Evaluations cannot support evidence-informed decision making if they do not provide the information needed by decision-makers. In this article, we reflect on our own difficulties evaluating the Geo-Referenced Infrastructure and Demographic Data for Development (GRID3) approach, an intervention that provides high-resolution demographic and geographical information to support health service delivery. GRID3 was implemented in Nigeria's northern states to support polio (2012-2019) and measles immunisation campaigns (2017-2018). Generalising from our experience we argue that Finagle's four laws of information capture a particular set of challenges when evaluating complex interventions: the weak causal claims derived from quasi-experimental studies and secondary analyses of existing data (the information we have is not the data we want); the limited external validity of counterfactual impact evaluations (the information we want is not the information we need); the absence of reliable monitoring data on implementation processes (the information we need is not what we can obtain) and the overly broad scope of evaluations attempting to generate both proof of concept and evidence for upscaling (the information we can obtain costs more than we want to pay). Evaluating complex interventions requires a careful selection of methods, thorough analyses and balanced judgements. Funders, evaluators and implementers share a joint responsibility for their success.
Keywords: Epidemiology; Health systems evaluation; Immunisation; Public Health; Qualitative study.
© Author(s) (or their employer(s)) 2023. Re-use permitted under CC BY. Published by BMJ.