1 of 24

Understanding preferences around PrEP modalities and implementation among Tanzanian AGYW: �a discrete choice experiment

Lila Sheira, MPH

8 December 2023

2 of 24

Acknowledgements

Youth Advisory Boards
Healthcare providers
Agatha Mnyippembe
Dr. Amon Sabasaba
Dr. Prosper Njau
Julie Allan
Lisa Richard
Janeth Msasa
HPON research assistants

Dr. Sandra McCoy
Dr. Will Dow
Dr. Julie Deardorff

Dr. Jenny Liu
UCSF Center for AIDS Research
National Institutes for Health

3 of 24

Motivation

Adolescent girls and young women (AGYW) remain vulnerable to HIV acquisition in East Africa
Oral PrEP is highly effective at HIV prevention, however there are several deterrents to AGYW engaging and retaining in PrEP care

Fears of being seen as promiscuous or living with HIV
Stigma from providers and low willingness to prescribe
Not wanting to be seen with PrEP medications

Newer formulations (injection and vaginal ring) show promise in addressing some barriers to care – and give AGYW choice

We are at an opportunity juncture to hear from AGYW how and where newer modalities may be implemented to address other barriers to care among potential users

I don’t have to tell this audience that adolescent girls and young women remain especially vulnerable to HIV acquisition across the continent, and in East Africa in particular. We also know how effective oral daily PrEP is—which is the primary prEP method currently available. There are numerous studies, however, documenting deterrants. AGYW don’t want to be seen as being promiscuous or living with HIV. We also know healthcare providers can be biased and have lower willingness to prescribe PrEP to AGYW.

There are newer formulations of PrEP which might address some of these barriers-mainly being seen with PrEP medications. With these newer modalities in regulatory review across many countries, we are at an opportunity juncture to understand what we might do differently to ensure success in the development of the implementation models and roll out of these newer modalities among AGYW.

4 of 24

Methods

Formative research with 3 youth advisory boards, healthcare providers, research team, and MoH partner
Discrete choice experiment

N=41 wards across Shinyanga and Mwanza provinces of Tanzania
Target sample size: 700 AGYW

Eligibility criteria: Tanzania National AIDS Control Programme Screening (NASCOP) for PrEP

Current timelines (6 months) for “current risk”
Ever for determining previous risk in lifetime

5 of 24

What options did we evaluate?

PrEP Type/Modality

Refill location

Screening Location

Price

Oral daily

Injectable

(CAB-LA)

Vaginal ring

Provider Identity

Standard health facility

Mobile Clinic

Pharmacy

Mobile Clinic

Pharmacy

CHWs

Younger/ older female

Younger/ older

male

Anyone

We build upon important research, restricting our sample to AGYW who were overwhelmingly currently screening + for PrEP and evaluating real PrEP products already determined effective and under regulatory review. This has a direct policy implication and relevance as we all prepare for the roll out of newer PrEP types. We worked closely with your YAB and stakeholder meetings to develop the major determinants of PrEP program characteristics most salient to AGYW, including:

1-Prep type: injectable PrEp or CAB LA, the vaginal ring, and the standard of care oral daily PrEP

2-the screening location where PrEP eligibility is assessed-a mobile clinic, a pharmacy, as well as the standard of care health facility

3-the refill location-including a mobile clinic, pharmacy, and community health workers

4-provider identity, including older younger males and females, as well as an anyone option and

5- price: ranging from free to 5000 Tanzanian Shillings

6 of 24

What is a discrete choice experiment?

A quantitative technique which aims to elicit stated preferences over a defined choice set
DCEs offer a number of hypothetical options known as attributes, in pairs, and through the options allows the researcher to understand which specific attributes are valued over others.

Which of these two programs would you prefer?
Injection	Pill
Health Facility	Pharmacy
Community Health Worker	Mobile Clinic
Anyone	Younger man
2000 TSh	4000 TSh

Alternative 1 Alternative 2

Choice task

Attributes

Attribute levels

Level 1 Level 2

7 of 24

Background characteristics

202 participants currently or ever married (29%)

	Ever risk (n=138)	Current risk (n=558)	Total (n=696)
Age (Median, IQR)	20.0 (18.0, 22.0)	20.0 (18.0, 22.0)	20.0 (18.0, 22.0)
Relationship Status
Single	40 (29.0%)	79 (14.2%)	119 (17.1%)
Steady partnership	85 (61.6%)	333 (59.7%)	418 (60.1%)
Engaged	4 (2.9%)	26 (4.7%)	30 (4.3%)
Married	9 (6.5%)	120 (21.5%)	129 (18.5%)
Ever Married¹	10 (7.8%)	63 (14.4%)	73 (12.9%)
Any live children	40 (29.0%)	223 (40.0%)	263 (37.8%)
Every attended school
Yes, currently	21 (15.2%)	52 (9.3%)	73 (10.5%)
Yes, but not currently	117 (84.8%)	495 (88.7%)	612 (87.9%)
Never attended	0 (0.0%)	11 (2.0%)	11 (1.6%)
¹Asked among those not currently married

259 Primary or less education (37%)

8 of 24

Interpreting DCE results

Multinomial logistic regression
Preference weight vs categorical variables

Preference weights use effects coding which give the preference relative to the mean effect of the given attribute

Categorical coding: PrEP pill (reference), ring, injection
Effects coding: mean (PrEP pill, ring, injection)

How we interpret it: strength of preference relative to the attribute level

Positive= more likely to be chosen
Negative=less likely to be chosen (but not negative preference)

Attribute importance: “weight” of attribute in decision-making

The interpretation of DCE are a bit particular so I will take a moment to explain. I used a multionomial logistic regression. Given we are interested in preference weights, I used effects coding. This differs from traditional categorical coding of variables. Preference weights give the preference for an attribute level, such as the vaginal ring, relative to the mean effect of the given attribute. While not so intuitive, you can think of it in comparison to categorical coding. So in standard categorical coding, we have a reference group and compare categories to that reference. Effects coding doesn’t have a reference group—so we can look at the preference for all attribute levels.

We interpret the pref level in terms of the magnitude and direction of the effect relative to the mean value of the level. A positive preference level means an attribute level was more likely to be chosen; a negative mean less likely (but not less preferred). The larger the magnitude, the more or less preferred the level was.

Further, I report attribute importance, which helps us understand the weight of each of the five attributes in decision-making.

9 of 24

Attribute levels

More preferred

Less preferred

PrEP modality and implementation preference among AGYW currently or ever at risk for HIV acquisition in Tanzania, N=694

10 of 24

Attribute levels

More preferred

Less preferred

PrEP modality and implementation preference among AGYW currently or ever at risk for HIV acquisition in Tanzania, N=694

Would you actually take up PrEP in this offered program?

89% said YES

11 of 24

Attribute Importance (Relative Importance)

I want to finish sharing the results by sharing the attribute importance. Again, the attribute importance or relative importance helps us determine the “weight” of each overall attribute (but not the levels) in decision-making

We get these scores by taking the sum of the range of each attribute preference weight across all attributes, and then the range of preference weights within each attribute. This quantifies the potential influence of each attribute relative to the total utility.

Here we see Modality type and cost are the biggest influencers of decision-making, with dispenser identity being somewhat important and screening and refill location accounting for <6% of total relative importance.

This further demonstrates that adding these newer modalities to the market and giving AGYW choice in their PrEP options is incredibly important

12 of 24

Discussion

Limitations

At what level of risk are the married AGYW truly at? They often meet >1 criteria (condomless sex, violence)
Perhaps AGYW at ADDOs not truly representative—AGYW here mostly in the 18-22 range. How do we find younger, at-risk AGYW?
Didn’t include dual prevention pill in primary DCE. However secondary questions about DPP found interest high, but still lower than injectable PrEP

Main Findings

PrEP awareness was very low 🡪 best kept secret!
Product type—specifically injectable PrEP—was highly influential on AGYW decision-making and was strongly preferred
Cost and provider identity were also influential
Lack of dominating preferences in refill or screening location suggests a diversity of approaches may be needed

There were a few limitations to our study. We saw about a third of the participants were married, which makes us wonder how much risk they were truly at for HIV. Nevertheless, these AGYW often met >1 of the criteria, and research shows that married women constitute a large proportion of incident HIV cases.

We also were not successful in recruiting AGYW in the 15-19 range as much—we need to work harder to get the younger AGYW.

We also didn’t evaluate the DPP as this is not under review in Tanzania. Nevertheless, we included a module separate from the DCE on the DPP and found that while interest was high, the preference for injectables was higher.

IN summary, we note that PrEP awareness is low and it remains the best kept secret. The product type was influential in decision making—and injectable PrEP was HIGHLY preferred.

Cost and provider identity were also important, underscoring the price sensitivity of AGYW and the desire for youth and girl friendly services. It is important to note that there were no dominating preferences in refill or screening location.

13 of 24

Key Takeaways

AGYW in Tanzania are enthusiastic about injectable PrEP
No clear winner for screening / refill location 🡪 need all the options
There is not a one size fits all preferred approach
Choice matters

Lila.Sheira@ucsf.edu

www.MalkiaKlabu.ucsf.edu

14 of 24

Lila.Sheira@ucsf.edu

www.MalkiaKlabu.ucsf.edu

15 of 24

Current criteria for screening for PrEP in Tanzania

Essential criteria (must meet all)

Adults/mature minors under 18 years who are at substantial risk for HIV acquisition
HIV negative using a rapid antibody test
Hepatitis B negative
Normal renal function
Willing to consent for and use PrEP as prescribed
No suspicion of acute HIV infection
Willing to provide detailed contact information for tracing
Any ONE of the following substantial risk in the past six months:

Vaginal or anal sex without a condom with more than one partner
History of a new sexually transmitted infection, or
Use of post exposure prophylaxis for sexual exposure

AGYW specific criteria (1 min)

Not used condom consistently with the latest partner in the last 3 months OR
Current or STI in the last 3 months OR
Receive cash/item for sex in the last 6 months OR
Victim of sexual violence in ongoing risk of repeated incidence OR
In sexual relationship with PWID

16 of 24

What about that extra question-would you actually use this program?

17 of 24

Eligible Sample

	Shinyanga	Kahama	Mwanza	Total
Eligibility
Eligible	133	213	351	696 (81.5%)
Ineligible	30	75	54	159 (18.5%)
Reason not eligible
Didn’t meet basic criteria	9 (30%)	15 (20.0%)	13 (22.6%)	36 (22.8%)
Didn’t meet sex hx criteria	19 (63.3%)	50 (66.7%)	19 (35.8%)	88 (46.8%)
Living with HIV or refused to say HIV status	2 (6.7%)	10 (13.3%)	22 (41.5%)	34 (18.1%)

18 of 24

Post-estimation predictions: �Interaction of cost and modality

19 of 24

Post-estimation predictions: �Interaction of provider identity and modality

20 of 24

Post-estimation predictions: �Interaction of screening location and modality

21 of 24

Post-estimation predictions: �Interaction of refill location and modality

22 of 24

Minnis, Alexandra M., Millicent Atujuna, Erica N. Browne, Sheily Ndwayana, Miriam Hartmann, Siyaxolisa Sindelo, Nangamso Ngcwayi et al. "Preferences for long‐acting Pre‐Exposure prophylaxis (PreP) for HIV prevention among South African youth: results of a discrete choice experiment." Journal of the International AIDS Society 23, no. 6 (2020): e25528.

Previous DCE among youth in South Africa about PrEP:

807 participants, 50% female, ages 19-22
Used potential product formulations not yet confirmed to understand hypothetical preferences

Potential contributions of our study:

Among AGYW in Tanzania—first in East Africa
Those who would screen + for PrEP
Updated for real PrEP products
Can help with roll out of PrEP potentially-policy interest

What do we already know?

So let’s take a step back and look at what do we already know. While numerous DCEs have been conducted to evaluate potential PrEP formulations as well as understanding desires among sub-groups such as PWID or sex workers, the one most pertinent to AGYW was conducted in South Africa a few years ago among 800 participants half of whom were female. The researchers evaluated differentiated care options, such as pharmacies, mobile clinics, or a health clinic, as well as hypothetical product formulations based on prEP products that were being evaluated or tested. They did not test on formulations which have already gone through Phase 3 trials.

We build upon this important research, restricting our sample to AGYW who were overwhelmingly currently screening + for PrEP and with real PrEP products under already under regulatory review. This has a direct policy implication and relevance.

23 of 24

Secondary results: sexual history

High levels of behaviors that put AGYW at risk for HIV:

49.9% report ever intergenerational sex
35% ever transactional sex, and 24.6% in last 6 months
Ever abuse is too high: 16% physical and 26% sexual abuse
1 in 6 AGYW said first sexual intercourse was non-consensual
High condomless sex independent of marital status: 83-90%
Ever STI symptom moderate: 16%
Sex with PLHIV: 4%

Some good news: ever testing for HIV was high too: 84%

24 of 24

Secondary results: PrEP knowledge & awareness (pre-DCE)