Clearing the Fog Surrounding LOINC

Beginning 15th March 2020, submissions of studies for New Drug Applications (NDAs), Abbreviated New Drug Applications (ANDAs), and certain Biologic License Applications (BLAs) will be required to include Logical Observation Identifiers Names and Codes (LOINC®) as specified by the Food and Drug Administration (FDA) Data Standards Catalog with some INDs (Investigational New Drug) having the same requirement from 15th  March 2021. These standards will allow for succinct interoperability between clinical data systems. 

While seemingly problematic from  the outset, LOINC provides a rich database for clinicians and researchers to draw from. This ensures that once standards are applied to their data, local or proprietary terms can be easily transferred to other institutions, allowing for a swift exchange of data.

Each LOINC term is comprised of up to six subsections, where the function of each subsection provides a unique identity for every test in the LOINC database. The components that make each term unique will be elaborated upon and a few examples provided of differing results that could be contained within each.  

The global scope of LOINC and the growing community that is becoming increasingly prominent in data standards governance makes this a fast-evolving and exciting area of clinical development.  

LOINC’s purpose

First established in 1994 by the Regenstrief Institute, LOINC codes are developed and maintained in a freely available database including both medical laboratory test names and clinical observations. The key issue that LOINC aims to remedy is that the aggregation of many different sources of data may inherently contain a myriad of ways to identify the same tests. 

While a specialist may be able to observe these and interpret that they all refer to the same measurement, a computer has no such knowledge. Local computer systems can apply test names that can be viewed as ambiguous when transferred across to other systems; the naming convention of one system may not necessarily be the same as another. They can use different methods of identifying the same test or measurement which can create difficulties with data exchange which could potentially impact the speed at which data can be interpreted by these systems. The key idea behind the inception of LOINC is to provide interoperability between different systems; allowing communication, exchange of data and use of the information that has been exchanged. The final result of this implementation is that clinical data can be available when and where it is needed without the complication caused by differing systems local codes being confused with each other.

LOINC term

A LOINC term can be considered to be a representation of a question about clinical phenomena that can be observed or measured. There are two essential pieces to a LOINC term: the LOINC Code and the LOINC Fully Specified Name (FSN).

The LOINC Code is a unique permanent identifier that acts as a computer processable representation of a LOINC term. The LOINC code itself is not intended to be understood by humans; the codes are assigned sequentially, so only infer the relative age in the database. Once these codes are published, they are never removed from the database and are permanently available. Linked to each code is the LOINC Name which is the human-readable text version of the LOINC code. 

The FSN’s fundamental purpose is to include the essential defining characteristics of a term and is most useful for mapping to LOINC. It is designed to include all information necessary to distinguish between clinically important differences and is essential to defining the uniqueness of each LOINC term. The FSN is comprised of five or six components that will be described in more detail in the next section of this paper. It makes use of abbreviations in naming that could be difficult to interpret for a clinician unfamiliar with the terminology. In addition, there are 2 other main LOINC Name types that can be used in other contexts, but are no substitute for the FSN in providing the defining characteristics of a term:
Long Common Name (LCN)
Short Name

The LCN can be thought of as the primary display name of a LOINC term; a more descriptive and easily interpretable name than the FSN that would be loaded into a database. It removes abbreviations and uses common names for a general display of a term that allows a reader to understand a test name without needing to read the FSN for the full details of the test.

The Short Name is mainly used for reporting purposes, e.g. in column headers or report labels in systems that have character length limitations. The target character limit for the Short Name is 40, although due to the nature and complexity of some tests requiring multiple components, this limit is not always practicable. Not all LOINC terms feature a Short Name and as a result, the Short Name cannot be used as an identifying key in any database.

In development is a fourth and fifth name type that aim to be more easily understood. These names are currently known as Display Name and Consumer Name and are in beta and alpha stages respectively at the time of writing. Display Name aims to be a provider-friendly name and Consumer Name is intended for the consumer. The criteria surrounding the consideration of what makes a name ‘provider friendly’ and ‘consumer-friendly’ are currently being investigated, hence the early stages of development for these names.


These are available for viewing when searching LOINC codes, but, as with the Short Name, not all LOINC codes currently have these names.

FDA inclusion

A LOINC working group recommendation for the submission of LOINC codes with new studies specifies that, while LOINC codes can be broadly categorised into laboratory codes and clinical codes, the FDA will only require Laboratory codes to be submitted with new regulatory applications.

The LOINC laboratory codes are generally considered to be most closely in line with the existing standards in the CDISC SDTM LB domain. A key difference between LOINC and the LB domain is that LOINC is pre-coordinated and CDISC LB is post-coordinated.

A pre-coordinated set of data is considered to be static, i.e., the information presented contains all the information necessary to be used in an application. A post-coordinated set of data has the components required and these can be compiled together as needed.

With this in mind, one LOINC term can represent a concept through the components described above. These are unchanging once the LOINC code is applied; the correct LOINC code can only be applied to one concept. In contrast, the LB domain contains several dimensions that are needed for accurate identification of a test and its results that are compiled as required.

As such, LOINC and the LB domain variables share characteristics in their results that can coincide with each other. While the LB.LBLOINC variable is currently already supported in the LB domain, its inclusion is currently permissible, but will be required for new studies after 15th March 2020.

The current mapping between the LOINC parts and the CDISC LB domain variables are shown below:

Table 1. A comparison of LOINC Parts to CDISC Laboratory (LB) Domain Variables

LOINC Part
(entire column represents a single LOINC code)

CDISC Standard Equivalent
(each row represents a single CDISC code)

Component

LBTEST/CD + other variables

Property

-

Time

MULTIPLE: Various Timing Variables

System

SPEC+LOC

Scale

-

Method

METHOD + other variables

Source: https://www.fda.gov/media/109376/download

However, these two systems are not always aligned in this way.

The LOINC components of Scale and Property do not have a comparable analogue in the LB domain.

Additionally, the LB.LBLOINC variable should be taken from the original result (LB.LBORRES) and not the converted form i.e. LB.LBSTRES(C/N). This is consistent with the working group recommendation that LB.LBLOINC is taken from the original laboratory result and that the LOINC code should be applied during the creation of the raw data.

Global adoption

LOINC version 2.  is available and features over 91,000 terms. This is the result of a continual input by the LOINC community that comprises over 81,000 registered users from 175 countries with over 20 translations made available.

There are major releases for LOINC twice per year in June and December that allow for a continually evolving database of codes. These could be requested due to a number of reasons e.g. new lab tests to take into consideration, new data models and even information that had previously been unstructured being transitioned to structured formats that need new LOINC codes to be added to accommodate these new formats.

Development is propelled forward by the LOINC community as new terms are added at the request of the users. As these new terms are continually added to the database, the collection becomes greater and encompasses more of an institution’s local test names. This allows for even greater adoption of the standard and enhances the usability of the standard as this occurs.

The only major prohibition stipulated for the use of LOINC is that it cannot be used to develop a new set of order or observation standards to provide the same function that LOINC does. This would defeat the purpose of creating the standard in the first place; the goal being to establish one standard with one code for these observations.

The working group established by the FDA may make additional recommendations as required for topics such as: requests for inclusion of LOINC codes for tests that are not routine, or inclusion of non-human lab LOINC codes.
Additionally, the complete alignment of the pre-coordinated data models from LOINC and the post-coordinated models from the CDISC LB domain may be considered for future developments.

LOINC’s goal to unify laboratory and clinical data results across institutions is enabling separate institutions to make fast comparisons of their results that would potentially be delayed and more difficult to decipher without the standard applied. This interoperability goal of LOINC thereby allows a patient’s data to be shared quickly and allows studies to progress at a faster and more efficient rate than without.

The rapidly increasing LOINC membership coupled with increasing number of requests for new terms across the globe signifies that this standard is becoming more widely adopted; with the increasing number of codes, the greater the ease at which the codes can be implemented by institutions into their own databases. This ensures that the LOINC database becomes more valuable as the database expands.

Development for LOINC is a continuous process and encompasses such enhancements as updates to the database and website allowing for faster searching and returning of term results, and improvements to allow for a more streamlined approach when requesting new terms.

LOINC’s active community engagement allows for suggestions to be provided by its members that can then be implemented to make processing or viewing of LOINC codes more efficient. LOINC’s development is further driven by a user survey for suggested improvements that is conducted twice annually.

In summary, LB.LBLOINC will become a required variable within FDA submission datasets, but the codes contained therein are not intended to be understood by humans. Instead the LOINC Names are designed to provide information that is interpretable by humans. The LOINC FSN provides the full definition of a term that is split into five or six component parts. It is recommended that sponsors do not derive mappings to LOINC for the creation of SDTM files. This could lead to serious data issues without confirming with the test performer due to the ambiguity surrounding lab test names. PHASTAR are currently preparing for compliance with this requirement by compiling and disseminating this information to the wider community.

For further information regarding LOINC, please see the references section of this paper, these will be able to provide a deeper insight into the intricacies surrounding LOINC as well as the integration of LOINC in tandem with other data standards.