Artemisinin combination therapies (ACT) are highly effective for preventing an uncomplicated P. falciparum malaria fever from progressing to severe disease, which can result in death [3,4]. Nearly all endemic countries have now adopted ACT as the first-line treatment for P. falciparum malaria [5]. Among children with severe malaria, effective case management with parenteral or rectal artemisinin or quinine is also recognized to be highly effective at preventing death [6-8]. More recently, a large scale trial of intravenous artemisinin has been shown to reduce mortality compared to quinine in African children with severe malaria [9]. However, the impact of prompt and effective treatment of uncomplicated malaria and case management of severe malaria on reducing child malaria deaths has not been quantified with placebo-randomized controlled trials for obvious ethical reasons.

In the absence of sufficient systems to monitor real-time trends in cause-specific child mortality in many developing countries, mathematical models are increasingly being used to estimate the impact of the scale-up of child survival interventions on child mortality. The Lives Saved Tool (LiST) is one such model developed by WHO and UNICEF’s Child Health Epidemiology Reference Group (CHERG) and based on the earlier work on effectiveness of interventions [10]. The model estimates child deaths prevented (within specific cause of death categories) due to intervention scale-up within a specified country as a function of three primary parameters: 1) the number of child deaths by cause projected to occur in each year (including population growth parameters over time); 2) the protective efficacy (PE) on cause-specific mortality (PE = 1 - relative risk (RR) * 100) for each intervention being scaled-up; and 3) increases in population coverage of each intervention [11].

To quantify the PE of case management of uncomplicated and severe malaria for preventing malaria deaths in children, we conducted a systematic literature review to identify randomized controlled trials using PubMed. Studies were limited to those implemented in areas of endemic P. falciparum malaria transmission. Where sufficient data from individual or community-randomized controlled trials could not be identified, we considered all available evidence, including observational and Delphi studies, in quantifying the PE of interventions reviewed here. Where possible, study quality was assessed according to the GRADE technique [12], adapted by the Child Health Epidemiology Reference Group as outlined elsewhere [11]. We assessed the quality of the data sources contributing to each CFR estimate as high, moderate, or low (Table ​(Table11).

A malaria death is defined as one death per parasitologically diagnosed clinical episode. It is recognized that non-lethal episodes of malaria frequently result in increased vulnerability to other infectious diseases and subsequent death due to another cause. Some have suggested that the likelihood of dying in the 8 weeks following discharge from a malaria diagnosis is equal to the likelihood of dying in the hospital [13]. However, lack of sufficient data precluded the inclusion of such deaths due indirectly to a P. falciparum infection.

Given the lack of placebo-controlled trials of treatment for malaria due to ethical reasons, we were unable to directly determine PE for prompt effective treatment of uncomplicated malaria for preventing malaria child mortality, as compared to no treatment. A meta-analysis was used to quantify the CFR for children with uncomplicated malaria that received prompt effective treatment. Results showed that death following uncomplicated malaria treated promptly with an effective antimalarial is very rare (0.068%, 95% CI 0.024 – 0.112). Because of a paucity of available data, we based the estimate of the natural mortality history of uncomplicated P. falciparum malaria on a previously published Delphi method, which found that experts estimated the risk of death among untreated children with uncomplicated P. falciparum malaria to be 5% for <2 year olds and 2% for 2-5 year olds. Combining these estimates, we estimated a very high PE of prompt ACT treatment of uncomplicated malaria compared to no treatment [99% (94-100%) for children younger than 2 years old, and 97% (86-99%) for children 2-5 years old].

For programmatic utility, we estimated a PE of effective case management of all children hospitalized with malaria. The CFR of 3.4% (95% CI 1.6-5.2) for all hospitalized children estimated by this method is very similar to that found by other authors [71,81,82], and the CFR of 13.6% for strictly defined malaria is slightly higher than the AQUAMAT trial [9], not surprising for observational studies compared to clinical trial conditions. As a proxy for mortality of untreated hospitalized malaria (natural history), we used mortality data from observational studies in the setting of chloroquine failure, during the era when chloroquine resistance was being established and chloroquine was still in widespread use. This resulted in the estimated CFR of untreated hospitalized malaria ranging from 13-21%. The higher bound (21%) is problematic, as the majority of the children in this sample who died had received chloroquine, and died despite respite receiving intravenous quinine after being admitted in critical condition, thus it may be a major underestimation. A qualitative study in Mali found that mortality for severe malaria (by maternal definition during interview) was 17.0% regardless of treatment-seeking [83], which falls in the middle of this range and lends some credence to its accuracy. Thus we believe that a PE of 82% for protective efficacy of effective case management with intravenous quinine is reasonable. Given that these studies all used intravenous quinine, the PE of intravenous artesunate for preventing mortality would be expected to be even higher.

There are numerous limitations to using the LiST model to predict the number of child deaths that could be prevented by scaling-up case management of uncomplicated and severe malaria. First, no placebo-controlled data exists and the historical literature is of insufficient quality to establish a CFR for children who go untreated for uncomplicated and severe P. falciparum malaria (natural history). For uncomplicated P. falciparum malaria, we relied on expert opinion from a Delphi survey to obtain an estimate of the probability of a child dying following uncomplicated P. falciparum malaria when left untreated, within the context of stable transmission. Such an estimate from a Delphi study is clearly subject to bias, although the direction and magnitude are unknown. For estimating the probability of a child dying from severe P. falciparum malaria when left untreated, we relied on a proxy estimated as the probability of a child dying from severe malaria after treatment with chloroquine, in the context of high chloroquine resistance, which ranged from 13.1-21.1%. While this was a helpful proxy, it likely underestimates the true untreated CFR and therefore underestimates the true PE of case management of children hospitalized with P. falciparum malaria. Second, while the majority of children with uncomplicated malaria are still not treated with an ACT, and many hospitalized children receive inadequate care, often with sub-optimal medication as well as ancillary treatment, many do receive some form of treatment. Thus hospitalized children who had received chloroquine likely had somewhat better outcomes than children who received no treatment. Third, our approach takes into account only a given episode of malaria, and does not take into account longitudinal impact of adequate or inadequate treatment, or the post-prophylactic effect of some antimalarials. If more reliable estimates of untreated mortality of uncomplicated and severe disease become available, it will be important to update the estimates. Finally, while it would be optimal to generate these estimates for children with stringently defined severe malaria, as CFR for severe malaria after treatment with intravenous artesunate has been well documented, no proxy for untreated mortality in severe disease currently exists. If these data become available, the protective efficacy of case management with intravenous artesunate in this subset of children should be estimated.

Case management coverage indicators in most sub-Saharan African countries lag far behind coverage indicators for malaria prevention, in part because the delivery of effective case management of both uncomplicated and severe disease is quite complex and involves not only procurement of the drugs, but supply chain management, health worker training and supervision, revision of payment schemes to remove financial barriers, extending this care to more peripheral levels of the health care system, and working to encourage use of these services by the population.

The currently recommended coverage indicator for prompt effective treatment is the proportion of children <5 years old with fever in the past two weeks who received an ACT within 24 hours from the onset of fever [84]. While an attempt has been made to standardize this indicator to be measured by most nationally-representative household surveys (e.g. Demographic and Health Survey, UNICEF’s Multiple Indicator Cluster Survey and Malaria Indicator Surveys), it is problematic for measuring trends in coverage over time for several reasons. First, as scale-up of prevention interventions such as ITNs increases, the proportion of fevers due to malaria is likely to decrease. Second, most countries have now adopted the policy of laboratory diagnosis [5], with either microscopy or rapid diagnostic tests, for all suspected malaria cases seeking treatment through the health system. For this reason when assessing trends in coverage for prompt effective treatment of malaria fevers, the proportion receiving treatment has actually gone down in most countries since 2005 as countries have switched from clinical diagnosis to laboratory diagnosis. Finally, most countries switched their first-line antimalarial from a monotherapy to an often less available ACT within the last 5-8 years [5], decreasing treatment with a recommended first line antimalarial in some countries over the past decade.