EDITOR’S NOTE: The following article is part of an occasional series on ICD-11.
ICD-11 has been in development since 2007 but will not be ready for use in the United States, by some estimates, until the year 2023 or later. ICD-11 is an important undertaking, as its precursor, ICD-10, is the most widely used code set in the world. The World Health Organization (WHO) has elected to modernize ICD so that it will be suitable for use “with electronic health applications and information systems.”
This will increase the value of ICD-11 for patient care, population health, and clinical research. The legacy features of ICD-10 and earlier versions of ICD will largely be preserved, but ICD-11 will have some of the features found in computer-friendly terminologies such as SNOMED CT.
The WHO entered into a collaborative agreement with the International Health Terminology Standards Development Organization (IHTSDO) in 2010. IHTSDO is the entity that owns and maintains SNOMED CT. Multiple parties are working together to create a version of ICD (ICD-11) that will differ significantly from prior versions of ICD. It will incorporate features and content from SNOMED CT.
This article will provide a high-level overview of the status of ICD-11, its potential benefits, and potential challenges that may be associated with its eventual implementation.
In order to understand ICD-11, it is important to have an understanding of SNOMED CT (an acronym for the Systematized Nomenclature of Medicine – Clinical Terms). SNOMED CT has some similarities to ICD-10, but these two code sets were developed for different purposes. They have unique structures and significant differences in their levels of complexity.
ICD-10-CM has over 65,000 codes that cover a wide range of healthcare concepts. Many of the codes in ICD-10-CM represent “classifications,” meaning they represent a group of similar concepts. SNOMED CT is significantly larger than ICD-10-CM, with over 300,000 concepts. Each concept can only have one meaning. Synonyms (called “descriptions” in SNOMED CT) are strictly defined but number over 600,000. SNOMED CT also has over 1,000,000 relationships between concepts referred to as “attributes.” In order to manage this level of complexity, SNOMED CT was formed through use a derivative of artificial intelligence known as description logics.
The size and complexity of SNOMED CT make it a powerful tool for clinical data capture, population health management, and medical research. It is also gradually evolving into a common international language for healthcare information systems, in particular for disorders and clinical findings. However, with some notable exceptions, efforts to implement SNOMED CT have faced significant headwinds in the U.S. and other countries. SNOMED CT requires a high level of sophistication to implement and use. The sheer number of terms and complex interrelationships has a tendency to impact its use at the point of care. In most cases, implementers have been required to customize SNOMED CT in a manner that allows for only some of its content to be displayed during use in clinical care. Only a limited number of healthcare organizations have been able to take full advantage of SNOMED CT’s ability to capture combinations of codes that represent even moderately complex clinical expressions (e.g., “doubtful multiple sclerosis”).
A subset of SNOMED CT is now required to be used as an alternative to ICD-9-CM codes in problem lists in electronic health records (EHRs) under the meaningful use of EHR incentive program in the U.S. SNOMED CT’s role is anticipated to expand in U.S. healthcare systems under the EHR incentive program.
ICD-11 Development Process Overview:
ICD-11 is being developed through a cooperative effort under direction of the WHO. The goal of the WHO/IHSTDO collaborative agreement was to form a modern terminology optimized for clinical information systems. Both parties intend to preserve the fundamental features and value of ICD and SNOMED CT. However, some of the differences between these two code sets (e.g., inclusion and exclusion rules in ICD and the use of description logics in SNOMED CT) were reportedly difficult to harmonize. For this reason ICD-11 will be significantly different from both ICD-10 and SNOMED CT. Once ICD-11 is released SNOMED CT will continue to have value as an independent code set.
ICD-11 will overcome some of the complexities inherent to SNOMED CT by providing what are referred to as “linearizations” that flatten the SNOMED CT hierarchies. However, as work continues it will be interesting to see how ICD-11 will address many of the important underpinnings of SNOMED CT, including strict concept representation, strict use of synonyms, and its broad range of relationship types between concepts.
An ICD-11 beta draft is available to the public, and the WHO has invited all stakeholders to participate in its development. Once the development of ICD-11 has been completed (scheduled for 2017), it will need to undergo a quality assurance process. Once this has been completed, ICD-11 will need to be vigorously tested in a variety of healthcare settings. Feedback from these test implementations will likely be used to refine ICD-11. The U.S. may or may not need to create a “clinical modifications” version of ICD-11, as the ICD-11 development process is including terms added to ICD-10 to form ICD-10-CM. The ICD-11 leadership team is encouraging the U.S. to incorporate features of ICD-11 into ICD-10-CM to help with the eventual migration from ICD-10-CM to ICD-11.
Implementation of ICD-11:
While it is difficult to forecast what future implementations of ICD-11 will look like, the incorporation of features from SNOMED CT has the potential to make this process challenging and resource-intensive. The level of use of SNOMED CT in the U.S. in information systems by 2025 could, in theory, make the migration from ICD-10-CM to ICD-11 relatively seamless, but this represents little more than speculation at this writing.
ICD-10-CM and SNOMED CT have fundamental differences that are being addressed within the ICD-11 development process. ICD-11 will not, by most estimates, be ready for use in the U.S. healthcare for at least 7-10 years. ICD-11 will be fundamentally different from both ICD-10/ICD-10-CM and SNOMED CT, and as a new code set, it will require vigorous testing in a variety of operational settings. The implementation of ICD-11 will be a significant undertaking for U.S. healthcare that will require system designers, implementers, and users to have an understanding of SNOMED CT, ICD-10, and features unique to ICD-11.
The leadership of the ICD-11 development process has stressed the need for U.S. healthcare systems to adopt ICD-10-CM as soon as possible. ICD-10 is an essential platform for migrating to ICD-11 and represents a marked improvement over ICD-9-CM. Once ICD-11 development and testing efforts have been completed, the U.S. will be in a position to consider the adoption of ICD-11, a code set that has the potential to markedly benefit healthcare and medical research.
Disclaimer: The information contained within this article represents the opinions of Michael Stearns, MD.
About the Author
Dr. Stearns is the CEO of Apollo HIT, LLC, a healthcare information technology and compliance organization. He played a role in the development of SNOMED CT.
Contact the Author
Comment on this Article