Nigeria accounts for 26% of the world’s malaria cases and 31% of malaria deaths [1]. The country is challenged with high levels of poverty, a large rural population with few transportation options, and a generally poorly functioning health system, particularly at the rural primary care level [2].

Uncomplicated malaria may progress to severe disease and death if it is not appropriately treated [3–5]. This process can occur quickly; therefore, early diagnosis and prompt, effective management are essential [6]. Initial malaria treatment is generally provided by primary health centres (PHC) and, in several countries in sub-Saharan Africa and beyond, by community health workers [7]. In Nigeria, community health workers, known as Community Oriented Resource Persons (CORPs), implement the integrated community case management (iCCM) guidelines and primary health centers (PHCs) implement the integrated management of childhood illnesses (IMCI) algorithm. Both, CORPs and PHCs serve as first point-of-care; for severe conditions, both providers are supposed to refer children directly to a hospital. According to guidelines, both of these providers treat children with uncomplicated malaria with oral artemisinin-based combination therapy (ACT), while children with suspected severe malaria are referred to health facilities with inpatient departments for comprehensive clinical management [6]. Referral facilities are few and often more difficult to access, particularly from remote rural areas.

In an observational study, the Community Access to Rectal Artesunate for Malaria (CARAMAL) project investigated the introduction of rectal artesunate (RAS) as pre-referral treatment for children with suspected severe malaria in Nigeria, the Democratic Republic of the Congo, and Uganda [8]. RAS products were recently pre-qualified by WHO [9], several years after a clinical trial found pre-referral RAS to reduce death or permanent disability when administered to children under the age of 6 years with suspected severe malaria who delayed referral to a higher level health facility [10]. In Nigeria, the CARAMAL study did not find a positive health effect of introducing pre-referral RAS in three Local Government Areas (LGA) of Adamawa State. The study also found that the case fatality rate (CFR) among children reporting with signs of severe malaria was higher in children attending a PHC (18.5%) compared to those attending a CORP (5.7%) [9].

Mortality from malaria is affected by several factors, including the Plasmodium species, transmission intensity and vector control measures, age and health status of a person, and the effectiveness of antimalarial treatment [11, 12]. In the context of introducing pre-referral RAS in Nigeria, this study focused on the latter two aspects. The severity of a child’s condition, approximated by the presence of clinical danger signs or caregiver’s perceptions, may determine and explain treatment seeking [13]. Conversely, prompt treatment seeking is also critical to prevent progression to severe disease and death [12]. In Nigeria, among children who sought professional care for fever, only 68% did so within 24 h of illness onset [14] and home treatments for malaria are reported to be poor [15]. Costs and distance to a health facility are barriers to pursuing appropriate treatment [16, 17].

While a prompt visit to a community-based provider, such as a CORP or PHC, can accelerate treatment initiation for common conditions, severely ill patients may need to be referred to a higher level provider for appropriate care. In the CARAMAL project, two critical aspects of a functioning continuum of care for severe malaria cases, namely referral completion and post-referral treatment, were often not implemented according to guidelines, as described elsewhere [18, 19]. However, neither of these system shortfalls seem to provide an explanation for the difference in CFR between the two groups of patients [9].

Understanding reasons for the differences in CFR between children who visited CORPs and those who visited PHCs as community level providers is crucial to strengthen primary healthcare and reduce child deaths. We investigate patients enrolled in the CARAMAL project, focusing on the type and severity of symptoms, home treatment and treatment seeking delay, while taking into account background characteristics of the patients.

Severely sick children need prompt diagnosis and effective clinical management to prevent long-lasting sequelae or death [6]. Diagnostic algorithms such as iCCM and IMCI should allow rapid triaging of severely sick patients close to a patient’s home, and ensure administration of pre-referral treatment and transfer to a higher-level health facility. In Nigeria’s Adamawa State, we found that children with suspected severe malaria who attended a PHC were three times more likely to die than those who visited a CORP (CFR 18.5% vs. 5.7%, [9]), a finding that could not be explained by differences in referral delay or post-referral antimalarial treatment [9]. We therefore investigated differences in background characteristics and treatment history of patients who attended a CORP or a PHC, respectively, to better understand the observed difference in CFR.

There were no significant differences in demographic characteristics of patients attending CORPs or PHCs. However, those attending PHCs lived in more urban areas, which may reflect socio-demographic differences not directly visible in our data. Children at PHCs also appeared to be more severely ill as reflected in the presence of CNS danger signs (90% vs. 74% in those attending CORPs). As the observational study design did not allow for an expert clinical examination, the presence of CNS symptoms was used as a proxy for severe disease, as previously done in similar situations [23–26]. Severity was also reflected in more children attending a PHC having ≥4 clinical danger signs compared to children taken to a CORP. While this analysis considered danger signs on which CORPs were specifically trained, the more thorough clinical training of health workers at PHCs may have biased the recognition of danger signs. Yet, while danger signs indicated that more severely ill children were rather brought to a PHC, this was not reflected in higher caregiver-perceived illness severity, though this study did not explore multidimensional aspects of perceived severity (e.g. emotions, knowledge about severe malaria, economic consequences in seeking or not seeking treatment) [13].

It is plausible that more children with specific signs of severity are rather brought to a PHC than to a CORP as caregivers assume higher professionalism and experience of providers at PHCs. Greater illness severity increased the likelihood of care seeking at health centres compared to community-health workers and self-care in Papua New Guinea [17]. Warsame et al. (2016) previously found that children attending community health workers, in settings comparable to ours, had higher rates of CNS symptoms compared to those attending other health providers such as trained mothers or traditional healers, suggesting higher perceived professional capacity [26]. In Burkina Faso, cases of severe disease often bypass community health workers in favour of a professional health worker [27]. CNS symptoms and difficulties in breathing as signs of severity in febrile children have also been associated with hospital attendance in rural Tanzania [28].

Initial home treatment (usually an antipyretic, rarely an antimalarial) and first attending another provider (mostly a drug shop/chemist/pharmacy) were more common among children brought to a PHC compared to those brought to a CORP. This pattern may reflect a staged escalation in the level of care if symptoms do not improve. As observed elsewhere [29], and also in our study, home treatment and attending medicine retailers did not facilitate access to appropriate medicines but more likely delayed the administration of appropriate treatment by a qualified healthcare provider. As a result, an uncomplicated malaria episode may have progressed to severe malaria, or urgent treatment of an already severe illness episode may have been delayed In Nigeria, lower quality providers are more often sought for children of lower socioeconomic status [30–32] and most drug sellers are not integrated in the formal health system and the continuum of care for severe malaria (for instance, they do not administer pre-referral RAS).

Two-thirds of the children in our study were not brought to the PHC or CORP on the day of onset of symptoms or the day after. Whether delay in attending a healthcare provider has contributed to the overall high CFR in Adamawa is unclear since we lack information on how quickly a particular episode became severe. Yet, it is known that treatment seeking delays increase the risk of progressing from uncomplicated to severe malaria [33] and mortality is highest within 24 h of onset of severe disease [12]. Distance and transport time may negatively affect the condition of a severely sick child especially if transport is strenuous. In Zambia, greater distance from the hospital increased the odds of death in children with severe malaria [34].

We did not find a significant difference in treatment seeking delay between children attending a CORP or PHC, though the retrospective assessment with a 28 days recall period may have limited how precisely time periods were recalled. Thus, some nuances in treatment delay may have remained undetected. During COVID-19 pandemic movement restrictions, delay in seeking treatment increased substantially among those going to a CORP but only marginally among those attending a PHC. CORPs experienced stockouts of commodities and delay in replenishment during the COVID-19 lockdown (own observations), which could have discouraged attendance. Additionally, over the course of the study, behavioural change communication campaigns were conducted alongside RAS roll-out to sensitize communities [35] and may have contributed to caregivers’ awareness about severe malaria and perception about health providers’ capabilities in treating their child’s illness.

Other factors may have influenced the choice of community-based providers without necessarily affecting the CFR. This includes the distance between a patient’s home and the nearest CORPs, PHCs, and referral facilities, as well as the distance of these health providers to one another. Druetz et al. showed that in rural Burkina Faso, consulting a community health worker for malaria was associated with increased distance to the nearest health facility [36]. Provider operating hours also affect health-seeking behaviours. We found that on weekends, patients rather attended CORPs than PHCs that did not provide 24-hour service. A different finding was published by Koce et al. who found that in Nigeria, the limited staff and operating hours of community-based providers (compared to referral facilities) resulted in patients directly seeking care from hospitals, which usually operate 24 h and have medical doctors [37].

Taken together, our findings suggest that, compared to those attending a CORP, children who attended a PHC may have been more severely ill, potentially aggravated by inadequate home management, delay in attending the health facility, and possibly a longer and more strenuous travel. As recognition of malaria and severe disease is often poor [38], caregivers may have underestimated the severity of the child’s condition even in the presence of danger signs. Combined with a low capacity of PHCs to manage severe illness, these factors may have contributed to a higher CFR among children attending a PHC; however, they are unlikely to be the only explanation. Other factors affecting the health outcome of children in this study were investigated in separate analyses. For example, the likelihood of death was higher among children who were treated with pre-referral RAS (AOR: 2.42, 95% CI: 1.25–4.70), whereas a higher proportion of children attending PHCs received RAS compared to those attending CORPs [9]. This finding may be related to incomplete referral and the quality of post-referral care rather than to the administration of RAS itself. For example, only 48% of children completed referral to a Cottage Hospital after administration of RAS by a CORP or PHC, though children attending a PHC were much more likely to complete referral [18]. While hospitalized, injectable antimalarial treatment was common but <4% of children were subsequently treated with the required oral ACT [19]. Several suspected viral outbreaks occurred in Adamawa State during the study period, including Lassa Fever and Yellow Fever [39, 40]. The symptoms of these diseases are often unspecific and overlap with severe malaria, while health providers at all levels in remote locations often lack diagnostic capacity for these infections [41]. Over-diagnosis of severe malaria in children with severe febrile illness is also a well-known issue [42, 43]. While all children included in this analysis had a positive malaria RDT, severe illness and eventually death may have resulted from a missed or inappropriately managed co-infection. Further support for this is that the time between visiting the community-based provider and death was 6.3 days in Nigeria [9], however, the risk of death due to severe malaria is highest early on. Our findings hence indicate that children in Adamawa State with a severe febrile illness indicative of severe malaria face a high risk of dying even if they attend formal primary healthcare services. Several factors along the continuum of care are likely to contribute to the high proportion of deaths in children attending a PHC, including the inadequate quality of treatment received at higher level referral facilities.

In the broader context, these findings underscore the necessity to not only improve access to primary healthcare services and prompt treatment initiation, but to implement health systems strengthening measures to improve referral systems and quality of care at all levels [44]. These considerations, based on findings in this publication and others emerging from the CARAMAL project, were presented at a technical consultation convened by the WHO with the aim to develop operational guidance for an effective implementation of pre-referral RAS in community-based health systems [45].

This study contributes to literature on care seeking for severely ill children, a population which is comparably small and for which published evidence is scarce [13]. While findings from this study may not necessarily be directly applicable to other settings, they point to community level service delivery issues that warrant careful attention when promoting the management of severe childhood illness in African settings.

In conclusion, this research provides context on treatment seeking for children with suspected severe malaria in a northern Nigerian setting, where children visiting a PHC were significantly more likely to die than those attending a CORP. In the context of weak healthcare systems, delay in treatment seeking, higher illness severity in children visiting PHCs, and the inability of PHCs to provide appropriate care, are all likely important drivers of high case fatality. Promoting prompt treatment seeking and strengthening PHCs, referral, and post-referral care are all critical to prevent deaths in children attending community healthcare services.