Evidence for the impact of SMC on malaria transmission is limited, but has critical implications for programme scale-up18. In this study, we have applied a household-level modelling approach to assess the effects of extended-age SMC on older household inhabitants in a pre-elimination setting in the Gambia. In adjusted models we observed a significant reduction in the incidence of clinical malaria in eligible children who received SMC compared to those who did not. We observed that older participants ineligible for SMC in households with higher coverage of SMC were significantly less likely to contract clinical malaria compared to households where SMC coverage was low. In a survey of the population in the late transmission period we also observed fewer asymptomatic infections in older participants ineligible for SMC in households with higher SMC coverage. There was no evidence of spatial clustering of these impacts.
Whilst the incidence of clinical malaria in children aged 0–9 years decreased with increasing household level SMC coverage and irrespective of their individual level SMC status, we could not reliably demonstrate this constituted an additional benefit of SMC in our adjusted models (STable 7 Supplementary File) neither did we detect an impact of household level coverage of SMC on asymptomatic infections in SMC-eligible children (STable 8 Supplementary File).
The impacts observed in this study broadly align with the results of a 2008–2010 pilot stepped wedge trial which experimentally increased the eligible age range for SMC in Senegal to 10 years6 – to our knowledge this is the only other source of data on clinical impact following SMC in this age group. The authors reported a decrease in the incidence of clinical malaria in eligible children (69%) and in older age groups (26%). The study did not consider the impacts of lower coverage or impact on asymptomatic infections, and was implemented with excellent research fidelity and high SMC coverage over the three years of implementation, where 84%-93% of children received 3 rounds of SMC. A key question for decision-making is whether similar impacts on malaria transmission are achievable in routinely delivered large-scale programmes. Following an era of efficacy and effectiveness trials of SMC initially demonstrating high programme coverage and large effects in study populations, emerging data from routine SMC programmes in West Africa suggests mixed, or reduced impact 8,9,29–31. In our study, coverage of children receiving 3 or more rounds of SMC was 49% overall and only 36% of children 5–9 years; however this was the first season in which the NMCP of the Gambia extended the eligible age for SMC, and the coverage may reflect the initial complexities in operationalising the change. It nonetheless suggests transmission impacts are detectable in the context of programmes with moderate implementation strength.
There is an extensive theoretical basis from both laboratory and field studies which suggest the infectious reservoir for falciparum malaria in sub-Saharan Africa is normally maintained by children up to 15 years of age 12,15,16, and the expected extended benefits of preventative interventions in this age group should be observable in non-eligible adolescents and adults18. We justify the use of a household-level models to assess impact of SMC on transmission based on previous research in The Gambia26 and similar low to moderate endemic settings27 which indicate malaria transmission is highly localised within household/family groups32 with negligible contribution from nearby households. This is supported by the null findings from our global Moran tests for spatial autocorrelation in household level SMC models in the study site. Based on this proposed framework, we hypothesised the mechanism through which herd impacts of SMC arise will be by first reducing infections in the target age group in the same household. We identified a significant reduction in clinical infections in children 0–9 years, but only decreased asymptomatic infections in children with 3 or more rounds of SMC which did not reach significance (Supplementary file 1 STable 5). Interestingly children who received only 1–2 rounds of SMC benefitted in terms of reduced incidence of clinical episodes but had increased risk of asymptomatic infections compared to children receiving no SMC, though overall prevalences were low (< 6.5%). Children who had received only 1–2 SMC round were also more likely to have received SP in the first half of the season (Supplementary file 1 STable 6) perhaps leading to a slightly elevated risk of infection in the following months of rising transmission due to naïve immunity - however given low prevalence estimates we cannot rule out the potential presence of sampling error. The evidence for the transmission-blocking potential of SP-AQ is conflicting; in vivo and in vitro studies have indicated exposure to SP may increase commitment of blood stage parasites to sexual differentiation resulting in increased gametocyte concentrations and potential increased transmission potential to feeding mosquitoes33–35 however detailed analysis of this phenomenon suggests impacts occur for a small number of drug classes under a narrow drug concentration window, and are unlikely to result in a net increase in transmission potential36. Pyrimethamine has also shown suppressive activity against oocyte production or successful development of sporozoites in the mosquito37,38, which may offset increased gametocytogenesis with use of SP. Field studies support a theory of increased gametocytogenesis following SP use or SMC but are also characterised by small sample sizes and incomplete accounting for potential confounding, and alone do not resolve this conflict39,40. Whilst gametocyte concentrations were not available for participants in our study, we did observe a trend of increased parasite concentrations in children with SMC, and in older participants in households with higher SMC coverage known to be positively associated with gametocyte density12, however infection rates were too small to confirm this relationship. Other studies have shown that densities of both asexual and gametocyte stages are nonetheless considerably lower in asymptomatic infections compared to clinical disease41,42. This current body of evidence highlights the mechanisms of action of SP-AQ SMC are complex and not yet fully elucidated; the net product of a trade-off between reduced clinical infection, increased asymptomatic parasitaemia and gametocytaemia, and potential suppressive effects on mosquito lifecycle stages may be a reduction in overall transmission potential.
An assessment of the transmission-reducing effects of SMC is most useful for policy if based on results of routinely delivered programmes. This setting better predicts real-life impacts under sub-optimal or heterogenous SMC coverage and in the presence of other nationally implemented malaria interventions. High intervention coverage and strict eligibility criteria for participants and clusters observed in trial settings often do not transfer with high fidelity or effectiveness for this reason. Nonetheless a limitation of our observational approach is determining the appropriate techniques to account for other exogenous predictors of transmission that may also differ by SMC coverage, not all of which can be measured. We included a priori fixed effects to statistical models to address this. Village ID was used as a proxy for underlying transmission potential for a community, clustering by ethnicity, and in the context of the wider INDIE study also allowed us to adjust for ongoing interventions that differed by community. We also included household-level baseline prevalence and indicators of individual or household level bed net usage - variation in treated bed net usage can predict mosquito biting rates, local intensity of malaria transmission, and/or potential to adhere to other protective behaviours, and may correlate with higher acceptability and adherence to SMC programmes. We examined other characteristics of households by SMC coverage group (Table 1 and STable 4 in Supplementary File) to identify potentially confounding characteristics. All else being equal, the household impact of SMC may be influenced by the number of children relative to older inhabitants and/or total household size. These parameters differed by SMC coverage level in this population. We therefore addressed these key sources of potential confounding by presenting both unadjusted models and models adjusted for the above features with correction of errors to account for clustering at person or household level. We excluded some households from our spatial analysis sample - which detected no geographical clustering of effects of household level models of SMC - however the vast majority of households (90%, 129/144) across the site were successfully enumerated and the missing (15 households) were spread amongst all communities (SFigure 1). Given P-values for Moran tests were close to 1 it is unlikely data from the 15 excluded households would significantly modify these. As a robustness test we defined coverage of SMC in two ways, as the percentage of children receiving any SMC, and the mean number of SMC rounds per child with both definitions showing similar patterns of impact. There may be some misclassification of SMC status as the NMCP did not have a policy of directly observed treatment (DOT) for SMC at the time, and the status for a third of children was based only on caregiver recall. However given the short length of the transmission season and the short length of time between our surveys and the SMC rounds themselves, we expect the levels of misclassification to be relatively low. Our study did not allow us to separate differential herd impacts of SMC coverage by age ( 0–4 years and 5–9 years), nor for a detailed assessment of additional herd impacts of household SMC coverage on eligible children irrespective of their own status. In the case of the second, malaria outcomes were relatively less common in this group and initially significant positive effects did not survive adjustment for key confounders. The possible impacts in this age group therefore remains to be established. Our study was conducted in a moderate/low transmission setting – future studies in higher burden settings will add to this growing evidence base.
In conclusion, we show in the context of a routine extended-age chemoprevention programme with SP-AQ that SMC is associated with a significant reduction in clinical malaria in direct recipients and with a significant reduction in clinical and asymptomatic infection in older household inhabitants. Our data demonstrate these effects are robust to adjustment, are household-specific with little negligible from nearby households. The results support findings of previous laboratory and simulation studies of the importance of children to the infectious reservoir for malaria, addressing a critical evidence gap in demonstrating herd effects in practice. They point to important additional benefits of SMC in reducing overall malaria transmission, highlighting such programmes are likely to be more cost-effective than currently estimated.