ICD9, ICD10 and SNOMED, a guest blog

Knowing that yesterday's blog about ICD-10 would raise questions about how HITSP will incorporate ICD-10 into its future work products as well as the role of SNOMED verses ICD-10 as a clinical vocabulary, I asked one of our HITSP Technical Committee Co-chairs to give me his perspective. Jamie Ferguson leads standards efforts for Kaiser and he gave me his permission to publish his personal thoughts on ICD9, ICD10 and SNOMED, which provide helpful background about the issues. Jamie wrote:

"On the point that SNOMED and ICD-10 have the potential to be competing vocabularies, Kaiser Permanente's position is that ICD-10-CM/-PCS and SNOMED-CT are not competing vocabularies. As CMS noted in its NPRM, ICD-10 is a hierarchical classification system for billing and administrative purposes, whereas SNOMED is a knowledge-based ontology for clinical documentation and clinical decision-making purposes. SNOMED-CT is recommended for the US private sector and mandated for federal agencies for clinical documentation purposes as a result of its selection in the CHI initiative of OMB's e-Gov program. A one-way authoritative mapping from SNOMED-CT clinical documentation to the US ICD-10 billing codes is a current project and has been requested to be made official before the ICD-10 compliance date, i.e. published by an authoritative source such as NLM and cited by CMS, in comments on the NPRM from several multi-stakeholder organizations. Also, perhaps most importantly, for clinically-relevant analysis and clinical decision making the value of inference based operations such as subsumptive queries should not be underestimated. Therefore, KP would advise HITSP to make use of an authoritative mapping of clinical documentation (SNOMED) to administrative classification (ICD-10) when it is available and to include it as needed for its scenario solutions.

A further note on this was submitted to HHS by Kaiser Permanente as part of our public comments on the NPRM:

According to CMS, the benefits of transitioning to the ICD-10 code sets will include: 1) more accurate payments for new procedures; 2) fewer rejected or improper claims, i.e., fewer supplemental information requests will be required to support the medical necessity of claims and the details of procedures to be reimbursed; and 3) better understanding of new procedures. These improvements are aimed at providing more consistency in how conditions and treatment are captured for billing purposes because of the increased level of detail and specificity in the new code sets and are appropriately within the scope of the ICD-10 code sets adoption (73 FR 49821-23).

However, CMS has also identified potential benefits that go beyond the primary administrative purposes of the new ICD-10 code sets. These include improved disease management by payers, and better understanding of health conditions and health care outcomes. More specifically, CMS envisions using the ICD-10 codes within claims data to support a broad array of population health research and quality initiatives, such as: outcomes analysis, quality assessments, bio-surveillance, chronic care management, registries (including immunization registries), etc. (73 FR 49821, 49823-25). For these purposes, the SNOMED-CT is preferred because it is a knowledge-based ontology capable of inferences and subsumptive queries (e.g., “find all disorders that include “kidney””) whereas ICD is not, which among other reasons make SNOMED CT the coding system of choice for clinical documentation. Moreover, SNOMED CT codes can also be used for billing and reporting purposes without being inappropriately manipulated as CMS suggests (73 FR 49803). In fact, inference-based query systems render them more useful than billing classification systems for these reporting and analytical purposes related to population health and quality programs.

There is a distinction between coding for billing, which is relatively simple and coding for clinical documentation and decision-making, which is complex by comparison. As a result, we developed our clinical systems based on the SNOMED CT clinical terminology, which was specifically designed to support clinical decision-making and interoperability. For these reasons, we strongly support the continued use of SNOMED CT for these clinically relevant purposes and urge CMS to refrain from mandating the exclusive use of ICD-10 code sets for clinical purposes. We also note that using ICD-10 code sets for some of the cited purposes may conflict with the existing recognized federal standard, such as coding for the chief complaint or reason for visit. Those standards require SNOMED-CT in certain EHR and personal health record (“PHR”) systems, for electronic laboratory reporting, and for bio-surveillance, including public health reporting.

CMS also mentions that adoption will lead to harmonization of disease monitoring and reporting worldwide. However, these ICD-10 code sets are uniquely U.S. versions, so whether such harmonization is achievable is questionable.

HHS and OMB recommend adoption of SNOMED CT as the preferred coding system for clinical documentation in the Consolidated Health Informatics initiative within OMB’s e-Gov program, and SNOMED CT is the required clinical terminology standard in certain recognized HITSP Interoperability Specifications such as IS-01. "

Thus a strategy of using SNOMED CT for clinical observations such as problem lists while using ICD-10 for billing makes a great deal of sense. I imagine that the National Library of Medicine and vendors will develop products that will help turn SNOMED clinical documentation into accurate ICD-10 billing codes to streamline workflow and ensure appropriate coding.