Epistemic communities in global health and the development of child survival policy: a case study of iCCM

Nearly all African countries have recently implemented some form of integrated community case management of childhood illness (iCCM), a strategy aimed at reducing child mortality by providing curative care for common yet fatal childhood illnesses. This case study describes the evolution of iCCM at the global level using the theory of epistemic communities first outlined by Haas, which explains how international policy coordination on technical issues takes place via transnational expert networks. We draw from in-depth interviews with global policy-makers (n = 25), a document review (n = 72) and co-authorship network analysis of scientific articles on iCCM. We find that members of the iCCM epistemic community were mainly mid-to upper-level technical officers working in the headquarters of large norm-setting bodies, implementing partners, funders and academic/research groups in global health. Already linked by pre-existing relationships, the epistemic community was consolidated as conflicts were overcome through structural changes in the network (including or excluding some members), changes in the state of technology or scientific evidence, shifting funding considerations, and the development of consensus through argument, legitimation and other means. Next, the epistemic community positioned iCCM as a preferred solution via three causal dynamics outlined by Haas: (1) responding to decision-makers' uncertainty about how to reduce child mortality after previous policies proved insufficient, (2) using sophisticated analytic tools to link the problem of child mortality to iCCM as a solution and (3) gaining buy-in from major norm-setting bodies and financial and institutional support from large implementing agencies. Applying the epistemic communities framework to the iCCM case study reveals the strengths and weaknesses of a focused policy enterprise with highly specialized and homogenous disciplinary origins, allowing for efficient sharing of complex, high-level scientific information, but possibly excluding voices with relevant methodological, operational or country-level perspectives.