Advancing the state of the art

Most industries advance over time to use more sophisticated data sets and technologies. This was seen in the payer-provider space over the last decade, as leading firms gradually advanced from claims data to EHR discrete data to complete EHR data. Those firms that were able to graduate to more advanced data sets and technologies were able to improve performance, thrive in value-based healthcare, and build or maintain leadership positions. Others that did not invest in advanced data and technologies have fallen behind.

The pharmaceutical industry is now facing the same progression that the payer-provider market followed over the last decade. Firms have graduated from claims data to EHR discrete data over the recent years. Leading firms are now considering a move toward complete EHR data and advanced technologies to meet emerging needs in regulatory and subgroup analytics.

 

Levels of data completeness

Data source Challenges
Level 1 Claims data Very low accuracy (30-50% recall), missing
interventions and outcomes
Level 2 EHR discrete data (problem list and medication list) Low accuracy (40-70% recall), missing most
interventions and outcomes
Level 3 EHR discrete and narrative data (lists and text) Requires advanced technologies and expertise

Data sets are selected based on the use case. As the use case moves from internal insight to regulatory support and peer-reviewed publications, the data sets used also need to change.

 

Complete EHR data is required for advanced data uses

Data source Claims EHR discrete EHR narrative
Completeness No intervention or outcome  No outcome Intervention and outcome
Accuracy  <50%  50-70% Up to 95%
Validation possible  No  No If not aggregated
Appropriate for
observational trial
 No  No Yes

Few teams have familiarity with comprehensive clinical data sets and the advanced technologies required to process them. Marketing and hype are often ahead of actual experience. If a company will not allow you to use and test their technology, including drill-down capabilities from trial to cohort to patient to data to technology, we recommend caution. Most pilots in real world evidence fail because of an over-promise, under-deliver approach.

We invite you to test our technology hands-on with synthetic data. We think you’ll be impressed with the state of the art.