Maternal RSV Vaccination in South Africa: Benefits and Risks Explored

Citation: Monoi A, Endo A, Procter SR, Leuba SI, Flasche S, Jit M, et al. (2026) The benefits and risks of maternal RSV vaccination on mortality in South Africa: A modeling study. PLoS Med 23(1): e1004625. https://doi.org/10.1371/journal.pmed.1004625 Academic Editor: Annettee Nakimuli, Makerere University College of Health Sciences, UGANDA Received: February 8, 2025; Accepted: November 21, 2025; Published: January 20, 2026 Copyright: © 2026 Monoi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability: This analysis uses data presented at the SAGE meeting in September 2024, which includes data shared by Pfizer from the Phase 3 MATISSE trial (NCT04424316). Pfizer did not participate in the analysis of such data nor did Pfizer have any role in the conclusions drawn from the analysis presented by SAGE. Data from the Drakenstein Child Health Study were obtained from Heather Zar (heather.zar@uct.ac.za) (personal communication); some of these data (on the total number of births) were available in the publication. Some additional data (on the number of births by gestational age and the number of neonatal deaths) was not published but was shared in confidence at the SAGE meeting. Data cannot be shared publicly because of ethical conditions with which study investigators are obliged to comply. Access to the project data is restricted to nominated investigators approved by the University of Cape Town Human Research Ethics Committee, as per the consent document. Interested, qualified researchers may request to access this data by contacting the Drakenstein Child Health Study (via lesley.workman@uct.ac.za) to submit a formal data use request and ensure required ethical approval received prior to use. Data analysis code used in this study is available at https://github.com/ayakamon/BR-RSV-MV. Funding: AM’s travel for collaborative visits for this project was supported by funding from the Bill & Melinda Gates Foundation (INV-069494, https://www.gatesfoundation.org/). MJ, SIL, and SRP were also supported by funding from the Bill & Melinda Gates Foundation (INV-069494, https://www.gatesfoundation.org/). This funder had roles in the study design, and data collection through its membership on the maternal RSV vaccine benefit-risk advisory group, but no role in data analysis, decision to publish, or preparation of the manuscript. AM was funded by the Japanese Ministry of Education, Culture, Sports, Science and Technology through the Doctoral Program for World-leading Innovative & Smart Education as part of the NU-LSHTM Joint PhD Programme for Global Health(https://www.mext.go.jp/en/policy/education/highered/title02/detail02/1373919.html). AE is supported by the Japan Science and Technology Agency (JPMJPR22R3, https://www.jst.go.jp/EN/), Japan Society for the Promotion of Science (JP22K17329, https://www.jsps.go.jp/english/), and Japan Agency for Medical Research and Development (JP223fa627004, https://www.amed.go.jp/en/). SF is funded by the Einstein Foundation Berlin as an Einstein BUA Strategic Professor (EPP-BUA-2022-697, https://www.einsteinfoundation.de/en/). HJZ was funded by the Bill & Melinda Gates Foundation (grants OPP1017641 and OPP1017579, https://www.gatesfoundation.org/) for the Drakenstein Child Health study. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: SAM institution has received funding from Pfizer, GSK, MSD, and AstraZeneca for research on RSV vaccines and RSV monoclonal antibodies. PB institution has received public-private partnership funding through the European Union IMI/IHI Respiratory Syncytial Virus Consortium in Europe (RESCEU) and Preparing for RSV immunisation and surveillance in Europe (PROMISE) projects (ceased in 2023). HJZ has received funding from Pfizer, MSD, and Sanofi for studies of RSV vaccines and monoclonal antibodies and serves on the Data Safety Management Board for Moderna RSV maternal vaccine and advisory boards to MSD and Pfizer. The other authors declare no conflict of interest. Abbreviations: CFR, case fatality rate; CI, confidence interval; Crl, credible interval; DCHS, Drakenstein Child Health Study; EMA, European Medicines Agency; HICs, high-income countries; LMICs, low- and middle-income countries; LRTIs, lower respiratory tract infections; MA, medically-attended; MCMC, Markov chain Monte Carlo; SAGE, Strategic Advisory Group of Experts on Immunization Introduction Respiratory syncytial virus (RSV) is a significant cause of pediatric morbidity and mortality worldwide, particularly in low- and middle-income countries (LMICs) [1]. It is estimated that globally, RSV leads to approximately 100,000 deaths among children under 5 years in a year [1]. The burden is concentrated in LMICs where over 97% of RSV-attributable deaths occur [1], and especially high among younger infants, who are at the greatest risk of developing severe disease [1–4]. Antillón and colleagues estimated that although children under 6 months of age represent only 10% of the under 5 years population in LMICs, they bear a disproportionate burden of disease, accounting for approximately 30% of hospitalizations and 38%−50% of deaths [3]. Prophylactics against RSV in infancy have recently been licensed, including a maternal RSV vaccine [5] and a long-acting monoclonal antibody [6]. The bivalent RSV prefusion F protein-based vaccine (RSVpreF) developed by Pfizer (Abrysvo, hereafter referred to as “RSVpreF”) has received approval for use in pregnant women from the United States Food and Drug Administrationand the European Medicines Agency (EMA) following the successful completion of the MATISSE trial [5] (hereafter referred to as “the trial”). The trial provided evidence of the vaccine’s efficacy in preventing RSV disease in infants, especially severe disease. Mathematical modeling studies suggested that an RSV vaccination programme could reduce RSV-associated mortality and be cost-effective, particularly in LMICs like South Africa [7–9]. However, in the trial, a non-significant imbalance in preterm birth rates was observed: overall this was not statistically significant with 5.7% (95% confidence interval (CI): 4.9, 6.5) of infants being born prematurely in the intervention arm versus 4.7% (95%CI: 4.1, 5.5) in the placebo arm [10]. The imbalance in preterm birth rates was most pronounced in South Africa [10,11]. Post-hoc analysis found that there is a statistically significant increase of preterm birth risk in the intervention arm in the South African component of the trial (relative risk between the intervention arm and the placebo arm is 2.06 (95%CI; 1.21, 3.51)) [12]. However, no imbalance in newborn and infant deaths between study arms was observed either in the whole trial or in South Africa (e.g., for the whole trial, 8 in the intervention arm versus 14 in the placebo arm with relative risk of 0.57 (95%CI: 0.24, 1.36)) [10,12,13]. Nevertheless, the observed imbalance in preterm birth rates has led to concerns about vaccine safety. Notably, a trial of RSVpreF3-MAT vaccine in pregnant women was terminated due to increased preterm birth rates in the intervention arm, accompanied by a non-significant numerical increase in neonatal deaths likely related to prematurity [14]. To inform national decision-making, it is crucial to balance the risks of maternal RSV vaccination against the benefits [15] in the local context. This study aimed to assess the potential impact of maternal RSV vaccination with RSVpreF in South Africa on mortality, should the excess risk of preterm birth associated with vaccination be substantiated and should preterm births translate into mortality. Given the low relative risk of preterm births recorded among the trial participants in Argentina, Chile, the United States, Taiwan, and Japan [13], similar analyses for those countries would result in the risks being very unlikely to exceed the benefits after the actual preventive effect of vaccination on RSV is included. We compare the vaccine benefits of reducing RSV-associated infant mortality against the potential risks of increased neonatal mortality due to preterm birth in the most pessimistic scenario. Discussion If vaccination is introduced at 24–36 GA weeks, the estimated benefit of maternal vaccination through reduction in RSV-associated infant mortality is unlikely to substantially outweigh the potential risk of increased neonatal mortality due to vaccine-associated preterm birth in South Africa. However, there is considerable uncertainty around our modeled estimate of vaccine-related neonatal mortality risk. We also estimate that if vaccination is introduced at 27–36 GA weeks, the mortality benefit is likely to outweigh the risk. These findings are based on the speculation that the observed increased preterm birth was related to RSVpreF as being connected with the increased risk of preterm birth, based on observations in the South African component of the trial, and that this increased preterm birth risk leads to neonatal deaths based on GA-specific risks derived from the DCHS. The findings have been presented at the SAGE (the Strategic Advisory Group of Experts on Immunization) meeting in September 2024, and SAGE ultimately recommended RSV vaccination in the third trimester of pregnancy, as defined by the local context which in most countries is 28 GA weeks onwards [24]. Again, the observed association may not be causal, i.e., vaccination may not have caused the preterm births. Our conclusions are largely dependent on a relatively small number of early preterm infants in the trial, and we find that the point estimate of the benefit-risk ratio could reverse if we exclude these early preterm infants in the trial. It is hard to compare these benefit-risk ratio with other vaccines’ ratios, which are context-specific [25–27]. In the scenario analysis, we ran one scenario in which the risk and benefit of vaccination were weighted equally. Meanwhile, we also ran another scenario in which the risk is weighted five times the benefit, given that debates regarding decision-making in vaccination; i.e., people may put more emphasis on the risk of vaccination than on the benefit [25,28,29]. The results are also subject to uncertainty surrounding limited data on mortality of early preterm infants. Furthermore, by applying the model-estimated neonatal mortality risks to the number of live births in the South African component of the trial, the findings indicate that expected excess deaths either in the benefit or the risk were too small to be detected among the small number of live births in the South African component of the trial, in which there was no increase in deaths based on vaccination. This analysis is based on limited data from a single country in the MATISSE trial. Given the limited number of participants in the single country, the observed numerical imbalance may simply be a Type I error, or could reflect a (currently unknown) biological mechanism. While our study cannot determine which is more likely, it extrapolates South African trial birth outcomes assuming the effect is genuine. Maternal RSV vaccination with restricted GA windows has been licensed by several regulatory authorities in high-income countries (HICs) [18,30]. Notably, licensure indications for RSV immunization vary among HICs; e.g., EMA allows for maternal vaccination from 24 GA weeks onwards. In South Africa, the maternal vaccine has been licensed for use in pregnant women between 28 and 36 GA weeks, and the NITAG has recommended the vaccination during this period, i.e., from the third trimester onwards. Although there remain challenges to practicing this in LMICs (e.g., there is considerable uncertainty around GA assessment [31], timing of attendance to antenatal care [32], etc.), our analyses indicate that with vaccination from 27 weeks onward, the benefits may outweigh the risks. Our analyses and the subsequent SAGE recommendation support decision-makers in LMICs in introducing maternal RSV vaccination in their countries. Also, post-licensure surveillance is needed to monitor the association carefully in order to address concerns about potential causality between preterm birth and vaccination. There is ongoing post-marketing surveillance to assess potential adverse outcomes including preterm birth among vaccinees in early-introducing countries (e.g., U.S., Argentina, U.K.) [33,34]. In the U.S., among 13 healthcare organizations, a target trial emulation with matched analysis of data from the first year of vaccine use found the preterm birth rate was 4.0% among vaccinated and 4.5% among unvaccinated pregnant individuals (RR: 0.90; 95%CI: 0.80–1.00) [35]. Moreover, a multisite phase IV study is planned in Africa that will evaluate preterm births (NCT06955728) [36]. GA-specific neonatal mortality estimates will need to be updated once we have results from the phase IV study. Our study design had several limitations. Firstly, our analysis focuses solely on mortality due to the impact of RSV and preterm birth. However, the burden of both conditions can extend beyond mortality. For instance, our previous analysis estimated that maternal vaccination would reduce RSV-associated hospitalizations in South Africa by 24.2% (95%CrI: 18.7, 28.6) and RSV-associated deaths by 27.4% (95%CrI: 21.6, 32.3) [8]. Both RSV-associated LRTI during early childhood and preterm birth have also been linked to long-term consequences. Although most of the total disease burden, as measured through disability-adjusted life years, is due to deaths, evaluating non-fatal and long-term outcomes would provide further refinement to the estimated benefit-risk ratio from vaccination. We also used estimates of reductions in RSV-associated infant deaths based on RSV disease burden in 2011–2016 in South Africa [4]. Also, we did not include analysis of the potential secondary benefits of prevention of severe RSV disease in infants through freeing up resources for other conditions (e.g., more availability of hospital beds, etc), enabling reduction of mortality from other treatable causes [37]. In addition, we estimated benefits using efficacy estimated from all births in the South African component of the trial (i.e., births born to mothers given intervention or placebo at 24–36 GA weeks). Meanwhile, for the 27–36 GA weeks analysis, we estimated risks using data from the subset of trial births: i.e., we used 27–36 GA weeks vaccinated (or given placebo) dataset for the 27–36 weeks analysis, while using 24–36 weeks dataset for the 24–36 weeks analysis because of limited data availability. This assumes that vaccine efficacy is consistent between mothers vaccinated at 24–36 weeks and those vaccinated at 27–36 weeks. Furthermore, in estimating the waning vaccine efficacy, we only considered the influence of time since birth. Hence potential influence of GA on vaccine efficacy is not captured in our analysis. Given that protection may be lower if infants are born within 14 days after vaccination [19,20] this study may overestimate or underestimate the benefits depending on the distribution of GA at vaccination in the population. We also assumed that uniform protection regardless of GA at birth when estimating the RSV-associated deaths averted through vaccination (benefit). Hence potential variation of protection between infants born term and preterm [19,20] is not captured in our analysis. This study may overestimate or underestimate the benefits depending on the distribution of GA at birth. Another limitation is that we used the baseline preterm birth risk from the placebo arm of the trial, which is substantially lower than the overall preterm birth risk in South Africa [38,39]. For instance, Ohuma and colleagues estimated that preterm birth rate in 2020 is 13 (95%CI: 9.2, 17.9) per 100 live births in South Africa [39]. In addition, another limitation is that our conclusions were very sensitive to the outcomes associated with early preterm infants, but as the South African cohort data aggregated the neonatal mortality risk before 28 weeks, we did not know the exact GA-specific neonatal mortality risk before 28 weeks and instead assumed a constant risk. Moreover, our estimates of GA-specific neonatal mortality are based on a study conducted between 2012 and 2015 of a population-based birth cohort of unselected pregnant women attending public health facilities in a low income area of South Africa [16]. Using more recent neonatal survival rates may change our conclusions, depending on whether RSV management or general neonatal care in the study setting has improved more rapidly. Lastly, given the advanced capacity for neonatal and pediatric intensive care in South Africa, case-fatality from severe RSV disease, as well as early preterm births, is likely lower than in some other LMICs; thus, the findings of this model based on South African trial data might not be generalizable to some other LMIC settings. It is currently not established nor understood whether the observed association between vaccination and preterm birth is genuine or causal [10,40].The numerical imbalance in preterm birth in the trial was only statistically significant in South Africa and occurred predominantly at peaks of the delta and omicron waves of SARS-CoV-2 [11]. A similar preterm birth imbalance was observed in a trial of another pre-F maternal vaccine also undertaken during the Covid-19 pandemic [12]; however, no imbalance in preterm births was observed in a pre-pandemic trial of another maternal RSV vaccine that enrolled over half of the participants in South Africa [41]. Moreover, in the MATISSE trial most infants were born more than 30 days after vaccination, and there was no temporal relationship or proposed biological mechanism between the vaccination and preterm birth [18]. Post-licensure surveillance is needed to clarify if RSVpreF and preterm birth are associated [11,14,42], however, it was out of the scope of our analysis. Establishment of safety monitoring was recommended by the World Health Organization (WHO) in countries where maternal RSV vaccine is to be introduced [20]. However, WHO position paper noted that the vaccine introduction should not wait until surveillance systems have been set up, while emphasizing the need for adequate funding, training and planning to support such activities. Our analysis did not consider some other key outcomes that are potentially important, including severe RSV disease associated with preterm birth [43], stillbirths, and other fetal deaths [10], or seasonal and other temporal variations in RSV incidence [4] and preterm birth risk [18]. Our study illustrates the potential importance of the observed imbalance in preterm birth following maternal RSV vaccination at broader GA vaccination windows. However, we also show that any potential risk could be largely mitigated by changing vaccine eligibility to begin in the third trimester. The first long-awaited maternal RSV vaccine has recently been recommended by WHO SAGE for use in the third trimester of pregnancy and will likely be globally available in the next few years. Post-marketing surveillance is important to obtain further evidence about its safety and effectiveness when used in real-world settings.