CovidPredict

Project summary

During the beginning of the COVID-19 pandemic in (March 2020) in the Netherlands, there was great uncertainty what was going to happen. Treatments were lacking and the clinical course of COVID-19 was unknown. For this reason, physicians and researchers from Amsterdam UMC and Maastricht UMC+ started a consortium to start the collection of data of COVID-19 patients. This initiative, coined CovidPredict, was in collaboration with the Dutch Association for Intensive Care and the National Intensive Care Evaluation Foundation (NICE) and shorty after its start, many other hospitals joined the initiative. At first, this initiative was two-fold: an automated ICU-patient data collection (COVID-19 ICU Decision Support), and a chart-reviewed data collection for both ward- and ICU-admitted patients (COVID-19 Clinical Course).  

Currently, the CovidPredict consortium is continuing the Clinical Course data collection. Data of patients with COVID-19 at any hospital department (excluding paediatrics) is collected through an adapted form of the World Health Organization ISARIC data collection form. Where possible, automated data extractions from electronic patiënt charts are used. 

Social impact

By making the COVID-predict data FAIR it is now possible to access and use the largest, domain transdescending, Dutch COVID-19 database which contains detailed, clinical information. This not only offers scientific opportunities but is also important to identify risk groups and evaluate the effect of corona policies. We can now answer questions such as "Which patients are extra vulnerable in the next wave?" "What is the effect of vaccinations on the course?" and "What is the optimal treatment for COVID-19"?  

By using the FAIR COVID-predict data, the future corona policies can be based on current data 

FAIR objectives

The main FAIR objective is creating a dataset of COVID patient-characteristics on a national scale, based on the WHO-CRF and ready for reuse on request. 

The main obstacles we encountered were: 

• the determination of the differences between the data collections (of UMCs) in:
  • variable definition
  • variable naming
  • answer options (e.g. different codings)
• writing the scripts needed for mapping these differences to the WHO standard 
• writing the scripts needed for uploading the institutional data to the national portal 
• the development of a standardized data sharing agreement

Read more

CovidPredict (official website)

FAIR elements involved

Findable

Accessible

Interoperable

Reusable

Key facts

• largest high granular covid database of admitted patients in the Netherlands 
• ongoing inclusions and scientific projects
• ongoing scientific output
• societal impact in many national newspapers

Data type

• Clinical data

Organizations

• Amsterdam UMC