Patient Records
Since the early 2000s, UMC has developed and deployed methods for effective exploratory analysis of temporal patterns in longitudinal observational health data to complement the analysis of individual case reports.
Analyses of temporal associations between drug prescriptions and adverse events in electronic patient records provide the basis for many pharmacoepidemiological studies. Such data may also play a role in the identification and preliminary assessment of early signals of suspected harm from medicines. As such, UMC has been active in developing and using methods to explore temporal patterns in patient records.
Read more:
Using data from patient records at UMC
Enhancing signal assessment
Data from patient records could be used to support and enhance the assessment of potential safety concerns, which may help contextualise findings from analyses of VigiBase and other collections of individual case safety reports. Temporal pattern discovery can help identify previously unrecognised confounding resulting from, for example, the underlying disease or treatment indication, and provide a basis for more precise analyses in this respect.
Read more:
Complementing signal detection
It may be possible to detect safety signals directly based on temporal patterns in longitudinal observational data. For this purpose, UMC has developed a self-controlled cohort analysis methodology, which has performed well compared with other methods proposed for the same purpose. We have also explored and developed principles for how clinical and epidemiological review may be performed for identifying potential safety signals in electronic patient records.
Collaborations
Several international collaborations have been important catalysts to UMC’s research in this area. We led the signal detection work-package within the European public-private partnership IMI PROTECT, which among other areas explored the use of electronic patient records for improved signal detection. We also participated as a methods contributor to the Observational Medical Partnership (OMOP), and supported the US FDA’s Sentinel Initiative in a comparative evaluation of different self-controlled methods for signal detection. At present, our research efforts in this area are focused on the Observational Health and Data Sciences Initiative (OHDSI) and the associated IMI European Health Data & Evidence Network (EHDEN), where we lead the pharmacovigilance use case.
