The increasing adoption of Health Information Systems (HIS) does not seem to have resolved the ongoing lack of ubiquitous, dependable and accurate patient information so as to effectively prevent medical errors. Previous work has identified multiple causes, including but not limited to improper or incomplete HIS implementation, incompatibility in healthcare standards, lack of proper data input and validation, and accelerating evolution of technology triggering instability of candidate solutions depending on it. This paper continues the research by describing high-level non-functional requirements that any solution should satisfy relying on current best-practice, and subsequently customizing an established international standard so as to define an evaluation framework that can be used to assess candidate HIS architectures. The ultimate aim of the research is to support selection of stable, sustainable long-term architectural solutions and thus to assist HIS strategic decision making and self-evolution supporting agility.