An Phung, a dedicated medical doctor from Vietnam, moved to the Netherlands to improve her research skills through a Research Master’s degree in Clinical Psychosocial Epidemiology following the track Health Systems and Prevention. Seeking to combine her clinical experience with research skills, she started a project with Datapoort to link and analyse data from Academisch Huisarts Ontwikkel Netwerk (AHON), HartNet Noord-Nederland, and Centraal Bureau voor de Statistiek (CBS).
My research evaluates the implementation of a Cardiovascular Risk Management (CVRM) program among patients who have undergone coronary intervention. By linking different data sources, I studied the frequency of measuring cardiovascular risks, changes in these risks, and medication prescription before and after the intervention
An Phung, MD
Master student Clinical Psychosocial Epidemiology
Bridging clinical practice and research
An’s inspiration for her research stems from her clinical experience in Vietnam, where she treated patients with chronic diseases. The medical doctor explains: “I work with many patients who have chronic diseases, and I manage their diet, alcohol consumption, and smoking habits. Besides my clinical experience, I decided to pursue a master’s degree in the Netherlands so I could improve my research skills.”
Reducing cardiovascular risk factors by linking data
Coronary artery disease (CAD) is the primary cause of mortality and economic burden in the Netherlands. Managing modifiable cardiovascular risk factors is key to reducing this burden. An: “My research evaluates the implementation of a Cardiovascular Risk Management (CVRM) program among patients who have undergone coronary intervention. By linking different data sources, I studied the frequency of measuring cardiovascular risks, changes in these risks, and medication prescription before and after the intervention.”
Connecting data
Datapoort played an important role in An’s research, enabling her to link and analyse data from various sources. She used three main data sources: patient data on coronary interventions from HartNet, cardiovascular risk management program data from AHON, and medication and mortality data from CBS. “Datapoort allowed me to seamlessly integrate and analyse patient data on coronary interventions from HartNet, cardiovascular risk management program data from AHON, and medication and mortality data from CBS. By joining information from these three different data sources that relate to the same individual, I was able to create a time sequence illustrating how patients were managed in CVRM programs before their coronary intervention, treated with coronary intervention, and continued their CVRM program after coronary intervention” the researcher explains.
Challenges
Due to legislation and regulations, one significant challenge was the inability to use the BSN number (personal identification number) for data linking on the individual level. Instead, An had to rely on personal characteristics such as date of birth, gender, and address at that time to pseudonymize the data. “This method introduced uncertainty, as some individuals could not be uniquely identified based on this information and therefore not linked to the other data sources, leading to data loss. For HartNet, this meant about 15% of the data could not be pseudonymized and linked. Improving this process would be good for future analyses,” the researcher explains. While the linking rate was higher, AHON has already expressed its willingness to explore better linking options, as they also aim for the highest possible data linking accuracy.
Another challenge was cleaning the data from AHON and HartNet. The data came in different formats, including open text fields, which needed organization before it could be uploaded into the secure research environment of CBS, which requires structured datasets with clear descriptions of variables and labels for values. An used the secured myDRE workspace for data cleaning from which the Data Manager could securely upload the data to the secured Datapoort-CBS microdata environment for linking and analyzing the data. An: “Especially the data sets from AHON required a lot of effort to prepare them for analysis. The lack of a clear codebook, a document that provides detailed information about the variables, resulted in spending a lot of time talking with AHON to clarify data elements. We received valuable assistance from the AHON Data managers and committee, who have extensive knowledge about the unique data.” HartNet documented their data in codebooks and has procedures implemented to provide a structured and validated dataset together with National Heart Registration (NHR, in Dutch: Nederlandse Hart Registratie).
One codebook for all Datapoort partners
To enhance support for future research projects, Datapoort could develop a unified codebook. An: “This would ensure that essential information is consistently captured, providing a clear structure and improving the quality of future research projects. It will also save a lot of time.” Additionally, providing data cleaning syntax would facilitate easier (re)use of variables in subsequent research, aligning with the FAIR principles of data management.
Over the past six months, AHON has developed a codebook and manual for current research applications, including a codebook and R-packages for cleaning steps. This manual provides valuable resources to streamline the research process.
Datapoort-CBS infrastructure
The procedures and requirements for starting a project to link medical records with CBS microdata will be documented, allowing other researchers to make use of the Datapoort-CBS infrastructure for their research questions.
Are you a researcher interested in using the Datapoort-CBS infrastructure for your research? Contact us to explore how we can assist you with your research questions.
Curious to read more about the research? Contact us