Handling ICU scores without standard OMOP concept IDs in the INDICATE Data Dictionary

[BorisDelange]

Handling ICU delirium and pain scores without standard OMOP concept IDs in the INDICATE Data Dictionary

Context

The INDICATE Data Dictionary includes several ICU clinical scores for delirium screening and pain assessment that do not have standard concept IDs in the OMOP Standardized Vocabularies (ATHENA). These scores are well-established, validated clinical instruments widely used in intensive care, but they are currently absent from the OMOP vocabulary.

Affected concept sets

Score	Full Name	Key Reference
3D-CAM	3-Minute Diagnostic Interview for CAM-defined Delirium	Marcantonio et al., Ann Intern Med, 2014
4AT	4 'A's Test for Delirium Screening	Bellelli et al., Age Ageing, 2014
BPS	Behavioral Pain Scale	Payen et al., Crit Care Med, 2001
CAM-ICU	Confusion Assessment Method for the Intensive Care Unit	Ely et al., JAMA, 2001
DDS	Delirium Detection Score	Otter et al., Neurocrit Care, 2005
DOS / DOSS	Delirium Observation Screening Scale	Schuurmans et al., Res Theory Nurs Pract, 2003
DRS-R-98	Delirium Rating Scale – Revised-98	Trzepacz et al., J Neuropsychiatry Clin Neurosci, 2001
ICDSC	Intensive Care Delirium Screening Checklist	Bergeron et al., Intensive Care Med, 2001
Nu-DESC	Nursing Delirium Screening Scale	Gaudreau et al., J Pain Symptom Manage, 2005

All these instruments are clinically validated and widely used in ICU practice.

Problem statement

Since these scores have no standard concept_id in the OMOP vocabulary, using concept_id = 0 means we are not using standard OMOP concepts for these clinical instruments. This is a deviation from the OMOP CDM standards, which limits interoperability and prevents these data from being included in standard network studies across the OHDSI community.

Proposed approach (transitional)

Following the OHDSI Custom Concepts convention and the Themis Working Group recommendations, we propose:

1. Set *_concept_id = 0 in the standard concept ID fields (e.g., measurement_concept_id), indicating that no standard OMOP concept exists.

2. Assign custom concept IDs > 2,100,000,000 in the *_source_concept_id fields (e.g., measurement_source_concept_id), following the OHDSI convention that custom concepts use IDs above the 2 billion threshold. We deliberately start at 2.1 billion to reduce the risk of collision with local custom concepts already used by data providers who may have started their own numbering at 2,000,000,000.

3. Define a dedicated INDICATE custom vocabulary (e.g., vocabulary_id = INDICATE) in the local VOCABULARY table, and register all custom concepts under this vocabulary in the CONCEPT table with:

   • standard_concept = NULL (non-standard)
   • Appropriate domain_id (likely Measurement)
   • Clear, unambiguous concept_name and concept_code

4. Define all custom concept IDs centrally in the INDICATE shared data dictionary to ensure all data providers use the same IDs — avoiding collisions between sites.

Example

concept_id	concept_name	domain_id	vocabulary_id	concept_class_id	concept_code
2100000001	3D-CAM score	Measurement	INDICATE	Clinical Observation	INDICATE_3DCAM
2100000002	4AT score	Measurement	INDICATE	Clinical Observation	INDICATE_4AT
2100000003	Behavioral Pain Scale score	Measurement	INDICATE	Clinical Observation	INDICATE_BPS
...	...	...	...	...	...

Important caveats

This is a transitional solution

• The range above 2 billion is a shared space: other data providers who have already performed OMOP ETL may use overlapping ranges for their own local custom concepts. The specific sub-range chosen by INDICATE (e.g., starting at 2,100,000,000) must be coordinated across all participating sites.
• Custom concepts are non-standard by definition: they can only be used in *_source_concept_id fields and will not be available in standard OMOP network studies without additional mapping.

Need for engagement with OHDSI

We strongly recommend engaging with the OHDSI Vocabulary Working Group to:

1. Request the addition of these ICU scores to the official OMOP Standardized Vocabularies. These are widely used, validated clinical instruments that would benefit the broader OHDSI community — not just INDICATE.
2. Discuss a long-term convention for ICU-specific concepts that are not yet represented in LOINC, SNOMED-CT, or other standard terminologies.

Key OHDSI resources

Resource	URL
Vocabulary GitHub repository	https://github.com/OHDSI/Vocabulary-v5.0
Vocabulary WG wiki	https://github.com/OHDSI/Vocabulary-v5.0/wiki
Custom Concepts documentation	https://ohdsi.github.io/CommonDataModel/customConcepts.html
Contact the Vocabulary Team	https://ohdsi.github.io/CommonDataModel/contactVocab.html
OHDSI Forums — Vocabulary Users	https://forums.ohdsi.org/c/vocabulary-users
Vocabulary office hours (1st Thursday, 9am EST)	https://forums.ohdsi.org/t/vocabularies-office-hours/23295
OHDSI Workgroups overview	https://www.ohdsi.org/workgroups/

Action items

• Finalize the list of custom concept IDs for the 9 scores in the INDICATE shared data dictionary
• Ensure all data providers reference the same concept ID mapping
• Open a request on the OHDSI Forums or Vocabulary GitHub to propose inclusion of these ICU scores in the standard vocabulary
• Attend OHDSI Vocabulary office hours to discuss this use case