Logical Observation Identifiers Names and Codes for Laboratorians: Potential Solutions and Challenges for Interoperability

The LOINC system resulted from a collaborative development initiative in 1994 at Regenstrief Institute, a health care and informatics research organization associated with Indiana University.¹⁴  Its creation spawned from needs related to the Indiana Network for Patient Care, the first citywide health information exchange in the United States, first organized by investigators at Regenstrief Institute.⁴  Although their computer system could receive data using standardized messaging protocols (eg, HL7) from institutions within their network, researchers found it difficult to interpret the data, as each institution used its own unique local codes to identify equivalent or the same tests and/or observations. The LOINC system, of which the first iteration was released in 1995, was developed to mitigate this problem by standardizing the codes that would be embedded in the data exchange messages, regardless of the originating institution. Regenstrief continues as the steward of the LOINC standard's development, maintenance, and dissemination.

Given the extensive growth of the LOINC database, Regenstrief Institute developed a Windows-based software program called Regenstrief LOINC Mapping Assistant (RELMA) to assist with code mapping, by providing a semiautomated search capability using the laboratory's test compendium terms. Once data have been imported in a RELMA-specific import format, RELMA uses an algorithm to search the LOINC database based on the laboratory's locally used test description/name and produce a ranked list of candidate LOINC terms for the user to evaluate. By counting the number of matches between words in the laboratory's test name and description and the words and synonyms in the formal LOINC term name, the software is able to narrow the potential candidate LOINC terms to be used and present them to the user of the software. Once a LOINC code has been assigned to each test in the compendium, the mapping table can then be exported to facilitate assignment within the LIS.

Regenstrief recognized that laboratory technology would advance over time, and consequently designed LOINC to allow new codes to be added as novel tests emerged. LOINC version updates and new releases occur biannually. The LOINC system encourages end users to submit requests for additional codes and provide comments and feedback about the curated database. Like the LOINC database, RELMA is freely available for download from the official LOINC Web site (https://loinc.org, accessed March 1, 2019). RELMA also offers flexible display options for viewing the details about a particular LOINC item, which can be daunting to review within an Excel spreadsheet, and provides functionality for users to propose and submit requests for new LOINC items to Regenstrief. (For more information, see http://loinc.org/submissions/new-terms, accessed February 25, 2019.) Limitations of RELMA have been described (eg, uses restrictive, full name, exact name matching).¹⁵ 

In the United States, the LOINC database has been included by the National Library of Medicine as one of the source vocabularies for the Unified Medical Language System metathesaurus. The LOINC system is currently used in more than 145 countries, and Regenstrief Institute continues to cultivate the LOINC community both within the United States and worldwide.^4,16,17  Additional language support, besides English, includes French, German, Greek, Italian, Spanish, Dutch, and Russian.¹⁸  Additional language support is under development.

Each record in the LOINC database contains a fully specified description composed of 6 major axes (also referred to as parts) that combine to uniquely identify each individual observation or measurement (Table 1). These major axes comprise the component (or analyte) being measured, the property being measured, the timing of the measurement, the system (usually sample type for laboratory measurements), the scale of measurement, and the method used to produce the observation. Of note, only the method axis is optional. Although each LOINC item is inclusive of this specific information, it lacks relevant information required by end users for the pragmatic use of the standard. For example, LOINC does not include details regarding the type of instrument used, testing priority, performing site, interpretation criteria, the individual responsible for the interpretation, or related diagnoses.

The current LOINC database contains more than 55 000 numeric codes specific for laboratory tests, each with a length of 3 to 7 characters. However, when the database exceeds 100 000 codes, 8 digits will be required. The December 2017 release of LOINC (version 2.63) also included a one-time increase to the database field size for certain components in order to accommodate future content development. For example, the database field length for the 6 major axes and short names expanded from a range of 15 to 50 characters to 255 characters. All LOINC codes end with a hyphen followed by a check digit.¹⁹  This final digit is an integer between 0 and 9 calculated from the prior digits using a mod 10-based (Luhn) algorithm.^20,21  The purpose of the check digit is to detect and reduce transmission of typographical errors common to human data entry.

Some coding schemes used in health care, such as the International Classification of Diseases standard, possess a structure that conveys relatedness between items, and a hierarchical structure to reflect increasing specificity. In contrast, LOINC codes have no inherent organization (sometimes referred to as nonsensical organization). New LOINC numbers are assigned sequentially as new entries are requested and subsequently added without regard for grouping similar concepts. For example, code 51487-7 refers to the observation of leukocytes (number/volume) in urine by automated count, and the next code in sequence, 51488-5, refers to the sixth cortisol specimen (mass/volume) in saliva postchallenge.²²  Lastly, if we look at a similar test to the first code, called leukocytes (number/volume) in urine by manual count, we see that its LOINC code, 24122-4, has no relationship to the first leukocyte urine test code, 51847-7. Because LOINC lacks a relationship paradigm such as the “is-a”–type relationships seen in Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT), the nonsensical codes prevents deriving any additional context, such as test similarity, from the code structure.

The LOINC code database contains a field for each record called ORDER_OBS (order observation) where the value for the code category is recorded. There are 3 code categories, orders, observations, or both, that are used to determine whether the LOINC item can be mapped to an order, an observation, or both. For system-to-system communication, this role is conveyed by its location in an HL7 message. For example, the LOINC codes 801-1 (sickle cell [presence] in blood by light microscopy) and 55456-8 (hemoglobin S in blood) are categorized as an observation-only LOINC code and an order-only LOINC code, respectively, whereas the code 718-7 (hemoglobin [mass/volume] in blood) is a LOINC code that can be mapped to an order and/or an observation (result).

The LOINC database comprises 4 domains, or class types: laboratory, clinical, attachments, and surveys. The laboratory domain is further subdivided into disciplines as shown in Table 2. The structure of the database supports organization into a multiaxial hierarchy. The LOINC codes can be organized by a multiaxial hierarchy based on the multiple axes/parts of each LOINC concept. The multiaxial hierarchy is generated using an automated process based on the individual class, component, system, and method hierarchies, which are maintained manually. However, these hierarchies are not meant to be interpreted as pure ontologies.

On February 17, 2009, the US Congress passed the American Recovery and Reinvestment Act. The act included an enactment of the Health Information Technology for Economic and Clinical Health (HITECH) Act to promote and expand the adoption of HIT. The HITECH Act was a critical step toward the creation of a nationwide, interoperable, and secure matrix of health information systems. The HITECH Act was also notable for its mandate of meaningful use (MU) for certifiable EHR adoption.²³ 

Meaningful use was one of the dominant driving forces of EHR interoperability in the United States.^24,25  Stage 2 MU and accompanying EHR standards and certification criteria regulations also established criteria whereby certified EHR technologies are able to support and enable improved quality, safety, efficiency, and patient engagement and reduced health disparities for the purpose of achieving better clinical outcomes.¹⁴  To meet the stage 3 MU requirements, providers and hospitals must use technology certified to the 2015 edition of HIT certification criteria. These criteria require HITs, such as LIS or EHR systems, to accommodate the use of, at a minimum, LOINC version 2.52 as the vocabulary standard for laboratory tests.^26,27  Consequently, LOINC coding of laboratory results such as public health reporting constitutes one of the criteria for an EHR to reach attestation that the system is certified to be MU compliant. The 2015 edition standard for the Common Clinical Data Set, which includes laboratory tests, requires that at least LOINC data set version 2.52 be used.

Although additional requirements for LOINC coding were not explicitly included with regard to computerized physician order entry or reporting of patient laboratory results in stage 3 MU, a rule by the Centers for Medicare and Medicaid Services (CMS) in 2015 clarified that certified EHR technologies must still include the required use of standards, including LOINC, associated with structured data capture.²⁶  The LOINC codes are required physician reporting of many clinical quality measures, as well as hospital reporting of health care–associated infections, and continuity of care documents in stage 3 MU.

Certification of EHR modules for LISs may also be achieved based on meeting a subset of the certification criteria.²³  Although there are vendor certification criteria for functionality supporting LOINC mapping to laboratory orders and results, such functionality may not exist in all laboratory interfaces between connected information systems. In other cases, the performing laboratory may “turn on” LOINC functionality for LOINC codes to be exchanged between systems. It is largely for these reasons, in addition to the role of LOINC in public health reporting, that laboratories in the United States started to devote attention and resources to adopting the LOINC standard. Table 3 lists several federal regulations in the United States that are pertinent to LOINC.

The Medicare EHR Incentive program (ie, MU) was transitioned to become 1 of the 4 components of the new Merit-Based Incentive Payment System (MIPS), which is a component of the Medicare Access and CHIP Reauthorization Act (MACRA).²⁸  The Advancing Care Information category within MIPS supersedes and replaces MU for eligible providers. Through the 2015 edition HIT certification criteria, the ONC provides the health IT foundation for MIPS and its quality category; furthermore, ONC encourages EHR vendors to certify their technology to the 2015 edition final rule, because this certification signals that the vendor's product can fulfill the reporting requirements included in MIPS, automating many of the processes needed to capture, calculate, and report quality measures to the CMS.²⁸ 

Clinical laboratories are not eligible to receive incentives as part of the MIPS, MACRA, and MU requirements, but those within eligible hospitals often provide LOINC-coded laboratory data in assisting their facility to receive its incentives. Although laboratories are not directly incentivized through these programs, there are often significant indirect impacts to their outreach business as their customers, whether eligible providers or eligible hospitals (for referral testing), want to receive their incentives and are requesting LOINC codes from the performing laboratory. As a result, many laboratories provide LOINC codes in their result messages and in their online test directories to customers so they can meet their reporting requirements. Additionally, all laboratories performing testing identifying reportable conditions are required by law to provide LOINC codes in their electronic laboratory reporting messages to public health agencies.

Assigning LOINC codes to the local laboratory tests is part of a pathology laboratory's duties, and although in many organizations LOINC codes are assigned to laboratory tests primarily because of regulatory requirements, this assignment should nonetheless be subjected to a defined process, similar to determining each test's reference range. Several years ago, laboratorians requested that LIS vendors should accommodate the use of LOINC and include LOINC in their baseline test dictionaries.²⁹  Indeed, in order to help meet MU criteria, most LISs now include an indexed field for LOINC codes in their test database. The LOINC codes, however, still need to be mapped in the test definition dictionaries in the LIS. Typically, each laboratory makes its own decisions regarding assignment of LOINC codes. If different teams are defining LOINC codes for the LIS and the EHR, there may be discrepancies even within the same institution. Coupled with the complexity of the LOINC database and multiple options for coding the same analyte or test, this local assignment of LOINC leads to variation in LOINC code mapping and accordingly a lack of complete semantic interoperability for exchange of laboratory test orders and results.^30,31 

Although significant focus has been placed on order codes, it is important to be aware that LOINC has separate codes for orders and results. For example, there is an order code for a complete blood count with differential that has distinct LOINC codes for the results (eg, an individual result code for percentage basophils). Despite having mapped their unique laboratory tests to LOINC, laboratories may still need to rely on creating their own internal names and codes for test results. For example, laboratories may need to use a test code to distinguish glucose measurements performed in a core laboratory from those in their satellite facilities.²⁹  However, all of these individual test codes will still need to be mapped to the appropriate, but identical, LOINC code. The LIS laboratory test definitions and dictionaries are dynamic and frequently changing, thus requiring a formal process for ensuring uniformity of coding among different instruments or methods and, if applicable, the capacity to accommodate different codes that are correctly applied depending on the specific methodology used for that particular result.³¹  There should also be a formal audit process and maintenance to keep up with the changes that occur with each LOINC release to ensure the accuracy and appropriateness of previous code selection as the database continues to mature.³¹ 

The assignment of LOINC codes in laboratories brings with it the need for LOINC expertise, or at the very least awareness of the functionality of the LOINC database; ideally, the task of accurately mapping local tests to LOINC requires an individual with expertise in both the complexity of laboratory testing and the multitude of coding options and underlying structure of the LOINC database, as well as the time required to be knowledgeable about new developments. In practice, laboratory personnel who are familiar with their laboratory's test menu sometimes lack sufficient knowledge about LOINC. As a result, the task of mapping may be delegated to information technology staff, who generally have limited or, in some cases, no experience with laboratory testing. In addition, lack of detailed information (ie, specimen type, units of measure) and confusion regarding the properties being measured (eg, mass concentration with quantity measured in mg/dL, substance quantity measured in mmol/L, or an arbitrary quantity measured in IU/mL) may create further mapping difficulties and result in substantial errors in the assignment of LOINC codes. Alternatively, laboratories may enlist a third party or consultant to assist in the LOINC mapping of its tests. Multiple entities provide such services. Even if a laboratory contracts out the task of LOINC mapping, the laboratory must still have involvement in or awareness of LOINC and how it is used for laboratory test coding in the laboratory. The LOINC codes may be assigned to meet requirements for insurance contracting, for voluntary MU incentives, or to meet customer needs as part of their outreach business. However, if LOINC codes are assigned only for the sake of meeting a regulatory requirement, and laboratorians do not see value in LOINC coding commensurate with the effort this task requires, then due diligence in ensuring that codes are assigned correctly may be lacking. In addition, the need for consideration of LOINC in change-control processes and understanding the potential implications of LOINC in other institutional systems (ie, EHR) still applies.

In an effort to decrease the effort required by individual laboratories to implement LOINC, Regenstrief, in collaboration with the National Library of Medicine, released a resource called The Common Lab Orders LOINC Value Set in 2010. This subset, derived based upon analyzing the frequency of millions of universally mapped codes from various health care institutions, is intended to cover more than 95% of commonly ordered laboratory tests and to provide laboratories with a “starter kit” of codes based on the most frequently ordered laboratory tests. In 2010, Version 1.0 of the Common Lab Orders was released.³²  In 2014, the Standards and Interoperability Framework Initiative included the aLOINC (“agreed-upon” LOINC) Order Code, which was launched with the goal of enhancing and expanding the LOINC standard to include more comprehensive coverage of commonly ordered laboratory tests in the ambulatory care and public health settings.³³  The development of aLOINC was complicated by the lack of a nationally representative data set from which to establish an empiric list of common order and result codes. In line with the experience noted in previous sections of this paper, the Standards and Interoperability working group addressed issues where LOINC mapping of laboratory orders was problematic for laboratories and recommended that training tools be provided to help laboratories select an appropriate LOINC code for mapping their local test menu.³³  In 2016, a new and significantly updated version of the LOINC Universal Lab Order Codes Value Set (version 2.0) was released based on the Standards and Interoperability Framework Initiative aLOINC, which contains 1523 unique LOINC codes.³² 

The Standards and Interoperability working group also made recommendations regarding the development of test codes for anatomic pathology and cytology, both orders and results. Although not a panel in the traditional sense, individual tests in anatomic pathology are nonetheless associated with multiple attributes pertaining to a single specimen, such as specimen collection, processing, and interpretation. Anatomic pathology reports for a single specimen may require multiple codes to adequately relay key specimen attributes and results; additionally, these codes would need to be linked together and recorded in a structured format. The working group also recommended that the ONC promote collaboration between the Food and Drug Administration and Regenstrief Institute to develop an additional requirement for instrument/device and/or reagent approval where the manufacturer would assign an appropriate existing LOINC code or submit a request for a new LOINC code from Regenstrief Institute; the LOINC code would then be provided in the manufacturer's package insert or in a database, so that the manufacturer's LOINC codes would be accessible in one location.³³ 

In 2013, Regenstrief Institute and the International Health Terminology Standards Development Organization, which maintains SNOMED CT, reached a long-term cooperative agreement forming the SNOMED CT Observable and Investigation Model Project. SNOMED CT is a standardized clinical terminology used in the electronic exchange of clinical health information.³⁴  The model project established a working group to develop, test, and deploy an ontology-based definitional structure of all observables.^35,36 

Many laboratory instrument and reagent vendors, aware of the immense challenges associated with interoperability, have begun to include LOINC definitions in their package inserts. Such package inserts are a logical source to provide information that can help laboratories map tests with an appropriate LOINC code, because these vendors are well equipped to identify the appropriate LOINC code for their specific laboratory test. Moreover, for the introduction of novel laboratory test technology, it makes sense for vendors to request new LOINC codes. Fortunately, many instrument vendors are collaborating to develop interoperability standards, as highlighted by the In Vitro Diagnostics (IVD) Industry Connectivity Consortium, which is working on appropriate LOINC codes for various analyzers.³⁷ 

The LOINC to Vendor IVD initiative was created in 2016 through a National Institutes of Health laboratory data semantic interoperability collaboration among the Industry Connectivity Consortium, Integrating the Healthcare Enterprise, Pathology and Laboratory Medicine, the Regenstrief Center for Biomedical Informatics, the Centers for Disease Control and Prevention, the National Library of Medicine, ONC, CMS, and the Food and Drug Administration.³⁸  Its purpose is to advance the interoperability of laboratory data by defining a format to facilitate the publication and exchange of LOINC codes for vendor IVD test results. The LOINC to Vendor IVD specification provides LOINC maps to IVD test results and includes important data attributes including the manufacturer, instrument model, unique device identifier, vendor transmission code, vendor specimen description, vendor result description, and test name, as well as optional data elements, which are helpful for laboratory professionals mapping to LOINC.³⁸  Their proposal uses JSON (JavaScript Object Notation), a widely used human-readable standard for data transport/exchange that possesses a simple syntax.

New advances in HIT include the development of FHIR, which was introduced by the HL7 health care standards organization as a proposed interoperability standard. The FHIR standard differs from previous HL7 standards in a number of ways, including the use of a modern Web-based suite of application programming interface technologies, so that current technologies (eg, XML) can be leveraged. Recently, LOINC has made its codes available programmatically via requests to LOINC's FHIR server, which is in the beta testing phase.³⁹  The LOINC to Vendor IVD specification on FHIR, FHIR service catalog for representing laboratory test compendium information, FHIR specification for genomic test reporting, and FHIR diagnostic report specifications all incorporate LOINC for encoding laboratory data exchanged.

There are well-founded criticisms of LOINC, including the complexity and resource-intensive process of selecting the most correct LOINC code for each laboratory test and mapping each local code to a standardized code. These challenges are further compounded by the increasing complexity and volume of LOINC codes and laboratory tests.

Although the assignment of appropriate LOINC codes to laboratory tests is integral to proper use, the process of accurately mapping a laboratory test to the correct LOINC code can be challenging.¹⁵  This is especially true when a laboratory is presented with numerous similar codes for a given test, as exemplified by the LOINC codes for urine leukocytes in Table 4; the options may include codes that are inaccurate for the actual laboratory test being performed (eg, a manual versus automated process, or a different specimen type). The LOINC code options presented in Table 5 for a variety of analytes (erythrocyte sedimentation rate, fibrinogen, fluoxetine, and triglycerides) have the same values for the first 5 parts/axes, but differ in the sixth axis, illustrating that the specification of a particular method type, or the presence or absence of a method type, can result in the designation of a different LOINC code. These variations in method type selection can contribute both to challenges in LOINC code selection and to potential mismatches when attempting to aggregate data. The LOINC codes may have substantive differences in the level of granular detail they convey, which can further confound interoperability and data aggregation. Hence, it is not surprising that some studies have shown up to 86% failure of LOINC (or perhaps failure to correctly use LOINC) to match the same tests between different institutions as a result of differences in local coding choices.⁴⁰  Moreover, newly developed laboratory tests (eg, molecular tests such as next-generation sequencing) may not have an appropriate or informative LOINC code in the current LOINC database.

The use of LOINC does not guarantee seamless interoperability of laboratory test names and codes in electronic interface messages between different information systems. The task of assigning a LOINC code can be daunting, as illustrated by the existence of more than 600 codes for serum glucose measurements alone (Figure 2). Given the typical size of an institution's testing compendium, the process of mapping thousands of codes is laborious and prone to human inconsistencies and/or errors. As a result, adopting quality metrics and defined procedures are critical to ensure accurate LOINC mapping.^31,41  Lin et al³⁰  evaluated the correctness (concordance) of mapping local laboratory result codes to their corresponding LOINC codes at 3 large institutions: a reference laboratory, a large nonprofit health care provider organization, and an informatics and health care research organization, all of which had experience and expertise in using LOINC. The authors manually reviewed 10% of more than 9000 laboratory tests mapped to 3669 distinct LOINC codes. Of the 884 tests manually reviewed for correct mapping, 4 were mapped to entirely unrelated LOINC codes and 36 contained at least 1 error in mapping with regards to the 6 LOINC code axes/parts, which the authors divided into 4 categories of systematic errors: (1) human errors, (2) mapping to different degrees of granularity, (3) lack of knowledge of the meaning of laboratory tests, and (4) lack of knowledge of LOINC naming rules.³⁰ 

The field for LOINC short names was created in 2002 with the goal of providing names with a maximum of 40 characters to fit within the field space of many laboratory reporting systems. All laboratory tests have a short name, but these names are not unique, are subject to change, and, according to the LOINC user guide for version 2.65, “should not be used as identifying keys in any database.”²¹  There are challenges in using the short or long names built into LOINC as EHRs and LISs. Arguably, the EHR and LIS should but cannot necessarily accommodate the long names as they can exceed character limits of the information system. The short names provide a different challenge in that they can be complex, even indecipherable, especially for our patient-facing colleagues, without an index of LOINC code abbreviations (eg, LOINC code 72647-1, short name albumin ser-plr fld-MCDiff, which refers to the mass concentration difference in grams per liter of albumin in serum and pleural fluid). The expectation of LOINC is that both the short and long names not be used for display to clinicians, despite requests from users to produce names that can be used for display in user interfaces. Rather, the laboratory is expected to link their own local preferred name to the LOINC codes for use in reports and displays. The LOINC committee also defined best practice with the recommendation that users update to the latest LOINC release “within 90 days of its publication” although many laboratories are likely unaware of and do not adhere to this recommendation.²¹ 

A more recent study comparing LOINC coding among 3 large academic medical centers (Cleveland, Ohio; Houston, Texas; Pittsburgh, Pennsylvania) in the United States for several hundred commonly ordered laboratory tests highlighted that these barriers to interoperability persist. The potential to erroneously aggregate results performed on different specimen types can occur, for example, when data from an assay performed on body fluids is combined with data from the same assay performed on serum/plasma; this is of particular concern when institutions select an LOINC code that inherently lacks a specified specimen type (eg, LOINC code system “XXX”).⁴²  These and other works and the authors' similar experiences suggest that it is imperative that quality measures such as audits be adopted to ensure accurate LOINC mapping.^15,31,41  To ensure best practice, a laboratory professional well versed in LOINC from the laboratory that is performing the testing should select the optimal LOINC test order and result codes, as this person best knows the nuances of how the testing is performed. Downstream or upstream individuals often do not have the complete test details available to them, particularly with regards to testing performed at reference laboratories. In addition, nonlaboratory physicians and technical team members typically do not have adequate information to choose the optimal LOINC codes correctly (unless the codes are provided by the performing laboratory). Ideally, scripting and automated methods will be developed to facilitate this task.^43,44  Changes to LOINC codes create challenges for maintenance in LISs, EHRs, and other systems, particularly when the LOINC code changes for a test that has already been mapped.

An evaluation of urinalysis data from one institution reiterated that relying on LOINC codes to exchange and aggregate data can lead to erroneous conclusions, and the potential for patient harm, even when the correct code is selected. To identify analyte thresholds for the development of a single, system-wide urinalysis to urine culture reflex algorithm, data were collected from multiple laboratories within the same hospital system. Two of the laboratories performing urinalysis were located only a short distance apart and shared the same LOINC codes, the same method, and the same instrument, suggesting that at the most granular level of detail these 2 laboratories were identical. However, a review of the LIS rules that converted the quantitative instrument results into an ordinal value stored in the LIS revealed that a value of 3000 bacteria/μL was being translated into few bacteria by one laboratory and many bacteria by the other laboratory. The aggregation of the ordinal results from these 2 laboratories obtained from the LIS/EHR would have led to erroneous conclusions about the optimal thresholds for a reflex algorithm. This example highlights a challenge not appreciated by those outside of the laboratory; there can be important and significant functional gaps between superficially identical-appearing data, and the ability to use a coding system for safely and accurately transmitting data in laboratory order and result interfaces may be so wide that it is not clear that they can necessarily be bridged. The use of a LOINC code to populate results from 2 laboratories into a single result line in an EHR or health information exchange can therefore pose a significant safety risk, for example in the quantification of viral load, which can vary considerably among laboratories, especially when one or both laboratories are not using an international standard.⁴⁵ 

Even with the ability to add additional LOINC codes, the current standard fails to precisely and reliably be used to communicate structured anatomic and molecular pathology data for clinical care, public health, quality improvement, or research.³⁶  The disciplines of anatomic pathology, especially in the context of a dynamic transformation to digital pathology, and genomic medicine often contain complex heterogeneous data that is not amenable to a one-analyte-one-code model.

Historically, most complex genomics and molecular diagnostics testing has been reported in a narrative text format that does not lend itself to coding data elements to LOINC or other terminology standards to leverage for analytics, precision medicine, and other use cases. The complexity of next-generation sequencing results would require a coding system that can communicate result codes applicable for an array of testing from hot-spot to full gene sequencing, identification of deletions/duplications, epigenetic changes, fusions, and other results for more than 22 000 coding genes, in addition to the testing of noncoding regions known to contribute to disease. Creation of LOINC codes for all of the different combinations therein would be a significant undertaking with the current LOINC model. The historic LOINC model, as it applies to common basic laboratory tests, is not scalable with the rapidly evolving acquisition, interpretation, and reporting of genomic data or the complexity of new genetic tests, methods, and variants, such that LOINC codes are not currently meeting the needs of many molecular laboratories.

The HL7 Clinical Genomics (CG) Work Group, in collaboration with Regenstrief Institute, has developed a new data model designed to address some of these challenges in genomics reporting and LOINC mapping of the data elements. The structured genetic data can be used to report simple, complex, structural, germline, somatic, and copy number variants, partial DNA sequencing, whole genome and exome studies, mosaicism, mitochondrial DNA variation, and pharmacogenomics studies, as indicated in the HL7 Version 2.51 Lab Results Interface Implementation Guide STU Release 3 from June 2018.⁴⁶  For each of the standardized data elements and sections contained within the clinical report, LOINC panels have been created to facilitate their easy integration into current LIS and EHR 2.51 messaging functionality in facilitating personalized medicine. Feedback from laboratorians is integral to shaping and modifying this evolving process to evaluate whether the data model and requests for LOINC codes can keep pace with the complexity and rapid rate of development of new genetic tests, methods, and variants and to scale with the rapidly evolving acquisition, interpretation, and reporting of genomic data.

There are limitations and challenges that diminish the utility of LOINC in CG for conveying orders and results of genomic testing methodologies such as next-generation sequencing or full-exome sequencing. This has ramifications for the expansion of personalized medicine, and contributes to the lack of HL7 interfaces available to use with most next-generation sequencing instruments and genomic software tools.⁴⁷  Guidance and specifications for reporting genomic data, as well as associated LOINC codes for many of the report elements, are available in the CG Implementation Guide developed by the HL7 CG Workgroup. In addition to the Implementation Guide for CG in version 2.51 format, FHIR resources for reporting CG results are also under development.

One of the main goals of LOINC is to facilitate the exchange and pooling of results for clinical care, outcomes management, and research, thereby improving interoperability among disparate information systems and reducing the fragmentation of data that often happens when test orders and results are coded with local systems.³⁹  The LOINC codes convey key features about an analyte's measurement defined by the 6 axes/parts, including the scale and kind of property measured, the time aspect of the measurement, and the system (specimen type), with the additional option to select either a methodless code or a code with the method specified. The process of mapping local codes to LOINC codes is facilitated by the code mapping tool RELMA, which like LOINC is also free of charge. Requests for new codes can be made to Regenstrief Institute, the curator of the LOINC database, through a simple process, and updates to LOINC are made biannually.

As a large, curated, and freely available system of standardized codes, LOINC has become the federally required code for conveying laboratory observations. Regulations that impact LOINC may change in concert with the changing political setting. The regulatory landscape has already largely moved beyond the previous quota-style LOINC requirements of MU stage 2. The Advancing Care Information category within MIPS supersedes and replaces MU for eligible providers. Support of LOINC is still required for vendor certification of EHR products to the 2015 edition final rule, and ONC encourages EHR vendors to obtain this certification because it signals that the vendor's product can fulfill the reporting requirements included in MIPS, and MU stage 3 for hospitals.

The LOINC system has notable limitations and is subject to valid criticisms. First, it can be challenging to map the appropriate code for laboratory tests, and the task of code mapping ideally requires a laboratorian with the time and resources who is versed in both LOINC and laboratory medicine, or, at the very least, an individual in the laboratory who has experience in information technology. Second, the use of LOINC does not guarantee seamless interoperability of laboratory test names and codes in electronic interface messages between different information systems. The implications of this fact are substantial, particularly when persons not familiar with the domain of pathology and laboratory testing are making decisions regarding management of laboratory data within and across systems and organizations. Third, even institutions with expertise in LOINC fail to correctly map all of their local laboratory result codes to the correct LOINC codes.³⁰  Finally, LOINC has been criticized for its limited applicability in molecular and anatomic pathology.⁴⁷ 

The LOINC system is one of the most widely available vocabulary standard for categorizing and coding medical laboratory observations in health care organizations and data sets. Use of LOINC has been endorsed by the US government for use in HIT as part of federal programs. Understanding of the challenges associated with the use of LOINC in real-world settings is crucial to its effective and safe deployment.