Abstract
The integration of different administrative data sources from a number of European countries has been shown useful in the assessment of unwarranted variations in health care performance. This essay describes the procedures used to set up a data infrastructure (e.g., data access and exchange, definition of the minimum common wealth of data required, and the development of the relational logic data model) and, the methods to produce trustworthy healthcare performance measurements (e.g., ontologies standardisation and quality assurance analysis). The paper ends providing some hints on how to use these lessons in an eventual European infrastructure on public health research and monitoring. Although the relational data infrastructure developed has been proven accurate, effective to compare health system performance across different countries, and efficient enough to deal with hundred of millions of episodes, the logic data model might not be responsive if the European infrastructure aims at including electronic health records and carrying out multi-cohort multi-intervention comparative effectiveness research. The deployment of a distributed infrastructure based on semantic interoperability, where individual data remain in-country and open-access scripts for data management and analysis travel around the hubs composing the infrastructure, might be a sensible way forward.