Gary Gaumer, Wu Zeng, Allyala Krishna Nandakumar

BACKGROUND Diverse opinions have emerged about the best way to scale up malaria interventions. Three controversies seem most important: (1) should the scale-up focus on a broader target of febrile illness (including infectious disease and pneumonia)? (2) should the scale-up feature a single intervention or be targeted to the situation? (3) should scale-up have a preference for one kind of delivery mechanism or another? METHODS A decision model of 576 nodes describes the patterns of access, treatment and outcomes of an episode of febrile illness for a child below 5 years. Incremental costs and outcomes relative to baseline (2010) are computed for particular scenarios for Ethiopia using data from the literature. Two perspectives define the relevant costs: society at large and financiers (government and donors) where the costs borne by households are not included.Findings Scaling up malaria interventions by one means or another is a very inexpensive way of saving young lives in poor countries. The low cost per life saved stems from two main reasons: the excessive baseline costs of presumptive use of antimalarial drugs for non-malaria cases, and the excessive costs of delayed treatment of pneumonia. A very limited policy of supplying antibiotics to facilities to eliminate stockouts would save 2100 lives, at a cost of only $615 a life. A much broader programme option, bundling malaria and pneumonia together for patients presenting with febrile illness [including rapid diagnostic test (RDT) for malaria, respiratory rate timers (RRTs) and free antibiotics], would save tens of thousands of young lives at and still cost society less than child fever management in the baseline situation! It is not clear that scale-up via community health workers (CHWs) is to be preferred to a facility-based intervention. The delivery through CHWs allows for a broader coverage of using RDT and RRT, but with limited effectiveness due to limited skills of CHWs in treating and managing patients.