Modeling the Potential Impact of Mentor Mother Programs on Vertical Transmission of HIV in Kenya

Abstract

Background

Peer support provided by “mentor mothers” (MM) may increase retention in care for pregnant and breastfeeding women (PBFW) living with HIV, which can lead to improved viral suppression and prevention of vertical transmission (PVT). The impact of MM on PVT may differ among PBFW subpopulations with different baseline risks of antiretroviral therapy (ART) interruption.

Method

We used a microsimulation model of HIV progression and care for PBFW living with recently acquired HIV in Kenya to evaluate an MM program that reduced the cumulative risk of ART interruption between initiation of antenatal care and 18 months of postpartum breastfeeding by 40%. We conducted this evaluation in 2 maternal populations: (1) newly positive (NP) pregnant women who initiate ART during antenatal care and are at greater risk of ART interruption and (2) known positive (KP) women who initiate ART prior to conception. We simulated live births, maternal deaths, and vertical transmission with and without MM services to determine the impact of MM.

Results

In the absence of MM services, infants acquired HIV at a rate of 609 vs. 459 per 10,000 live births in the NP vs. KP populations. MM services reduced vertical transmission by 15.1% among NP women, compared to a 6.5% reduction among KP women.

Conclusions

MM services can reduce vertical transmission among PBFW living with HIV, with greater potential for PVT among NP women compared to KP women. More research evaluating MM programs in these 2 maternal populations will help refine estimates of their PVT impact.

The global expansion of maternal antiretroviral therapy (ART) coverage from 49% in 2010 to 84% in 2023 has significantly decreased vertical (mother-to-child) transmission of HIV (VT) [1], as viral suppression substantially lowers the risk of VT [2]. Continued progress toward the elimination of mother-to-child transmission will require not only further improvements in maternal ART coverage rates but also additional reductions in ART interruption among pregnant and breastfeeding women (PBFW). UNAIDS estimates that, globally, 19% of infants who acquired HIV in 2023 had mothers who initiated ART but were later lost to follow-up [1], indicating substantial potential for prevention of vertical transmission (PVT) in this maternal subpopulation. Numerous interventions for improving retention among PBFW have been implemented in areas with high HIV burden [3]. Of these, peer support programs have showed particular promise, with “Mentor Mother (MM)” programs being one of the most widely implemented. MM are peer advocates who operate through PVT programs to provide support to PBFW living with HIV during antenatal care and breastfeeding. Research to date suggests that MM programs improve rates of retention in care [4–7], when core components are implemented with high fidelity [8]. However, empirical evidence showing that MM decreases VT is lacking, due to the significant additional resources needed to track VT outcomes, especially among mother–infant dyads not retained in care.

Modeling can address this research gap by translating the MM improvement in retention to a corresponding reduction in VT. While some modeling studies have done this for select African settings [9–11], none have investigated whether the PVT impact of MM meaningfully differs for PBFW subpopulations with different baseline retention rates—information that could help prioritize MM implementation. In response, we used simulation modeling to evaluate how MM programs impact PVT in 2 key maternal populations: newly positive pregnant women diagnosed and initiated on dolutegravir (DTG)-based ART upon presenting to antenatal care, and known positive pregnant women living with HIV who initiate DTG-based ART prior to conception. We focus our analysis on Kenya, a high HIV burden country with a commitment to MM program implementation where differences in retention in care between newly positive and known positive PBFW have been observed [12, 13].

METHODS

Overview

We used a previously described microsimulation model of HIV progression and PVT care for pregnant women living with HIV in Kenya to evaluate the vertical transmission impact of enrolling in an MM program upon entering antenatal care [14]. The model is specifically parameterized for 25-year-old PBFW with recently acquired HIV, a population for whom viral suppression patterns have been well-characterized [15]. Table 1 shows key model parameters (base-case values and sensitivity analysis ranges). The model updates age and key clinical characteristics such as CD4 cell count, viral load (VL), and ART regimen status in monthly time cycles. Pregnancy, modeled as 9 monthly time cycles (39 weeks gestation), is confirmed at the first antenatal clinic (ANC) visit at 5 months gestation [17] and concludes in a full-term, singleton live birth. The newborn is explicitly represented in the model as a new individual who may acquire HIV during the peripartum or postpartum periods, either through a single peripartum risk (inclusive of VT that occurs in utero and at delivery) or through an ongoing monthly risk during breastfeeding. Women may die from HIV, maternal or other-cause mortality. Infants have ongoing all-cause mortality risk until HIV acquisition, after which we cease tracking infant outcomes.

Table 1.

Base Case and Sensitivity Values for Key Model Parameters

Parameter	Base-Case Value (Sensitivity Ranges)		References
Viral suppression on dolutegravir-based ART
ART effectiveness: virologic suppression (%)a			Kinuthia et al [15] Dorward et al [16]
Month 16 of ART	91.6 (84.0–94.7)
Month 33 of ART	87.6 (80.4–92.0)
Adherence to viral load testing schedule (% attending each recommended test)	50 (0–100)		Assumptionb
Antenatal and HIV Care
Months gestation at first antenatal clinic visitc	5 (3–8)		Aksunger et al [17]
Months duration of breastfeeding	18 (12–24)		Glaubius et al [18]
Interrupted ART (%, from first antenatal clinic visit to end of breastfeeding)a,d	Known Positive	Newly Positive	Humphrey et al [13] Chammartin et al [19]
	8.5 (3.4–13.6)	23.5 (9.4–37.6)
Impact of MM program: reduction in cumulative ART interruption at end of breastfeeding (%)	40 (20–60)		Assumption (see Appendix 1.7.1)
Peripartum Vertical Transmissione
Probability in month of delivery (%), mother on ART	Known Positive	Newly Positive	Mandelbrot et al [2] Townsend et al [20]
VL < 50 copies/mL	0.0 (0.0–0.0)	0.5 (0.3–0.8)
50 < VL < 400 copies/mL	0.3 (0.2–0.5)	1.4 (0.7–2.1)
400 < VL < 10 000 copies/mL	2.2 (1.1–3.3)	2.4 (1.2–36)
VL > 10 000 copies/mL	9.2 (4.6–13.8)	9.2 (4.6–13.8)
Probability in month of delivery (%), mother not on ART and HIV acquired preconceptionf			Rollins et al [21]
CD4 < 200 cells/mL	37
200 < CD4 < 350 cells/mL	27
CD4 > 350 cells/mL	15
Postpartum (Breastfeeding) Vertical Transmission			Dugdale et al [22]
Monthly probability (%), mother on ART
VL < 50 copies/mL	0.06 (0.05–0.07)
50 < VL < 1000 copies/mL	0.39 (0.34–0.44)
VL > 1000 copies/mL	0.78 (0.67–0.89)
Monthly probability (%), mother not on ARTf	0.89

See Supplementary Appendix for additional details.

Abbreviations: ART, antiretroviral therapy; VL, viral load; MM, mentor mother program.

^aReported as a cumulative risk for ease of interpretation; modeled as a monthly risk. In the scenario where MM improves ART effectiveness, these parameters are set to their upper bound.

^bDue to gaps in the testing cascade, suboptimal VL monitoring is assumed (see Supplementary Appendix 1.5.4).

^cGestational age is measured in months due to the monthly time cycle of the model. Corresponding ages in weeks are: month 5 = weeks 17.4–21.7; month 3 = weeks 8.8–13; month 8 = weeks 30.4–34.7.

^dFor known positive women, the corresponding monthly risk of ART interruption also applies to the time period between ART initiation and first ANC visit.

^ePeripartum vertical transmission parameters reflect combined in utero and intrapartum transmission.

^fSensitivity analyses were conducted by varying on-ART transmission risk relative to these values, which were kept constant.

We used this model to construct 2 cohorts of pregnant women living with recently acquired HIV who present to antenatal care: a newly positive (NP) Cohort who initiates dolutegravir-based ART at the first ANC visit, and a known positive (KP) Cohort that has been on DTG-based ART for 6 months at their time of conception (Figure 1). HIV acquisition occurs 1 month prior to conception for all NP women and 1 year prior to conception for all KP women. Key maternal events are also synchronized within each cohort, including conception, the first ANC visit, delivery, and the conclusion of breastfeeding at 18 months postpartum. Women have a monthly risk of ART interruption that is higher for NP women, following empirical studies of NP versus KP women in Kenya (Table 1) [12, 13]. In the KP Cohort, we model the first ANC visit as an opportunity for 100% of those with interrupted ART to re-engage, after which they resume being at risk for ART interruption. We modeled 2 strategies for each cohort: (1) a standard of care (SOC) reflecting ART interruption without MM services, and 2) an implementation of MM services that reduces the cohort's monthly risk of ART interruption, as described in the following section. All simulations were programed in Matlab R2021a and results were analyzed using R 4.4.3.

Figure 1.

Simulation and maternal timelines for the newly and known positive cohorts. The simulation runs for 40 time steps (middle panel). The maternal timelines indicate months since conception (upper and lower panels), with Mentor Mother (MM) starting at month 5. Simulation time steps 23–28 not shown. ANC, antenatal clinic.

MM Program

While existing MM programs have been implemented in different ways, we focus on the potential for MM programs to increase retention in care, which can reduce VT through the improvement in viral suppression that arises from reduced ART interruption. Women enter the MM program upon initiating ART (NP women) or continuing or resuming ART (KP women) at the first ANC visit, and based on a synthesis of empirical studies of MM impact on retention in PVT care [4–7], we assume MM reduces each cohort's cumulative risk of ART interruption by the conclusion of breastfeeding by 40% (Table 1 and Supplementary Appendix 1.7.1). However, we assume the MM program has no impact on the probability of ART initiation or reinitiation.

ART Effectiveness and VL Monitoring

We define ART effectiveness as the probability that the regimen successfully reduces and maintains VL below 1000 copies/mL (see Supplementary Appendix 1.3.2 for VL dynamics). Reflecting both the pharmacologic characteristics of the regimen and the individual's adherence, this estimate is informed by data from Kenya and South Africa (Supplementary Table A5) [15, 16]. Effective ART also (1) promotes CD4 cell count recovery, which in turn reduces the risk of acute clinical events and HIV-related death, and (2) decreases VT risk, particularly when initiated earlier in the maternal timeline (Table 1). We model a monthly risk that poor adherence renders ART ineffective, at which point VL rebounds to pretreatment levels [23], CD4 cell count decline resumes, and VT risk increases. These same changes occur when women experience ART interruption.

We assume that women in care are offered VL monitoring at intervals determined by age- and PBFW-based national guidelines (Supplementary Appendix 1.5.4). Given existing gaps in the VL testing cascade among PBFW [24–28], each time a VL test is indicated, we assume that there is only a 50% probability that the VL test occurs. If a routine VL test detects VL ≥ 200 copies/mL, this prompts adherence counseling, per Kenyan guidelines [29]. These individuals then have a 50% probability of re-establishing VL declines consistent with effective ART (“resuppression,” Table 1).

Health Outcomes and Sensitivity Analyses

For each strategy, we projected maternal viral suppression below 1000 copies/mL, live births, maternal deaths, and peripartum versus postpartum infant infections. To standardize the strategies for potential differences in live births, we quantified the PVT impact of the MM program as the relative (percent) reduction in infant infections per 10 000 live births compared to SOC. We conducted one-way sensitivity analyses for key model parameters, and also investigated 2 alternative MM scenarios: (1) an MM program with ART interruption reduction effects ending early, at 6 or 12 versus 18 months postpartum, and (2) an MM program that not only reduces ART interruption but also improves adherence, which we implement by using a higher level of ART effectiveness (Table 1, see Supplementary Appendix for calculations).

Ethics

Data used to inform model parameters in this manuscript were publicly available aggregate data, for which consent is not required.

RESULTS

Maternal Viral Suppression

Under SOC, the KP Cohort had a high rate of viral suppression at conception due to their prior time on ART. The NP Cohort was viremic at conception but reached similarly high viral suppression levels within 3 months of initiating ART at the first ANC visit (Figure 2). In both cohorts, viral suppression declined over time due to the accumulation of ineffective and interrupted ART. This decline was steeper in the NP Cohort due to their greater ART interruption risk (Figure 3). In the KP Cohort, re-engagement in ART at the first ANC visit for those with interrupted ART led to a small increase in the proportion with viral suppression, after which viral suppression continued to decline. Although ART effectiveness was the same in both cohorts, their different timing of ART initiation relative to maternal events and different ART interruption risks resulted in the KP Cohort having higher viral suppression under SOC at conception, 9 months postpartum, and 18 months postpartum, whereas the NP Cohort had slightly higher viral suppression at delivery.

Figure 2.

Viral suppression below 1000 copies/mL in the 2 maternal cohorts under each strategy, over maternal time. The x-axis indicates months since conception. The y-axis between 15% and 50% not shown. Color indicates cohort. Key events occur at the same maternal time point for both scenarios but are only marked on the standard of care (SOC, solid) lines. MM, Mentor Mother (dashed lines); ANC, antenatal clinic.

Figure 3.

ART interruption in the 2 maternal cohorts under each strategy, at key maternal time points. After the first antenatal clinic visit, both cohorts have 0% ART interruption. ART interruption accumulates over the course of the remainder of antenatal care and postpartum. Colors indicate standard of care (SOC) versus Mentor Mother (MM) strategies. Top panel: newly positive cohort; bottom panel: known positive cohort.

Upon its initiation at each cohort's first ANC visit, the MM program increased the percentage of women with viral suppression at each subsequent time point by improving retention and thus decreasing ART interruption (Figure 2). This impact was larger for NP than for KP women, as the NP Cohort had higher ART interruption under the SOC at each maternal time point, and thus more women who could benefit from MM (Figure 3). In both cohorts, the number of women who could benefit from MM increased at later maternal time points, as ART interruption was zero at the first ANC visit but accumulated over subsequent months (Figure 3).

Live Births and Maternal Deaths

Under SOC, the NP Cohort had fewer live births and more maternal deaths than the KP Cohort, due to their later initiation of ART (Supplementary Table 1). Through its impact on ART interruption, MM support reduced cumulative mortality among NP women at 18 months postpartum by 3.1%. In contrast, MM had only a very small mortality reduction among KP women (Supplementary Table 1).

Vertical Transmission

Figure 4 A shows how viral suppression in the 2 cohorts translated into absolute numbers of peripartum and postpartum infections per 10 000 live births. Under SOC, both peripartum and postpartum VT were higher in the NP Cohort than in the KP Cohort (150 more total infant infections per 10 000 live births), consistent with the NP Cohort's greater degree of ART interruption at each maternal time point (Figure 3 and Table 1). The improvements in viral suppression facilitated by MM support led to a corresponding reduction in VT for both cohorts, with more peripartum and postpartum infections averted in the NP Cohort than in the KP Cohort. The relative reduction in total VT due to MM support in the NP Cohort was over double that of the KP Cohort (15.1% vs 6.5%; Figure 4B).

Figure 4.

Peripartum, postpartum, and total vertical transmission in the 2 maternal cohorts under each strategy. Panel A: number per 10 000 live births. Panel B: percent reduction in vertical transmission due to the Mentor Mother (MM) program, compared to the standard of care (SOC). NP, newly positive; KP, known positive.

Sensitivity Analyses

In one-way sensitivity analyses, the relative reduction in VT due to MM support increased substantially with higher MM program effect, earlier gestation at first ANC visit, and greater baseline ART interruption (Supplementary Figure 1). Across low and high settings of these parameters, the relative reduction in total VT ranged from 4.1% to 22.3% in the NP Cohort and from 2.7% to 9.6% in the KP Cohort, with high MM program effect yielding the maximum reduction in VT in both cohorts. The ART effectiveness and breastfeeding parameters had lesser influences in both cohorts, as did the peripartum and postpartum VT risks (Supplementary Figures 2–3). Shortening the MM program to conclude at 12 months reduced PVT impact slightly, whereas shortening the program to 6 months postpartum further reduced relative reductions to 11.8% (NP) and 5.0% (KP) (Supplementary Table 2). In the sensitivity analysis where MM improved adherence in addition to reducing ART interruption, the relative reduction in VT increased in both cohorts, but more significantly in the KP Cohort (from 6.5% to 11.1%) than the NP Cohort (from 15.1% to 18.6%) (Supplementary Table 3).

DISCUSSION

We used a simulation-based model to evaluate the potential impact of MM services on PVT outcomes among Kenyan PBFW living with HIV who have initiated ART. Our analyses revealed 4 key insights that have important implications for future MM program implementation and research.

First, the PVT benefit of an MM program is likely to be substantially greater among NP women than among KP women. This finding reflects our empirically based assumption that ART interruption rates in the absence of an MM program are higher among NP women, leaving more room for improvement via MM, and is consistent with our sensitivity analysis showing that the relative reduction in VT due to MM is very sensitive to baseline ART interruption. These findings highlight the value of PVT programs having reliable retention data, collected using standard measures [30] that can be used to prioritize MM roll-out to higher-need populations. While retention data from Kenya show differences between NP and KP women [12, 13], additional research is needed to investigate the consistency of this finding across high HIV burden countries.

Second, the reduction in ART interruption associated with MM services is one of the most influential parameters but also one of the most uncertain. Our base-case assumption that MM reduces ART interruption by 40% was derived from a synthesis of data from studies conducted throughout Africa, across which the effect size varied widely. Much of this variability is likely due to heterogeneous approaches to implementation of MM across diverse settings. Most of the evidence came from observational studies that varied by location, baseline retention rates, and mixture of KP versus NP PBFW. Future implementation research could help determine which MM strategies are most effective and whether this effectiveness varies across groups with different baseline retention rates (eg, NP vs KP).

Third, PBFW populations with greater ART interruption, such as NPs and younger women [31], may stand to benefit the most from MM programs with longer-term effects. While we found that shortening the duration of the MM program effect to 6 months postpartum decreased its PVT potential in both cohorts, the absolute reduction was greater for NP women than for KP women. However, additional MM evaluation studies with multiple follow-up time points throughout postpartum are needed to help clarify how long the effects of MM last and whether the duration of effect depends on how the MM program is implemented. Our results suggest that these studies should be prioritized in PBFW populations with higher risk of ART interruption.

Fourth, if MM programs also improve rates of viral suppression by improving ART adherence, their potential PVT impact would increase, particularly among KP women. While some pre-DTG era studies found that, among those on ART, MM support was associated with a higher rate of viral suppression [4, 9], this finding was not consistent across all studies [6]. The current widespread use of potent DTG-based ART may leave minimal room for improvement in ART effectiveness. We thus assumed that the MM program did not improve rates of viral suppression among those on ART in our base-case analysis. However, in a scenario analysis where MM increased the effectiveness of DTG-based ART, we found that the relative reduction in VT among KP women almost doubled compared to the base-case. This is because KP women have less ART interruption than NP women even in the presence of MM support, which means a higher proportion of them benefit from the increased effectiveness of DTG-based ART. Further studies of the MM program's impact on viral suppression in the current era of DTG-based ART would help clarify the likelihood of this potential MM impact among KP women.

Three simulation studies have previously evaluated the PVT potential of maternal retention support programs (Supplementary Appendix 1.7.2) [9–11]. Most similar to our study was a pre-DTG cohort model of PBFW living with HIV who initiated PVT care in Malawi, Nigeria, and Zimbabwe [9]. This study projected that over 12 months postpartum, retention support would reduce VT from 12.3% to 9.9%, a 19.6% relative reduction that is slightly greater than our NP estimate, despite a shorter program duration and more modest reduction in ART interruption due to the MM intervention. Among the multiple differences in model structures and assumptions, a major driver of this larger impact may have been their stronger reduction in VT risk associated with retention, resulting from modeling VT risk as depending only on time since ART initiation and not on viral suppression.

The remaining 2 simulation studies both modeled the impact of MM in a population-representative setting. A program evaluation of MM in Mozambique projected a 7.9% relative reduction in VT, from 14.0% to 12.9%, assuming that MM would reduce the 29% of women not retained in care by 66% over 12 months postpartum [10]. This percent reduction would increase when analyzed among the subpopulation in our study, ie, PBFW living with HIV who initiate ART. Another study evaluating multiple PVT interventions in Zambia estimated that MM alone could reduce VT by 35%, a large effect likely driven by 1) a higher maternal ART coverage rate than the Mozambique study (>90% vs 69%) and 2) a very strong direct reduction in VT risk due to MM that is optimistic given the existing MM data (Supplementary Appendix 1.7) [11].

While our study expands upon these prior studies by disaggregating NP and KP women and by examining a range of MM program effects and durations, it shares their common limitation that models rely on assumptions and parameter values derived from available data. Although we used sensitivity analyses to explore the impact of the wide variation in MM program implementation and evaluation studies, our base-case results reflect the uncertainty inherent in the empirical MM data. Due to limited population-based data on DTG-based ART, we also assumed that ART effectiveness was the same for NP and KP women. However, future empirical research may reveal that NP women have lower adherence and viral suppression rates, given the challenges of managing a new HIV diagnosis along with a new pregnancy. In addition, while our use of a uniform timeline of HIV and maternal events for each cohort allowed us to efficiently highlight key contrasts, we do not represent the full variety of timelines occurring in PBFW living with HIV. Finally, we note that our estimates of MM benefit are limited to PBFW who have initiated ART, as our model does not include the estimated 23% of infants with new infections in Kenya who were born to the 11% of mothers who did not initiate ART in 2023 [1]. The 40% of VT attributed to ART interruption in Kenya indicates that MM programs have significant potential for PVT [1], but we do not estimate the national impact of MM in Kenya.

While real-world studies are limited in their ability to determine outcomes among PBFW and their infants who are lost to follow-up, a strength of our study is that we are able to model outcomes regardless of retention status. In doing so, we were able to demonstrate significant reductions in VT, whereas in real-world settings, increased retention of high-risk PBFW who would have otherwise been lost to follow-up (ie, those most likely to transmit HIV to their infants whether or not they are retained in care) might result in apparent paradoxical increases in VT [4]. Therefore, we caution programs implementing MM services to carefully monitor and interpret VT rates, and to potentially expect initially worrisome VT rates as MM successfully work to retain high-risk PBFW in care.

CONCLUSION

We present a novel modeling evaluation of an MM program for PBFW living with HIV in Kenya that shows a difference in PVT benefit for newly positive versus known positive women, due to their different baseline risks of ART interruption during pregnancy and breastfeeding. Our analysis highlights directions for future research that can help identify MM implementation strategies with the greatest potential for PVT benefit.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.