March 9-12, 2021
Slides from each presentation will be provided below after the event. Unless noted, recordings and downloads on this page are only available to LOINC Conference registrants.
Introduction to LOINC
LOINC is the universal standard for identifying health measurements, observations, and documents. It is now ubiquitous in health data systems worldwide and is an essential ingredient of system interoperability. This tutorial presents an overview of LOINC and its use around the world, discusses the LOINC concept model and data structures, and describes the resources available for implementing LOINC. If you are new(ish) to LOINC, this session should be your starting point.
Advanced LOINC Concepts
David Baorto, PhD, MD
This session will review LOINC's transition from a naming convention 25 years ago toward a richer semantic model involving LOINC Parts. We will cover in detail how these "LEGO bricks" known as Parts are used to build LOINC terms. We will review how these reusable pieces have many other applications within LOINC, including those related to descriptions and name building, and also can be used to link to resources outside of LOINC, such as external terminologies and taxonomies. Finally, we will touch briefly on the evolution of the various ways that Parts can be leveraged to create collections of LOINC terms (hierarchies and ontologies), as well as future directions. While basic knowledge of LOINC is recommended, this session is suitable for both novice and experienced LOINCers.
LOINC Searches and LOINC Groups
With nearly 95,000 terms, searching the LOINC database can be intimidating but does not have to be. We'll show you tips and tricks to help you find appropriate terms to fit your needs. This presentation will demonstrate all the features available in the new SearchLOINC—now available as a public Beta. We'll also highlight LOINC Groups from methodology to practical use cases.
LOINC, Real-World Data, Clinical Research data and the HL7 Vulcan project
Regulatory authorities such as the FDA want to start "Real World Data" (RWD), i.e. data that come from regular healthcare and not from interventional clinical studies, not only in post-marketing research, but also e.g. as a replacement for placebo arms where providing placebo to subjects is regarded as unethical, which is more and more the case. RWD usually uses LOINC codes for exact definitions of the tests performed or observations obtained. In clinical research, the use of LOINC codes is however still seldom. For example, CDISC, the Clinical Data Interchange Standards Consortium only supports LOINC codes for submissions where this is mandated by the FDA (lab data). For all other domains, LOINC is still regarded as "not-invented-here" and CDISC keeps developing its own controlled terminology for these domains.
In the last year, we extended the by CDISC published LOINC-CDISC-LB (laboratory) mapping, a mapping that was developed due to the FDA requirement (and that took almost 3 years development time), containing 2,400 mappings for the 1,400 most popular laboratory LOINC codes, with over 5,000 additional mappings, resulting in a total set of over 7,700 mappings. Unfortunately, such mappings are necessary as CDISC still refuses to recognize the LOINC code as the unique identifier for a test - it uses its own combination of post-coordinated variables and controlled terms for this. Also, CDISC decided to move microbiology tests from the lab domain (LB) to a new MB (microbiology) domain, with the consequence that microbiology test results now fall outside the FDA mandate to provide the LOINC code for each tests. Especially in COVID-19 times, this seriously hinders the use of RWD in clinical research in the battle against COVID-19. Therefore, we generated our own mappings for the newly developed LOINC codes that are related to the Corona virus and made these available as a RESTful web service..
But this can and must be only the beginning. Also vital signs data in healthcare are characterized by their LOINC code. So, in order to be able to use vital signs RWD in clinical research, we developed 635 mappings between LOINC codes for vital signs and corresponding CDISC (post-coordinated) controlled terminology. Also these have been made available as a RESTful web services. At the moment we are developing similar mappings for Electrocardiography. These will enable to automatically generate FDA-submission datasets starting from Electronic Health Records (EHRs) that are used as RWD.
The better way of course is to use the LOINC code as unique identifier for each type of observation for FDA submissions, but this would require a major redesign of the CDISC submission standards, and as LOINC is "not invented here", there is no drive to do so at CDISC. This also means that there is no support from CDISC for our work, maybe out of fear for one day having to give up their own controlled terminology in favor of LOINC coding.
To further boost the use of LOINC and RWD, we also developed an algorithm and software to fully automatically generate CDISC electronic Case Report Forms (eCRFs) on CDISC-ODM-XML format starting from LOINC panel codes. Medical doctors usually think in panels when ordering laboratory tests, so it makes sense to have a way to generate eCRFs automatically from LOINC panel codes. This software is now available as a RESTful Web Service too. Finally, we will explain the extremely important role of LOINC in the new HL7-Vulcan project. The Vulcan project is an "accelerator" project dedicated to connecting clinical research and healthcare. It currently is still in its "community building phase" with very little first results. Our own work in this field is however already far ahead, and we hope that we will be allowed to contribute to this extremely interesting project. The current construct of Vulcan however is that only large companies also willing to pay a considerable amount of money can contribute.
Mapping LOINC document type codes: an experience in managing clinical documents for the long-term preservation
Maria Teresa Chiaravalloti
Italian National Council of Research
University of Calabria, Italy
University of Calabria, Italy
As LOINC Italia we are collaborating with a company which operates for long term preservation of clinical documents in order to map their document type codes to LOINC. In the presentation we would like to describe the steps of this work, the challenges and the results obtained until now. As the work is still in progress, this presentation aims at sharing our experience and also receiving suggestions from LOINC worldwide experts and users.
Identifying relevant data elements for Exome Testing and its coding with LOINC, SNOMED CT and Genomic Standards
Gloria González Gacio, PhD
Mireia Rodríguez Naqué
Vicky Bérez Baldrich
Toni Mas Mota
In recent years, exome sequencing has become a widely used method for identifying the molecular basis of genetic disorders across various medical specialties, when other alternative methods fail to detect causal gene mutations for diseases of Mendelian inheritance.
Currently, both Spanish Private laboratories and Public Hospitals have implemented exome testing as the first approach for genetic disorders. Most of our customers are using Exome Solutions as SOPHiATM (SOPHiA Genetics) with relevant gene content. SOPHiA Clinical Exome Solution and whole Exome Solution cover the coding regions of 4490 genes associated with most inherited diseases, and more than 19.000 genes enabling an exome-wide investigation, respectively.
The Exome studies generate complex reports containing different type of relevant unstructured data that must be codified to integrate into EHR. These issues highlight the urgent need to unify standards in variant annotation, with consistent reporting within the genomic context, to enable accurate data-driven care medicine.
In this presentation we want to show our experience in laboratories of genomics mapping to standard codes the variables related with Exome reports and other DNA massive studies. The data sets include variables related to the patient, laboratory test and results. Also, we would like to show the mapping of these data elements compared to other approaches such as the one from Mayo Clinic.
LOINC in Real-World Clinical Document Exchange: Findings and Opportunities
John D'Amore, MS
Diameter Health, Inc.
Sandi Mitchell, RPH, MSIS, FASHP
Veterans Affairs, DQT Team Leader
We will presentreal-world LOINC utilization among major EHRs using the C-CDA documents and other standards. This includes a review of highest volume codes, common issues related to vocabulary, code selection in respect to LOINC axes (i.e. Property, Method, Scale and System) and usage of interpretations, structured responses, values and UCUM. In addition, specific data related to COVID-19 code and response utilization will be presented. This presentation surveys usage across 500 different source systems representing a wide variety of information exchange in the United States.
A survey of licensed Patient Reported Outcome Measures in the process of submission to LOINC
Berlin Institute of Health, Charité
Naveen Moses Raj Rajkumar
Berlin Institute of Health, Charité
Background: Patient Reported Outcome Measures (PROMs) are Self-report questionnaires that patients complete on their health, prognosis and quality of life. The information collected from PROMs helps in researches, in monitoring patients’ progress allowing constant communication between physicians and patients.
Logical Observation Identifiers Names and Codes (LOINC) is a common language (set of identifiers, names and codes) for identifying health measurements, observations and documents. The use of LOINC with PROMs facilitate the collected data to make it further reusable for extended use across health information systems enabling interoperability.
Objective: To investigate the review of 99 Self-report measures included from various sources by Linton et al. for their presence in the LOINC panel. The Self-report measures that are not present in the LOINC panel are categorized into licensed and license free, the copyright owners of the licensed instruments were provided with detailed information about LOINC and asked permission for instruments to be added to LOINC panel. The copyright owners are asked to take part in a survey and to give their reasons for allowing or not allowing the instruments to be added to LOINC panel.
Results: The survey contributes an elaborate explanation on the various reasons for allowing or refusal to add the instruments to LOINC panel by the copyright holders. The reasons for allowing or refusal are categorized on several themes such as intellectual property rights, scoring methods, organizational issues etc.
Conclusion: This report provides the users and researchers on the availability of the instruments in LOINC panel and their reason for their absence. The report provides on the various barriers existing in the usage of licensed instruments particularly to be interoperable for access across systems and aggregation of useful data.
A comprehensive US Ontology for better management and use of LOINC datasets
James R. Campbell, MD
Professor of Medicine, University of Nebraska Medical Center
This presentation will describe the conceptual framework and operational features of an integrated ontology unifying the semantics of SNOMED CT, LOINC and RxNORM. The conceptual model capitalizes upon the shared work of the Regenstrief Institute, SNOMED International and the National Library of Medicine. The ontology's conceptual model for LOINC will be discussed in detail in support of use cases for LOINC in clinical decision support and population health employing Electronic Health Record datasets.
Unsupervised Refinement of a LOINC based COVID-19 Knowledge Graph
School of Informatics and Computing, IUPUI
Sunandan Chakraborty, PhD
School of Informatics and Computing, IUPUI
Research in response to COVID-19 has produced thousands of academic articles, constituting a wealth of information. Knowledge graphs are a promising method of organizing and understanding this growing collection of scientific knowledge but have received less attention than other information extraction techniques. Recently constructed knowledge graphs have successfully identified relationships between entities of interest in COVID-19 literature, but few attempts have been made to refine these relationships. While all knowledge graphs draw relationships between entities, a refined knowledge graph will quantify some aspect of those relationships. This presentation discusses COVID-19 knowledge graph construction using LOINC, and refinement of the graph using unsupervised clustering.
The presentation will begin with a brief introduction to knowledge graphs and their previous applications to COVID-19 literature. We will discuss the dataset used for our analysis, the COVID-19 Open Research Dataset (CORD-19), which includes a large collection of full text academic articles. Our application of the LOINC coding system for knowledge graph construction will be discussed, with emphasis on the usefulness of terminology standardization and the use of RELMA for entity tagging. We will discuss relation extraction using an Open Information Extraction method, and how we refined relations using the unsupervised clustering of BERT sentence embeddings. We will discuss our resultant knowledge graph, and how it could be used for comprehending the ever-growing collection of COVID-19 literature.
LOINC in FHIR forms
Clem McDonald, MD
Chief Health Data Standards Officer
National Library of Medicine, National Institutes of Health
Abstract to follow.
Payer Usage of LOINC: Automated process to pull Covid-19 lab results from network providers
UnitedHealthcare uses our EMR Integration Services Layer (EISL) Lab Chase Service (LCS) to pull missing lab results from our network providers. We leverage LOINC to target specific observations for clinical quality measures and other purposes. In this presentation we will show how we reacted quickly to the creation of LOINC pre-release terms for SARS-CoV-2 (COVID-19) testing to gather data on our members. Results can be queried from labs in response to claims or periodically by member rosters.
Implementing LOINC based TEST_TYPE for HCSRN common data model
Henry Ford Health System - Department of Public Health Sciences
Kaiser Permanente Northwest - Center for Health Research
We will present our process for creating the infrastructure to include SARS-CoV-2 testing into the 19-member HCSRN's (Health Care Systems Research Network) VDW (Virtual Data Warehouse) common data model.
The VDW requires grouping similar lab tests under one TEST_TYPE to facilitate combining data from 17 health care systems that have lab results available for analysis. We decided to create categories of SARS-CoV-2 testing that would accommodate equivalent findings. i.e., all qualitative nucleic acid testing was assigned to one TEST_TYPE regardless of probe/target (S-gene, N-gene, rdrp or ORF). Virus detection by NAA or immune assay will not be combined and have separate TEST_TYPES. We have also mapped antibody testing. The grouping method to assign LOINC codes to one TEST_TYPE will be described.
Driving Adoption with Standardization: A Vision for Disease-Specific Patient-Reported Outcomes Measures
LaVar Edwards, MS
Vanderbilt University Medical Center, SyTrue
Double Lantern Informatics
Robert A. Swerlick, MD, FAAD
Mary-Margaret Chren, MD, FAAD
Vanderbilt University Medical Center
In the pursuit of improving outcomes and providing care that is simpler to deliver and easier to assess, an electronic medical record must be more than just a repository of information. It must also be a tool that providers and their patients can use to make better health outcomes more attainable. We propose to engage patients to participate in their care through standardized assessments of disease impact, and seamlessly integrate these assessments into medical record. But assessing outcomes using patient input is difficult if the data is not structured and standardized.
The Skindex instruments (Skindex-29, Skindex-16, Skindex-Mini) are three validated patient-reported outcomes measures (PROMs) used to capture the impact of skin disease on the quality of life (QoL) of patients. These tools provide reproducible quantitative assessments of the impact of skin disease symptoms, and their emotional and functional effects on patients.
A vision for improving the availability of tools that can measure impact of skin disease on QoL and positively impact patient care experiences relies on structured and standardized data. This team embarked on the process of assigning LOINC terms to each instrument in the fall of 2020. Having successfully assigned terms to all three tools, the team can transmit the data between systems and aggregate the results. We will continue the standardization process for our vision for the future of PROMs and add them to the standard libraries of REDCap, Epic and Cerner, initially.
Keys for IVD Vendors to Fast Track LOINC Adoption
Pam Banning, MLS(ASCP), PMP(PMI)
3M Health Information Systems
In vitro diagnostic (IVD) vendors do not have the same starting point as laboratories with an information system. How to start? Where to start? Learn the best implementation tactics to adopt LOINC from package inserts and technical information while keeping the best foot forward with future maintenance of an IVD manufacturer compendium. Resources and the appropriate time to use each (the LOINC Implementation Guides, master file management, LOINC In Vitro Diagnostics (LIVD) formatting for clients' usage) will be discussed. Make every effort count!
Gender Harmony: Time for clear observations of sex and gender
Robert McClure, MD, FAMIA, FHL7
President, MD Partners, Inc.
Caroline Macumber, MS, PMP, FAMIA
Andrea J.C.L. Downey, BSc, MNRM, PMP
President, The Project Consortium Leadership Practice | Health Portfolio
The representation of patient gender identity and clinical sexual characteristics have been hampered by "social norm" expectations and data representation bias for centuries. The advent of health information technology perpetuated these biases in many systems, often limiting data capture to a single field with variable labeling that did not support recognition of gender identity and sex differences that could change over time or co-exist with different values supporting different contextual uses. This has resulted in systematic difficulties for patients where accurate documentation is critical and restricted the delivery of proper clinical care even in environments striving to be responsive. The presentation authors have just completed an HL7 informative ballot: "Gender Harmony - Modeling Sex and Gender Representation, Release 1" that proposes a logical model that clarifies an improved approach for representing clinical and demographic sex and gender identity information. LOINC has many of the observations needed to support the proposed model but some modification may be in order. This presentation will give background on the current situation and the clinical importance of timely and accurate representations of sex and gender identity. We will then review the HL7 model including the model elements necessary to support distinguishing different sex and gender documentation requirements. We will identify aligned LOINC concepts, how to use LOINC to support the model, and areas that may need additions and improvements. We will also discuss enhancements to existing clinical models in order to adopt the proposed solution, including HL7 (FHIR, C-CDA, V2), DICOM, USCDI and others.
Broader value of LOINC terms in real-world data/evidence data sets
Horst Donner, PhD
Ani John, PhD, MPH, BSN
The use of real-world data / evidence in development and registration of diagnostic products is constantly increasing within the diagnostic industry. The use of this data is of special importance in the current situation where many diagnostic tests are provided to the public using an emergency use authorization based on high medical need and pandemic restrictions making a classical evidence generation (traditional clinical studies) difficult to perform.
LOINC information is a very important attribute in the analysis of real-world data (RWD) and real-world evidence (RWE) since LOINC information has been used as generic identifier of diagnostic assay procedures in various RWD/RWE databases. Especially electronic patient records often display LOINC terms together with diagnostic test results to provide sufficient information for the documentation of the health status.
Since the LOINC term is shared among manufactures of assays using similar detection principles it's use in registration purpose is questioned by regulatory authorities due to a lack of manufacturer specific information.
To overcome this obstacle it might be of great value to the society in case it would be possible to modify the LOINC term in a way that it additionally uniquely identifies the manufacturer of the assay the code relates to. Whether this is feasible and how this could be managed might be of interest for the LOINC committee and community as it may broaden the use of this code and may contribute to speed and efficiency of evidence generation which serve the diagnostic industry, health care professionals and organizations, authorities, and last but not least the patients.
Language Intelligence: The Future of Semantic Interoperability between LOINC, ICD-10, and unstructured EHR via layered hierarchical ontologies
Adam Tomkins, PhD
AI Engineer, Sorcero
Senior Project Leader, Smart Diagnostics, Roche
Xavière Pincemin, MD
Senior Clinical Scientist, Roche
The present-day universe of health data is composed of a wide array of diverse and disparate data sources. A broad range of healthcare, demographics, and other varieties of information relevant to clinical providers are housed in these data sources. For example, electronic health records and claims data hold unique promise and challenges. When leveraged for Language Intelligence (LI), artificial intelligence-enhanced natural language understanding that incorporates layered ontologies and specific lexicons, these challenges take center stage.
Whether structured data or free-text, the data required for LI are most often individualized, person-based records sourced from individual cases, that reside in EHR/EMR systems, with 80% of their data in unstructured medical notations. Aspects of these data sources not bound by HIPAA may be shared more widely, but often lack the specificity necessary for the purposes of LI.
In order to organize this diverse information in a proprietary way, layered ontologies that bridge between existing conceptual coding formats like LOINC and ICD-10 must be designed and developed, always evolving to satisfy the requirements of new contexts. The driving force behind this evolution arises from a vast array of semantic interoperability issues, making the integration of data held on these very different and often scattered systems a dynamic and ongoing challenge.
LI offers new ways to meet the challenges of semantic interoperability: hierarchical ontologies. This presentation intends to explore the manner in which domain specificity can enable semantic interoperability in workflows previously resistant to enhancement by LI. The resulting potential for novel integrations of disparate technologies can better serve healthcare providers, clinicians, and knowledge workers of all kinds in their efforts to improve patient outcomes.
The future of LI applied to semantic interoperability: facilitating the construction of uniquely specific ontologies for workflows critical in healthcare, and applying those ontologies to challenging, rewarding use cases. This presentation will explore the results of a hierarchical approach to semantic interoperability, with demonstrations including entertaining results that are only possible through Language Intelligence technology.
Update from Office of the National Coordinator for Health IT (ONC)
Presenters to be announced
More information to follow.