BackgroundInnovation theory has focused on the adoption of new products or services by individuals and their market-driven diffusion to the population at large. However, major health sector innovations typically emerge from negotiations between diverse stakeholders who compete to impose or at least prioritise their preferred version of that innovation. Thus, while many digital health interventions have succeeded in terms of adoption by a substantial number of providers and patients, they have generally failed to gain the level of acceptance required for their integration into national health systems that would promote sustainability and population-wide application. The area of innovation considered here relates to a growing number of success stories that have created considerable enthusiasm among donors, international agencies, and governments for the potential role of ICTs in transforming weak national health information systems in middle and low income countries. This article uses a case study approach to consider the assumptions, institutional as well as technical, underlying this enthusiasm and explores possible ways in which outcomes might be improved.MethodsLiterature review and case study analysis.ResultsThe two systems considered have had considerable success in terms of gaining and maintaining government support and addressing the concerns of providers without compromising their core elements. In Uganda, the system has flourished in spite of severe resource constraints, using a participatory approach that has encouraged a high level of community engagement. In China, concern with past failures generated the political will to build a high quality surveillance system, using the latest technology and drawing on a highly skilled human resource base.ConclusionsBoth example stress the importance of recognising the political, social and historical context within which information systems have to function. Implementers need to focus as much on the perceptions, attitudes and needs of stakeholders as on the technology. Implementers should distinguish between factors which may influence engagement at an institutional level and those aimed at supporting and supervising individuals within those institutions. Finally, we would suggest that designing interoperability into systems at the outset, rather than assuming that this can be achieved at some point in the future, may prove far easier in the longer term.