Malaria accounts for an estimated 16% of child mortalities in sub-Saharan Africa [1]. Inequities in access to treatment can lead to increased morbidity and mortality rates in populations distant from health centres due to further delays in treatment [2, 3]. In Zambia, treatment-seeking behaviour is perhaps the largest barrier to ensuring prompt access to anti-malarials [4]. Expanding malaria treatment services through community case management (CCM) of malaria has shown to be an effective method of improving individual access to malaria treatment [5], but information regarding the impact of CCM on the health system remains understudied.

Long wait times and staff shortages are associated with poor health systems throughout sub-Saharan Africa [6–8], a problem to which malaria contributes directly with its estimated 165 million incident cases in 2013. There are simply too many malaria cases for the number of health workers on the continent [9]. The HIV/AIDS pandemic, outmigration of trained health staff, and a lack of funding to health systems contribute to the problem of an inadequate health system, which in turn leads to challenges in reducing mortality caused by preventable and treatable infectious diseases [10]. This shortage of health workers is felt in Zambia [11], and most prominently in rural areas [12].

Malaria and other illnesses such as pneumonia and diarrheal disease can be safely addressed by community health workers (CHW) with 1–2 week trainings to administer simple diagnostics and medicines [13]. In Zambia, CCM of malaria through CHWs has been implemented in various settings following a policy shift to allow CHWs to conduct malaria rapid diagnostic tests and treat those testing positive with artemether-lumefantrine [14, 15]. Expanding on this policy in Southern, Central and Western Provinces of Zambia, a reactive case detection (RCD) system has been implemented wherein CHWs perform case investigations for incident malaria cases suspected of being locally acquired [16]. RCD is based on the premise that incident malaria may be representative of local malaria transmission, likely in the immediate vicinity of the incident case’s household [17]. While CCM refers to CHW efforts to test and treat symptomatic community members who seek care for their illness, RCD is an additional service whereby CHWs follow-up incident cases to their household, and then test and treat around the incident case. CCM with RCD is increasingly viewed as a key element of the Zambian Government’s national elimination efforts especially in areas in the final hurdle of identifying residual foci and stomping out local remaining infections (Malaria elimination strategy—Zambia 2016). This article assesses the impact of CCM and RCD on the malaria burden at the health system level.

This is the first assessment of which we are aware examining how CHWs conducting CCM and RCD may be affecting the broader health system. These data suggest that CHWs performing malaria CCM and RCD not only expand access to malaria treatment but also shift a significant portion of the malaria case management case load to CHWs. Months where CHWs reported conducting CCM and RCD were associated with a 50% increase in confirmed malaria infections after accounting for temporal and seasonal trends. This increase likely reflects an improvement in the capacity of malaria surveillance rather than an actual rise in malaria transmission as test positivity rates did not appear to increase during the time period included in this analysis (Fig. 2b). Further supporting this argument is the fact that malaria incidence measured at the health centre is only a fraction of the malaria burden [24]. By equipping CHWs to conduct CCM and RCD, the surveillance ‘net’ is wider, more granular and more sensitive and is thus able to detect more malaria infections than a surveillance system consisting of health facilities only. Improving the quality and capacity of a surveillance system in the context of malaria elimination is especially important; maximizing the number of malaria infections detected and treated will maximize the probability of eliminating existing parasite reservoirs, identifying importation and stopping onward transmission [25]. Similarly an improved surveillance system is crucial for documenting the maintenance of zero transmission and ultimately qualifying for WHO elimination certification. CCM alone did not provide as much benefit as RCD alone, and malaria elimination programmes should examine the feasibility of conducting RCD to improve surveillance.

Interestingly, the gains in malaria surveillance were made only when taking into account the malaria infections found during CCM and RCD. This strongly suggests that the CCM with RCD system that was implemented in Southern Province provides a tangible benefit in pursuing malaria elimination. Contrary to these findings other studies have found little benefit to RCD [26]. One of the main criticisms of RCD is that the RDTs currently used have difficulty detecting subclinical infections. Certainly RCD and malaria surveillance in general would benefit from improved diagnostics to identify asymptomatic infections, however these analyses suggests CCM and RCD provides a benefit with the current tools available.

Not only have community efforts improved access to malaria treatment, disaggregating malaria surveillance at the health post level improves the understanding of heterogeneity of malaria transmission within health centre catchment areas. The spatial refinement of surveillance is an important precursor to targeting malaria interventions such as indoor residual spraying, larviciding and/or insecticide-treated mosquito net distributions to geographical areas within health centre catchments.

Months when CHWs reported being in the community doing CCM and RCD were associated with a 6% decrease in the total outpatient attendance at the health facility. There is reason to believe that the decrease in outpatient attendance is likely due to fewer individuals with fever seeking care at health facilities. With a mean monthly outpatient attendance of 823 (95% CI: 798–848), a 6% decrease translates to roughly 49 fewer outpatients received at each health centre per month. Assuming each outpatient takes 10 min of time from healthcare providers at the health centre, those 49 fewer outpatients per month leads to approximately 8 person-hours of time saved per health centre per month when CHWs are conducting malaria community case management. The decrease in outpatient attendance and person-hours saved at the health centre did not account for increased time spent by CHWs conducting RCD. Volunteer CHWs could potentially become overburdened by RCD, particularly in areas where malaria transmission is more intense. When people being tested for malaria during CCM and RCD were included in the measure of outpatient attendance, there was only an association for CCM suggesting that the lower attendance may be due to individuals seeking and receiving care in the community rather than going to the health facility.

These analyses have some limitations. Firstly, it should be noted that the use of IRR in the analysis is due to the negative binomial regression model, but the incidence also includes prevalence cases found through RCD. Second, routinely reported malaria case data often has inherent problems; reporting completeness and treatment-seeking behaviour are often noted as most problematic [23]. To mitigate the issue of reporting completeness, we measured the level of reporting completeness and conducted analyses on time periods with at least 60% reporting completeness. To account for the issue of treatment-seeking behaviour at health facilities, confirmed malaria infections standardized to total outpatient attendance. Including a measure of time also helped to account for variation in treatment-seeking behaviour due to seasonality. A second limitation of this analysis was that not all suspected malaria cases are confirmed at health centres using either a malaria rapid diagnostic test or microscopy. This issue was addressed by including malaria testing rate as a covariate in the analyses of confirmed malaria infections to account for those facilities that would appear to have higher confirmed malaria infections, but in reality just have higher testing rates. Finally, the interventions were scaled gradually, limiting the ability to do a more robust analysis that might incorporate a time by intervention interaction. The associations seen here may perhaps be caused by temporal drift including varying coverage and/or effectiveness of vector control interventions such as ITNs, though measures of time were incorporated through the use of a sinusoidal function and categorical year. Despite these limitations, these findings strongly support the case for implementing community-led malaria surveillance and interventions.

Community implemented CCM and RCD are associated with decreased outpatient attendance at health centres and increased capacity of malaria surveillance systems, respectively. Scaling these programmes will likely lead to tangible benefits for both targeting malaria specific interventions and resources to areas with higher case loads and expanding the capacity of a currently overburdened health system. Augmenting the CHW skillset with the ability to treat diarrhea and pneumonia would likely lead to even larger gains in terms of expanding the access to care, easing the burden on health facilities, and ensuring continued use of community-based care in a malaria elimination environment. These gains need to be weighed against the additional workload placed on volunteer CHWs.