Introduction

Over the last few years, healthcare has been undergoing dramatic transformation due to the rapid growth of three important innovations: electronic health records (EHRs), genomics and big data. The Affordable Care Act has substantially increased the volume of patients demanding care and is transforming the way healthcare organisations must provide that care. In particular, healthcare providers will be held more accountable for their ability to meet important quality measures such as adequate control of chronic disease, broad preventive healthcare and significant reductions in hospitalisations.1 Similarly, the Health Information Technology for Economic and Clinical Health Act has led to a dramatic increase in the adoption of EHR technology throughout the country.2 The development of mandated meaningful use requirements and interoperability standards has propelled the growth and availability of EHRs and other forms of health information technology (IT).1 This influx of new regulations and digital records will encourage large amounts of clinical and quality data to be managed, shared and applied across diverse organisations and institutions.3

At the same time that EHRs are providing increased access to clinical information, the sequencing of the human genome over a decade ago has catalysed new discoveries explaining the genetic contributions of a patient’s susceptibility to disease.4 Precision medicine is broadly defined as the application of patient-specific health and genomic information for highly targeted and effective methods of clinical diagnosis, management and treatment.5^,6 This genomic data can come from many sources, from single and multigene tests to the sequencing of exomes and entire genomes, all of which provide valuable insight into the human clinical condition.4 Precision medicine has the potential to leverage health IT in ways that could dramatically improve public and population health, bringing practical genomic information exchange into sharp focus.7

A major challenge for healthcare providers is, however, the rapid and dramatic increase in the volume and complexity of data that must now be collected, organised and evaluated.2 For EHRs, the large amount of clinical information includes medical history, laboratory results, imaging files and other data collected during the patient appointment.2 For genomics, this involves descriptive metadata along with the sequenced genome of a patient, which (at roughly six billion base pairs in size) will require significant investments in storage, analysis and dynamic interpretation.4 These challenges will be compounded exponentially when scaled to the patient population affected by EHR adoption and genomics. It is widely predicted that current EHRs alone will not be capable of handling the volume and complexity of genomic information central to the practice of precision medicine.8^–10 Indeed, no universally accepted approach exists for describing clinically relevant genomic findings, and a very few EHRs today even attempt to report both clinical and genomic data at the point of care.11^–13 As a result, understanding the current state of precision medicine interoperability will be a key first step towards effectively annotating results and communicating genetic test information across different health IT systems.14^,15 The goal of this study is to analyse the genetic testing registry (GTR) in the context of relevant health IT nomenclatures and standards in order to understand the practical challenges in addressing interoperability for precision medicine.

Methods

Results

Discussion

Precision medicine is expected to play a key role in transforming healthcare, and interoperable health IT provides the critical infrastructure around which precision medicine can be applied. To the author’s knowledge, this is the first study to assess the current state of precision medicine interoperability by analyzing GTR data with existing interoperability standards. This study is timely given the announced U.S. Precision Medicine Initiative (PMI), also known as the NIH All of Us Research Program, and the rapid convergence of health IT, genomics and big data analytics.7^,20^,21

There were a large number of registered genetic tests for a diverse set of genes focused primarily on diagnosis, mutation confirmation and/or risk assessment. When broken down by laboratory setting, academic institutions focused primarily on the diagnosis or confirmation of mutations, while companies reported a much more diverse set of registered purposes. This likely reflects the differing priorities and varied stakeholders involved for these settings. Companies, for example, develop tests for a broad set of stakeholders (including directly to consumers) consistent with the diverse reported test purposes, while tests at academic hospitals focus heavily on helping physicians, addressing the clinical diagnostic needs of their patient populations.22

While tests for germline mutations that could be passed to offspring predominated, the expansion of registry submission criteria will likely lead to a growing volume of genetic tests for somatic mutations as well.16 The relatively small volume of tested genes in the registry likely reflects the current lack of evidence supporting the clinical validity and utility of most genes in the human genome; furthermore, unlike analytical validity, clinical validity and utility remain optional entries in the NIH GTR.2^,16 Even at this early stage of precision medicine, however, several laboratories have begun offering genomic sequencing services and evaluating large panels of genes.17 Oncology is an example of one important area for precision medicine, and where an understanding of the human genome has guided not only disease risk assessment and diagnosis, but also selection of the most effective treatments for patients as well.7^,23^,24 As new guidelines and standards for identifying, classifying and assessing evidence for genomic data are developed, the breadth of clinically relevant genes will likely expand considerably over time.25^,26

The successful application of precision medicine in practice will require health IT capable of processing large volumes of genomic data and presenting relevant results to physicians at the point of care.2^,4^,27 While the largest number of laboratories came from the Academic/Hospital setting, the largest volume of actual tests originated from companies; in particular, there were twice as many academic/hospital labs than company laboratories, yet those companies registered twice as many genetic tests. Prior studies have shown different types of physicians order different genetic tests, and this study similarly showed that different organisations focus on different types of assays.22 Effective practical adoption of precision medicine will require a strong understanding of the diverse backgrounds and behaviours of stakeholders, ranging from patients being tested to providers ordering tests to the labs building new technologies.22^,28

One major purpose of the NIH GTR is to help healthcare providers to make informed decisions about the need to order genetic tests for patients.16 Genomic clinical decision support largely depends on the ability to connect genetic information with relevant clinical conditions at the point of care.27^,28 While the majority of genes were successfully mapped to both HGNC-approved gene symbols and NCBI RefSeq identifiers, a majority of genetic tests did not have any SNOMED CT code assigned to them, reflecting a critical gap in core information needed for the practice of precision medicine. The voluntary nature of the GTR is likely a major contributor to the poor degree of clinical mapping. In particular, not only is submission to the NIH GTR optional for organisations, but critical data fields (e.g. clinical codes, clinical validity and clinical utility) are currently the optional components of each submission as well. A required mapping of medical and clinical terms through a mandatory registry submission process would make the NIH GTR a more valuable resource to help physicians make sense of the overwhelming volume of genomic information that may soon be integrated into clinical care.2^,4^,27^,29

Currently, multiple genetic tests map to a single SNOMED CT code, obligating physicians to spend time deciding among multiple options for the same clinical indication. The presence of incomplete or confusing clinical mappings for genetic tests is likely due to current uncertainty around which standards should be used to map genetic data with other types of medical information.29^–31 The U.S. Federal Government’s Precision Medicine Task Force, for example, is responsible for recommending the set of standards to be used for exchanging data for the million (or more) patients expected to participate in the National PMI. Yet even with hundreds of relevant standards available, from Fast Healthcare Interoperability Resources to HL7 Clinical Genomics standards to the Global Alliance for Genomic Health, surprisingly only one standard (HL7’s Family Health History/Pedigree) has been recognised by the task force as mature enough for practical use in the PMI.32^,33 Strong multidisciplinary leadership capable of addressing the critical technical, regulatory and interoperability gaps will be needed so that the vision of precision medicine can become a practical reality.

Our study had several limitations. First, the results may not be generalisable since the GTR is a voluntary registry that may not capture every laboratory offering genetic testing services, and selection bias is thus possible. However, this NIH-based registry currently represents the most comprehensive attempt at creating a centralised resource of genetic tests and laboratories for the healthcare community, and will likely become more complete over time with the growing focus on precision medicine.16 Second, the dynamic nature of genomic medicine means that any categories used in our study to describe genetic tests will likely change as the field evolves. However, our study provides a solid starting point for gaining useful insight into the current state of precision medicine, and language describing the field will begin to stabilise as standards are adopted, guidelines are developed and policies and regulations are put in place.7^,31^,34^–36 Finally, there is a wealth of available standards and formats that could be applied to precision medicine, but the analysis of any single standard would not be able to adequately address every major issue. The primary purpose of our study was to take a data-driven approach to assessing the challenges and opportunities of precision medicine through the lens of health IT interoperability. As precision medicine evolves from assessing genetic tests to applying sequenced genomes, informatics approaches can be used to provide valuable insight into the wealth of diverse data describing all aspects of healthcare IT.27

In conclusion, the practice of precision medicine enabled by interoperable health IT has the potential to dramatically improve healthcare. However, this will first require the comprehensive but responsible adoption and implementation of appropriate standards, terminologies and formats across all aspects of the precision medicine pipeline.