HealthDataNavigator Learn from best practice examples to identify success factors and to improve data availability

Lessons from EuroREACH Diabetes Case Study

Data Comparability 

Performance assessment is based on comparisons, whether over time or between different health systems or health service providers. When using person-level data the comparisons will be meaningful only in case similar study populations are evaluated according to the same methodology and identical indicators are used.

The EuroREACH case study avoided any selection bias of study populations by using national databases, which ensured the representativeness for the three countries involved. However, the national demographic differences had to be taken into account, because the study populations of diabetes patients were considerably older in Estonia and Finland as compared to Israel. This was solved by stratification of analysis by gender and major age groups (0–17, 18–44, 45–64 and over 65 years) whenever relevant.

Whenever measuring health outcome, a number of patient-related factors may have major impact. Administrative health service data does not contain information on patient-related health factors such as lifestyle, but it contains information of concomitant diseases, which can be used for risk-adjustment if considered relevant. Read more

Data Linkage

One of the strengths of administrative person-level data systems is that it includes personal identifiers to allow linkage between data sources or over time. For research purposes any personal identifier should be pseudonymized.

The availability of the unique personal identifier is an unconditional prerequisite to evaluate health system performance with the help of person-level data. Unique personal ID is not used or even legally allowed in some countries, which seriously undermines the opportunities for trans-European performance evaluation.

Within the three countries included in this study, namely Estonia, Finland and Israel, the legal framework in place made it possible to deterministically link individual-level data so that long-term follow-up became feasible.

For this case study no data linkage was required or done with Estonia and Israeli data, because the health service databases of the Estonia Health Insurance and the Maccabi Healthcare Services included all data necessary for this international comparison. Read more

Data Protection and Data Governance 

One of the major lessons learned in the case study was how to handle the issues of privacy and data protection, which are topical questions in health data analysis. We avoided these problems by using anonymous data and by performing the data analysis nationally in-house by the institutions owning the data.

In order to mitigate any legal, security, ownership, governance, confidentiality and privacy concerns we decided neither to create a separate central data warehouse nor to share the original data.

In addition to avoiding the legal and privacy concerns, the approach taken allowed the data holders to retain physical and logical control of their data and eliminated the need to create, maintain, and secure access to central data repositories.

icon plus References