Malaria is a major public health concern in Africa. In regions where transmission is highly seasonal, seasonal malaria chemoprevention (SMC) has been widely implemented to prevent morbidity and mortality in children at risk of severe malaria. SMC involves administration of sulfadoxine–pyrimethamine plus amodiaquine at 4-week intervals, given during the peak transmission season to clear existing parasitaemia and prevent new infections.¹ Since its recommendation by WHO in 2012, SMC has expanded to 18 African countries, treating 1·2 million children during 2013–14 and a total of 53·0 million children by 2023,^2,3 resulting in reductions in incidence of childhood malaria of 60–88%.^1,4

High coverage of SMC, strong vector control, and rapid case management have enabled countries in Sahelian and sub-Sahelian Africa to substantially advance malaria control, prompting many to establish new goals for elimination. However, progress towards goals during the past 5 years has stalled,³ requiring enhanced coverage of core interventions and new interventions to accelerate transmission reduction. One promising approach is mass drug administration (MDA), which involves administering antimalarials to all individuals in a defined geographical area at a frequency and duration adapted to the local malaria epidemiology and goals. To effectively affect transmission, MDA needs to achieve high coverage of the target population (ie, ≥80%),⁵ which requires optimised delivery methods and strong community engagement.⁶ These requirements might be more achievable in countries already implementing SMC, where there is existing infrastructure of door-to-door delivery and community and health-system acceptance of chemoprevention.^7,8

The effectiveness of MDA depends on the antimalarial regimen used. In regions with dominant Plasmodium falciparum, dihydroartemisinin–piperaquine is an attractive choice due to its good safety profile, its long prophylactic period, and low prevalence of artemisinin resistance in Africa.^9,10 However, dihydroartemisinin–piperaquine is not effective against mature gametocytes,^11,12 the parasites responsible for human-to-mosquito transmission. Single, low-dose primaquine, a gametocytocidal agent, is safe and associated with the near-complete prevention of human-to-mosquito transmission.^13–15 Although empirical data on whether the human-to-mosquito transmission-blocking activity of primaquine translates into population-level effects on transmission is scarce, its addition to dihydroartemisinin–piperaquine could offer greater benefits than dihydroartemisinin–piperaquine alone, including the potential to reduce the spread of drug-resistant parasites.^16,17

In 2022, WHO updated their guidelines recommending MDA in areas with high coverage of standard malaria interventions and low malaria-importation risk.¹ The guidelines recommend MDA for transmission reduction in settings with very-low-to-low transmission, broadly defined by WHO as parasite prevalence less than 10% or annual malaria incidence of less than 250 cases per 1000 population on the basis of eight cluster-randomised controlled trials showing a substantial but short-term effect.^1,18,19 In settings with moderate-to-high transmission, defined by WHO as parasite prevalence 10% or more or annual malaria incidence of more than 250 cases per 1000 population,¹ MDA is recommended for rapid reduction of disease burden but not for transmission reduction due to scarce evidence on its short-term or long-term benefits.

In southeast Senegal, malaria transmission ranges from low to high (ie, 50–500 cases per 1000 population per year) and is highly seasonal, with most cases occurring between July and December. In this region, the national malaria programme Programme National de Lutte contre le Paludisme (PNLP) implements standard malaria-control interventions, including routine distribution of insecticide-treated bednets; health-facility case management; and SMC to children aged 3–120 months, except those with reported acute illness or fever, known hypersensitivity to SMC drugs, history of antimalarials in the previous 3 weeks, or history of sulfonamide-containing drugs in the previous 10 days. In many remote villages in southeast Senegal, there have been ongoing efforts to scale up proactive community case management of fever through the Prise en Charge à Domicile Plus (PECADOM+) model. In the PECADOM+ model, community health workers, known as dispensateur de soins à domiciles (DSDOMs), conduct household visits once per week to test people with suspected malaria and treat those with a positive diagnosis during the high transmission season, regardless of age. Despite scale-up of these interventions, progress in transitioning southeast Senegal to pre-elimination, defined by the PNLP as an annual incidence of less than 5 cases per 1000 population, has been slow. Therefore, the programme needs an accelerator intervention to aggressively increase elimination margins and meet the national malaria-elimination goal by 2030.

We aimed to assess the safety, coverage, and effect of three cycles of MDA with dihydroartemisinin–piperaquine plus single, low-dose primaquine on P falciparum incidence and prevalence in southeast Senegal, a setting with highly seasonal, low-to-moderate transmission where malaria-control measures have been scaled up.

Between Sept 1 and Oct 25, 2020, 523 villages were geolocated and screened for eligibility; 111 met the inclusion criteria (figure 2). Of these 111, 60 villages were randomly selected and assigned to the intervention arm or control arm. Baseline malaria incidence was highly variable across villages (coefficient of variation 1·04). Village-level factors included in the stratified, constrained randomisation approach were balanced between arms (table 1). Overall, coverage of pyrethroid–piperonyl butoxide bednets was high and similar between arms (table 1). Overall, 687 (80·3%) of 856 children younger than 10 years reported receiving the most recent cycle of SMC during the pre-intervention year and 610 (71·1%) of 858 children younger than 10 years reported receiving all three cycles during the pre-intervention year. 459 (19·9%) of 2305 participants reported sleeping away from their home in the past 15 days.

In the intervention arm, 8931 residents were screened for the first MDA cycle, 9571 were screened for the second, and 9703 were screened for the third. Distribution coverage of all three doses of dihydroartemisinin–piperaquine was 6057 (73·6%) of 8229 participants in the first cycle, 6836 (78·8%) of 8673 participants in the second cycle, and 7065 (81·3%) of 8690 participants in the third cycle. Distribution coverage of all three doses of single, low-dose primaquine was 6286 (78·6%) of 7999 participants in the first cycle, 6949 (82·1%) of 8462 participants in the second cycle, and 7199 (84·0%) of 8575 participants in the third cycle. Distribution coverage of all three doses of dihydroartemisinin–piperaquine was 2618 (81·0%) of 3232 children younger than 10 years in the first cycle, 2904 (83·1%) of 3493 children younger than 10 years in the second cycle, and 3046 (85·6%) of 3560 children younger than 10 years in the third cycle versus 3438 (68·8%) of 4996 children aged 10 years or older in the first cycle, 3932 (75·9%) of 5180 children aged 10 years or older in the second cycle, and 4017 (78·4%) of 5127 children aged 10 years or older in the third cycle. Crude coverage of all three doses of dihydroartemisinin–piperaquine was 6057 (67·8%) of 8931 participants in the first cycle, 6836 (71·4%) of 9571 participants in the second cycle, and 7065 (72·8%) of 9703 participants in the third cycle.

At the village level, median distribution coverage of all three doses of dihydroartemisinin–piperaquine was 74·7% (IQR 68·9–78·8) in the first cycle, 78·6% (76·3–81·6) in the second cycle, and 81·5% (76·5–85·8) in the third cycle (appendix p 6). Six (20·0%) of 30 intervention villages in the first cycle, 12 (40·0%) in the second cycle, and 18 (60·0%) in the third cycle reached the WHO target of at least 80% coverage (appendix p 6). The main reasons for non-participation were absence (range 13·5–21·1% across doses and cycles) and illness (5·1–7·5% across doses and cycles; appendix p 7). Absences were similar by sex (1·11:1 male:female ratio) and age (median 16 years, IQR 8–26 vs 12 years, 6–26). Refusals were rare (range 1·4–1·8% across doses and cycles) and were mostly among male participants (218 [67·1%] of 325 participants who refused any one cycle) with a median age of 21 years (IQR 14–30; appendix p 7).

In the control arm, 3492 children aged 3–120 months were screened for the first SMC cycle, 3489 were screened for the second, and 3487 were screened for the third. Distribution coverage of all three doses of SMC was 3187 (92·2%) of 3457 children in the first cycle, 3158 (91·8%) of 3442 children in the second cycle, and 3139 (91·4%) of 3434 children in the third cycle. Crude coverage of all three doses of SMC was 3187 (91·3%) of 3492 children in the first cycle, 3158 (90·5%) of 3489 children in the second cycle, and 3139 (90·0%) of 3487 children in the third cycle. At the village level, median distribution coverage of all three doses of SMC was 93·9% (IQR 87·9–98·2) in the first cycle, 92·9% (87·2–97·4) in the second cycle, and 93·3% (85·2–96·5) in the third cycle (appendix pp 6–7). The main reasons for non-receipt were absence (range 6·7–7·5% across doses and cycles) and illness (1·0–1·7% across doses and cycles; appendix p 7). Refusals were low across cycles (0·6–0·8% across doses and cycles; appendix p 7).

Mean village-level malaria incidence during the transmission season of the pre-intervention year (ie, July to December, 2020) was 181 cases per 1000 population in the intervention arm and 204 cases per 1000 population in the control arm. Mean village-level malaria incidence during the transmission season of the intervention year was 93 cases per 1000 population in the intervention arm and 173 cases per 1000 population in the control arm. Mean village-level malaria incidence during the transmission season of the post-intervention year was 126 cases per 1000 population in the intervention arm and 146 cases per 1000 population in the control arm (table 2; appendix p 8). Analyses of incidence data (appendix pp 8–10) indicated that neither SMC or MDA were optimally timed to cover the entire transmission season.

During the intervention year (ie, July to December, 2021), the adjusted effect of MDA was 55% (95% CI 28 to 71). Prespecified subgroup analyses showed an interaction by age; the adjusted MDA effect was 58% (34 to 73) in participants aged 10 years or older and 45% (10 to 66) in participants younger than 10 years (p_interaction=0·012). The adjusted MDA effect was 56% (32 to 71) in low-transmission settings and 52% (−22 to 81) in moderate-transmission settings (pinteraction=0·87). Analyses restricted to villages with DSDOMs at baseline showed that the effect of MDA was 60% (36 to 76). In the post-intervention year (ie, July to December 2022), the adjusted MDA effect was 26% (−17 to 53). There was no interaction by transmission intensity (p_interaction=0·98). An interaction by age was observed (p_interaction=0·0065), although differences in effect estimates were minimal. Analyses restricted to villages with DSDOMs at baseline showed the effect of MDA was 30% (−14 to 57). Only two villages in the intervention arm reached the threshold for pre-elimination (incidence of less than 5 cases per 1000 population) during the intervention year and none in the post-intervention year. None of the control villages reached pre-elimination status in either year. We assessed incidence during the full calendar year (ie, January to December) in the post-intervention year, as specified in the protocol, and found similar results to analyses restricted to the transmission season (appendix p 11).

Surveys conducted at the end of the transmission season in the pre-intervention and intervention years (ie, December, 2020 and 2021), showed mean village-level microscopy-confirmed parasite prevalence reduced from 6·1% (95% CI 2·8 to 9·4) in 2020 to 1·8% (0·8 to 2·9) in 2021 in the intervention arm and from 6·7% (4·0 to 9·4) to 4·7% (2·5 to 6·9) in the control arm (adjusted MDA effect 62%, 95% CI 22 to 80; table 3). The effect of MDA was 76% (42 to 90) in children younger than 10 years and 51% (−14 to 79) in children aged 10 years or older (p_interaction=0·17). Post-hoc analyses showed mean village-level microscopy-confirmed gametocyte prevalence decreased from 2020 to 2021: from 3·2% to 0·6% in the intervention arm and from 3·1% to 1·0% in the control arm (adjusted MDA effect 39% [95% CI −98 to 81]).

By PCR, mean village-level parasite prevalence decreased from 17·9% (95% CI 11·3 to 24·4) to 4·5% (2·5 to 6·4) in the intervention arm and from 19·9% (11·8 to 28·1) to 8·3% (4·8 to 11·7) in the control arm (adjusted MDA effect 47%, 95% CI 3–71). The effect of MDA was 71% (35 to 87) in participants younger than 10 years and 33% (−27 to 65) in children aged 10 years or older (pinteraction=0·050). The effect of MDA on microscopy-confirmed and PCR-confirmed parasite prevalence did not differ between low-transmission and moderate-transmission settings (pinteraction=0·73; table 3).

In both active and passive pharmacovigilance systems, the frequency of adverse events decreased with each cycle and no serious adverse events or anaemia were detected in either arm. Active surveillance showed that more participants reported any adverse event in the intervention arm than in the control arm (table 4). Among children aged 10 years or older, the proportion of participants reporting an adverse event across all cycles did not differ between arms (table 4). In the intervention arm, common adverse events were gastrointestinal issues, fever, and other (ie, dizziness, cough, skin rashes, pruritus, jaundice, muscle and joint pain, convulsions, and influenza-like illness; table 4). Of the 498 adverse events in the intervention arm, 319 (64·1%) were mild, 164 (32·9%) were moderate, and 15 (3·0%) were severe. Severe adverse events were fever (n=7), headache (n=4), drowsiness (n=1), vomiting (n=1), diarrhoea (n=1), and loss of appetite (n=1). All adverse events occurred within 3 h after drug intake, resolved within 72 h, and did not require hospitalisation. Results from passive surveillance were similar (appendix p 12).

Genotypic analysis of mutations associated with antimalarial resistance showed no differences between arms or time periods (appendix p 13). There were no pfdhps K540E, pfdhps A581G, or PfK13 C580Y mutations (appendix p 13).

Our two-arm, open-label, cluster-randomised controlled trial showed that after the pre-intervention year, malaria incidence and parasite prevalence reduced in both arms, likely in part due to distribution of pyrethroid–piperonyl butoxide bednets. During the intervention year in the MDA arm, malaria incidence was reduced by 55%, microscopy-confirmed parasite prevalence by 62%, and PCR-confirmed parasite prevalence by 47%. Subgroup analyses showed reductions in incidence were observed in both low-transmission and moderate-transmission settings and in participants of any age. Consistent with other trials,^9,28,29 MDA with dihydroartemisinin–piperaquine plus single, low-dose primaquine was generally safe and well tolerated. During the post-intervention year, when villages returned to standard of care, incidence was lower in intervention villages than in control villages. However, the strength of this evidence was weak. Pre-elimination status (ie, less than 5 cases per 1000 population) was reached in only two MDA villages during the intervention year and none in the post-intervention year. Overall, our findings suggest that although MDA is effective in rapidly reducing malaria burden for a short-term reduction in transmission, its sustained effect was weak and did not accelerate areas to pre-elimination.

The direction of our estimates are consistent with a 2022 cluster-randomised controlled trial conducted in The Gambia,²⁹ which evaluated the effect of three MDA cycles with dihydroartemisinin–piperaquine plus ivermectin in a moderate-transmission setting across two transmission seasons. The trial showed an odds ratio of 0·78 (95% CI 0·43–1·41) regarding malaria incidence during the second transmission season and an odds ratio of 0·30 (0·16–0·59) regarding PCR-confirmed parasitaemia 2 months after the final round of MDA. Unlike our trial, The Gambia trial did not assess whether effects persisted into the post-intervention year. On the basis of our findings and a previous consensus modelling study,²³ the effects of MDA are likely transient and would eventually return to pre-MDA levels without equally aggressive follow-up measures. For MDA to accelerate malaria elimination in low-to-moderate transmission settings, it might need to be sustained for several years until malaria cases reach sufficiently low levels (ie, less than 5 cases per 1000 population), at which point more targeted interventions, such as focal MDA, could be introduced.

One potential reason we might not have observed a sustained effect of MDA could be insufficient statistical power, as the trial was designed to detect a 50% intervention effect in the follow-up year. Another possibility could be the unexpectedly high variability in baseline malaria incidence, which was differentially measured in villages without a DSDOM before the trial, potentially biasing effect estimates towards the null. Restricting our analysis to those with DSDOMs at baseline showed a slight increase in the intervention effect. The previous consensus modelling study²³ suggested that additional years of MDA or other transmission-reducing measures are needed to maintain or further reduce transmission. We were unable to secure funding for additional intervention years, reflecting the challenges that programmes might have in securing sustained funding to avoid rebound and effectively drive elimination.

Our incidence data suggest that neither SMC or MDA were optimally timed. Had both campaigns started 1 month later and been extended to four cycles, more of the transmission season would have been covered, potentially enhancing the effectiveness of both campaigns. The suboptimal delivery of MDA might also partly explain its weak sustained effect in the post-intervention year, as it likely resulted in incomplete clearance of the transmission reservoir.

Our trial had several strengths, including rigorous safety monitoring. By the final MDA cycle, 60% of villages achieved the WHO target of at least 80% coverage⁶ for all three doses, despite the COVID-19 pandemic. High intervention coverage was likely attributable to community acceptance of SMC, support from key administrative and health authorities, and repeated community-sensitisation campaigns. However, absences were common during the campaign, especially among adolescents and young adults, who often go undetected but contribute substantially to trans mission.³⁰ To further sustain MDA gains, future campaigns should consider additional strategies to reach these groups.

Beyond the potential underpowering to detect sustained effects, our trial had other limitations. First, the trial was open-label and not placebo-controlled, which could explain the higher number of adverse events in the intervention arm among participants aged 10 years or older who do not routinely receive chemoprevention. However, this likely did not bias our malaria outcomes, which were supported by diagnostic methods. Second, the superior effect of MDA over SMC in children younger than 10 years should be interpreted with caution, as SMC was not optimally timed to align with the transmission season. Third, due to budget constraints, we were unable to conduct a survey in the post-intervention year, limiting our ability to understand MDA effects on parasite prevalence during this time. Fourth, our trial was likely underpowered to detect subgroup effects. However, effect estimates were consistent across low-transmission and moderate-transmission settings with highly overlapping confidence intervals, suggesting that differences were likely minimal. Finally, the trial design did not permit separate estimation of the effect of adding single, low-dose primaquine.

Our findings might not be generalisable to most settings using SMC where the malaria burden is mostly moderate to high. However, when these settings reach low-to-moderate malaria transmission and coverage of standard malaria-control interventions is high, MDA with dihydroartemisinin–piperaquine plus single, low-dose primaquine could safely and rapidly reduce transmission. Effects are unlikely to be sustained after discontinuation, and sustained funding for additional years of transmission-reducing interventions will be required to accelerate towards and achieve elimination.