A patient-centered, combination intervention to support adherence to HIV pre-exposure prophylaxis during pregnancy and breastfeeding: a randomized pilot study in Malawi

Abstract

Background:

Daily oral pre-exposure prophylaxis (PrEP) can reduce HIV incidence in pregnant and breastfeeding women, but adherence is essential.

Methods:

We conducted a pilot randomized trial to evaluate an intervention package to enhance antenatal and postnatal PrEP use in Lilongwe, Malawi. The intervention was based on patient-centered counseling adapted from prior PrEP studies, with the option of a participant-selected adherence supporter. Participants were locally eligible for PrEP and randomized 1:1 to intervention or standard counseling (i.e., control) and followed for six months. Participants received the intervention package or standard counseling at enrollment, one month, three months, and six months. Adherence was measured via plasma and intracellular tenofovir concentrations and scored using a published algorithm. Our primary outcome was retention in care with concentrations consistent with 4–7 doses/week.

Results:

From June to November 2020, we enrolled 200 pregnant women with median gestational age of 26 (IQR:19–33) weeks. Study retention was high at three months (89.5%) and six months (85.5%). In contrast, across the two timepoints, 32.8% had adherence scores consistent with 2–5 doses/week and 10.3% had scores consistent with daily dosing. For the composite primary endpoint, no substantial differences were observed between the intervention and control groups at three months (28.3% vs. 29.0%, probability difference [PD]:−0.7%, 95%CI:−13.3%, 11.8%) or at six months (22.0% vs. 26.3%, PD:−4.3%, 95%CI:−16.1%, 7.6%).

Conclusions:

In this randomized trial of PrEP adherence support, retention was high, but less than one-third of participants had pharmacologically confirmed adherence of ≥4 doses/week. Future research should focus on antenatal and postnatal HIV prevention needs and their alignment across the PrEP continuum, including uptake, persistence, and adherence.

INTRODUCTION

Throughout pregnancy and breastfeeding, women in sub-Saharan Africa are at elevated risk for HIV acquisition.^1,2 Pre-exposure prophylaxis (PrEP) is an effective, evidence-based intervention to prevent HIV and is an important part of combination HIV prevention in antenatal and postnatal settings.³ Although the options for PrEP are expanding, at this time, daily oral emtricitabine/tenofovir disoproxil fumarate (FTC/TDF) has the most established track record for use in pregnancy and breastfeeding.⁴ When taken consistently, daily FTC/TDF has been shown to dramatically reduce HIV incidence across a number of clinical trials.⁵ Daily oral PrEP has been recommended by the World Health Organization for pregnant and postpartum women at high risk for HIV acquisition,³ and formally incorporated into national HIV guidelines in many sub-Saharan African countries.⁶ PrEP delivery has been integrated into existing maternal and child health platforms as part of large-scale HIV prevention initiatives.^7,8

Despite encouraging progress, challenges remain. While PrEP use during pregnancy and breastfeeding appears acceptable,^9–11 persistence diminishes in the months following PrEP initiation.^7,8 Although data are still limited, the available studies suggest that pharmacologically confirmed adherence may also be low in this population.¹² This gap may be particularly important for pregnant and breastfeeding women. Pharmacokinetic/pharmacodynamic modeling suggests that longer lead times may be needed to reach protective tenofovir concentrations in the lower female genital tract.¹³ Additionally, studies suggest that tenofovir levels are reduced during pregnancy,^14–16 potentially requiring higher rates of adherence to maintain efficacy.

Interventions have been proposed to support PrEP in other groups; however, few behavioral approaches have been specifically tailored to pregnant and breastfeeding women. Integration of such services within maternal and child health platforms offers important efficiencies in service delivery. Moreover, by adopting an HIV status-neutral approach^17,18—one tailored for women living with HIV and those at-risk for newly acquiring HIV—interventions can optimize training of healthcare staff in a single adherence support strategy.

To address individual and social barriers to adherence, we designed a theory-based combination intervention to support antiretroviral use during pregnancy and breastfeeding.¹⁹ Our overarching goal was to develop cross-cutting interventions that foster an HIV status-neutral approach to prevent vertical HIV transmission, capable of supporting HIV services for treatment or prevention. This intervention was evaluated in a pair of randomized pilot studies, separately enrolling HIV-negative women starting PrEP and women with living with HIV initiating antiretroviral therapy.²⁰ In this report, we describe the clinical findings among women on PrEP in antenatal and postnatal settings.

METHODS

Study design and procedures

This pilot randomized trial was designed to assess the acceptability, fidelity, and clinical outcomes associated with our combination intervention for antiretroviral adherence. A full description of the study protocol has been published elsewhere.²⁰ Below, we highlight specific details from the PrEP component, the focus of the current report.

Candidates were eligible for screening if they were 18 years or older, pregnant, had a negative HIV test in the past three months, and expressed willingness to initiate and continue daily oral PrEP for the duration of follow-up. Criteria for PrEP eligibility were adopted from Malawi Ministry of Health guidelines,²¹ including: known positive or unknown partner HIV status, multiple sexual partners, diagnosis of a sexually transmitted infection, use of postexposure prophylaxis, use of shared injection equipment, or unspecified concern for HIV acquisition. Women were excluded if they had a history of known renal disease or reported history of intimate partner violence or social harms. Those who met eligibility criteria provided written informed consent and underwent laboratory-based screening for HIV and hepatitis B, and elevated serum creatinine. Women with any of these conditions were excluded from further participation.

After screening, eligible women were enrolled and randomly allocated 1:1 to receive either the combination adherence package (intervention) or standard counseling (control). The study statistician (KRM) independently generated randomization assignments using a permuted block design. Randomization assignments were placed in sequentially numbered, opaque, sealed envelopes by study staff not otherwise involved in study procedures. Envelopes were opened sequentially with each new enrollment. Throughout enrollment, allocations were documented and monitored for quality assurance according to standard practice.²²

Study visits were scheduled at one, three, and six months after enrollment. The schedule of evaluations has been detailed elsewhere and included rapid HIV testing, renal monitoring, and drug concentration measurement.²⁰ In line with local research practices, participants were reimbursed for transportation at each clinic visit (equivalent of $10 USD). All participants were dispensed daily oral FTC/TDF for PrEP over the course of study follow-up. Originally, we planned to refill PrEP prescriptions on a monthly basis; however, due to the COVID-19 pandemic, this was changed to align with the three scheduled study visits. PrEP was not widely available to women in Malawi over the course of the study. When follow-up was complete, however, women who wished to continue PrEP were referred to local programs providing such services.

Intervention group

The Tonse Pamodzi adherence intervention was designed through a multi-step process that included formative research, expert consultation, and iterative development and refinement with stakeholders and end-users—a process that has been described elsewhere.¹⁹ We sought to address individual, relationship, and structural barriers to antiretroviral adherence, in ways that would enhance patient engagement and social support. The combination intervention comprised two parts. The first was integrated Next Step Counseling (iNSC), a semi-structured approach informed by motivational interviewing.

iNSC is designed to help participants identify relevant needs and generate their own solutions for improving, facilitating, and sustaining well-being, sexual health, and drug adherence. This approach was originally developed to support PrEP adherence in clinical trials for men who have sex with men.^23,24 To date, its use among pregnant and breastfeeding women has been limited and has not been formally evaluated. We selected iNSC for this study to include mental health and wellbeing in discussions of HIV prevention, center counseling flexibly on issues women found most relevant at a given session, and promote participant identification of most appropriate “next step” in their movement from a current situation to one that offered greater opportunities for wellbeing and protection of health.

Participants assigned to the intervention group received iNSC at four study visits (enrollment and at one, three, and six months). All sessions were facilitated using a specific process to guiding the discussion that included several “steps”: (1) introduce discussion and invite participant to an open conversation, (2) frame the discussion in terms of focus (wellbeing, sexual health, adherence), (3) explore the participant’s current situation and factors that facilitate and challenge them across the socioecological model, (4) tailor the remainder of the conversation to focus on improvements to current situation or movement from current situation to a better one, (5) identify, from the participant’s perspective, what would need to change or happen in any of the socioecological levels of influence for movement towards an improved situation to be possible, (6) strategize with participant on how that need could be met, (7) ask participant to agree to try out one of those strategies, and (8) close the session checking in on action-plans. Typically, counselors implemented steps 1 through 7 focused on mental-health and wellbeing and then completed steps 3 through 8 focused on PrEP use or sexual health more generally. Each session lasted approximately 15–40 minutes. In an effort to minimize contamination of the control group, iNSC sessions were conducted by dedicated counselors in a separate building on the hospital grounds.

Four counselors were trained to provide iNSC at study visits through a series of webinars and in-person workshops. This was guided by an interactive workbook developed specifically for this study. Training included role-playing sessions where interactions were recorded and then reviewed as a group. These activities were led by two iNSC experts (KRA, KF) who certified counselors that demonstrated competency in the counseling approach. Once the trial began, the team continued to monitor performance. We developed worksheets to guide counselors through the main components of the session. On a regular basis, our iNSC experts met with counselors to review their experiences, answer questions, and conduct targeted training as needed. All sessions were audio-recorded for later assessments of fidelity to the structured iNSC approach.

The second part of our intervention was the optional engagement of an adherence supporter identified by the participant, called an omukhulupilira (“trusted one” in Chichewa). The adherence supporter—who could be the woman’s partner, family member, or friend—attended an orientation session within a month of their selection. They helped to reinforce adherence behaviors at home and were invited to accompany the woman at follow-up visits and participate in specific parts of the iNSC sessions, after the participant and counselor identified strategies, so the omukhulupilira could assist with implementation of the plan. Although the identification of an adherence supporter was considered optional, it was encouraged by clinic staff. Other than the orientation session, there were no required interactions between the omukhulupilira and the study team.

Control group

In line with the standard of care at the time, participants allocated to the control group received standard counseling about preventing vertical HIV transmission, safe obstetrics, and newborn care, based on guidelines from the Malawi Ministry of Health. In addition, participants received targeted education about HIV treatment and prevention. This included the basics of HIV transmission and prevention, stages of HIV infection, the role of ART and PrEP in preventing vertical HIV transmission, as well as the mechanisms of action, dosing, and potential side effects of specific antiretroviral drugs. These were delivered at each study visit by study nurses not otherwise associated with the study intervention.

Study setting

Women were recruited from the antenatal clinic at Bwaila District Hospital in Lilongwe, Malawi. At the time of the study, PrEP—in the form daily oral FTC/TDF—had been recently approved for use by the Malawi Ministry of Health, including for pregnant and breastfeeding women.²¹ However, access was initially limited to pilot programs, particularly those focused on adolescent girls and young women. Although such services were not routinely available at Bwaila District Hospital, we received permission from hospital leadership to prescribe and dispense daily oral PrEP within maternal and child health services. This built upon our formative work examining perspectives on PrEP from different stakeholders, including antenatal/postnatal women, their partners, and frontline health providers.⁹

Retention and adherence outcomes

Women presenting for study visits were classified as retained at three and six months. Those who missed study visits were classified as not retained for that specific visit. If women returned later, they could be considered retained at the later study visit. To assess PrEP adherence, we measured tenofovir concentrations in plasma and intracellular tenofovir diphosphate concentrations in upper layer packed cells at three and six months. Results from these two assays were used to categorize PrEP adherence over the prior 28 days according to a published algorithm (supplemental information A).²⁵ This was used a proxy for adherence behaviors between study visits (i.e., enrollment to three months, three to six months). Tenofovir concentration adherence scores range from 0 to 5, with each category representing a specific adherence pattern: low number or no doses in the interval (score 0), a few doses in the interval (score 1), 1–2 doses per week (score 2), several doses early in the interval followed by a stop 1–2 weeks leading up to sampling visit (score 3), 4–5 doses per week (score 4), and approximately daily dosing (score 5). This algorithm simultaneously measures adherence over the span of days (plasma) and weeks (intracellular); however, it was developed in non-pregnant populations and there is evidence that tenofovir concentrations are lower in pregnancy. For this reason, we conducted a sensitivity analysis where score thresholds were lowered by 30% for women who were pregnant at three or six months. This reduction was based on findings from IMPAACT 2009,¹⁴ which measured intracellular tenofovir diphosphate levels in red blood cells, and studies examining plasma tenofovir during PrEP and antiretroviral therapy.^15,16 Drug concentration testing was performed at the University of North Carolina at Chapel Hill by the Clinical Pharmacology and Analytical Chemistry Core of the Center for AIDS Research. Women retained in the study with “functional” PrEP adherence (defined as scores 4 or 5) met our primary clinical outcome. Secondarily, we assessed the components of this composite outcome—retention and functional PrEP adherence among women retained in the study—separately at each assessment point.

Safety outcomes

At each study visit, we collected a recent medical history and conducted a physical examination. At three and six months, we performed serum creatinine testing to longitudinally assess renal function. In line with Malawi HIV guidelines for individuals taking PrEP, repeat HIV testing was performed every three months, using rapid antibody assays. If a woman tested HIV-positive, PrEP was discontinued and the diagnosis confirmed by HIV RNA concentrations. At each visit, we screened for social harms and intimate partner violence; for those who reported a new event, we worked with local agencies to ensure safety and mitigate risk. However, unless they specifically asked to withdraw, they were not discontinued from the study. Clinical and laboratory abnormalities were graded according to the U.S. National Institutes of Health Division of AIDS (DAIDS) Adverse Events Grading Tables.²⁶ We graded serum creatinine levels according to ranges defined by the upper limit of normal criteria. Because of physiologic changes that occur during pregnancy and following delivery, we did not use adverse events thresholds based on baseline values. Grade 3 or higher events were reported to the protocol team as serious adverse events and relatedness to study agent was documented.

Analytic approach

For our main clinical endpoint, we compared the proportion of women retained with functional PrEP adherence at three and six months between the intervention and control group using an estimated probability difference (PD), analogous to a risk difference, and a corresponding Wald confidence interval (CI). Using the same approach, we also compared the two components of the composite outcome—retention in care and functional PrEP adherence among those retained in care—between randomization groups. In sensitivity analyses, we used inverse probability of treatment weights to estimate a PD and corresponding Wald CI that accounted for imbalances in baseline covariates that were associated with retention and PrEP adherence outcomes. This included socioeconomic status, financial dependence on partner, gestational age at enrollment, intercourse frequency in the 30 days prior to enrollment, perceived HIV risk at enrollment, and number of lifetime sex partners. For our safety outcomes, we compared incidence between intervention and control groups using Fisher’s Exact Tests. A type I error of 0.05 (i.e., 95% CI) was applied with no adjustment for multiplicity given the pilot nature of this study. Because <2% of data were missing, we used a complete-case approach for all analyses. All analyses were conducted using Windows SAS version 9.4 (SAS Institute, Cary, NC, USA).

Sample size considerations

As a pilot trial, this study was not powered to detect a priori differences in the above-stated outcomes at the expected levels.²⁷ Instead, the sample size was determined based on operational considerations such as projected recruitment, time, and resources. The target sample size of 200 (100 per arm) was deemed adequate to characterize various study endpoints, including the adherence and retention outcomes that are the focus of this analysis.

Ethical review

Our study protocol was reviewed and approved by the National Health Services Research Committee (Lilongwe, Malawi) and the University of North Carolina at Chapel Hill Institutional Review Board (Chapel Hill, NC, USA). The trial was registered at clinicaltrials.gov (NCT04330989).

RESULTS

From June 2020 to November 2020, we approached 255 HIV-negative women about the study. Of these, 200 were enrolled and randomly allocated 1:1 to either the control arm (n=100) or the intervention arm (n=100). A CONSORT diagram for this trial is displayed in Figure 1. Baseline characteristics are shown by study arm in Table 1. Among the 100 participants in the intervention arm, 56 selected an omukhulupilira who subsequently completed the orientation session. People chosen for this role included a partner (n=41), other family member (n=14), or friend (n=1).

Figure 1.

Abbreviations: PrEP= oral pre-exposure prophylaxis with daily emtricitabine/tenofovir, M6= Month 6

Table 1:

Baseline characteristics of HIV-negative pregnant women enrolled from June 2020 to November 2020

		Intervention group (n=100)	Control group (n=100)	Total (n=200)
Age at enrollment (years)	Median (Q1, Q3)	25 (22, 28)	24 (21, 28)	24 (21, 28)
Primary school complete	Yes	48 (48%)	56 (56%)	104 (52%)
Electricity in home	Yes	17 (17%)	36 (36%)	53 (27%)
Running water in home	Yes	27 (27%)	30 (30%)	57 (29%)
Partner is primary source of money	Yes	66 (66%)	57 (57%)	123 (62%)
Time to clinic (minutes)	Median (Q1, Q3)	40 (30, 53)	40 (30, 45)	40 (30, 45)
	< 30 minutes	1 (1%)	1 (1%)	2 (1%)
	30 to 59 minutes	75 (75%)	85 (85%)	160 (80%)
	60+ minutes	24 (24%)	14 (14%)	38 (19%)
Gestational age at enrollment (weeks)	Median (Q1, Q3)	28 (20, 33)	26 (17, 34)	26 (19, 33)
	First trimester	8 (8%)	16 (16%)	24 (12%)
	Second trimester	40 (40%)	43 (43%)	83 (42%)
	Third trimester	52 (52%)	41 (41%)	93 (47%)
Gravidity	No prior pregnancies	27 (27%)	19 (19%)	46 (23%)
Number of living children	Median (Q1, Q3)	1 (0, 2)	1 (0, 2)	1 (0, 2)
	No living children	32 (32%)	29 (29%)	61 (31%)
	1–2 living children	46 (46%)	55 (55%)	101 (51%)
	3+ living children	22 (22%)	16 (16%)	38 (19%)
Perceived risk of acquiring HIV in next 12 months	No chance at all	21 (21%)	19 (19%)	40 (20%)
	Small chance	21 (21%)	21 (21%)	42 (21%)
	Moderate chance	26 (26%)	20 (20%)	46 (23%)
	Great chance	32 (32%)	40 (40%)	72 (36%)
Number of lifetime sex partners	1 lifetime sex partner	31 (31%)	18 (18%)	49 (25%)
	2–3 lifetime sex partners	63 (63%)	61 (61%)	124 (62%)
	4+ lifetime sex partners	6 (6%)	21 (21%)	27 (14%)
Primary partner age (years) ^	Median (Q1, Q3)	29 (25, 35)	29 (26, 35)	29 (25, 35)
Relationship length (years) ^	Median (Q1, Q3)	3 (1, 6)	2 (1, 5)	2 (1, 6)
Residing with primary partner ^	Yes	85 (88%)	87 (88%)	172 (88%)
Married to primary partner ^	Yes	91 (94%)	92 (93%)	183 (93%)
Number of sexual intercourse acts in the past 30 days ^	Median (Q1, Q3)	8 (3, 12)	5 (2, 10)	6 (2, 12)
Condom use with primary partner in past 30 days *	Never	71 (87%)	77 (87%)	148 (87%)
	Sometimes	9 (11%)	8 (9%)	17 (10%)
	Consistent	2 (2%)	4 (4%)	6 (4%)
Primary partner HIV status^	HIV-negative	73 (75%)	73 (74%)	146 (74%)
	HIV-positive	4 (4%)	6 (6%)	10 (5%)
	Partner never tested for HIV	3 (3%)	0 (0%)	3 (2%)
	I don’t know	17 (18%)	20 (20%)	37 (19%)

^{^}Among women who reported at least one sex partner in the past three months (n=196)

^*Among women who reported at least one sex act in the past 30 days (n=171)

Retention and adherence outcomes

Overall, 179 (89.5%) of 200 women completed a three-month visit and 171 (85.5%) of 200 women completed a six-month visit. Retention did not differ between intervention and control groups at either three months (90.0% vs. 89.0%; PD:1.0%, 95%CI −7.5%, 9.5%) or six months (88.0% vs. 83.0%; PD:5.0%, 95%CI:−4.7%, 14.7%).

Nearly all participants retained in care had tenofovir concentration levels reported: 178 at three months and 170 at six months. The distribution of PrEP adherence scores among women retained in care are shown in Figure 2. Overall, 57 (32.0%) of the 178 women retained at three months met our definition of functional adherence and no substantial difference was observed between intervention and control groups (31.5% vs. 32.6%, PD:−1.1%, 95%CI:−14.8%, 12.6%). At six months, 48 (28.2%) of the 170 women retained in care had drug concentration levels consistent with functional PrEP adherence. Again, similar proportions were observed in the intervention and control groups (25.0% vs. 31.7%, PD:−6.7%, 95%CI:−20.3%, 6.8%). A substantial proportion of women had drug concentrations consistent with a low number or no doses of FTC/TDF at three months (34.3%) and at six months (49.4%). Score distributions were largely unchanged in sensitivity analyses, where thresholds were lowered by 30% for women who were pregnant (supplemental information B).

Figure 2.

Distribution of PrEP adherence scores based on tenofovir-based algorithm proposed by Corneli, et al. (21) according to study visit and randomization group

When retention and functional adherence were considered together in our composite outcome (n=199 at each assessment point), we did not observe differences between the intervention and control groups at three months (28.3% vs. 29.0%, PD:−0.7%, 95%CI:−13.3%, 11.8%) or at six months (22.0% vs. 26.3%, PD:−4.3%, 95%CI:−16.1%, 7.6%). Results from our adjusted analyses did not meaningfully differ from our unadjusted analyses (Table 2).

Table 2:

Retention in care and PrEP adherence based on tenofovir drug concentrations by study group

Outcome	Time point	n *	Intervention	Control	Unadjusted probability difference (95% CI)	Adjusted probability difference (95% CI)^
Retained in care
	At Month 3	200	90/100 (90.0%)	89/100 (89.0%)	1.0% (−7.5%, 9.5%)	5.1% (−3.7%, 14.0%)
	At Month 6	200	88/100 (88.0%)	83/100 (83.0%)	5.0% (−4.7%, 14.7%)	7.4% (−2.7%, 17.4%)
Functional PrEP levels among women retained in care (score of 4 or 5)
	At Month 3	178	28/89 (31.5%)	29/89 (32.6%)	−1.1% (−14.8%, 12.6%)	−4.5% (−20.0%, 11.0%)
	At Month 6	170	22/88 (25.0%)	26/82 (31.7%)	−6.7% (−20.3%, 6.8%)	−9.7% (−34.4%, 4.9%)
Retained in care with functional PrEP levels (score of 4 or 5)
	At Month 3	199	28/99 (28.3%)	29/100 (29.0%)	−0.7% (−13.3%, 11.8%)	−1.8% (−16.2%, 12.7%)
	At Month 6	199	22/100 (22.0%)	26/99 (26.3%)	−4.3% (−16.1%, 7.6%)	−5.5% (−18.0%, 6.9%)

^*One specimen misplaced at Month 3 and one specimen misplaced at Month 6; outcomes set to missing

^{^}Model adjusted for baseline imbalances in gestational age (continuous, in weeks), socioeconomic status (measured as presence of electricity in the home), financial dependence on partner (yes vs. no), number of lifetime sexual partners (continuous), intercourse frequency in the 30 days prior to enrollment (continuous), and perceived HIV risk at enrollment (no or low risk vs. moderate or high risk)

Safety

Safety outcomes are shown in Table 3. At three and six months, no participants had serum creatinine elevations at grade 2 or higher. Social harms due to study participation were infrequently reported: two participants at three months (both in the intervention arm) and three participants at six months (two in the control group, one in the intervention group). Four of these five participants reported emotional harm from partners, family members, and/or friends; financial and physical harms were reported as well. When specifically asked about intimate partner violence, only one participant (intervention group) reported an issue at three months. At six months, five women in the intervention group and one in the control group reported experiencing intimate partner violence. These incidents were emotional, physical, and/or sexual in nature, but none were associated with study participation.

Table 3:

Safety outcomes by study group

	n	Intervention group n/N (%)	Control group n/N (%)	Fisher’s p-value
Elevated serum creatinine (grade 1 or higher)
3 months	179	0/89 (0.0%)	0/90 (0.0%)	--
6 months	171	0/89 (0.0%)	0/83 (0.0%)	--
Intimate partner violence
3 months	179	1/90 (1.1%)	0/89 (0.0%)	1.0
6 months	171	5/89 (5.7%)	1/83 (1.2%)	0.2
Other social harms
3 months	179	0/90 (0.0%)	2/89 (2.2%)	0.2
6 months	171	1/88 (1.1%)	2/83 (2.4%)	0.6
Obstetric Complications
Maternal Sepsis	188	2/95 (2.1%)	0/93 (0.0%)	0.4
Preeclampsia/Eclampsia	188	2/95 (2.1%)	0/93 (0.0%)	0.5
Antepartum Hemorrhage	188	1/95 (1.1%)	0/93 (0.0%)	1.0
Postpartum Hemorrhage	188	0/95 (0.0%)	2/93 (2.2%)	0.2
Adverse Birth Outcomes
Preterm birth (< 37 weeks)	188	5/95 (5.3%)	6/93 (6.5%)	0.7
Stillbirth/Intrauterine fetal death	188	7/95 (7.3%)	2/93 (2.2%)	0.2
Miscarriage/Spontaneous abortion	188	1/95 (1.1%)	1/93 (1.1%)	1.0
Neonatal Death*	178	3/88 (3.4%)	2/90 (2.2%)	0.7
New maternal HIV infection	171	0/89 (0.0%)	1/83 (1.2%)	0.5

^*Among women who reported a live birth

Among the 188 participants with available birth outcomes data, 11 (4.2%) were preterm deliveries (i.e., <37 weeks) and 2 (1.1%) were spontaneous abortions. Nine stillbirths (4.8%) were observed: seven in the intervention group vs. two in the control group (p=0.2). Additional obstetrical complications were rare and did not differ by randomization group. Five neonatal deaths were reported: three in the intervention group and two in the control group (p=0.7).

Over follow-up, one participant acquired HIV (incidence rate: 0.78/100 person-years). This individual was diagnosed at her six-month study visit and was in the control group; her drug concentrations were consistent with few to no doses per week (score 0). In HIV genotype testing, no mutations were associated with antiretroviral drug resistance.

DISCUSSION

We successfully conducted a pilot randomized trial of PrEP adherence support in antenatal and postnatal settings. Overall, we demonstrated >85% retention over a window that included pregnancy, delivery, and postpartum. Among those in care, approximately 30% demonstrated functional PrEP adherence according to published pharmacological thresholds. We did not observe differences in retention and adherence between our intervention and control groups, though the pilot trial was not powered for such comparisons. This study provides insights about the conduct of PrEP research in this target population, with lessons to inform future work in this important and emerging field.

PrEP has been integrated into maternal and child health platforms, but persistence and adherence remain areas of concern. Among approximately 2,000 pregnant women who initiated daily oral PrEP in Kenya, only 39% returned after one month.⁷ Among 1,014 pregnant women initiating PrEP in South Africa, 66% returned at one month and 58% returned at three months.⁸ In this pilot trial, we demonstrated the ability to rapidly enroll and retain pregnant women into a behavioral intervention trial to enhance PrEP adherence. We were encouraged by comparatively high rates of retention at three and six months, despite the complexity of a randomized trial design and follow-up that spanned pregnancy, delivery, and postpartum. These findings support the feasibility of PrEP delivery in antenatal and postnatal settings, a current implementation gap.²⁸

We incorporated tenofovir drug concentrations—an objective proxy for adherence—into our primary clinical outcome. Few trials have measured drug concentrations longitudinally to confirm adherence during pregnancy and breastfeeding. PrEP in Pregnant and Postpartum Women (PrEP-PP), an implementation study in South Africa’s Western Cape, evaluated adherence via intracellular tenofovir diphosphate in dried blood spots. In an analysis of 687 samples from 382 women, 31% had drug concentrations associated with recent PrEP uptake of 2–6 doses/week. Only 2% met the threshold consistent with approximately 7 doses/week.¹² Although different assays and adherence algorithms were used, our estimate for 2–5 doses/week (score of 3 or 4) was similar at 33%. However, we observed a higher proportion with drug levels consistent with daily dosing (score of 5; 10%). These proportions were not meaningfully different when we adjusted score thresholds for pregnancy.

Our study has important lessons for future work. Amico and colleagues previously described a Mutuality Framework that considers dynamics between the individual, community, and researchers in the context of study-provided PrEP.²⁹ Within this framework, four such dynamics were derived from thematic analyses; in each, the relationship between constituencies can influence key behaviors around study-provided PrEP. At time of trial design, we assumed that individuals would be generally willing to initiate PrEP, they would be persistent, but adherence to daily dosing might vary. (In retrospect, this scenario approximated the dynamic of alignment within the Mutuality Framework.) Given our findings, however, it is reasonable to question this original assumption. Because our study was among the first to provide daily oral PrEP to eligible pregnant women in Malawi, dynamics such as distrust or uncertainty may also have been prevalent. Distrust is associated with concerns and misconceptions about safety, including from members of the community, and this may negatively influence PrEP uptake. It may be more frequent when there are structural or economic motivators to trial participation (such as transport reimbursements provided by this study) and often results in high retention but low adherence. Uncertainty, on the other hand, can lead to an oscillation between PrEP use and non-use and thus variable persistence in PrEP use. Such themes emerged in an ancillary, mixed-methods study of trial participants,³⁰ lending further support to these alternate dynamics. Correctly matching outcomes with the local context, as described by the Mutuality Framework, is essential in future research. Design features, such as options to decline or discontinue PrEP, may prove valuable for PrEP-related behaviors across a wider care continuum.³¹

Another important consideration is the motivation for PrEP use during pregnancy and breastfeeding. Unlike antiretroviral therapy, which is initiated and continued for life, HIV prevention is only needed during periods of elevated risk for HIV exposure. When designing this study, we reasoned that the high HIV incidence across pregnancy and breastfeeding would sufficiently motivate participants to adhere to PrEP over six months. In reality, however, women had their own perceptions of HIV risk over time and these evolving “seasons of risk” can drive adherence behaviors.³² Numerous contextual factors may have contributed to the fluctuating perceptions of HIV risk, including varying coital frequency over the course of pregnancy and breastfeeding,³³ local traditions (e.g., customs of staying with extended family following childbirth), or public safety measures during the COVID-19 pandemic. As potential for HIV exposures decreased—or ceased altogether—PrEP use may appropriately decrease or stop as well. In HPTN 082, periods of higher HIV risk among adolescent girls and young women was associated with high PrEP adherence (as defined by drug concentrations).³⁴ It is possible a similar dynamic was at play in our trial. Further, because women in this trial were otherwise healthy—and because PrEP was relatively new in Malawi at the time of this trial—perceptions of HIV risk also had to be sufficiently high overcome the risk threshold for other adverse events, including side effects and fetal drug exposure.⁹

The iNSC component of our intervention—which is designed to engage and empower its clients—could have also led participants in the intervention group to differentially evaluate their individual HIV risk or objectively reduce their risk. For example, women undergoing iNSC could have developed sexual health protection plans that obviated the need for PrEP. This could include intervention to increase partner HIV testing—either through couples approaches or via HIV self-testing³⁵—that better characterizes a women’s perception of HIV risk. If HIV risk was deemed low, or if the perceived importance of PrEP changed over time, the intervention itself could have precipitated early cessation of PrEP, an unanticipated outcome of our approach. A more nuanced and relevant approach to PrEP assessment could include “prevention-effective adherence”—that is, measuring adherence only in those periods when HIV risk is subjectively or objectively high.³⁶ Such outcomes should be considered as part of future work.

Despite the strengths of this study, we recognize some limitations. First, numerous factors may been associated with adherence during pregnancy, including male partner engagement, social support, and mental health factors.^37–39 Although participants were randomly assigned to study arms, the sample size was small and groups may have been imbalanced according to important covariates. Second, as a pilot trial, we instituted specific eligibility criteria that could limit the generalizability of our findings. For example, although women experiencing intimate partner violence have poorer adherence outcomes—and may thus benefit from this type of behavioral intervention—we elected not to include them in this pilot study. Instead, we obtained valuable, reassuring safety data to support the use of the intervention in future phase 3 studies. Third, most participants who stopped PrEP over the course of the study did so silently, without reporting it to study staff. For this reason, we do not have information about the circumstances surrounding discontinuation. Finally, this report focuses narrowly on clinical outcomes, such as adherence and safety, and is only one part of our broader pilot evaluation. Already, we have learned that the intervention was highly acceptable for participants via mixed methods.⁴⁰ We are also evaluating fidelity to the iNSC approach, an indicator of feasibility, via audits of audio-recorded sessions. Together, these components will inform future work, as we seek to further refine the study intervention for broader use and evaluation.

In summary, few interventions have been shown to effectively improve PrEP adherence in sub-Saharan Africa and none among pregnant and breastfeeding women. Results from this rigorously conducted pilot randomized trial provide valuable insights into PrEP adherence in antenatal and postnatal settings. As national programs expand the reach of PrEP to these important populations,⁴¹ behavioral interventions can play an important role in tailoring services to individual needs. As PrEP modalities and policies rapidly evolve, patient engagement, empowerment, social support, and adherence remain highly relevant and should be considered an implementation priority.