The Unforeseen Medical and Business Challenges Found When Integrating Abstracted Data from Disparate EHR Systems While Building the Curecloud Direct-to-Patient Multiple Myeloma Registry

Introduction

The CureCloud™ is being created by the Multiple Myeloma Research Foundation (MMRF) as a follow-up project to the CoMMpass Trial. The CureCloud direct-to-patient (CC-DTP) registry's goal is to accumulate over 5,000 well-characterized myeloma patients with linked data from EHR, myeloma-specific 77-gene NGS panels, exome genomic panels, and patient reported data. COTA, Inc. and MMRF have formed a partnership to populate the CC-DTP with real-world data (RWD). COTA manages the record retrieval and abstraction of data from various contributing provider institutions.

Methods

After obtaining IRB approval, electronically-signed consent forms were obtained from patients during the online registration process for the CC-DTP registry. During an initial 24 patient pilot, COTA presented the consent forms to 54 provider organizations to obtain medical records for abstraction. A request package containing the signed consent, IRB certificate of action, and overview of the CC-DTP registry was prepared and submitted to each of the institutions using fax or direct mail. Metrics tracked during the record retrieval process included time to delivery and fees for record release. Additionally, all returned data was screened for the presence of structured, coded medical information prior to a manual abstraction process into the CC-DTP data dictionary format.

Results

Of the 54 institutions contacted in the pilot, 80% of record returns were in under 30 days, but 20% took over 40 days, and half of those (or a total of 10%) took over 60 days. The median retrieval time across all sites was 25 days (range of 5 to 87). In two instances, a processing fee request in excess of $300 was initially tied to the release of the medical record, and human intervention was required to have the fees waived. MMRF and COTA recorded that six of the 54 (11%) institutions that received a medical record request rejected the use of electronically-signed patient consent forms and five of the six institutions required their own authorization to be used in place of the IRB-approved patient consent. At 10 of the 54 institutions, the information requests were rejected in favor of hand-signed documents before patient records were provided. Additionally, for all retrieved records, the only coded information included were ICD-10 codes for the primary diagnosis. All other ICD10, LOINC, RxNORM, SNOMED or any other coded vocabulary were absent from the records. In no cases were any adverse events coded with MedDRA.

Discussion

Despite a significant investment in electronic health solutions in the United States, many of the institutions approached as part of the CC-DTP pilot were unable to provide all of the needed data in a timely, cost-free, and coded manner. The four major challenge areas were: 1) extensive administrative delays (slow processing times), 2) rejection of electronic patient signatures on consent forms, 3) unexpected high fees to obtain EHR record abstracts, and 4) virtually all structured, coded medical information was stripped from provided abstracted EHR records. These challenges not only delayed the gathering of clinical data but forced a series of complex and costly work-arounds. The goal of building a fully linked registry complete with coded EHR data remains challenging. Future studies to explore obstacles of record retrieval and limited codified data should be conducted. A comprehensive, research-focused data abstraction process for coded data exchange is being put into place and will be needed to support future RWD projects of this nature.