Opportunities and challenges of real-world data in heart failure

The rapid increase of real-world data (RWD) has led to a great potential benefit to contribute to improvements in healthcare. However, the sheer increase of healthcare data collected cannot automatically be translated to clinical practice. The aim of this thesis was to assess the potential of RWD in heart failure by investigating the opportunities RWD provides, but also what the challenges are within RWD.

Opportunities that were identified were linkage of electronic health records (EHRs), the ability to study patients in a real-world setting and techniques to analyse large quantities of data as opportunities in heart failure. It was possible to identify differences in risk factors for heart failure between men and women, as well as expand the knowledge on healthy lifestyle factors in the general population. Trends for heart failure medication over 15 years of follow-up were described, indicating room for improvement. Furthermore, it was shown that beta-blockers are associated with improved mortality in patients with heart failure with reduced ejection fraction who were aged 80 years or older.

Challenges that were observed were quality of routine healthcare data, a lack of consensus on phenotyping diseases in RWD, harmonising and standardising data and incomplete data collection. Yet, it this thesis it was shown that these issues could be tackled and data could be harmonised across different European countries. Furthermore, an algorithm was created to predict subphenotypes of heart failure, a feature often missing in EHRs, and validated these results in an independent cohort. Last, using a machine learning model distinct heart failure with preserved ejection fraction clusters were identified, one of the heart failure subphenotypes with heterogeneity complicating evidence based medicine. The clusters that were found could form a basis for tailoring trial design to individualised drug therapy.

This thesis shows that real-world data have significant potential, but still need enhancement of its quality, more efficient processes for their collation and increased interoperability.