Oncology
ConventionsConventions
TODO: Brief overview of the use and intent of these conventions
Cancer diagnoses are recorded within the CONDITION table, ideally with ICDO3 concepts. Specific attributes or details about that condition which are not contained within the concept are attached to these condition records as MEASUREMENTS, referred to as Condition Modifiers, which are described below.
Under Construction
(Placeholder)
What we are calling ‘Condition Modifiers’ are observations/findings/attributes about a cancer and are stored as Measurements using the Cancer Modifier vocabulary.
A linkage between the Condition Modifier and the respective condition is made leveraging the two columns in the Measurement table that serve as a foreign key. This same mechanism is used for procedure modifiers as well. The first (meas_event_field_concept_id) indicates which table we are referring to, the second (measurement_event_id) indicates which row.
There are concepts for every potential PK that a measurement can potentially modify.
Vocabulary ID : CDM Domain Id : Metadata Concept Class Id : Field
For condition modifiers, we are pointing to the primary key of the condition table, which is condition_occurrence.condition_concept_id (https://athena.ohdsi.org/search-terms/terms/1147127)
An example condition modifier record within the measurement table would contain:
OHDSI Oncology WG SELECT
...
1147127 as meas_event_field_concept_id -- static for all condition modifiers
temp.condition_occurrence_id as measurement_event_id -- the condition_occurrence_id to which this Cancer Modifier refers to Complete
Condition Modifier
The date of initial diagnosis can be represented in the OMOP CDM by mapping to the correct concept in the Cancer Modifier vocabulary and represented as a MEASUREMENT, or rather, a cancer modifier. This can be an important data point from the source to distinguish this particular date as the most reliable diagnosis date of a patient’s cancer and can, depending on various factors, substantially vary in a non-trivial way from other diagnosis dates in the CONDITION_OCCURRENCE table for the same disease occurrence.
There are specific circumstances in which there is a higher level of certainty that the date contained within a condition record is an accurate representation of the initial diagnosis date. Some examples would be from tumor registries or manually curated or verified data. In those cases, to distinguish these records as the source of truth, we assign a condition modifier of ‘Initial Diagnosis’ (Concept_id = 734306) to that condition record, recorded as a measurement.
If the source data record in question is not confirmed to reliably represent the initial diagnosis date, such as from typical EMR records, the modifier is not used.
| Metadata | Value |
|---|---|
| Concept Name | Initial Diagnosis |
| Domain ID | Measurement |
| Vocabulary ID | Cancer Modifier |
| Concept ID | 734306 |
| Concept code | init_diag |
Adding this condition modifier would look something like:
INSERT INTO MEASUREMENT
SELECT
...
734306 as measurement_concept_id -- "Initial Diagnosis" ******
1147127 as meas_event_field_concept_id -- static for all condition modifiers
temp.condition_occurrence_id as measurement_event_id -- the condition_occurrence_id to which this Cancer Modifier refers to Finding conditions which have the “initial diagnosis date” flag (modifier) can be done by:
SELECT *
FROM CONDITION_OCCURRENCE co
WHERE CONDITION_OCCURRENCE_ID IN (
SELECT measurement_event_id
FROM measurement
WHERE measurement_concept_id = 734306 -- Initial diagnosis
AND meas_event_field_concept_id = 1147127 -- Condition_Occurrence
)Complete
Condition Modifier
Circumvent the concept mappings and map directly to the standard concepts within the ‘Cancer Modifier’ vocabulary by concatenating your source data and joining on the concept code.
The concept code has 4 components that the ETL developer will need to concatenate to match. Those components are:
The concept code format is : [c/p]-[version]th_AJCC/UICC-[TNM Value]
For example : c-7th_AJCC/UICC-T1c2
Above: TNM value of T1c2, clinically staged, defined by the 7th version of AJCC/UICC
To find the correct standard concept, you need to join to the concept code, as shown here:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'c-7th_AJCC/UICC-T1c2'There are multiple methods of navigating concept hierarchies for TNM depending on the use case.
Using M1a for example:
Find all concepts that fall under “M1a”, regardless of version or clin vs. path
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1635100 -- AJCC/UICC M1a CategoryFind all concepts that fall under “M1a”, regardless of version but ONLY path
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1634312 -- AJCC/UICC Pathological M1a CategoryFind all concepts that fall under “M1a”, are pathological only, and are only using the 7th version
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1634268 -- AJCC/UICC 7th Pathological M1a CategoryLastly, you can search for the concept directly given the structured concept codes. For the same example as above:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'p-7th_AJCC/UICC-M1a' -- Pathological + version + codeComplete
Condition Modifier
Circumvent the concept mappings and map directly to the standard concepts within the ‘Cancer Modifier’ vocabulary by concatenating your source data and joining on the concept code.
The concept code has 4 components that the ETL developer will need to concatenate to match. Those components are:
The concept code format is : [c/p]-[version]th_AJCC/UICC-Stage-Stage Group
For example : c-7th_AJCC/UICC-Stage-3A1
Above: Stage Group of 3A1, clinically staged, defined by the 7th version of AJCC/UICC
To find the correct standard concept, you need to join to the concept code, as shown here:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'c-7th_AJCC/UICC-Stage-3A1'There are multiple methods of navigating concept hierarchies for Stage Group depending on the use case.
Using Stage Group 3A1 for example:
Find all concepts that fall under “3A1”, regardless of version or clin vs. path
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1635412 -- AJCC/UICC Stage 3A1 CategoryFind all concepts that fall under “3A1”, regardless of version but ONLY path
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1633303 -- AJCC/UICC Pathological 3A1 CategoryFind all concepts that fall under “3A1”, are pathological only, and are only using the 7th version
SELECT descendant_concept_id
FROM concept_ancestor
WHERE ancestor_concept_id = 1635016 -- AJCC/UICC 7th Pathological 3A1 CategoryLastly, you can search for the concept directly given the structured concept codes. For the same example as above:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'p-7th_AJCC/UICC-Stage-3A1' -- Pathological + version + codeComplete
Condition Modifier
Map directly to the concepts in the Cancer Modifier vocabulary. The standard concepts for grade are broken down into 1-4 as well as High, Intermediate, and Low.
A few specific grading systems (e.g. Nottingham, FIGO) are represented as standard concepts in the Cancer Modifier vocabulary. Map to these concepts if they match the granularity of your data. If you have a specific need for a different grading system, you can request that it be standardized by creating an issue.
For example: Grade 1
Above: Any tumor that has been assigned a Grade 1 in any grading system
To find the correct standard concept, you need to join to the concept code, as shown here:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'Grade-1'Another example: Gleason Score 8
Above: A tumor assigned a Gleason Score of 8 using the Gleason grading system
To find the correct standard concept, you need to join to the concept code, as shown here:
SELECT concept_id
FROM concept
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND concept_code = 'Gleason-Score-8'To find all standard Grade concepts in the Cancer Modifier vocabulary:
SELECT b.concept_id, b.concept_code, b.concept_name
FROM (
SELECT *
FROM concept c
INNER JOIN concept_ancestor ca
ON c.concept_id = ca.ancestor_concept_id
WHERE vocabulary_id = 'Cancer Modifier'
AND concept_class_id = 'Staging/Grading'
AND standard_concept = 'S'
AND lower(concept_code) LIKE '%grade%'
) a
INNER JOIN concept b
ON a.descendant_concept_id = b.concept_idUnder Construction
Condition Modifier
Caveat
TODO
HIJKLM
NOPQ
Under Construction
Condition Modifier
Caveat
TODO
HIJKLM
NOPQ
Under Construction
(Placeholder)
Caveat It is currently under discussion as to whether or not records of metastases will be represented as both a MEASUREMENT and a CONDITION, and once it has this documentation will be updated, but we are sure that we need the MEASUREMENT representation which is detailed below. Additionally, there are concept mappings being worked on by the Vocabulary team to de-standardize NAACCR and map those concepts to our destination representation within the Cancer Modifier vocabulary.
The standard Metastasis concepts can be found within:
| Metadata | Value |
|---|---|
| Domain ID | Measurement |
| Vocabulary ID | Cancer Modifier |
| Concept Class ID | Metastasis |
There is an overarching concept of ‘Metastasis’ (https://athena.ohdsi.org/search-terms/terms/36769180) . For one, this is important as it is a destination standard concept for sources where the data is vague to the point that you only know metastasis occurred. A secondary usage of this would be to find all of the standard concepts in which fall under it.
TODO
To find all concepts in that hierarchy:
SELECT descendant_concept_id -- The various concepts under the metasis hierarchy
FROM CONCEPT_ANCESTOR
WHERE ancestor_concept_id = 36769180When it is known which CONDITION the metastasis occurrence refers to, it is represented linked similarly to that CONDITION record using the Cancer Modifiers mechanisms.
For example, a record of ‘Metastasis to Lung’ would look like:
INSERT INTO MEASUREMENT
SELECT
...
36770283 as measurement_concept_id -- "Metastasis to Lung" ******
1147127 as meas_event_field_concept_id -- static for all condition modifiers
temp.condition_occurrence_id as measurement_event_id -- the condition_occurrence_id to which this Cancer Modifier refers to Under Construction
Caveat
These will be implemented similarly to the Metastasis approach, as Cancer Modifiers or MEASUREMENTS, but we are currently waiting on a revision of the concepts within this class of the Cancer Modifier vocabulary before mapping and providing examples as there is an absence of hierarchical relationships.
HIJKLM
NOPQ
Under Construction
Treatment representation is currently being finalized and this documentation will be updated accordingly
Treatment records are similar to conditions in that the concepts themselves do not contain all of the relevant information and there is a need for ‘modifiers’ to append additional information to these records, which we refer to as ‘Treatment Modifiers’
Under Construction
Representation:
(TODO)
Examples: - type - site - margins - topography - intent - …
Under Construction
Scope
Includes but not limited to:
Representation: (TODO)
(TODO)
Examples:
Under Construction
Representation:
(TODO)
Examples: