Diagnosis/Procedure Code Formatting

[NickKramer87]

As a data analyst, I want to see diagnosis and procedure codes that are in the same format as real CA hospital data so that they are interchangeable.

Proposed Subtasks:

1. Alter the output of the Synthea database generator so that all codes can be displayed in ICD-10 format (not SNOMED).

Acceptance Criteria:

1. Diagnosis and procedure codes are the same for real data and synthetic data.

[TravisHaussler]

As of the close of Phase 2 we have:

• Diagnosis are basic mapped using the first 1-1 map for each SNOMED code (real map is 1-many based on additional patient criteria)
• Procedure are not mapped at all, but passed through as SNOMED codes and also truncated to fit the field lenght (i.e. they are unusable at this point). This is because a basic map for procedure codes similar to the one for diagnosis codes could not be found.

There are some vague leads on how to achieve this task properly, but they involve finding a way to access the JSON maps used in this project (we hit a wall accessing the AWS repo for their project on last attempt):

• https://github.com/CMSgov/beneficiary-fhir-data

Some discussion on why the procedure map is so difficult:

• https://forums.ohdsi.org/t/snomed-ct-to-icd-10-pcs-mapping/10512

[rileeki]

Quick update: Synthea won't provide the JSON maps for us because of licensing issues (per their answer to my question on the discussion board.

However, they provided some guidance for how we can create those files ourselves.

The mapping files are structured like this:

{
  "70536003": [ // synthea SNOMED code
    {
      "code": "XXXXX", // first mapped CPT code
      "description": "description of code",
      "weight": "10" // all weights are summed and then individual weights are used to drive the distribution of codes so this one is twice as likely to be used as the next one
    },
    {
      "code": "YYYYY", // second mapped CPT code
      "description": "description of code",
      "weight": "5"
    },
    ...
  ],
  ...
}

And, to just get the code running, we can create empty map files:

If you create a set of empty mapping files (just "{}") and then run the exporter, it will generate a missing_codes.csv file that will give you a list of all of the codes that need to be mapped in each of the files.

I think the next step is probably to create the empty map files and run Synthea with the bfd export option on and see what happens.

[rileeki]

@masonium This is the issue we talked about on Wednesday.

Summary

• Synthea outputs procedures and diagnosis codes in SNOMED-CT format.
• The California Inpatient data we are trying to emulate uses ICD10-CM codes for diagnoses and ICD10-PCS codes for procedures. (Note: when we move to other claim types like Emergency Department and Ambulatory Services, the data will use CPT codes for procedures... more on that later.)
• Travis has implemented a translation of the SNOMED diagnosis codes to ICD10-CM using a mapping from UMLS.
• We have not been able to find a mapping of SNOMED procedures to ICD10-PCS.
• You might have more success doing your own research on SNOMED to ICD10... While this post is long, don't consider it well-informed or comprehensive.
• See Travis's notes above, too!

Introduction to Billing Code Systems

CPT® (Current Procedural Terminology)

The CPT coding system describes how to report procedures or services. The CPT system is maintained and copyrighted by the American Medical Association. Each CPT code has five digits. The AMA CPT Editorial Panel reviews and responds to requests for additions to or revisions of the CPT.

HCPCS (Healthcare Common Procedures Coding System)

HCPCS codes are used to report supplies, equipment, and devices provided to patients. A limited number of procedures not otherwise contained in the CPT system are also found here. HCPCS is alphanumeric and is administered by the Centers for Medicare and Medicaid Services (CMS) in cooperation with other third party payers.

CMS includes two levels in its Healthcare Common Procedures Coding System: HCPCS Level I is the CPT coding system; HCPCS Level II is usually referred to as HCPCS codes, described above.

ICD-10-CM (International Classification of Diseases, 10th revision, Clinical Modification)

Healthcare professionals use these codes to report diagnoses and disorders. The ICD-10-CM is maintained by the National Center for Health Statistics (NCHS). The ICD-10-CM replaced the 9th revision (ICD-9-CM) on October 1, 2015.

UMLS (Unified Medical Language System®)

The UMLS, or Unified Medical Language System, is a set of files and software that brings together many health and biomedical vocabularies and standards to enable interoperability between computer systems.

You can use the UMLS to enhance or develop applications, such as electronic health records, classification tools, dictionaries and language translators.

I don't know much about this, except that Travis and I both got free UMLS licenses to be able to access a mapping from SNOMED-CT to ICD10-CM. We haven't been able to find a similar mapping for ICD10-PCS codes.

SNOMED

• SNOMED CT can be characterized as a multilingual thesaurus with an ontological foundation.

• SNOMED International has created maps from SNOMED-CT to other codes systems. I haven't looked at this much, but they do seem to have a mapping to CPT procedure codes.

ICD10-PCS

ICD10PCS, often spelled “ICD-10-PCS”, is a vocabulary in the USA for coding hospital-based medical procedures. It is currently implemented as a standard vocabulary, with OMOP-generated hierarchical relations to Procedure concepts from SNOMED CT vocabulary and other relationships mimicking SNOMED CT's internal model.

I'm unfamiliar with these codes. Usually I've encountered procedure codes as CPT/HCPCS. But the California data format has procedure codes in ICD10-PCS, so that's what we're looking for!

More links that might be useful

• Achieving Logical Equivalence between SNOMED CT and ICD-10-PCS Surgical Procedures (2017 paper)

  Surgical procedures are coded in SNOMED CT in the electronic health record and in ICD-10-PCS in administrative systems. We compared the logical definitions of SNOMED CT concepts to the ICD-10-PCS axial components to identify overlap and gaps.

• SNOMED and ICD: Aligning to Standards

  This post will give an overview of the SNOMED terminologies and ICD-10 classifications as we look at ways to normalize data between the two sets.

• WHO-FIC Information Sheet

  Through the ongoing collaboration between WHO and IHTSDO, ICD and SNOMED CT are being strengthened by joint efforts to improve the process of encoding clinical and administrative data and support evidence based care. A comparison of the two systems is found on page 2.

• OHDSI

  The Observational Health Data Sciences and Informatics (or OHDSI, pronounced "Odyssey") program is a multi-stakeholder, interdisciplinary collaborative to bring out the value of health data through large-scale analytics. All our solutions are open-source. OHDSI has established an international network of researchers and observational health databases with a central coordinating center housed at Columbia University.

• SEMI-AUTOMATIC INCORPORATION OF ICD-10-PCS CONCEPTS INTO A COMPREHENSIVE PROCEDURE HIERARCHY BASED ON SNOMED-CT IN THE OMOP COMMON DATA MODEL (2019 Poster)

  We present the process of automatically placing leaf-level ICD-10-PCS concepts in SNOMED-CT hierarchy using manually matched attribute sets as part of our efforts to consolidate procedures from different vocabularies into a singular hierarchical model. There are key differences in the construction of SNOMED-CT and ICD-10-PCS concepts. As a result, equivalence mapping was not considered an optimal way of constructing a combined hierarchical system. Placing ICD-10-PCS leaf-level concepts into existing SNOMED-CT hierarchy was chosen instead

  • referenced in this forum post: https://forums.ohdsi.org/t/snomed-icd10-procedure-relationships-in-vocabulary/3777/5