Multisite evaluation of automated electronic case report form data entry from electronic health records

Multicenter clinical trials often abstract data from the electronic health record (EHR) onto a case report form (CRF) via an electronic data capture (EDC) system. The abstraction process is manual, time-consuming, and error prone. We evaluated scaling automated CRF completion from one institution to other sites in a multicenter trial.

We exported a REDCap project with embedded EHR mapping for a completed platform trial from one institution and delivered it to two other study sites. The receiving sites determined whether additional data elements could be mapped for their institution. We measured the proportion of data entry that could be automated, the extent of agreement between the human- and automation-entered data, and the staff effort required to set up automated CRF completion.

It took approximately 26 and 15 h to set up automation and to map data from the EHR systems at the two receiving institutions, respectively. For 20 total participants at the two receiving institutions, out of 4404 fields with human-entered data, using CDIS could have prevented 764 data entry errors that persisted after monitoring and would have saved 17 total hours or 51 min per participant of manual data entry time.

With these initial estimates of the configuration and data re-mapping time required to scale automated CRF completion and impact on data quality, investigators planning multicenter trials are better positioned to determine when benefits of automation outweigh the expense of manual data abstraction for clinical trials.