Individual-Level Linkage of Primary and Secondary Data from Three Sources for Comprehensive Analyses of Low Birthweight Effects

Abstract

Aim of the study The linkage of primary and secondary data is becoming an increasingly popular approach in healthcare research, but involves some challenges for all involved parties, for example due to data protection requirements. The aim of this article is to systematically outline the methods used and experiences made during a cohort study in the field of pediatric health care research (EcoCare-PIn) that involved access to and linkage of three different data sources. Particular focus is placed on the necessary regulatory measures with regard to data access and data linkage as well as on data validation to ensure a correct linkage. Methods While complying with all relevant data protection requirements, the study realized an individual-level linkage of a) pseudonymized administrative health insurance data from a statutory health insurance on Saxon children born between 2007 and 2013, b) primary data collected via postal questionnaires from parents/caregivers and c) medical data from kindergarten- and school-entry-examinations of Saxon health authorities. The fundamental principle of the concept of data linkage was to strictly separate the sites of data collection and data analysis, which was realized through the involvement of a trust center. Results Challenges especially pertained to the extensive regulatory pre-requirements for data access as well as to data protection requirements while performing the study. Technical aspects and data validation also required a considerable share of attention and resources. A number of validation routines were applied to avoid incorrect data linkage and to ensure the high quality of the final dataset. Data validation included both plausibility checks within the primary data and consistency checks of information given in primary and secondary data. Conclusion The linkage of primary and secondary data on the individual level offers great opportunities for using the strengths of different data sources synergistically and overcoming some of their limitations. Statutory health insurance data and medical data from kindergarten- and school-entry-examinations of Saxon health authorities are examples of already existing data sources that can complement cost-consuming primary data collections by valuable data sets and open up opportunities for longitudinal analysis.

COVID-19: Symptoms Clustering and Severity Classification Using Machine Learning Approach

“…However, this may be a limitation of the secondary data source. It may be more beneficial for reliability to link primary and secondary data [34] with open-source data as the secondary data.…”

Zusammenhang zwischen sozioökonomischem Status und Entwicklungsstand – Datenlinkage von Ergebnissen der Kita- und Schulaufnahmeuntersuchung mit einer sächsischen Geburtskohortenstudie

“…Für die Untersuchung wurden die Daten der Fragebogenerhebung der BMBF-geförderten Geburtskohortenstudie EcoCare-PIn 1 , [8]) mit Daten der Kinder-und Jugendärztlichen Dienste (KJÄD) verknüpft [9]. Aus allen 13 Gesundheitsämtern Sachsens standen die Ergebnisse für die Entwicklungsbereiche Sprache, Feinmotorik und Grobmotorik der KU (für die Geburtsjahrgänge 2007-2009) und der SU (für die Geburtsjahrgänge 2007 bis 2008) pseudonymisiert zur Verfügung.…”

“…Nach Verknüpfung der Daten wurden 615 Kinder in die Auswertung eingeschlossen, für die Daten aus der KU, der SU und der Befragung vorlagen; zum Zustandekommen der Fallzahl vgl. [9].…”