Data Standards in Eye Care
The availability of large-scale digital health data has enabled a wide range of innovations and advancements in artificial intelligence (AI), predictive analytics, and public health surveillance. However, standardized representation of data is needed to fully realize these benefits. Data standards are important for clinical interoperability (e.g., sharing/transmitting information between different clinical information systems) as well as for research applications (e.g., aggregation of data into central repositories or enabling distributed/federated learning).
In this article, we provide a brief overview of some of the applications for data standards and an overview of current efforts related to data standards in ophthalmology by various organizations and workgroups.
Today’s healthcare requires gathering information from numerous systems, including diagnostic devices, laboratories, clinical notes, and pharmacy systems, and communicating with various members of the healthcare team, payors, and patients. A large healthcare system exchanges more data than the entire NASDAQ per day, and deals with data of higher degree of complexity. To make the right information available at the right time for patient care, the information needs to flow efficiently with a high degree of fidelity. Such a flow of information has the potential of making patient care better, safer, and more efficient. Data standards are essential in building the framework for such a data exchange.
Standards allow data from different institutions and different EHRs to be represented in the same standardized way and aggregated/harmonized into a single common format. Subsequent analyses can then be performed on the harmonized data. Having standardized data representation offers several benefits: aggregation into larger samples provides greater statistical power for observational studies and increased capability to study rare diseases that may not be well-represented at any given site, and the ability to facilitate rigor/reproducibility by applying analyses to any dataset that is also mapped to the same data model.
Public health efforts require the collection of geographically disperse health information in order to detect trends and identify challenges in the area of public health. The Center for Disease Control and Prevention initiated the National Health and Nutrition Examination Survey (NHANES) in 1960 to assess the health status of adults and children in the US. In addition to questionnaires, the NHANES program has also measured vision and ophthalmic imaging (visual fields and retinal imaging) over many years. Standardizing the reporting and communication of ophthalmology measurements and imaging results will help similar regional and national health surveillance efforts.
Examples of Standards
Health data standards can be broadly understood as two types: standards that deal with the meaning of concepts, and standards that facilitate the exchange of information. Terminologies code concepts within a domain and describe their relationship with each other. Examples include International Statistical Classification of Diseases and Related Health Problems (ICD) or Current Procedural Terminology (CPT). But heathcare and health research deal with the relationship among the concepts described by these domains. The structure of representing these concepts can be complex and create barriers to data exchange if each system organized them in a different way. Detailed Clinical Data Models describe a system of unambiguously representing the form of the data and linking them to coded elements in the structure. They provide a higher level of standard for data exchange and further facilitate interoperability and aggregation of data. An example is OMOP Common Data Model, initially developed to facilitate analysis of claims-based data from disparate sources.
Data interchange standards facilitate the sending and receiving of health information across different systems. HL7 standards are widely used for communicating electronic health information. FHIR, or Fast Healthcare Interoperability Resource, is a more recent standard developed by HL7 that uses web standards for communication, with concepts that are mapped to existing health standards with the capability for extension. This is particularly useful for communicating health data to and from devices and applications outside the healthcare setting. DICOM, or Digital Imaging and Communications in Medicine, was originally developed by the American College of Radiology and the National Electronic Manufacturers Association to develop standards for exchange of medical images. This standard has been adopted by ophthalmology with specifications made in the form of DICOM supplements to describe specific standards related to ophthalmic imaging and diagnostics.
Examples of standards include:
- International Statistical Classification of Diseases and Related Health Problems (ICD)
- Current Procedural Terminology (CPT)
- Systematized Nomenclature of Medicine (SNOMED)
- Logical Observation Identifiers Names and Codes (LOINC)
- Digital Imaging and Communications in Medicine (DICOM)
- Health Level Seven International (HL-7) / Fast Healthcare Interoperability Resource (FHIR)
- Observational Medical Outcomes Partnership (OMOP)
Several groups in the eye care community are working to advance data standards. Activities include gap analyses to understand where current standards may be lacking in representation of ophthalmic data elements, identifying and evaluating use cases, and working on implementation of standards for data harmonization and/or clinical interoperability efforts.
- American Academy of Ophthalmology Subgroup on Interoperability Standards for Workflow
- SNOMED Eye Care Clinical Reference Group
- OHDSI/OMOP Workgroup in Eye Care and Vision Research
- DICOM Working Group 9 (WG-09): Ophthalmology
- Eyes on FHIR Initiative
- National Eye Institute (NEI) source code repository
Case Examples/Domain Specific Developments
Some of the efforts above are organized into specific domain areas, such as specialty-specific use cases. For example, there are working groups active in the following areas:
- Pediatric Ophthalmology