Aktuellt‎ > ‎

Webinar 6 sept 2016

skickad 27 feb. 2018 00:35 av Mikael Hoffmann

"Kombinera data från flera datakällor:  meta-analys jämfört med att poola data"

Combining multiple health care databases from different countries is valuable when analysing rare exposures and outcomes in pharmacoepidemiology. Multi-database studies, however, pose several challenges with regard to data management and analysis.
Experience from the CARING (CAncer Risk and INsulin analoGues) project is presented. In this project, data from national health registers in Denmark, Finland, Norway and Sweden were combined with data from the United Kingdom Clinical Practice Research Datalink.
We describe how the use of a common data model facilitates the use data from different countries with different drug and diagnosis code systems. We also compare different analytic approaches: pooled individual-level data analysis and country-level aggregate meta-analysis, and discuss when these methods are best used.


Senior researcher Morten Andersen, CPE Karolinska Institutet
Post doc researcher Nils Ekström, CPE Karolinska Institutet
Database manager Zoltan Thinsz, CPE Karolinska Institutet