pharmacy collection was least preferred compared to government clinic pick-up (-0.70, 95% CI=-0.90, -0.51). There was no significant difference in preferences for frequency of use when PrEP was more effective, but when it was less effective, participants showed a preference for less frequent dosing. Participants favoured a method that had no side effects or discomfort (0.51, 95% CI = 0.36, 0.67) compared to moderate side effects or discomfort, although this preference was not as strong compared to other characteristics. While there was a slight preference for less frequent refills compared to monthly refills, no discernible difference was found between 3- or 6-month refill intervals (0.28; 95% CI = 0.13, 0.42 versus 0.30, 95% CI = 0.15, 0.46).
Figure 2. Mean estimates of PrEP preferences for all participants (n = 450 pregnant and breastfeeding women in East London, Cape Town, SA, and Gaborone, Botswana)
Main effects by site
Across all sites (Fig. 3), oral PrEP was strongly preferred to vaginally inserted or implanted PrEP. However, combination prevention methods were preferred in Cape Town (1.37, 95% CI = 0.80, 1.95) and East London (0.53, 95% CI = 0.23, 0.83) compared to methods that prevent only HIV, with the strongest preference observed in Gaborone (1.88, 95% CI = 1.32, 2.44). In East London, the difference in preference for type of combination prevention was less pronounced. Conversely, East London participants were indifferent to injectable or oral PrEP (0.24, 95% CI=-0.07, 0.55), while in Gaborone, a preference for injectable PrEP was shown (0.44, 95% CI = 0.10, 0.77). Clinic pick-up for PrEP was favoured in East London and Gaborone, with a negative preference between pharmacy pick-up (-0.45, 95% CI=-0.74, -0.16) and community pick-up (-0.50, 95% CI=-0.75, -0.25) in East London compared to Gaborone (-1.10; 95% CI=-1.52, -0.68). There was no difference in choice in Cape Town between clinic pick-up and community delivery (-0.08, 95% CI= -0.42, 0.26). Effectiveness of PrEP was prioritized over frequency of use in East London and Gaborone, while in Cape Town, dosing frequency had greater significance than effectiveness. Discomfort or side effects were less important in Cape Town and Gaborone, but women in East London significantly preferred no mild discomfort/side-effects (0.68, 95% CI = 0.42, 0.94) compared to moderate effects. Although PBFW in Cape Town and East London slightly preferred infrequent refills over monthly refills, the difference was not significant. In Gaborone, less frequent refills were preferred, with no difference between 3-month (0.64, 95% CI = 0.33, 0.95) and 6-month intervals (0.56, 95% CI = 0.24, 0.87).
Figure 3. PrEP delivery preferences by study site (Cape Town, East London, and Gaborone)
Interactions by pregnancy or postpartum status and maternal age
Participants across different perinatal periods and age groups exhibited similar PrEP delivery preferences (Supplemental Figs. 2 and 3). Pregnant women were less likely to opt for moderately effective, frequently used PrEP (-0.71, 95% CI=-0.89, -0.53) in relation to highly effective daily PrEP. Additionally, younger women (< 25 years) were more hesitant toward vaginal insertion of orally administered PrEP (-2.01, 95% CI=-2.57, -1.45).
Latent Class Analysis
Latent class model analysis identified three classes that effectively described PBFW’s PrEP delivery preferences (Fig. 4). The largest group (43%) fell into Class 1, where PBFW prioritized combination prevention and PrEP dosing frequency: ‘comprehensive delivery seekers’. Class 3 was the second largest group (32%), where preferences were primarily driven by avoidance of vaginal insertable products and where participants prioritized combination prevention, pickup location, and dislike of implants. This group was assigned a qualitative label comprising ‘vaginal insertion avoiders’. Last, 25% of participants fell into Class 2, favouring physical and physiological aspects of PrEP, such as pickup locations and side effects, as well as showing a strong rejection of implants: ‘physical and physiological prioritizers’.
Figure 4. Latent class model analysis identifying three classes that effectively describe pregnant and breastfeeding women (n = 450) in PrEP Choice’s PrEP delivery preference
In Class 1, ‘comprehensive delivery seekers’, women strongly favoured combination prevention over HIV-only prevention, particularly for STIs, HIV, and pregnancy prevention (1.65, 95% CI = 1.13, 2.18). They also had a significant aversion to less effective PrEP compared to effective PrEP, regardless of dosing frequency (-1.17, 95% CI=-1.55, -0.79 and − 0.76, 95% CI=-1.23, -0.28). Compared to other groups, there were less pronounced preferences for vaginally inserted (-0.40, 95% CI=-0.69, -0.12) or implanted (-0.35, 95% CI=-0.63, -0.08) PrEP than for oral PrEP. Similarly, there was no strong preference for refill frequency, pick-up location, nor discomfort/side effects.
In Class 2, ‘physical and physiological prioritizers’, women prioritized government clinic pickup over other options and particularly strongly opposed pharmacy pick-up (-1.19, 95% CI=-1.74, -0.64). This group showed a negative preference for PrEP implants (-0.84, 95% CI=-1.33, -0.36) compared to oral PrEP, with no difference in preference for injections (0.12, 95% CI= -0.29, 0.52) nor vaginal insertion (-0.41, 95% CI=-0.84, 0.02). They more strongly preferred no side-effects/discomfort (0.51, 95% CI = 0.17, 0.84) opposed to moderate side effects. Although they were indifferent to characteristics relating to refill frequency, they showed no real preference for combination prevention options or dosing effectiveness/frequency compared to their relevant baseline characteristics.
PBFW in the Class 3 group, ‘vaginal insertion avoiders’, were very strongly motivated by avoiding vaginal insertion (-3.69, 95% CI=-4.80, -2.58). Unlike Class 2, they showed a positive preference for any combination prevention over HIV-only prevention. Additionally, although less pronounced than Class 2, they displayed a negative association with community delivery (-0.60, 95% CI=-1.04, 0.16) or pharmacy pick-up (-1.20, 95% CI=-1.66, -0.73) compared to government clinic pick-up. Similarly, attributes relating to refill frequency and dosing effectiveness concerning dosing frequency were not of significant concern for PrEP choices.