The purpose of this track is to develop methods to use the HL7 FHIR standard to retrieve relevant data from Real World Data sources – specifically for this track, Electronic Health Record (EHR) systems - and ultimately transform that data into a format suitable for submission to pharmaceutical regulatory agencies. The destination format for this track is the SDTM (Study Data Tabulation Model) standard, created by the Clinical Data Interchange Standards Consortium (CDISC) standards development organization, which optimized for clinical research and regulatory uses and is the data standard used for regulatory submissions of study data to the US Food and Drug Administration.
Background and Scope
Real World Data can be considered data created in the “real world” of everyday experience, such as a routine patient visit to a healthcare provider, as opposed to data created under clearly defined protocols typical of controlled clinical trials. The primary purpose for such data, collected for a purpose other than use in a clinical trial, is in support of clinical care of patients and knowledge for their healthcare providers. However, large amounts of such information could potentially be used for the secondary purpose of supporting clinical research to analyze the data and generate supporting evidence for, as an example, a new indicated use for an already approved pharmaceutical drug or safety-related analyses.
Many sources of RWD exist, but for the current phase of work, the scope of the track is firmly limited to the use of Electronic Health Record (EHR) systems as sources of RWD. Additionally, broad use case is currently limited to the use of EHRs for retrospective analysis of data (to generate evidence for new indications, comparisons, and/or safely) and preparation of such data for submission to governmental regulatory bodies covering pharmaceutical approvals such as the United States Food and Drug Administration (FDA). The use of EHRs as a mode of direct data collections for traditional prospective clinical trials (sometimes called “electronic source data” or “eSource” activities) is not currently in scope. However, we consider it highly likely that types of solutions developed for eSource and for RWD will have significant overlap.
The challenge faced by the track is: How, with the aide of HL7 FHIR, can we most efficiently and comprehensively migrate data and bridge the many syntactic and semantic gaps from the healthcare data sphere to the research and regulatory sphere?
January 2022 starts the second year of this connectathon track. In the first year of this work, to better understand the end-to-end challenges faced when moving information from a healthcare setting to a research and regulatory setting, we focused on the specific data domain of concomitant medications (other medications taken by people at the same time as a medication of interest). Specifically, we focused on what it takes to find, retrieve, and transform, relevant medication information for patients from an EHR data source to an acceptable SDTM format while not losing critical supplemental information in the process.
Using available test data set of fictional patients, we initially retrieved medications information from the following resources: MedicationStatment; MedicationRequest. Subsequently, we expanded to allow retrieval from additional resources, MedicationDispense and MedicationAdministration, since we found that different EHRs implement and use the suite of these 4 FHIR resources (referred to as Medication[x]) in highly variable ways. We explored the challenges of differing levels of “certainty” in EHRs regarding the actual ingestion of a drug, how best to package results of data queries in FHIR for transport and transformation, and the use of SDTM SUPPxx supplemental domains to represent additional supplemental when such information semantically different from that typically stored in the core domains of SDTM.
Finally we identified key attributes of the Medication[x] Resources that we felt to be critical for implementation in EHRs if such data is to support clinical research and submissions. Some of these findings resulted in new proposal submissions to the US Core Data for Interoperability (USCDI) requirements created by the US Office of the National Coordinator for Health IT (ONC). Specific attributes such as Status and Patient for MedicationRequest and MedicationAdministration are now in queue for potential inclusion in USCDI version 3.
Plan for January 2022 Connectathon Track
In the new year, we aim to move beyond the focus on a single data domain and, instead, take a practical and realistic, query-based approach. Specifically, we intend to begin with a data query that can ask a (very small) question, based on some existing study or protocol, which calls on very specific information from a range of different data domains. This will allow us to increase out breadth of understanding of the challenges and solution. Another Vulcan track, Schedule of Activities, is working with a public domain study protocol H2Q-MC ZZT(c), https://wiki.ihe.net/images/4/47/Lzzt_protocol_redacted.pdf, which will serve as the protocol and will also create a cross-cutting synergy between the two tracks. Additionally we will be exploring revising the technical approach by working with a software developer who has worked on this set of issues (see his background and coding approach here: https://sourceforge.net/projects/fhirloinc2sdtm/ ).
Test the design of a Resource/set of Resources
Submitting Work Group/Project/Accelerator/Affiliate/Implementer Group
Scott Gordon, Lauren McCabe
Track Lead Email(s)
FHIR R4 is current primary focus, also US Core. However, since many other implementations exist, we are not limiting this exclusively. Further, as R5 moves forward in balloting, we will keep an eye on relevant changes.
Specification(s) this track uses
Artifacts of focus
Expanding to multiple resources. Medication[x]; Probably Observations, Patient, etc.
Vulcan members, Regulators, EHR Vendors, EDC Vendors, Academic Medical Centers, Pharma Companies. Additionally:
Artifacts from Previous Tracks
Draft of decision algorithms based of Medication[x] Status field
Draft identifying all elements of all Medication[x] resources which we recommend be implemented and utilized consistently (where relevant) in all EHRs in order to make concomitant medication determination more reliable. (There are absolutely strong pure healthcare justifications as well for these elements being present).
Useful artifacts for January 2022 Track
SDTM Data Set in CSV format:
SPECIAL NOTE: Please be aware that the CSV files, exported from the .xml file tables, has all dates/times represented in the required ISO 8601 date format (ie, 2022-01-25). We noted that when loading the CSV files into Microsoft Excel, Excel automatically converts some dates to local format and not others. Attempts to manually convert columns to YYYY-MM-DD resulted in dates that were just a year (ie, "2013") to be converted into very incorrect dates. We didn't have sufficient time to figure the magic words to make Excel do what we wanted .
The data in the actual CSV files is in the correct format.
OUTPUT ARTIFACTS from January 11-12, 2022
Evaluation of minimal sufficient FHIR elements needed to allow LZZT study from EHR data (potential basis for Profiling). This sheet has multiple uses, for now if you filter on the "In LZZT FHIR bundle" column for "Y", you will see just the elements present in the Bundle.
LZZT_FHIR_Elements was created by
SDTM Review Comments
Post-Connectathon developed mapping overview table Jozef Aerts,