Impact of introducing RTS,S/AS01(E) malaria vaccine on mortality in young

# Impact of introducing RTS,S/AS01E malaria vaccine on mortality in young children in Ghana, Kenya, and Malawi: an observational evaluation of a cluster-randomised implementation programme *The Lancet 2026* Articles Impact of introducing RTS,S/AS01 malaria vaccine on E mortality in young children in Ghana, Kenya, and Malawi: an observational evaluation of a cluster-randomised implementation programme Victor Mwapasa*, Kwaku Poku Asante*, Paul Milligan, Samuel Akech, Abraham Oduro, Don P Mathanga, Ari Fogelson, Titus K Kwambai, Mary J Hamel, Atupele Kapito-Tembo, Thomas Gyan, Nelli Westercamp, Rafiq N A Okine, Kerryn A Moore, Eliane Pellaux-Furrer, Christopher C Stanley, Daniel Ansong, Simon Kariuki, Patricia Njuguna, Tisungane Mvalo, Paul Welaga, Lucas Otieno, Paul Snell, David Schellenberg, Alfred Chimala, Edwin A Afari, Philip Bejon, Kenneth Maleta, Tsiri Agbenyega, Robert W Snow, Madalitso Zulu, Eliezer Odei-Lartey, Jobiba Chinkhumba, Aaron M Samuels, on behalf of the Malaria Vaccine Programme Evaluation partners† Summary Lancet 2026; 407: 1796–808 Background Malaria vaccines have been added to immunisation schedules in 25 sub-Saharan African countries, with See Comment page 1760 the expectation that deaths in young children would be prevented. The introduction of the RTS,S/AS01 malaria E *Contributed equally vaccine (RTS,S) in Ghana, Kenya, and Malawi in 2019 was evaluated over 4 years to show the impact on mortality in young children and to monitor severe malaria admissions, vaccine uptake, and safety. Favourable evidence on safety †Partners and their affiliations are listed at the end of the Article and impact on severe malaria admissions during the first 2 years contributed to WHO’s recommendations on malaria For the French translation of the vaccines. Here, we report the primary analysis of the impact on mortality at 46 months. abstract see Online for appendix 1 Methods Clusters of administrative units (districts in Ghana, subcounties in Kenya, and groups of immunisation School of Global and Public clinics in Malawi), each with an estimated annual birth cohort of about 4000 children, were randomly assigned 1:1 to Health, Kamuzu University of introduce the RTS,S malaria vaccine in 2019 (implementation areas), or to implement later (comparison areas). RTS,S Health Sciences, Blantyre, Malawi (Prof V Mwapasa PhD, was delivered in a four-dose schedule, at age 6, 7, 9, and 24 months in Ghana and Kenya, and at age 5, 6, 7, and Prof D P Mathanga PhD, 22 months in Malawi. Surveillance for post-neonatal mortality in children younger than 5 years was established A Kapito-Tembo PhD, throughout by a network of 26 000 local reporters who notified deaths in their community. The families were then C C Stanley PhD, visited at home by study sta to confirm details and complete a verbal autopsy. Surveillance for severe malaria and Prof K Maleta PhD, J Chinkhumba PhD); Kintampo other conditions was strengthened in 18 sentinel hospitals serving part of the study area and maintained for 46 months. Health Research Centre, Uptake of RTS,S and other vaccines was monitored by the Expanded Programme on Immunisation in each country Research and Development and independently through three household coverage surveys, at baseline, and at about 18 months and 30 months after Division, Ghana Health Service, introduction of RTS,S. The primary outcome of this impact evaluation was mortality due to any cause, except injury, in Kintampo North Municipality, Ghana (K P Asante PhD, children eligible to receive three doses of RTS,S. Mortality rate ratios were estimated by comparing the ratio of deaths T Gyan PhD); London School of among vaccine-eligible age groups to deaths in non-eligible age groups between implementation and comparison Hygiene & Tropical Medicine, areas. This evaluation is registered on ClinicalTrials.gov (NCT03806465) and is complete. London, UK (K P Asante, Prof P Milligan PhD, A Fogelson MSc, K A Moore PhD, Findings 158 clusters (66 in Ghana, 46 in Kenya, and 46 in Malawi) were selected and randomly assigned; 79 areas P Snell PhD); Kenya Medical served as implementation areas and 79 as comparison areas. By the end of the 46-month evaluation period, Research Institute-Wellcome 1 289 504 children had received the first dose of RTS,S, 1 158 850 had received the second dose, 1 068 039 had received Trust Research Programme, a third dose, and 436 527 had received a fourth dose. Coverage assessed in 2022 was 82·8% (95% CI 80·7–84·9) for Nairobi, Kenya (S Akech PhD, Prof R W Snow FMedSci); Global the first dose, 71·1% (68·8–73·5) for the third dose, and 39·9% (36·9–42·9) for the fourth dose. Excluding deaths due Health Center, Centers for to injury, there were 5576 deaths in implementation areas versus 6152 in comparison areas in children eligible to Disease Control and have received the third dose of RTS,S, and 7534 versus 7044 deaths among non-eligible children. The mortality rate Prevention, Kisumu, Kenya (S Akech); Navrongo Health ratio was 0·87 (95% CI 0·77–0·97; p=0·016). Research Centre, Research and Development Division, Ghana Interpretation Introduction of the RTS,S malaria vaccine in routine immunisation programmes was associated with Health Service, Ghana a significant reduction in mortality in young children, averting about one in eight deaths, in areas with moderate (A Oduro PhD); Malaria Alert Centre, Kamuzu University of coverage of three doses of the vaccine and low uptake of the fourth dose. These results highlight the urgency to Health Sciences, Blantyre, accelerate the deployment of malaria vaccines in areas of Africa where malaria continues to be a leading cause of Malawi (Prof D P Mathanga, child mortality. C C Stanley, A Chimala BSc, J Chinkhumba); Malaria Branch, Division of Parasitic Diseases Funding WHO; Gavi, the Vaccine Alliance; the Global Fund to Fight AIDS, Tuberculosis and Malaria; and Unitaid. and Malaria, National Center for Emerging and Zoonotic Copyright © 2026 Published by Elsevier Ltd. All rights reserved, including those for text and data mining, AI training, Infectious Diseases, Centers for and similar technologies. Disease Control and Prevention, Kisumu, Kenya 1796 Articles (T K Kwambai PhD, Research in context A M Samuels MD); Department of Immunizations, Vaccines, Evidence before this study measure the effects of malaria interventions on total deaths (of and Biologicals, WHO, Geneva, Malaria is a leading cause of death in young children in all causes), the impact has often been greater than expected Switzerland (M J Hamel MD, sub-Saharan Africa where, according to WHO estimates, more based on severe malaria case fatalities alone, because some E Pellaux-Furrer MSc); Malaria Branch, Division of Parasitic than 560 000 deaths were caused by the disease in 2023—the deaths attributed to other causes might not have occurred in Diseases and Malaria, National vast majority in young children. Two malaria vaccines, the absence of malaria co-infection. That RTS,S introduction Center for Emerging and RTS,S/AS01 and R21/Matrix-M, have been recommended by averted one in eight deaths, even though only 71% of children Zoonotic Infectious Diseases, WHO for the prevention of Plasmodium falciparum malaria in received the first three doses of RTS,S and 40% the fourth dose, Centers for Disease Control and Prevention, Atlanta, GA, USA children, prioritising areas of moderate and high malaria underscores the importance of malaria as a cause of death in (N Westercamp PhD, transmission. We searched PubMed in September, 2025, with young children in the regions where the vaccine was introduced A M Samuels); Global Malaria no language restrictions, for studies published since and confirms the need for additional malaria control measures, Programme, WHO, Geneva, Jan 1, 2000, that included primary research on malaria vaccine even where there is good coverage of existing malaria Switzerland (R N A Okine MBChB, effects on mortality using the terms (“malaria vaccine” OR interventions. P Njuguna MD, “RTS,S/AS01” OR “R21/Matrix-M”) AND “mortality” AND The study also used robust methodology. Randomisation was Prof D Schellenberg PhD); “child”. Our search yielded only two previous studies, which used to determine implementation and comparison areas. Murdoch Children’s Research reported efficacy against mortality. One was a clinical trial in Community surveillance for mortality was established on a Institute, Infection and Immunity, New Vaccines, VIC, which children were given RTS,S/AS01 doses just before the large scale in three countries, utilising a network of over Australia (K A Moore); Agogo onset of the malaria season in Burkina Faso and Mali; efficacy 26 000 local reporters who notified deaths in their Malaria Research Centre, against all-cause mortality of five doses of RTS,S/AS01 over communities, which were followed up by household visits to Agogo, Ghana (D Ansong PhD, 3 years (among children receiving at least one dose) was 53·4% complete a verbal autopsy. The uptake of other interventions T Agbenyega PhD); Kwame Nkrumah University of Science (95% CI 14·0–74·8), and the efficacy of six or seven doses given that could influence mortality was carefully monitored, and Technology, Kumasi, over 5 years was 44·5% (2·8–68·3). The second study was a showing that coverage of other routine vaccines, use of Ghana (D Ansong, non-randomised cohort-monitoring study, which reported an insecticide-treated nets, and access to health care were similar T Agbenyega); Centre for Global 8% reduction in all-cause mortality over a 12-month period in in both implementation and comparison areas throughout the Health Research, Kenya Medical Research Institute, Kisumu, children who received at least one dose of RTS,S/AS01 through evaluation. The observed impact on mortality can therefore be Kenya (S Kariuki PhD); PATH routine immunisation services. Data reported previously from confidently attributed to the effects of the malaria vaccine. Center for Vaccine Innovation the present study showed no evidence of the safety signals that and Access, Nairobi, Kenya had been observed in the phase 3 trial, indicating these signals Implications of all the available evidence (P Njuguna); University of were not causally related to the vaccine. There is an extensive body of evidence showing that current North Carolina Project-Malawi, Lilongwe, Malawi malaria vaccines are safe and effective. Malaria vaccines have (T Mvalo FC Paed, M Zulu MBBS); Added value of this study been introduced in 25 countries—Ghana, Kenya, and Malawi C K Tedam University of Using a cluster-randomised design, our study shows the in 2019; Central African Republic, Côte d’Ivoire, Burkina Faso, Technology and Applied community-level impact of malaria vaccine introduction on Liberia, Cameroon, Benin, Sierra Leone, South Sudan, Sciences, Navrongo, Ghana child mortality. During the 4 years following the introduction Mozambique, Niger, Chad, the Democratic Republic of the (P Welaga PhD); Kenya Medical Research of the RTS,S/AS01 vaccine through routine child Congo, Sudan, and Nigeria in 2024; Burundi, Uganda, Mali, Institute-United States Army immunisation programmes in Ghana, Kenya, and Malawi, Guinea, Ethiopia, Togo, and Zambia in 2025; and Guinea-Bissau Medical Research Directorate- there was a 13% reduction in the number of deaths (of all in 2026, with more countries planning introductions. In many Africa, Kisumu, Kenya causes) among children who were eligible to have received (L Otieno MBChB); School of areas where malaria vaccines are most urgently needed, vaccine Public Health, University of three doses of the vaccine. delivery is often constrained by weak health systems, mistrust, Ghana, Accra, Ghana The true burden of malaria mortality is difficult to measure and conflict. Operational research is needed to address these (E A Afari FGCP); Kenya Medical directly. Many deaths occur at home before medical diagnosis is implementation challenges; however, our findings show that Research Institute-Wellcome Trust Research Programme, possible, and, even in a hospital, it can be difficult to establish substantial reductions in deaths among young children are Kilifi, Kenya whether malaria was a direct or contributory cause of death or possible even when only moderate levels of malaria vaccine (Prof P Bejon FMedSci); Centre an incidental infection. In the past, when it has been possible to coverage can be achieved. for Tropical Medicine and Global Health, Nuffield Department of Clinical Medicine, University of Oxford, Introduction programme that supported the implementation of RTS,S Oxford, UK (Prof P Bejon, In October, 2021, WHO recommended the use of the over 4 years, from 2019 to 2023, to show the safety, Prof R W Snow); Kintampo Health Research Centre, RTS,S/AS01 malaria vaccine (RTS,S) to prevent feasibility, and impact of RTS,S vaccination in routine use. E Research and Development Plasmodium falciparum malaria in young children living in The measurement of impact was considered essential Division, Ghana Health Service, areas of high and moderate malaria transmission.1 The given the need for additional immunisation visits and the Kintampo North Municipality, recommendation followed a review of the extensive recommendation for a programmatically challenging Ghana (E Odei-Lartey MBA) evidence related to the safety and ecacy of the vaccine, fourth dose at about 2 years of age. The evaluation of the including data from the first 2 years of the Malaria Vaccine implementation programme would involve measuring Implementation Programme (MVIP) in Ghana, Kenya, the impact on all-cause mortality overall and in girls and Malawi.2 The MVIP was a WHO-coordinated compared with boys, and showing whether safety signals Articles Correspondence to: that had been observed in the phase 3 trial were causally Here, we report the primary analysis of the impact of Dr Kwaku Poku Asante, associated with the vaccine.3–5 RTS,S introduction on all-cause mortality (step 2) and Kintampo Health Research WHO developed a Framework for Policy Decision,6 describe coverage of the four-dose schedule, as well as Centre, Research and with expert advice, which proposed a two-step process the overall impact on hospital admissions for severe Development Division, Ghana Health Service, PO Box 200, that would allow a WHO policy recommendation on the malaria. Kintampo North Municipality, RTS,S vaccine as soon as a favourable risk–benefit ratio Ghana was established. Hence, a recommendation for vaccine Methods kwakupoku.asante@ use could be made if safety concerns were resolved, Study design, participants, and randomisation kintampo-hrc.org provided that severe malaria or mortality data were The design of the MVIP followed a prospective, cluster- or consistent with a beneficial impact of the vaccine. randomised approach to evaluate the introduction of the Prof Paul Milligan, London School of Hygiene & Tropical Primary analysis of the impact on mortality and RTS,S vaccine through the routine immunisation Medicine, London WC1E 7HT, UK assessment of uptake of the fourth dose would follow at system. Clusters corresponded to health service delivery paul.milligan@lshtm.ac.uk the end of the evaluation and, in the second step, the units, and randomisation ensured fair allocation of vaccine recommendation could be adjusted, if necessary. limited vaccine supplies, and strengthened the rigour of By April, 2021, 24 months after the vaccine was first the evaluation, and is described in detail elsewhere.7 introduced, sucient data had accrued to evaluate the Briefly, the Ministries of Health in Ghana, Kenya, and safety signals and the impact on hospital admissions for Malawi each selected subnational evaluation areas that severe malaria. The data were reviewed by WHO and had year-round malaria transmission, with an estimated informed the recommendation on RTS,S in October, 2021 malaria prevalence of at least 20%, and which had an (step 1), and are described by Asante and colleagues.7 annual cohort of at least 240 000 surviving infants. There was no evidence of the safety signals (excess cases Ministry of Health representatives in these regions, of meningitis and of cerebral malaria, and excess deaths assisted by an independent statistician, randomly among girls) observed in the phase 3 trial, after careful assigned (1:1) administrative units (clusters), with an monitoring specifically to detect these signals. This estimated annual birth cohort per cluster of finding indicated that the signals that had been observed approximately 4000, to introduce RTS,S as a four-dose were not causally related to the vaccine. Hospital schedule starting in 2019 or to delay introduction until admissions for severe malaria were reduced by 32% in after a WHO recommendation on wider use. There were the context of average coverage of the first three RTS,S 158 clusters: 66 districts in Ghana, 46 subcounties in doses of 63%. Kenya, and catchment areas of 46 groups of immunisation A B C Tanzania Benin Uganda Ghana Zambia Côte d’Ivoire Kenya Togo Malawi Mozambique 0 100 km Sentinel hospital Implementation Implementation, hospital catchment area Tanzania Comparison Comparison, hospital catchment area Figure 1: Maps of evaluation areas in Ghana, Kenya, and Malawi Areas shaded green are clusters randomly assigned to RTS,S implementation beginning in 2019, while areas shaded pink, randomised to implement later, served as comparison areas. The 158 clusters comprised 66 districts in Ghana, 46 subcounties in Kenya, and the catchment areas of 46 groups of immunisation clinics in Malawi. Mortality surveillance was maintained throughout all 158 clusters. Sentinel hospitals, indicated by red dots, served part of the evaluation areas. Diagonal hatching indicates the 77 clusters out of the 158 that were defined as sentinel hospital catchment clusters, based on a review of paediatric admissions before randomisation. RTS,S=RTS,S/AS01E malaria vaccine. 1798 Articles clinics in Malawi (figure 1 and figure 2A). The clusters Constrained randomisation in each country aimed to that implemented the four-dose schedule later served as balance implementation and comparison areas with comparison areas for the evaluation. Vaccine doses were respect to the number of infants surviving to 12 months, scheduled to be given at 6, 7, 9, and 24 months of age in malaria parasite prevalence, vaccine coverage (pentavalent Ghana and Kenya, and 5, 6, 7, and 22 months of age in dose 3 and measles dose 1), and the number of hospitals, Malawi (figure 2B; appendix 2 pp 7–8), administered health centres, and dispensaries. An independent See Online for appendix 2 through the routine childhood immunisation service in statistician used constrained randomisation to balance each of the three countries. implementing and comparator areas in each country with A 66 clusters (districts) 46 clusters (sub-counties) 46 clusters (groups of selected in Ghana selected in Kenya immunisation clinics) selected in Malawi 33 clusters randomly 33 clusters randomly 23 clusters randomly 23 clusters randomly 23 clusters randomly 23 clusters randomly assigned to assigned to delay assigned to assigned to delay assigned to assigned to delay introduce RTS,S introduction until introduce RTS,S introduction until introduce RTS,S introduction until on April 30, 2019 Feb 20, 2023 on Sept 13, 2019 March 7, 2023 on April 23, 2019 Nov 29, 2022 (implementation (comparison (implementation (comparison (implementation (comparison areas) areas) areas) areas) areas) areas) Doses of RTS,S given Children eligible for Doses of RTS,S given Children eligible for Doses of RTS,S given Children eligible for to children at 6, 7, 9, dose 1 if aged to children at 6, 7, 9, dose 1 if aged to children at 5, 6, 7, dose 1 if aged and 24 months 6–11 months on or and 24 months 6–23 months on or and 22 months 6–11 months on or Children aged after Feb 20, 2023 Children aged after March 7, 2023 Children aged after Nov 29, 2022 8 months and 12 months and older 6 months and older older when RTS,S was when RTS,S was when RTS,S was introduced were not introduced were not introduced were not eligible eligible eligible 479 200 had dose 1 387 221 had dose 1 423 083 had dose 1 447 610 had dose 2 341 357 had dose 2 369 883 had dose 2 431 836 had dose 3 294 085 had dose 3 342 118 had dose 3 198 092 had dose 4 95 196 had dose 4 143 239 had dose 4 Surveillance continued for 46 months Surveillance continued for 46 months Surveillance continued for 46 months (April 30, 2019, to Feb 28, 2023) (Sept 13, 2019, to July 12, 2023) (April 23, 2019, to Feb 28, 2023) B Surveys at baseline and about 23 and 38 months after RTS,S introduction Malawi Mortality and hospital- based surveillance RTS,S introduction April 23, 2019 RTS,S introduction in comparison areas Nov 29, 2022 Surveys at baseline and about 18 and 34 months after RTS,S introduction Ghana Mortality and hospital- based surveillance RTS,S introduction April 30, 2019 RTS,S introduction in comparison areas Feb 20, 2023 Surveys at baseline and about 21 and 32 months after RTS,S introduction Kenya Mortality and hospital- based surveillance RTS,S introduction Sept 13, 2019 RTS,S introduction in comparison areas March 7, 2023 2019 2020 2021 2022 2023 Figure 2: Random assignment to implementation and comparison areas (A), and evaluation timelines (B) RTS,S=RTS,S/AS01E malaria vaccine. Articles respect to the number of infants surviving to 12 months, autopsies were done to identify deaths caused by injury, malaria parasite prevalence, Expanded Programme on recognising the unreliability of determining other causes Immunisation coverage (pentavalent vaccine dose 3 and of death through verbal autopsies. In Ghana, community measles vaccine dose 1), and number of hospitals, health key informants were trained through the routine health centres, and dispensaries. A community representative in system to report adverse events in the community each country selected the final allocation at random from following public health interventions such as vaccination, a list of approximately balanced randomisation options. using a community register. For this evaluation, Further details are reported by Asante and colleagues.7 7310 community key informants were further trained to Data were collected on all children aged 1–59 months ascertain and report deaths in children younger than resident in the evaluation areas who were reported to 5 years. When a community key informant was notified have died during the evaluation; all children aged of a death, they visited the family as soon as possible to 1–59 months who were admitted to sentinel hospitals confirm the event, complete a notification form, and during the evaluation period; and on children aged report the event to a district verbal autopsy coordinator. 5–48 months, and their caregivers, who were visited Community health ocers in the Community-based during household surveys. Caregiver-reported sex of the Health and Planning Services zones assisted community child was recorded as male or female. We did not use any key informants to ensure that deaths were reported. data on ethnicity. In each case, data were collected after District coordinators then visited the home to complete a individual informed consent from a parent or guardian. verbal autopsy after seeking parental consent. Deaths in Consent procedures and ethics approvals are described children aged 1–59 months occurring in hospitals were in appendix 2 (pp 18–19). The WHO Ethics Review cross-checked to ensure that community key informants Committee approved the master protocol4 (reference had reported them, and verbal autopsies were then 003592 RTSS MVIP 14.07.2017) and country-specific arranged for any deaths that had not been reported. protocols, which were also approved by ethics committees In Kenya, surveillance was built on an existing system and regulatory authorities in each country. Community whereby deaths in children younger than 5 years were engagement was designed to provide sucient notified by community health volunteers who were information to allow parents and caretakers to make supported and managed by community health assistants. informed decisions as to whether to bring age-eligible When a death was reported, the lead community health children to receive RTS,S vaccine or to opt out. Consent volunteer visited the home about 2 weeks after the death for administration of the vaccine was implied by caregivers to complete a notification form, which the community bringing their child for immunisation. For household health assistant checked, scanned, and uploaded to a surveys, parents or legal guardians were provided with database. For deaths among children aged 5–48 months, information about the survey, and signed consent sought field teams were notified to organise a visit to seek for their child to participate. Parental or legal guardian consent and complete a verbal autopsy, 1–3 months after consent was also sought before verbal autopsy interviews the death. Initial sta training for mortality surveillance when a child had died. In Ghana, collection of sentinel was arranged for each county community strategy focal hospital data required parental or legal guardian consent, person, subcounty focal people, and two designated whereas in Kenya and Malawi the investigators obtained community health assistants in each of the subcounties. consent waivers to collect de-identified medical This team was responsible for training an additional information from hospitalised children. In all 611 community health assistants who then trained three countries, individual consent was obtained for the 15 530 community health volunteers. storage or shipment of cerebrospinal fluid samples for In Malawi, mortality surveillance was built on national multipathogen PCR testing at a reference laboratory. A registration, whereby village headmen, assisted by their Technical Working Group in each country, comprising clerks, were required to register deaths in their representatives from the national immunisation and community, and health facility sta registered deaths in malaria programmes, the national regulatory authority, their facility. 51 full-time research evaluation assistants civil society, media organisations, and WHO and other (at least one in each cluster), 2861 village reporters each immunisation and malaria partners, advised on the covering three or four villages, and village headmen and design and conduct of the evaluation. their clerks, were trained in death registration. When See Online for appendix 3 The study protocol is available online (appendix 3). family members or village reporters notified the village This evaluation is registered on ClinicalTrials.gov headman of a death, the headman, with the support of (NCT03806465) and is complete. their clerk, completed a death registration form and recorded particulars of the deceased (including names, Procedures date of birth, age, date and place of death, and place of Mortality surveillance residence) into village death registers. Evaluation Surveillance for mortality was established throughout assistants then visited the home 14–28 days after the the clusters selected, building on and substantially death to confirm details and complete a verbal autopsy strengthening existing reporting systems. Verbal questionnaire, after seeking consent. When a death of a 1800 Articles child younger than 5 years occurred at a health facility, who were born after Aug 30, 2018, in Ghana; in children trained sta completed a death registration form with aged at least 10 months who were born after Sep 13, 2018, information from facility records and close relatives of in Kenya; and in children aged at least 8 months who the deceased. Village reporters were expected to meet were born after Oct 23, 2019, in Malawi). Deaths were their village headmen each week to ensure that any included if they occurred between April 30, 2019, and deaths were notified, and evaluation assistants visited Feb 28, 2023 (Ghana), between Sep 13, 2019, and village headmen, village reporters, and facility sta July 12, 2023 (Kenya), and between April 23, 2019, and regularly for supervision and to extract data from Feb 28, 2023 (Malawi). By the end of the evaluation the registration forms into tablet computers. Further details oldest children eligible to have been vaccinated were on mortality surveillance are given in appendix 2 aged 53·9 months (Ghana), 57·9 months (Kenya), and (pp 12–13). 52·2 months (Malawi). Other key outcomes were coverage of the fourth dose of the malaria vaccine, and Surveillance in sentinel hospitals hospital admissions for severe malaria (reported Surveillance for severe malaria and other conditions was previously).7 Secondary outcomes were cerebral malaria, maintained in 18 hospitals (eight in Ghana, six in Kenya, meningitis, the number of admissions with a positive and four in Malawi; figure 1), which served part of the malaria test, and the total admissions for any cause. evaluation areas. Sentinel referral hospitals were selected that had a high volume of paediatric admissions and Statistical analysis were able to collect and analyse cerebrospinal fluid for It had been estimated that 23 clusters in each diagnosis of meningitis and cerebral malaria. Details implementation and comparison area in each country have been described previously.7 Briefly, laboratory would allow 80% power to detect a 10% reduction in capacity was strengthened, and hospital sta were all-cause mortality over 46 months in children eligible to supported to undertake thorough clinical investigat ions receive three doses of RTS,S, assuming a mortality rate using standardised methods. All hospitalised children of 8·5 per 1000 child-years and a coecient of variation aged 1–59 months were evaluated using a standardised among clusters of 0·1.4 Mortality rates during the first clinical and diagnostic algorithm, which included signs year of evaluation were lower than anticipated, but it was of severe malaria, such as cerebral malaria and estimated that there would be 90% power to detect a meningitis (appendix 2 p 42). Cases of severe malaria, 10% reduction in a pooled analysis; it was therefore those meeting the criteria for cerebral malaria, planned that the primary analysis would be of all meningitis, the number of admissions with a positive countries combined. malaria test, and the total admissions for any cause were The analytical approach, using a double ratio of event documented, and each child’s place of residence, date of counts to estimate incidence rate ratios, was defined birth, sex, and clinical and laboratory details were before data collection.8 The methods have been described reported by the caregiver and verified from the child’s previously.7–11 Surveillance was maintained in all children health record. aged 1–59 months, and therefore included children Quality assurance measures were instituted for eligible for RTS,S, children who were too old to be eligible mortality and hospital surveillance procedures and when RTS,S was introduced, and children who were too reporting systems through monthly data reviews young to be eligible. Events (deaths and hospital (appendix 2 p 15). admissions) in implementation and comparison areas were classified according to whether the child would have Household surveys been eligible for vaccination or not. Due to the use of In each country, household surveys in implementation randomisation, the ratio of event counts in eligible to and comparison areas were conducted by the research non-eligible children is expected to be the same in team in each country using stratified two-stage cluster implementation and comparison areas if vaccine sampling at baseline (before or soon after the introduction introduction has no eect; therefore, we used the relative of RTS,S), and approximately 18 months (midline) and dierence in this ratio between implementation and 30 months after RTS,S introduction (endline) to estimate comparison areas to estimate the incidence rate ratio coverage of RTS,S and other vaccines, to monitor use of associated with vaccine introduction. Using data for non- insecticide-treated nets and care-seeking behaviour, and eligible groups in each cluster in this way improves (at baseline and endline) to measure malaria parasite statistical eciency and reduces confounding.10 In this prevalence by rapid diagnostic test (appendix 2 pp 13–15). estimator, person-time denominators eectively cancel out (appendix 2 p 35). Outcomes To account for stratification by country, incidence rate The primary outcome for impact evaluation was mortality ratios were obtained by weighting the logarithm of country of any cause, apart from injury, in children eligible to estimates by the inverse of their variance. Heterogeneity have received three doses of RTS,S during the 46-month across country strata was assessed using Cochrane’s Q test evaluation period (ie, in children aged at least 10 months and the I² statistic. Impact was estimated in two groups: Articles (1) among children from when they were eligible for the introduced, and events that occurred in children eligible first dose of RTS,S (5 months of age in Malawi, 6 months to receive the vaccine in implementation clusters after the of age in Kenya and Ghana), and (2) among children vaccine was introduced in comparison areas, were also 1 month after they were eligible to receive a third dose excluded. The age and timing of deaths is illustrated in (8 months in Malawi, 10 months in Ghana and Kenya), the appendix 2 (p 29). Additional analyses estimated the latter being considered primary for analyses of impact. impact on mortality in girls and boys, and by age group. For each analysis, events occurring in children who were Vaccine coverage estimates, and ratios of coverage, were just too old, by up to 2 months, to be eligible were excluded calculated using survey-weighted Poisson regression with to avoid including vaccinated children in the non-eligible robust standard errors, with interaction terms to assess group. Events in children who would have been eligible to heterogeneity across country strata. Analyses were done receive RTS,S in comparison clusters after the vaccine was using Stata versions 17, 18, and 19. Pooled estimates Ghana Kenya Malawi Implementation Comparison Implementation Comparison Implementation Comparison Implementation Comparison areas areas areas areas areas areas areas areas Clusters, n 79 79 33 33 23 23 23 23 Mean surviving infants per cluster 4596 (1690·0) 4727 (1648·3) 3898 (1725·3) 4052 (1753·1) 5509 (1649·0) 5467 (1579·1) 4684 (1211·2) 4956 (1164·9) in 2018, n (SD) Children aged 12–23 months, n 2493 2564 1079 1176 644 694 770 694 Home-based vaccine record 2129/2493 2256/2564 926/1079 1062/1176 553/644 588/694 650/770 606/694 (85·4%) (87·7%) (85·6%) (89·7%) (86·2%) (85·3%) (84·4%) (85·8%) BCG 1954/2129 2103/2256 860/926 1000/1062 533/553 572/588 561/650 531/606 (92·5%) (93·7%) (92·0%) (94·6%) (95·5%) (95·2%) (86·7%) (87·1%) Penta-3 2110/2301 2233/2419 875/926 1015/1062 584/630 613/674 651/745 605/683 (91·1%) (94·4%) (93·3%) (96·0%) (91·8%) (91·4%) (87·5%) (88·5%) OPV-3 2000/2295 2119/2411 852/926 973/1062 536/622 564/667 612/747 582/682 (90·5%) (93·2%) (90·5%) (92·4%) (85·4%) (84·9%) (81·2%) (85·5%) PCV-3 2077/2291 2217/2405 879/926 1015/1062 553/622 602/660 645/743 600/683 (90·8%) (94·8%) (94·0%) (96·0%) (88·8%) (91·6%) (86·7%) (87·8%) Rota-2 2087/2294 2228/2405 877/926 1025/1062 563/624 607/659 647/744 596/684 (91·4%) (95·2%) (94·3%) (96·9%) (89·3%) (91·9%) (87·3%) (86·7%) Measles-1 2029/2310 2144/2425 841/926 971/1062 553/632 609/677 635/752 564/686 (88·0%) (88·9%) (90·2%) (90·8%) (85·5%) (90·6%) (85·4%) (82·5%) Fully vaccinated 1651/2129 1773/2256 734/926 847/1062 438/553 476/588 479/650 450/606 (74·7%) (78·0%) (77·2%) (80·3%) (76·4%) (81·0%) (73·1%) (74·8%) Vitamin A in the previous 6 months 1221/2301 1217/2429 494 /926 529/1062 290/638 308/688 437/737 380/679 (50·0%) (49·4%) (50·9%) (48·4%) (43·1%) (45·6%) (58·0%) (54·7%) Children aged 24–35 months (Ghana and 2344 2280 896 891 665 652 783 737 Kenya), 21–32 months (Malawi), n Measles-2 1537/2272 1498/2229 744/896 730/891 315/624 312/619 478/752 456/719 (69·1%) (67·3%) (82·8%) (80·2%) (48·1%) (5o·8%) (63·4%) (66·0%) Children aged 5–48 months, n 8789 8783 3881 3887 2430 2518 2478 2378 Slept under insecticide-treated nets 6858/8788 6870/8782 2521/3881 2453/3887 2094/2430 2218/2518 2243/2477 2199/2377 the previous night (87·4%) (84·8%) (63·4%) (59·8%) (85·6%) (88·9%) (90·3%) (91·9%) MUAC <13·5 cm 1224/8789 1251/8783 623/3881 665/3887 145/2430 161/2518 456/2478 425/2378 (11·3%) (13·1%) (17·1%) (17·7%) (5·8%) (6·6%) (18·5%) (17·6%) Malaria positive by RDT 2081/8754 1812/8746 805/3881 831/3887 642/2400 502/2488 634/2473 479/2371 (24·2%) (18·5%) (20·7%) (20·4%) (26·4%) (18·6%) (28·1%) (16·7%) Sought care for fever* 582/799 536/763 209/287 178/253 185/252 183/256 188/260 175/254 (73·0%) (72·6%) (74·5%) (76·6%) (71·6%) (71·8%) (71·9%) (67·7%) Data are n/N (%) unless otherwise specified. Percentages are survey weighted, hence may not equal the value calculated from numerator and denominator. Baseline surveys were conducted from Feb 25 to March 18, 2019 (Ghana), July 15 to Dec 16, 2019 (Kenya), and May 6 to June 19, 2019 (Malawi). Coverage of routine vaccinations in the first year of life was estimated in children aged 12–23 months. In Ghana, estimates are based on the home-based record only; in Kenya and Malawi, estimates are based on documented status from the home-based record, or caregiver recall if a home-based record was not available. Coverage of the second dose of measles vaccine was estimated in a 12-month age range, starting 6 months after the recommended age for the second dose, which was 18 months in Ghana and Kenya, and 15 months in Malawi. Fully vaccinated was defined as having received BCG at birth, three doses of OPV (excluding the birth dose), three doses of pentavalent vaccine, and one dose of measles vaccine. Children who received IPV at 14 weeks in place of OPV-3 were considered to be fully vaccinated for polio. IPV=inactivated polio virus. Measles-1=first dose of measles vaccine. Measles-2=second dose of measles vaccine. MUAC=mid-upper arm circumference. OPV-3=third dose of oral polio vaccine (excluding birth dose). PCV-3=third dose of pneumococcal vaccine. Penta-3=third dose of pentavalent vaccine. RDT=rapid diagnostic test. Rota-2=second dose of rotavirus vaccine. *Care-seeking for fever was assessed in children who had fever in the 2 weeks before the survey, according to the caregiver (this was 34·0% in implementation areas and 28·1% in comparison areas, pooled). Table: Characteristics of children in implementation and comparison areas at baseline 1802 Articles Role of the funding source Expanded Programme on Immunisation, coverage of the The funders of the study did not contribute to the study second dose of measles-containing vaccine in children design, data collection, data analysis, data interpretation, aged 2 years, receipt of vitamin A in the previous or writing of the report. 6 months, and, among children aged 5–48 months, use of insecticide-treated nets, prevalence of malnutrition Results (middle upper arm circumference <13·5 cm), and the Of the 158 clusters selected and randomly assigned, proportion seeking care for fever (appendix 2 p 20). The 79 areas served as implementation areas and 79 as proportion of children aged 5–48 months who tested comparison areas. National immunisation programm es positive for malaria by rapid diagnostic test at baseline introduced RTS,S through routine child vaccination was similar in both implementation and comparison services in April, 2019, in Ghana and Malawi, and in areas in Ghana. However, this indicator was higher in September, 2019, in Kenya. The study profile and implementation areas than in comparison areas in Kenya timelines are shown in figure 2. and Malawi. Results of household surveys conducted at baseline are During the 46 months of the evaluation, a total of presented in the table. When monitored during the 1 289 504 children received the first dose of RTS,S, evaluation, implementation and comparison clusters 1 158 850 received the second dose, 1 068 039 received the remained similar to one another with respect to coverage third dose, and 436 527 received the fourth dose in the in children aged 1 year of vaccines given through the implementation areas, through the routine Expanded 80 40 0 Penta-3 RTS,S-1 RTS,S-2 RTS,S-1 MCV-1 MCV-2 RTS,S-4 Figure 3: Coverage of RTS,S and other childhood vaccines by dose in implementation and comparison clusters at midline and endline household surveys Vaccination coverage was estimated in surveys about 18 months after introduction of RTS,S (midline) and about 30 months after introduction (endline) in RTS,S implementation and comparison areas. The error bars are 95% CIs. Penta-3 was given at 14 weeks, RTS,S-1 at 6 months (Ghana and Kenya) and 5 months (Malawi), RTS,S-2 at 7 months (Ghana and Kenya) and 6 months (Malawi), RTS,S-3 at 9 months (Ghana and Kenya) and 7 months (Malawi), MCV-1 at 9 months, and MCV-2 at 15 months (Malawi) or 18 months (Ghana and Kenya). Penta-3, RTS,S doses 1–3, and MCV-1 coverages were estimated among children aged 12–23 months. MCV-2 and RTS,S dose 4 coverages were estimated among children aged 30–40 months in Ghana, 30–36 months in Kenya, and 28–39 months in Malawi. Coverage is according to home-based record (HBR) or caregiver recall if HBR was not available. Tests of interaction showed coverage of RTS,S dose 3 was similar across countries at midline (F=1·09, p=0·34), but at endline coverage differed with respect to dose 3 (F=7·44, p=0·0007) and dose 4 (F=17·88, p<0·0001), with coverage higher , , , in Ghana. MCV-1=first dose of measles-containing vaccine. MCV-2=second dose of measles-containing vaccine. Penta-3=third dose of pentavalent vaccine. RTS,S-1=first dose of RTS,S/AS01 malaria vaccine. RTS,S-2=second dose of RTS,S/AS01 malaria vaccine. RTS,S-3=third dose of RTS,S/AS01 malaria vaccine. E E E RTS,S-4=fourth dose of RTS,S/AS01 malaria vaccine. E )%( egarevoC 80 40 0 E Malawi (implementation area) F Malawi (comparison area) Penta-3 RTS,S-1 RTS,S-2 RTS,S-1 MCV-1 MCV-2 RTS,S-4 )%( egarevoC 80 40 0 C Kenya (implementation area) D Kenya (comparison area) )%( egarevoC A Ghana (implementation area) B Ghana (comparison area) Midline Endline Articles Programme on Immunisation (appendix 2 pp 9–11). ranged from 46·9% (95% CI 33·8–65·0, Kenya) to 50·6% When assessed at about 30 months after RTS,S (41·6–61·5, Malawi) and 73·3% (68·6–78·3, Ghana), and introduction, coverage of the first, second, and third for the fourth dose, ranged from 21·6% (15·1–31·0, doses increased compared with the coverage estimated Malawi) to 22·8% (11·5–45·2, Kenya) and 42·5% about 18 months after the introduction of RTS,S (35·5–51·0, Ghana; appendix 2 p 26). (appendix 2 pp 23–25), from 75·8% (95% CI 73·4–78·2) During the 46 months after RTS,S introduction, to 82·8% (80·7–84·9) for the first dose, from 71·5% 33 664 deaths were reported among children aged (69·0–74·0) to 78·5% (76·4–80·7) for the second dose, 1–59 months (appendix 2 pp 30, 54). A total of 31 712 deaths and from 63·8% (61·1–66·5) to 71·1% (68·8–73·5) for were included in analyses: 15 808 in implementation the third dose (figure 3; appendix 2 p 23). Coverage of the areas and 15 904 in comparison areas. The analysis fourth dose was 39·9% (36·9–42·9). excluded 64 deaths from implementation areas and In implementation areas, coverage of the first dose of 54 from comparison areas because parents did not measles-containing vaccine, scheduled at 9 months of consent to data collection or there were missing data on age, was higher than the coverage of the first dose of date of birth or sex. A further 957 were excluded from RTS,S, showing missed opportunities for RTS,S. Coverage implementation areas and 877 from comparison areas of the fourth dose of RTS,S was lower than coverage of the because they occurred (in either area) in age groups that second dose of measles-containing vaccine (figure 3). In would have been eligible for RTS,S when it was February, 2023, Ghana adjusted the timing of the fourth introduced in comparison areas, or because they were dose from 24 months to 18 months, which was associated above the upper age limit for eligibility by less than with improved uptake (appendix 2 p 10). 2 months and thus might have received RTS,S (appendix 2 Coverage of the third and fourth doses of RTS,S was pp 30–31). similar in girls and boys in each country, but tended to be Among children eligible to have received the third dose higher in children from households with a higher wealth of RTS,S, excluding deaths due to injury, there were ranking than in those from lower wealth ranking 5576 deaths in implementation areas versus 6152 in households (appendix 2 pp 24–26). In Malawi, coverage comparison areas, whereas there were 7534 versus of the third and fourth doses was lower in clusters that 7044 deaths among non-eligible children. The mortality had higher malaria prevalence at baseline. Coverage of rate ratio comparing implementation and comparison the third and fourth doses of RTS,S tended to be lower in areas was 0·87 (95% CI 0·77–0·97; p=0·016; figure 4). children who did not use an insecticide-treated net There was no evidence that impact on mortality diered compared with those who did, but a substantial by country (interaction test χ²=1·45, 2 df, p=0·48). From proportion of non-users of insecticide-treated nets were the age when children would have been eligible to receive reached. In endline surveys, 498 (survey-weighted the first dose of RTS,S (ie, from 5 months in Malawi and estimate 37·4%) of 1402 children aged 12–23 months in 6 months in Kenya and Ghana), there were 7282 deaths Ghana, 56 (5·7%) of 880 in Kenya, and 136 (19·7%) of (excluding deaths due to injury) in implementation areas 694 in Malawi did not sleep under an insecticide-treated and 7865 in comparison areas, giving a mortality rate net. Coverage of the third dose of RTS,S among non-users ratio of 0·89 (0·81–0·97). Among age groups eligible to Events (eligible/non-eligible children), n Incidence rate ratio (95% CI) Implementation area Comparison area Mortality due to any cause except injury 5576/7534 6152/7044 0·87 (0·77–0·97) Mortality (girls) 2581/3563 2834/3335 0·85 (0·75–0·97) Mortality (boys) 2995/3971 3318/3709 0·87 (0·76–0·99) All-cause admissions to sentinel hospitals 11 738/11 752 12 338/11 149 0·92 (0·81–1·05) Admissions positive for malaria 3865/4933 4948/4914 0·83 (0·73–0·95) Severe malaria admissions 1457/1705 1853/1773 0·78 (0·64–0·97) Severe malaria (subset) 1261/1341 1562/1369 0·77 (0·63–0·94) Admissions with tracer conditions 5831/4236 5490/4018 1·04 (0·90–1·20) 0·5 1·0 2·0 Favours implementation areas Favours comparison areas Figure 4: Estimates of the association of RTS,S introduction with all-cause mortality (excluding deaths due to injury) and with admissions to hospital for any cause and with severe malaria Incidence rate ratios refer to children eligible to have received the third dose of RTS,S. Deaths were captured through mortality surveillance throughout the evaluation areas. Hospitalisation data were captured through sentinel hospital surveillance in part of the evaluation areas (hatched areas in figure 1). Severe malaria followed standardised case definitions; the subset refers to patients with severe malaria in whom lumbar puncture was performed in cases of coma to rule out meningitis. Tracer conditions are hospitalisations due to conditions unlikely to be influenced by RTS,S introduction (admissions from any cause, excluding those with a positive malaria test, with anaemia, or meningitis). RTS,S=RTS,S/AS01 malaria vaccine. E 1804 Articles receive RTS,S, 1163 (8%) of 15 499 deaths were due to 4914 among non-eligible children. The incidence rate injury (646 [8%] of 8276 in boys and 517 [7%] of 7223 in ratio was 0·83 (0·73–0·95, p=0·0065). Details of girls), comprising 98 (6%) of 1561 deaths in Ghana, 634 estimates for all hospital outcomes are provided in (10%) of 6273 in Kenya, and 431 (6%) of 7665 in Malawi. appendix 2 (pp 47–48). Including deaths of all causes, there were 8031 deaths in Estimates of the reductions in severe malaria implementation areas and 8633 in comparison areas, admissions were: 8·7% (95% CI –21·9 to 31·6) from the among children eligible for the first dose of RTS,S, age when children were eligible for the first dose until with 7777 and 7271 in non-eligible children, respectively they were eligible to have received the third dose; 39·6% (mortality rate ratio 0·88 [0·80–0·97]). The number of (22·2 to 53·1) during the 6 months from being eligible to deaths averted by RTS,S introduction during the have received the third dose; 20·4% (–2·0 to 37·9) over 46-month period was 993 (233–1692). We estimated that the next 8 months until the age scheduled for the fourth the number of child-years at risk among children eligible dose; 15·5% (–7·7 to 33·7) during the 12 months for the first dose of RTS,S in comparison areas was following the age eligible for the fourth dose; and 0·2% approximately 3·02 million giving an underlying (–39·1 to 28·4) more than 12 months after being eligible mortality rate in comparison areas of 2·86 per 1000 child- for the fourth dose (appendix 2 pp 40–42). years (0·77 in Ghana, 3·12 in Kenya, and 5·11 in Malawi; appendix 2 pp 27–42). Discussion Estimates of mortality rate ratios in girls and boys Our results show that introduction of the RTS,S malaria eligible to have received the third dose of RTS,S vaccine over a period of nearly 4 years through routine (appendix 2 p 33) were 0·85 (95% CI 0·75–0·97) for girls immunisation services in Ghana, Kenya, and Malawi and 0·87 (0·76–0·99) for boys (figure 4); the ratio averted an estimated one in eight deaths in children (girls:boys)of these mortality ratios was 1·00 (0·90–1·12, eligible to receive the vaccine. p =0·96). Our findings are based on a large-scale evaluation interaction The relative impact on mortality, from the age when involving 158 clusters with an average annual cohort of children were eligible for the first dose until they were 4000 surviving infants per cluster, where community eligible to have received the third dose, was 7·2% (95% CI surveillance for mortality was established, with over –4·9 to 17·8); during the 6 months from being eligible to 26 000 local reporters participating. Randomisation was have received the third dose, the impact was 12·4% used to assign clusters to either the introduction of (2·6 to 21·2); over the next 8 months until the age RTS,S in 2019 or to delay introduction of the vaccine. scheduled for the fourth dose, the impact was 14·0% Careful monitoring of the uptake of other childhood (1·9 to 24·7); during the 12 months following the age vaccines, the use of insecticide-treated nets, and care- eligible for the fourth dose, it was 15·2% (–2·8 to 30·2); seeking behaviour for febrile illnesses showed that these and in the last age group (≥34 months in Malawi, variables were well balanced at baseline and remained ≥36 months in Ghana and Kenya), impact was 2·0% similar in both implementation and comparison areas (–26·0 to 23·7; appendix 2 pp 38–39, 41–42). throughout the evaluation. These features increase the A total of 57 960 children aged 1–59 months admitted to confidence that the observed impact on mortality can be sentinel hospitals were included in analyses (appendix 2 attributed to the eects of the malaria vaccine, rather p 55). Of these, 21 419 (40·1%) had a positive malaria test than to confounding factors. and 7926 (15·6%) met the case definition for severe Our evaluation was conducted in the context of malaria (appendix 2 p 55). Among children eligible to delivering recommended childhood vaccines and malaria have received the third dose of RTS,S, there were control measures, including insecticide-treated nets, 11 738 hospital admissions from implementation areas targeted indoor residual spraying, and case management.7 (1457 [12·4%] with severe malaria) and 12 338 admissions The significant impact of RTS,S introduction on all-cause from comparison areas (1853 [15·0%] with severe mortality highlights that a substantial malaria burden malaria), whereas among non-eligible children there remained before introduction of the vaccine despite were 1705 admissions with severe malaria from relatively good coverage of the other interventions. implementation areas and 1773 from comparison areas, Vaccine coverage was broadly similar across countries, giving an estimated incidence rate ratio for severe and there was no evidence that impact (rate ratios for malaria admissions of 0·78 (95% CI 0·64–0·97) in mortality or severe malaria) diered by country. implementation versus comparison areas. There was no The evaluation areas were selected on the basis of evidence that impact against severe malaria diered by estimated malaria prevalence of at least 20% before the country (interaction test χ²=1·46, 2 df, p=0·48). The start of the evaluation, generally corresponding to areas incidence rate ratio for total admissions, of any cause, of moderate to high malaria transmission (appendix 2 was 0·92 (0·81–1·05). There were 3865 admissions with p 34);12 areas of sub-Saharan Africa with moderate and a positive test for malaria among children eligible for the high transmission account for most of the global malaria third RTS,S dose from implementation areas and burden.13 Countries currently introducing the malaria 4948 from comparison areas, compared with 4933 and vaccine have a malaria burden that is at least as high or Articles higher than that measured in the evaluation areas, and malaria in the implementation areas versus the therefore impact is likely to be at least as great as we comparison areas, and no evidence of an increase in cases observed, or greater, provided similar levels of vaccine of cerebral malaria or of meningitis, indicating that the coverage can be achieved. That this impact was achieved excess cases of these conditions that had been observed in when only 40% of children received the fourth RTS,S the phase 3 trial were not causally related to RTS,S vaccine.7 dose is reassuring, considering that there had been a Hospital surveillance continued throughout the evaluation concern that the public health impact of RTS,S might period, and the updated estimates provide no evidence of rely on achieving high coverage of four doses.3 an excess risk of cerebral malaria or meningitis (appendix 2 RTS,S was introduced in the context of routine p 36), and no new safety signals were identified.14 vaccination and scale-up, with the added challenges of Updated estimates for severe malaria show a decreasing delivery during the COVID-19 pandemic. Coverage of the impact of RTS,S as children age (appendix 2 p 40), which is initial three doses of RTS,S improved during the latter consistent with the expected gradual decline in protection part of the evaluation, but was lower in children from following the third dose, and the lower coverage of the households with lower wealth ranking, based on fourth dose, and the expected decline in protection after the household assets, than in children from households with fourth dose. The lower point estimate of overall impact higher wealth ranking, and was lower in areas of Malawi against severe malaria among children eligible to have with higher malaria prevalence. It is to be expected that received RTS,S, compared with the analysis after 24 months further improvements in coverage of the initial doses and of programme implementation7 (22% compared with 32%), of the fourth dose, including in households with lower reflects the greater proportion of older children in the wealth ranking and areas of highest malaria prevalence, current analysis (appendix 2 p 43). The point estimates of would lead to a correspondingly greater impact. impact on all-cause mortality did not decline as rapidly with It was evident from the household surveys that there increasing age as did those for severe malaria (appendix 2 were missed opportunities for administering RTS,S, as p 41). Although the confidence intervals are wide, this more children received the first dose of measles- finding is consistent with previous observations that, containing vaccine (scheduled at 9 months) than received among deaths in children aged 1–59 months, the proportion the first dose of RTS,S. Coverage of the second dose of attributed to malaria increased with age.15 Although the measles-containing vaccine was also higher than that of ecacy of RTS,S declines over time, the vaccine provides the fourth dose of RTS,S. In Ghana, moving the protection at important periods when children are most scheduled age of the fourth dose from 24 months to susceptible to malaria. coincide with the second dose of measles-containing Coverage of the initial three doses of RTS,S and the vaccine at 18 months, which happened in the final month fourth dose was similar in girls and boys. In the phase 3 of our evaluation, improved uptake of the fourth dose trial, exploratory analysis showed an interaction between (appendix 2 p 10). However, we were not able to evaluate sex of the child and vaccine group, apparently reflecting a the impact of this change. deficit of deaths among girls in the control group or an The percentage of children not using an insecticide- excess in the RTS,S group.2 However, the trial was not treated net (children aged 12–23 months in endline designed to measure ecacy against mortality, and this surveys) was 37·4% in Ghana, 5·7% in Kenya, and could have been a chance finding. Earlier data from our 19·7% in Malawi. Of these, 73·3%, 46·9%, and 50·6%, evaluation were sucient to exclude an interaction of the respectively, had received the third dose of RTS,S. Thus, magnitude observed in the phase 3 trial.9 The primary although coverage of RTS,S tended to be higher among analysis of impact on mortality undertaken at the end of users of insecticide-treated nets than non-users, an this evaluation shows a substantial impact on deaths of appreciable number of children who were not using nets all causes in both girls and boys, and no evidence that the received RTS,S, increasing the proportion of children impact diered according to the sex of the child. with access to at least one form of malaria prevention. A cohort-monitoring, phase 4 study was conducted in The introduction of RTS,S was not associated with an two areas in each country, adjacent to but outside our increase or decrease in uptake of other vaccines or use of evaluation area,16 primarily to evaluate safety signals in insecticide-treated nets. Rather, the additional health children receiving RTS,S, especially potential immune- visits for RTS,S immunisation provide opportunities for mediated diseases, which is a particularly challenging task. delivering other vaccines and child health interventions, The study found no evidence of the safety signals that had which could be exploited. been observed in the phase 3 trial. The study reported an Hospital surveillance was strengthened to monitor the excess of all-cause hospital admissions associated with incidence of severe malaria, including cerebral malaria RTS,S in some analyses, but caution is needed when and meningitis. Primary analysis of these outcomes was comparing estimates with those from our evaluation, as the undertaken in 2021, 24 months after RTS,S was introduced cohort-monitoring study was not randomised, and eects in Ghana and Malawi, and 19 months after it was were estimated among a select group of children who were introduced in Kenya. These findings showed a significant followed up individually through home visits, which could reduction in the number of children admitted with severe have altered patterns of health-seeking behaviour. 1806 Articles A limitation of our evaluation is that, as an impact (Abraham Hodgson PhD). Kenya: Centre for Global Health Research, assessment, estimates of individual personal protection Kenya Medical Research Institute, Kisumu (Bella Ondieg MSc, Brian Seda BSc, Dorcas Akach BSc, Gordon Orwa BSc, associated with the receipt of the initial three doses and of Isabella Nyang’au MSc, Oscar Odunga BSc, Francis Gumba BSc); the fourth dose could not be made. However, case–control KEMRI-Wellcome Trust Research Programme, Nairobi studies were conducted to estimate individual personal (Cynthia Khazenzi BSc, Eda Mumo MSc, Monica Musa BSc, protection.17 Recruitment concluded early in 2025, and Morris Ogero PhD, Mike English FMedSci). Malawi: Kamuzu University of Health Sciences, School of Public Health, Blantyre the results will be published separately. (Harrison Msuku BSc, Vincent Samuel Phiri MSc); University of Deaths might have been under-reported or missed, North Carolina Project-Malawi, Lilongwe (Hillary M Topazian PhD, despite eorts to standardise procedures and the Jonathan Juliano MD, Lusungu Msumba MPH). UK: Centre for Tropical implementation of quality assurance measures. However, Medicine and Global Health, Nueld Department of Clinical Medicine, University of Oxford, Oxford (Mike English FMedSci) by using data on non-eligible groups of children, we Contributors controlled for potential confounding due to dierences in KPA led the evaluation in Ghana, AMS led the evaluation in Kenya, reporting rates between implementation and comparator and DPM led the evaluation in Malawi. DS, PM, MJH, AMS, NW, PB, clusters. In Malawi, our estimates of mortality in vaccine- RWS, SA, SK, TKK, LO, KPA, DPM, AO, PW, TG, DA, EAA, VM, TM, eligible children were consistent with rates derived from and AK-T designed the evaluation. AC, AK-T, AO, AMS, DA, EAA, JC, KM, KPA, LO, MZ, NW, PW, SA, SK, TA, TG, EO-L, CCS, TKK, TM, and UNICEF subnational estimates of mortality rates VM were involved in data collection. PS was involved in curating the (appendix 2 p 37). In Ghana and Kenya, our estimates data. AF, KAM, and PM analysed the data. Authors who contributed to were somewhat lower; this finding might reflect that verifying the data and analyses were: AK-T, NW, PW, AF, and KAM for recent reductions in mortality in children younger than household surveys; DA, SA, TM, AF, KAM, and PM for hospital data; and AO, AMS, VM, AF, KAM, and PM for mortality data. AMS, VM, 5 years, during the period of our evaluation, are not fully KPA, and PM primarily wrote the manuscript, with contributions from captured by the UNICEF estimates. Even if there was AF, MJH, DS, KAM, RNAO, and PN. PM, AO, AMS, DA, DPM, DS, under-reporting in Ghana and Kenya, it is unlikely that EAA, KM, KAM, MJH, NW, PB, PN, PW, RWS, SA, CCS, TM, RNAO, there would be a systematic dierence in reporting EP-F, and VM were involved in reviewing and editing the paper. All authors had full access to all the study data, had final responsibility for between eligible and ineligible children, so a bias in the the decision to submit for publication, and agreed to be accountable for estimate of impact on mortality is unlikely. all aspects of the work. The significant reduction of deaths in young children Declaration of interests following RTS,S introduction, with only moderate KPA, AO, MJH, NW, PN, DA, SK, TM, DS, LO, PB, TA, and AMS coverage of three doses of the vaccine and low uptake of previously participated in RTS,S malaria vaccine clinical trials as the fourth dose, underscores the importance of malaria principal investigators or co-investigators before or after WHO’s RTS,S malaria pilot implementation programme. The authors were not as a cause of death in young children in the regions where involved in the RTS,S vaccine delivery programme. All other authors the vaccine was introduced. It also supports the need for declare no competing interests. additional malaria control measures, even where there is Data sharing good coverage of existing malaria interventions. Our Anonymised data from the mortality and hospital surveillance and results highlight the urgent need to accelerate the coverage surveys are available through Harvard Dataverse, at https://doi. deployment of malaria vaccines in areas where malaria org/10.7910/DVN/PXVN6W. Requests for access will be reviewed by a data access committee. continues to be a leading cause of child mortality. Acknowledgments Malaria Vaccine Programme Evaluation partners The MVIP was funded through WHO with support from Gavi, the WHO: WHO, Geneva, Switzerland (Pedro Alonso MD, Vaccine Alliance, the Global Fund to Fight AIDS, Tuberculosis and Katherine L O’Brien MD); WHO Country Oce, Nairobi, Kenya Malaria, and Unitaid. We would like to thank Gavi, the Vaccine Alliance, (Adam Haji MD, Josephine Njoroge MD); WHO Country Oce, Accra, the Global Fund to Fight AIDS, Tuberculosis, and Malaria, and Unitaid Ghana (Peter O Tweneboah MPH); WHO Country Oce, Lilongwe, for their generous support of the pilot introductions and evaluations. In Malawi (Boston E Zimba MD, Esther Chirwa MD); Department of Ghana, we would like to acknowledge the Ministry of Health and Ghana Immunizations, Vaccines, and Biologicals, WHO, Geneva, Switzerland Statistical Service, the Ghana Health Service Director General Expanded (Cynthia Bergstrom MPH); Vaccine-Preventable Diseases Unit, WHO Programme for Immunisation and National Malaria Elimination AFRO Regional Oce, Brazzaville, Congo (Mgaywa GMD Magafu PhD); Programme of Ghana Health Service, PATH, directors of the regional Tropical and Vector-Borne Diseases Unit, WHO AFRO Regional Oce, health service, regional health directorates, regional verbal autopsy Brazzaville, Congo (Jackson Sillah MD). Ghana: Kintampo Health coordinators, district verbal autopsy coordinators, clinical coordinators, Research Centre, Research and Development Division, Ghana Health clinical and laboratory monitors, sentinel hospitals, community key Service, Kintampo North Municipality informants, community members, Sunyani Regional Hospital, (Sulemana Watara Abubakari PhD, Dennis Adu-Gyasi PhD, St Elizabeth Hospital, St Francis Xavier Hospital, Winneba Specialist & Augustine Sarfo BSc, Francis Agbokey MSc, David Dosoo PhD, Trauma Hospital, Cape Coast Teaching Hospital, Keta Municipal Abdul Razak Nuhu MSc, Stephany Gyaase MSc); Ghana Statistical Hospital, Ho Teaching Hospital, and Nkwanta South Municipal Service, Accra (Peter K Peprah PhD); University of Health and Allied Hospital. In Malawi, we would like to acknowledge the Ministry of Sciences, Ho (Gregory Amenuvegbe MPH, Fred Binka PhD); School of Health (Expanded Programme on Immunisation and National Malaria Public Health, University of Ghana, Accra (Dora Dadzie PhD); Dodowa Control Programme), the Ministry of Local Government, Wellcome Trust Health Research Centre, Research and Development Division, Ghana Malawi, the Malawi National Statistics Oce, the Malawi Epidemiology Health Service, Dodowa (Elizabeth Awini PhD); Komfo Anokye Teaching and Research Unit, the National Registration Bureaux, district health Hospital, Kumasi (Justice Sylverken MD); Health Research Centre, management teams (Nsanje, Chikwawa, Phalombe, Machinga, Balaka, Research and Development Division, Ghana Health Service, Navrongo Mangochi, Lilongwe, Mchinji, and Ntchisi), chiefs in all the districts, (Aaron Kampim MPH); Noguchi Memorial Institute for Medical and PATH. In Kenya, we would like to acknowledge the Kenya National Research, University of Ghana, Accra (Kwadwo Koram PhD); Research Bureau of Statistics, the Expanded Programme on Immunisation, the and Development Division, Ghana Health Service, Accra Articles Division of National Malaria Programme, county directors of health, 5 RTS,S Clinical Trials Partnership. Ecacy and safety of RTS,S/AS01 county and subcounty focal people for community strategy, community malaria vaccine with or without a booster dose in infants and health assistants and community health volunteers (western Kenya children in Africa: final results of a phase 3, individually counties), sentinel hospitals in Homa Bay, Vihiga, Kakamega, and randomised, controlled trial. Lancet 2015; 386: 31–45. Bungoma, Busia County referral hospitals, and the Jaramogi Oginga 6 WHO. Proposed framework for policy decision on RTS,S/1S01 Odinga Teaching and Referral Hospital. From the Malaria Vaccine malaria vaccine. 2019. https://www.who.int/publications/m/item/ Implementation Programme Advisory Group, we would like to thank: proposed-framework-for-policy-decision-on-rts-s-as01-malaria- vaccine-2019 (accessed Jan 6, 2026). Nick Andrews, Graham Brown, Patrick Kachur, Corine Karema, 7 Asante KP, Mathanga DP, Milligan P, et al, and the Malaria Vaccine Dafrossa Lyimo, Eusebio Macete, Kim Mulholland, Kathleen Neuzil, and Programme Evaluation Partners. Feasibility, safety, and impact of Peter Smith (Chair). We would like to thank the data safety monitoring the RTS,S/AS01 malaria vaccine when implemented through board members, including Esperança Sevene, Jane Achan, E national immunisation programmes: evaluation of cluster- Charles Newton, Larry Moulton, and Cynthia Whitney (Chair). randomised introduction of the vaccine in Ghana, Kenya, and We are grateful to Annick Janin and Mayuko Takamiya for their Malawi. Lancet 2024; 403: 1660–70. administrative support during the manuscript writing process. SA was 8 Milligan P. Statistical analysis plan for the malaria vaccine pilot supported at the initiation of the pilots by the Initiative to Develop evaluation. 2021. https://cdn.clinicaltrials.gov/large-docs/65/ African Research Leaders (IDeAL) Wellcome Trust award (number NCT03806465/SAP_002.pdf (accessed Jan 6, 2026). 107769). Funding to Mike English supported the establishment of the 9 Milligan P, Moore K. WHO guidelines for malaria: systematic Clinical Information Network (CIN)–Senior Wellcome Fellow (number reviews, background papers and other unpublished evidence 097170). We acknowledge the support of the Wellcome Trust to the Kenya considered in the development of recommendations. 2021. https:// Africa Asia Programme (number 092654 and 203077) that provides core www.nitag-resource.org/sites/default/files/2022-05/Milligan-Moore- support to CIN, and the support to RWS as part of his Wellcome Trust 2021-statistical-report-rtss-pilot-evaluation.pdf (accessed Jan 6, 2026). Principal Fellowship (number 212176). KAM was supported by an 10 Ma X, Milligan P, Lam KF, Cheung YB. Ratio estimators of Australian National Health and Medical Research Council (NHMRC) intervention eects on event rates in cluster randomized trials. Early Career Fellowship (grant number APP1160936). The content of this Stat Med 2022; 41: 128–45. Article is solely the responsibility of the authors and does not necessarily 11 Milligan P, Fogelson A. Statistical report, Malaria Vaccine Pilot represent the ocial views of the NHMRC. The findings and Evaluation (MVPE), analysis of data to month 46. 2024. https://www. who.int/publications/m/item/statistical-report--malaria-vaccine-pilot- conclusions in this report are those of the authors and do not necessarily evaluation-(mvpe)--analysis-of-data-to-month-46 (accessed Jan 6, 2026). represent the ocial position of the US Centers for Disease Control and 12 WHO. WHO guidelines for malaria. 2025. https://www.who.int/ Prevention. MJH is a sta member of WHO; the views expressed herein publications/i/item/guidelines-for-malaria (accessed Jan 6, 2026). do not necessarily represent the decisions, policy, or views of WHO. 13 WHO. World malaria report 2025. 2025. https://www.who.int/ References teams/global-malaria-programme/reports/world-malaria- 1 WHO. WHO recommends groundbreaking malaria vaccine for report-2025 (accessed Jan 6, 2026). children at risk: historic RTS,S/AS01 recommendation can 14 WHO. Report of the Meeting of the WHO Global Advisory reinvigorate the fight against malaria. 2021. https://www.who.int/ Committee on Vaccine Safety, 15–16 May 2023. Wkly Epidemiol Rec news/item/06-10-2021-who-recommends-groundbreaking-malaria- 2023; 98: 345–54. vaccine-for-children-at-risk (accessed Jan 6, 2026). 15 Ogbuanu IU, Otieno K, Varo R, et al, and the CHAMPS 2 WHO Malaria Vaccine Implementation Programme Programme consortium. Burden of child mortality from malaria in high Advisory Group. Full evidence report on the RTS,S/AS01 malaria endemic areas: results from the CHAMPS network using vaccine. 2021. https://cdn.who.int/media/docs/default-source/ minimally invasive tissue sampling. J Infect 2024; 88: 106107. immunization/mvip/full-evidence-report-on-the-rtss-as01-malaria- 16 Haine V, Oneko M, Debois M, et al. Safety of RTS,S/AS01E malaria vaccine-for-sage-mpag-(sept2021).pdf (accessed Jan 6, 2026). vaccine up to 1 year after the third dose in Ghana, Kenya, and 3 WHO. Malaria vaccine: WHO position paper – January 2016. Malawi (EPI-MAL-003): a phase 4 cohort event monitoring study. Wkly Epidemiol Rec 2016; 91: 33–51. Lancet Global Health 2025; 13: e995–1005. 4 WHO. The Malaria Vaccine Pilot Evaluation: an evaluation of the 17 Milligan P, Fogelson A. Case-control studies of the eectiveness of cluster-randomised pilot implementation of RTS,S/AS01 through the RTS,S/AS01 malaria vaccine against hospital admission with routine health systems in moderate to high malaria transmission severe malaria, and associations with safety signals: statistical settings in sub-Saharan Africa: a post-authorization observational report. 2026. https://zenodo.org/records/17018739 (accessed study. Master protocol. 2020. https://classic.clinicaltrials.gov/ Jan 6, 2026). ProvidedDocs/65/NCT03806465/Prot_ICF_000.pdf (accessed Jan 6, 2026). 1808 --- [PDF原文](https://sci-net.xyz/storage/8010893/c4fcd77efa9911586507447f7aa09103569411bab2d0c8cddc2f1ff0985434ea/Impact-of-introducing-RTS-S-AS01E-malaria-vaccine-on-mortality-in-young-children-in-Ghana.pdf) DOI: 10.1016/S0140-6736(26)00248-5