Leveraging an Innovation Model to Facilitate ICD-11 Implementation

Abstract

The 11th Revision of the International Classification of Diseases (ICD-11) with its informatics-based infrastructure has transformed an antiquated classification system into a suite of 21st century computer applications. This manuscript proposes an innovation model to facilitate the implementation of ICD-11 by the US. The model introduces ICD-11 Comprehensive Clinical Linearization, Evolution and Response, or C-CLEAR, a fully coded comprehensive clinical linearization and syntactical rules for combining these codes. These enhancements can be incorporated into electronic coding tools that enable clinical reporters to transmit complex clinical concepts expressed in detailed natural clinical language by means of standardized clusters of ICD-11 stem and extension codes. The model can support rich clinical data captures such as condition acuity and severity, as well as pharmacological treatments. This approach shows promise to accelerate ICD-11 implementation with minimal disruption and maximal net benefits but will require vetting, testing and input from expert stakeholders.

Keywords: ICD-11, ICD-10, ICD-10-CM, International Classification of Diseases, ontology, morbidity, interoperability, health information exchange, episode of care, value-based healthcare

Introduction

In 2007, the World Health Organization (WHO) began a revision and restructuring of the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) to transform this classification system into a flexible clinical and research friendly structure aligned with advances in information technology. The 11th revision was endorsed by the World Health Assembly at its 72nd meeting in 2019 for implementation beginning in January 2022.¹

With the release of ICD-11 and its associated architecture, the National Committee on Vital and Health Statistics (NCVHS) began its stakeholder engagement around adoption of ICD-11 for morbidity in the US. Their work in 2019 and 2022 resulted in a set of recommendations to the Department of Health and Human Services (DHHS), ^2-4 including that DHHS should conduct research to evaluate the impact of different approaches to implementing ICD-11.³  Listed as the most important action for this recommendation was an assessment to determine whether ICD-11 can fully support morbidity data collection without the development of a US clinical modification (CM), and if not, which areas might require a CM version (or US-specific extension codes).³ These recommendations provided a framework from which to create an innovation model to streamline ICD-11 implementation in the US.

In response to the NCVHS recommendations, this manuscript proposes an innovation model to facilitate a seamless transition to ICD-11 by the US. The model introduces ICD-11 Comprehensive Clinical Linearization, Evolution and Response (C-CLEAR), a fully coded comprehensive clinical linearization along with syntactical rules for combining these codes that can translate detailed natural clinical language into standardized coded patient records that exploit the significant advantages of ICD-11.

Background

It was not until 2015, or 25 years after it was endorsed by the WHO, that the US implemented ICD-10-CM for morbidity data collection. While ICD-10-CM was seen as an improvement over the 9th Clinical Modification of ICD,^5,6 a system used since 1979, the implementation was considered highly disruptive and time-consuming, and added significant financial and administrative burdens on physicians and other healthcare providers.^7-10

More than four years have passed since the 11th revision was endorsed by the World Health Assembly.¹ WHO also has ceased updates to ICD-10.³ As of February 2023, 64 Member States are in different stages of ICD-11 implementation.¹¹ Compared to ICD-10, Harrison et al.¹² noted in their review that, “ICD-11 is a different and more powerful health information system, based on formal ontology, designed to be implemented in modern information technology infrastructures, and flexible enough for future modification and use with other classifications and terminologies.”

New in ICD-11, all clinical concepts are included in the Foundation, which is a medical knowledge base organized in a poly-hierarchy (in which an entity can descend from more than one branch or parent) that identifies relationships or connections among the entities.¹³ Foundation entities of interest are extracted based on use case to form a subset (called a linearization) from the Foundation in the form of a single hierarchy of entities and a corresponding code set. A linearization is the means by which ICD-11 would be accessed by most users, and in the US it ideally would provide backward compatibility to ICD-10-CM.¹⁴ Should each specialty move forward with its own linearization, this would tend to reinforce silos rather than promote integrated information systems and would not readily support use cases that require access to comprehensive and precise information across several or all clinical domains.

The WHO has provided a linearization or tabular list of codes, that is, ICD-11 for Mortality and Morbidity Statistics (MMS), as a potential system for countries to implement or transition in cases where an ICD-10 modification exists. Developed countries that have been using customized modifications of ICD-10 (e.g., ICD-10-CM in the US, -AM in Australia and parts of Asia, and -CA in Canada) are finding that MMS has gaps.¹⁵ The application of ICD-11 codes differs substantially from the ICD-10 coding process. ICD-10-CM consists of tens of thousands of precoordinated codes from which a coder selects the best match to suit the situation. Although MMS has many fewer individual stem codes than ICD-10-CM, a coder can express a clinical concept of interest by using a single stem code if that is sufficient or can form a post-coordinated cluster of stem and extension codes that are combined to capture a more complex concept. Code clusters can represent millions of different clinical scenarios, far surpassing the extent of any library of precoordinated codes. Theoretically, ICD-11 can deliver code expressions for most or all such clinical concepts in clinical modifications of ICD-10. Hindrances include the gaps in MMS, general unfamiliarity among stakeholders with “post-coordination” (clustering and adjoining codes to describe a clinical scenario), and the absence of a sanctioned syntax to provide robust discipline and consistency in the formation of post-coordinated code clusters. A further hindrance is widespread recollection of the transition from ICD-9 to ICD-10 with its fanfare and promised benefits, which failed to materialize in the minds of many observers.

Recognizing the NCVHS recommendations and ICD-11’s potential, the authors created a prototype innovation model with a volunteer group of professionals representing medicine, informatics, healthcare data, computer technology, analytics, performance evaluation, economics, and payment. This group developed the innovation model described in this manuscript as a novel approach that could facilitate implementation of ICD-11 by taking full advantage of ICD-11’s informatics-based infrastructure and architecture and thereby streamlining transition.

Methods

Comprehensive Code Set

C-CLEAR is an expansion of MMS in which every ICD Entity available in the ICD-11 Foundation is assigned a code. The Foundation is a multidimensional collection of all WHO-Family of Classifications (WHO-FIC) entities.¹⁶ An ICD entity represents a concept, such as a disease, disorder, sign or symptom, or extension code and is assigned a unique Uniform Resource Identifier (URI).¹⁷

To retain MMS as a common basis for C-CLEAR, all MMS blocks and codes were retained. Each Foundation URI in Chapters 1 through 25 that was included in an aggregated “other specified” code ending in the letter Y was assigned a sequential code in its appropriate series. These new C-CLEAR codes were demarcated with a terminal subscript “underscore CCL” (__CCL). For example, ICD-10-CM has a specific code representing the ICD-11 index term “Hyperplasia, maxillary.” In MMS, this clinical entity does not have its own code and is lumped into a Y (other specified) code.

ICD-11 MMS

DA0E.0 Major anomalies of jaw size 

DA0E.00 Micrognathia 

DA0E.0Y Other specified major anomalies of jaw size

whereas C-CLEAR has a specific code for maxillary hyperplasia by expanding on MMS and thus providing a one-to-one map back to ICD-10-CM.

ICD-11 C-CLEAR

DA0E.03__CCL Hyperplasia maxilla

Foundation URI: http://id.who.int/icd/entity/1336634664

Other specialty linearizations of ICD-11 also have placed new specialty-specific codes based on Foundation URIs in the series nested under the appropriate clinical concept in MMS. The advantage of C-CLEAR is that it has already incorporated all such potential codes, making it optimal and user-friendly for all coders regardless of specialty or clinical perspective. Offering C-CLEAR codes for each entity addresses the limitations of MMS where “other unspecified” codes mark the points where details are truncated. If all such C-CLEAR modifications were eliminated, the result would be MMS.

Composite Linearization

An academic concern might be that providing access to all clinical concepts does not conform to a restriction that is expected for linearizations, which is a single hierarchy that permits each child entity in the linearization to have only one parent. In other words, only one hierarchical set of relationships can be viewed at a time and all other valid branches or pathways that connect concepts in the Foundation are ignored. For example, MMS classifies salmonella pneumonia as a type of infectious or parasitic disease, while in the Foundation, it is both a type of infectious disease and a type of pneumonia. Similarly, in MMS, amebic abscess of the liver is classified as an infectious or parasitic disease but not as a disease of the liver although both are included in the Foundation's poly-hierarchy of clinical concepts.

To augment MMS and overcome this limitation, a composite linearization was created. All the hierarchical relationships residing in the Foundation, and unique C-CLEAR codes for all concepts, are made accessible to users according to their needs. The codes and relationships available in MMS are set as default values. However, alternative parent-child pairs and logical pathways are available whenever those are easier or clearer representations of the patient’s situation from the perspective or specialty of the clinician describing the patient.

Going back to our previous examples, an infectious disease specialist might focus on treatment options for amebic abscess or salmonella pneumonia as well as other manifestations of those bacteria in the patient (e.g., other organs or body systems). Meanwhile, the hepatologist and pulmonologist might address the respective body systems and organs holistically, including the abscess or the pneumonia, with secondary mention of the underlying external causes being addressed by the infectious disease specialist.

Clinical Language Syntax

Another component of the C-CLEAR innovation model is its clinical language syntax. MMS imposes the rules of a statistical classification on ICD-11’s richly expressive ontology. In contrast, C-CLEAR’s syntax is designed to enable clinical reporters to indicate the intended ancestry of each stem code used in a cluster, starting with a primary stem code that best captures the clinical reporter’s condition of interest. It then enables the clinical reporter to diverge from the reference MMS taxonomy by introducing a method of designating where and how a stem code’s ancestry deviates from MMS’s taxonomy. This feature of C-CLEAR enables a clinical reporter to communicate his or her clinical message utilizing the linguistic power of the entire ICD-11 ontology.

Another feature enables C-CLEAR to create uniquely ordered clusters of codes for complex concepts, mimicking the one-to-one relationship of a clinical concept to a single pre-coordinated code. C-CLEAR and its syntax follow ICD-11 MMS conventions including the use of stem codes as clinical concepts and extension codes as modifiers. Each unique C-CLEAR code can be mapped to a single ICD-11 Foundation URI. 

Results

Over the past year, the authors and their collaborators have made progress with the conceptual logic and an instantiation of the proposed innovation model. This includes the creation of C-CLEAR and its syntax.

We also have created clinical scenarios demonstrating the capability of C-CLEAR and its syntax to provide clinically credible representations of the detailed evolution of patients’ health status. These are intended to capture the clinical justification or appropriateness of medical interventions, document important changes in patients’ health in response to medical care, and provide representations superior to anything to date based on either ICD-9-CM or ICD‑10-CM.

Table 1 illustrates a simple medical scenario that compares the descriptive power of ICD-9-CM, ICD-10-CM, and C-CLEAR. The table depicts the six stages of the clinical progression of a female patient who first presents with aortic valve insufficiency due to aortic dilation and eventually is referred for a cardiac valve operation. The issue addressed here is how clearly the patient scenario is captured by each of the disease classification systems.
 

1. ICD-9-CM and ICD-10-CM. From ICD-9-CM, one can conclude this is a patient with aortic valve disease and thoracic aortic dilation. In ICD-10-CM, we know more specifically her condition was nonrheumatic aortic insufficiency (with the incremental details italicized in the table). Later, the patient had hyperpotassemia due to adverse effects from an antihypertensive agent, described in ICD-10-CM as hyperkalemia due to an ACE inhibitor. This complication apparently resolved. By stage four of this vignette, the patient had developed congestive heart failure (CHF). It is unclear why an aortic valve replacement was ultimately indicated and why it was recommended in stage six rather than in stage four or stage five, all of which appear identical in the coded data.
    
2. ICD-11 C-CLEAR. From C-CLEAR one learns that this patient had chronic mild aortic valve insufficiency due to thoracic aortic dilation that progressed from mild to moderate, after which it was treated with valsartan. This treatment resulted in hyperkalemia, so her medication was changed to benazepril. Her hyperkalemia resolved, but she developed chronic New York Heart Association (NYHA) Class II CHF as a complication of her aortic valve insufficiency. Treatment with furosemide resulted in lessening of her heart failure to NYHA Class I. However, her underlying aortic valve insufficiency progressed from moderate to severe, with a marked worsening of her chronic heart failure to NYHA Class III. The indications and timing for an aortic valve replacement are now made clear.

Precoordinated ICD-9-CM and ICD-10-CM codes, while lacking in important clinical detail, are easily interpretable representations. Because each accessible concept is represented by a single code, the challenge for coders is to identify which of a plethora of codes comes closest to the concept a clinical reporter wishes to convey. In contrast, C-CLEAR codes and syntax permit clinical reporters to convey nuanced clinical information in single, unique, systematically organized clusters of codes. However, these computer-friendly codes and clusters are not readily decipherable by a general clinical audience.

Fortunately, clinicians can generate and decipher C-CLEAR clusters with only rudimentary knowledge of the coding system itself. To make this possible, coding tools based on those created by the WHO to support ICD-11 MMS are being created. These enhanced coding tools will be able to translate ‘natural clinical language’ into C-CLEAR coded clusters. These clusters can support sophisticated analyses of the evolution of the health of individuals and populations and of the appropriateness, quality, and cost-effectiveness of diagnostic and therapeutic interventions and the care provided by healthcare practitioners and organizations. These tools also will be capable of transforming C-CLEAR clusters back into natural clinical language to allow clinicians to determine how well their clinical information has been captured, to revise their original input as needed to improve C-CLEAR coding, and to generate documentation consistent with coded data submitted for reporting, evaluation, and reimbursement.

Discussion

Accurate diagnoses are an essential element in providing appropriate and timely care to patients. However, as illustrated in Table 1, a wide range of clinical states can exist within single diagnostic categories. Curing diseases and reducing patient burden from diagnosed conditions are both essential elements of high-quality medical care. Similarly, patient-centered quality measurement, analytics, and fair payment of healthcare providers all require knowledge about each patient’s diagnoses, general health status, and related functional and socioeconomic factors as addressed in ICD-11, along with detailed information about the progression and regression of individual diagnosed conditions.

Furthermore, risk-adjustment based solely on diagnosed conditions without clinical details regarding the severity and complexity of these conditions can be anemic at best, or even misleading when systematic biases are present. This is a fatal flaw in the current data used for quality comparisons, performance evaluations, and alternative payment models.

Finally, the rationale and appropriateness for medical treatment and management as embodied in clinical guidelines require details available only in patient records. The adoption of ICD-11 could upgrade standard claims databases from catalogs of diagnoses, procedures, and costs to genuine clinical and research tools with new applications to monitor, improve, and pay for healthcare.¹⁸ Moreover, melding the EHR data and standardized claims data could eliminate current administrative redundancies.

The US is investing heavily in developing and supporting information technology innovation models. Several government agencies have established programs and provided funding for projects to accelerate the next generation of interoperable health information technology. For example, the Centers for Medicare & Medicaid Services (CMS) established the CMS Innovation Center to support development and testing of innovative healthcare payment and service delivery models.¹⁹ In 2020, the Centers for Disease Control and Prevention (CDC) launched the data modernization initiative intended to modernize core data and surveillance infrastructure across the federal and state public health landscape.²⁰ The Office of the National Coordinator for Health Information Technology (ONC) Leading Edge Acceleration Projects (LEAP) in Health Information Technology (IT) provides funding for projects that support the adoption of health IT and the promotion of nationwide health information exchange (HIE). ONC recently issued a Special Emphasis Notice stating that it “is critical that the field of health care innovate and leverage the latest technological advancements and breakthroughs far quicker than it currently does to optimize real-time solutions, especially in areas which are ripe for acceleration.”²¹ Our proposed innovation model would complement and enrich these initiatives.

The next logical step is to pilot test the innovation model. The NCVHS August 3^rd meeting discussed the need to pilot test ICD-11 options prior to implementation.²² In addition, WHO has indicated interest in pilot testing.²³ We welcome the opportunity to vet our approach among clinical and classification experts and to test C-CLEAR’s codes, architecture, syntax, and associated electronic coding tools for their stated purposes.

It is hoped this system would:

1. facilitate efficient user-friendly coding that completely and accurately captures the clinical information clinical reporters wish to convey,
2. produce output that is intuitively obvious to clinicians,
3. support the generation of coded ICD-11 data directly from EHRs including free text, and
4. enable analysts to manipulate these coded data to support important use cases.

C-CLEAR and its syntax will also serve as a framework for upgrading the Episode Grouper for Medicare (EGM),²⁴ which was developed in response to a provision in the Affordable Care Act²⁵ that directed CMS to create a public sector episode grouper with standard definitions of clinical conditions and procedures to support analyses, reimbursement, and other applications by all healthcare stakeholders. Work is currently underway to create a standard set of clinically nuanced episodes of care that will take full advantage of enhanced ICD-11 data capabilities to support the exchange, interpretation, and application of information among healthcare providers and other stakeholders. This upgraded episode grouper could facilitate a wide variety of extremely useful applications and replace some ineffective, inefficient analytic and operational applications that appear to be creating as many or more problems than they were designed to solve.

Conclusion

Unlike the transition to ICD-10, the adoption and implementation of ICD-11 would represent a major advance in medical informatics. The transformation of a collection of words into an architecture and syntax, a global language of sorts, enhances available information beyond diagnosed conditions to include how clinical progression within diagnoses and progressive interactions among diagnoses in different states affect a patient’s overall health status. ICD-11 also comes with a suite of 21st century computer applications that can be enhanced to support easy adoption, meaningful data sharing, and improved patient care.

When ICD-11’s informatics-based infrastructure is utilized to its fullest extent within the proposed innovation model, there is great potential to support clinically useful evaluations of the evolution of the health status of individual patients and populations and the contribution of alternative healthcare services to health and well-being. For example, C-CLEAR and its syntactical rules for combining these codes could conceivably become a universally applicable translator of parochial terms into a language that retains the important clinical details required to accurately monitor each patient’s clinical pathway. In addition, interaction with EHRs to locate and code clinical details, add key information to claims, increase interoperability, and invigorate many applications such as quality reporting and value-based payments may be possible. Using the model in such a manner could transform an inefficient healthcare payment and disjointed service delivery system into a cost-effective, coordinated, patient-centered healthcare ecosystem. And finally, the proposed approach shows promise of a faster and smoother transition to ICD-11, reduced administrative burden, seamless electronic healthcare information exchange, increased interoperability of electronic health information, and facilitation of a wide range of applications to foster the evaluation and improvement of the quality and cost-effectiveness of healthcare.

And finally, a next step is for the US and other countries should be to rigorously test ICD-11 linearizations for their ability to meet the many demands these countries have for accurate information in clinical care, clinical research, and secondary data use cases related to public health and policy. Specifically, the US needs to develop and embrace an approach that will justify an expensive Federal mandate to adopt ICD-11 via legislation or regulation. This research program should pilot test the implementation process by integrating ICD-11 into realistic health information technology environments and informing the industry with guidance and lessons learned on “how to” adopt ICD-11.

Furthermore, resulting data sets should be used to address the question of “why” adopt ICD-11. For example, with ICD-11, researchers could simulate potential net benefits to be expected in important use cases such as accurately and reliably measuring efficiency and quality of care. While the US should not rush to adopt ICD-11 merely because its developers wish it would, it should not remain stuck on the aging ICD-10-CM if it can do better, nor should it postpone meaningful reforms to accommodate stakeholders that are prospering despite its inability to achieve a sustainable, high-value healthcare system. 

Acknowledgements

The authors thank the following individuals for their contribution during manuscript preparation: Charles Hobson, Denise Love, Greg Wozniak, and John Martin

References

1. “The 72^nd World Health Assembly resolution for ICD-11 adoption | WHO.” May 27, 2019. https://www.who.int/publications/m/item/eleventh-revision-of-the-international-classification-of-diseases-adoption-wha72

2. “Subcommittee on Standards – ICD-11 Evaluation Expert Roundtable Meeting.” August 6-7, 2019. National Committee on Vital and Health Statistics. https://ncvhs.hhs.gov/meetings/subcommittee-on-standards-icd-11-evaluation-expert-roundtable-meeting/

3. “Recommendation to HHS Secretary on Preparing for Adoption of ICD-11.” November 25, 2019. National Committee on Vital and Health Statistics. https://ncvhs.hhs.gov/wp-content/uploads/2019/12/Recommendation-Letter-Preparing-for-Adoption-of-ICD-11-as-a-Mandated-US-Health-Data-Standard-final.pdf

4. “Updated recommendations for immediate action on ICD-11 to HHS Secretary.” September 10, 2021. National Committee on Vital and Health Statistics. https://ncvhs.hhs.gov/wp-content/uploads/2021/09/NCVHS-ICD-11-recommendations-for-HHS-Sept-10-2021-Final-508.pdf

5. “Transitioning to ICD-10 | CMS.” February 25, 2015. CMS. https://www.cms.gov/newsroom/fact-sheets/transitioning-icd-10

6. “ICD-10-CM - International Classification of Diseases, ICD-10-CM/PCS) Transition | CDC.” November 6, 2015. CDC. https://www.cdc.gov/nchs/icd/icd10cm_pcs_background.htm

7. Libicki, Martin and Irene Brahmakulam. March 2004. “The costs and benefits of moving to the ICD-10 code sets.” Rand Science and Technology. https://www.rand.org/content/dam/rand/pubs/technical_reports/2004/RAND_TR132.pdf

8. Deloitte. 2010. “ICD-10 implementation for health care providers: The business imperative for compliance. https://www.healthit.gov/sites/default/files/facas/us_lshc_icd-10implementationforhealthcareproviders_0810.pdf

9. Fiegl, Charles. “Organized medicine urges CMS to halt ICD-10 switch.” January 7, 2013. American Medical News. Accessed April 18, 2023. https://amednews.com/article/20130107/government/130109986/7

10. Nachimson Advisors, LLC. February 12, 2014. “The cost of implementing ICD-10 for physician practices – updating the 2008 Nachimson Advisors study.” https://docs.house.gov/meetings/IF/IF14/20150211/102940/HHRG-114-IF14-Wstate-TerryW-20150211-SD001.pdf

11. World Health Organization “ICD-11 2023 release is here.” February 14, 2023. https://www.who.int/news/item/14-02-2023-icd-11-2023-release-is-here

12. Harrison, James E., Stefanie Weber, Robert Jakob, and Christopher G. Chute. 2021. “ICD-11: an International Classification of Diseases for the twenty-first century.” BMC Medical Informatics and Decision Making, 21 (S6). https://doi.org/10.1186/s12911-021-01534-6

13. Tu, Samson W. Olivier Bodenreider, Can Çelik, Christopher G. Chute, Sam Heard, Robert Jakob, Guoquian Jiang , Sukil Kim , Eric Miller , Mark M. Musen , Jun Nakaya, Jon Patrick, Alan Rector, Guillermo Reynoso, Jean Marie Rodrigues, Harold Solbrig, Kent A Spackman, Tania Tudorache, Stefanie Weber, and Tevfik Bedirhan Üstün. 2015. “A content model for the ICD-11 revision.” https://getinthepicture.org/resource/content-model-icd-11-revision

14. Chute, Christopher G. and Can Çelik, 2022. “Overview of ICD-11 architecture and structure.” BMC Medical Informatics and Decision Making, 21 (Suppl 6), 378. https://doi.org/10.1186/s12911-021-01539-1

15. Fung, Kin W., Julia Zu, Shannon McConnel-Lamptey, Donna Pickett, and Olivier Bodenreider. 2023. “A practical strategy to use the ICD-11 for morbidity coding in the United States without a clinical modification.” Journal of the American Medical Informatics Association. July. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10531107/  

16. WHO-FIC Content Model Reference Guide, World Health Organization (WHO) 2021 https://icd.who.int/browse11. Licensed under the Creative Commons Attribution-NoDerivatives 3.0 IGO licence (CC BY-ND 3.0 IGO)

17. Ibid.

18. Fenton, Susan H., Kathy L. Giannangelo, and Mary H. Stanfill. 2021. “Preliminary study of patient safety and quality use cases for ICD-11 MMS.” Journal of the American Medical Informatics Association. September. https://doi.org/10.1093/jamia/ocab163

19. “CMS Innovation Center Homepage | CMS Innovation Center.” 2015. Cms.gov. 2015. https://innovation.cms.gov

20. “Data Modernization Initiative Basics | CDC.” 2023. www.cdc.gov. January 9, 2023. https://archive.cdc.gov/#/details?url=https://www.cdc.gov/surveillance/data-modernization/basics/index.html

21. “Leading Edge Acceleration Projects (LEAP) in Health Information Technology (Health IT) Notice of Funding Opportunity (NOFO) | HealthIT.gov.” April 10, 2023. www.healthit.gov. https://www.healthit.gov/topic/onc-funding-opportunities/leading-edge-acceleration-projects-leap-health-information

22. “Workgroup on Timely and Strategic Action to Inform ICD-11 Policy – ICD-11 Expert Roundtable Meeting.” August 3, 2023. National Committee on Vital and Health Statistics. https://ncvhs.hhs.gov/wp-content/uploads/2023/10/2023-August-3_NCVHS_ICD-11-Expert-Roundtable-Meeting-Summary-FINAL-508.pdf

23. WHO-FIC ICD-11 Implementation or Transition Guide, World Health Organization (WHO) 2019  https://icd.who.int/en/docs/ICD-11%20Implementation%20or%20Transition%20Guide_v105.pdf. Licenses under the Creative Commons Attribution (CC BY-NC-SA 3.0 IGO)

24. Tompkins, Christopher P. 2016. Episode Grouper for Medicare (EGM): Design Report. https://i11forum.org/EGM_Final_Report.pdf

25. The Patient Protection and Affordable Care Act. 2010. §§3003, 3007. https://www.govinfo.gov/content/pkg/PLAW-111publ148/pdf/PLAW-111publ148.pdf