Under Construction

Representation


Diagnosis

The Cancer Diagnostic Model in the OMOP vocabulary is comprised of: * Cancer diagnosis: a condition that is a combination of histology (morphology) and topography (anatomic site). * Diagnostic schema: a group of cancer diagnoses with similar diagnostic features. * Diagnostic modifiers: a set of diagnostic features that define a group of cancer diagnoses. These include stage, grade, laterality, genomic biomarkers, and other attributes.

Cancer Diagnosis

  • Cancer diagnosis is stored in the CONDITION_OCCURRENCE table.
  • Diagnostic modifiers are stored in the MEASUREMENT table. They are explicitly linked to the cancer diagnosis record in CONDITION_OCCURRENCE via the columns MEASUREMENT.modifier_of_event_id and MEASUREMENT.modifier_of_field_concept_id: MEASUREMENT.modifier_of_event_id contains value of the respective condition_occurrence_id. MEASUREMENT.modifier_of_field_concept_id contains the concept for the condition_occurrence_id field (1147127).

Treatments

The Cancer Treatment Model in the OMOP vocabulary is comprised of: * Cancer treatments: a higher-level procedure or regimen concept that raises above the transactional and observational details of treatment planning, treatment delivery, and treatment billing. Cancer treatment concepts attempt to capture the overall modality of a therapeutic or diagnostic intervention.
* Treatment modifiers: a set of features that refine the description of a cancer treatment. These include: ‘Number of Fractions’, ‘Radiation Primary Treatment Volume’, ‘Total Dose’ and other attributes of a radiation therapy treatment; ‘Lymph nodes examined’, ‘Surgical Margins’ and other attributes of a surgery.

Cancer Treatment

  • Cancer treatment is stored in the PROCEDURE_OCCURRENCE and EPISODE table.
  • Treatment modifiers are stored in the MEASUREMENT table. They are explicitly linked to the cancer treatment record in PROCEDURE_OCCURRENCE via the columns MEASUREMENT.modifier_of_event_id and MEASUREMENT.modifier_of_field_concept_id: MEASUREMENT.modifier_of_event_id contains value of the respective procedure_occurrence_id. MEASUREMENT.modifier_of_field_concept_id contains the concept for the procedure_occurrence_id field (1147084).

Episodes

Clinically and analytically relevant representation of cancer diagnoses, treatments, and outcomes requires data abstraction. Examples include ‘Disease First Occurrence’, ‘Disease Recurrence’, ‘Disease Remission’, ‘Disease Progression’, ‘Treatment Regimen’, and ‘Treatment Cycle’.

-TODO-

  • The EPISODE.episode_object_id column refers to a concept identifier in the Standardized Vocabularies describing the disease, treatment, or other abstraction that the episode describes. Episode entries from the ‘Disease Episode’ concept class should have an episode_object_concept_id that comes from the Condition domain. Episode entries from the ‘Treatment Episode’ concept class should have an episode_object_concept_id that comes from the ‘Procedure’ or ‘Regimen’ domain.
  • The relationship between a disease episode and treatment episodes can be represented by the self-referencing foreign key column EPISODE.episode_parent_id. This allows for the attribution of cancer treatment to a cancer diagnosis to help support the use case of calculating time from diagnosis to treatment.
  • The relationship between a ‘Disease First Occurrence’ disease episode and a subsequent ‘Disease Progression’ episode can be represented by the self-referencing foreign key column EPISODE.episode_parent_id. This enables support for the use case of calculating interval time to progression.
  • A treatment EPISODE can be delivered at regular intervals, cycles or fractions. The parent-child relationship between a treatment episode and its constituent treatment cycles can be represented by the self-referencing foreign key column EPISODE.episode_parent_id.

  • Episode modifiers Episode modifiers are very similar to condition and procedure modifiers. For example, a ‘Disease First Occurrence’ of ‘Carcinoma of breast’ can have stage, grade, and other attributes of the ‘Carcinoma of breast’. A ‘Treatment Regimen’ of ‘External beam, NOS’ can have ‘Number of Fractions’, ‘Radiation Primary Treatment Volume’, ‘Total Dose’ and other attributes of ‘External beam, NOS’.

  • Connection between episodes and lower-level events (e.g. conditions, procedures, drugs) are represented in the linking EPISODE_EVENT table. In this table: event_id contains value of the respective lower level event (e.g. condition_occurrence_id) and ;episode_event_field_concept_id field contains concept for the corresponding field (e.g. concept ID 1147127 for ‘condition_occurrence_id field’).

  • Episode population may be accomplished by multiple strategies:
    • Directly from Tumor Registries: ETL Disease Episodes, Disease Episode Modifiers, Treatment Episodes and Treatment Episode Modifiers directly from tumor registry data. Tumor registries natively capture many of the abstractions and detailed modifiers that the Oncology CDM Extension strives to instantiate. A standard ETL has been written capable of ETLing NAACCR-formated tumor registry data.See the documentation of the NAACCR ETL here
    • Directly from non-registry sources: ETL Disease Episodes, Disease Episode Modifiers, Treatment Episodes and Treatment Episode Modifiers directly from EHR systems, ancillary clinical systems, Oncology Analytic Platforms and custom clinical data repositories. Some Oncology EMRs (Elekta MOSAIQ, Flatiron OncoEMR), the Oncology modules of standard EMRs (EPIC Beacon), ancillary clinical system (Pathology LIMS capturing CAP Cancer checklists via mTuitive CAP eFRM/xPert for Pathology), Oncology Analytic platforms ((Flatiron OncoAnalytics, COTA, TEMPUS) and custom clinical data repositories natively capture and/or curate the abstractions and detailed modifiers that the Oncology CDM Extension strives to instantiate.
    • Indirectly via Algorithmic Derivation on top of the OMOP CDM: Run algorithms on top of the OMOP standardized clinical data tables. For example, machine learning, rules-based compendium-aided temporal logic, and natural language processing. ETL Disease Episodes, Disease Episode Modifiers, Treatment Episodes and Treatment Episode Modifiers from the outputs of the algorithms.

Schema

MEASUREMENT

The Oncology CDM Extension adds two new fields (modifier_of_event_id and modifier_of_field_concept_id) to enable the MEASUREMENT table to represent modifiers.

Field Required Type Description
measurement_id Yes integer A unique identifier for each Measurement.
person_id Yes integer A foreign key identifier to the Person about whom the measurement was recorded. The demographic details of that Person are stored in the PERSON table.
measurement_concept_id Yes integer A foreign key to the standard measurement concept identifier in the Standardized Vocabularies. These belong to the ‘Measurement’ domain, but could overlap with the ‘Observation’ domain (see #3 below).
measurement_date No date The date of the Measurement.
measurement_datetime Yes datetime The date and time of the Measurement. Some database systems don’t have a datatype of time. To accommodate all temporal analyses, datatype datetime can be used (combining measurement_date and measurement_time forum discussion)
measurement_time No varchar(10) The time of the Measurement. This is present for backwards compatibility and will be deprecated in an upcoming version
measurement_type_concept_id Yes integer A foreign key to the predefined Concept in the Standardized Vocabularies reflecting the provenance from where the Measurement record was recorded. These belong to the ‘Meas Type’ vocabulary
operator_concept_id No integer A foreign key identifier to the predefined Concept in the Standardized Vocabularies reflecting the mathematical operator that is applied to the value_as_number. Operators are <, <=, =, >=, > and these concepts belong to the ‘Meas Value Operator’ domain.
value_as_number No float A Measurement result where the result is expressed as a numeric value.
value_as_concept_id No integer A foreign key to a Measurement result represented as a Concept from the Standardized Vocabularies (e.g., positive/negative, present/absent, low/high, etc.). These belong to the ‘Meas Value’ domain
unit_concept_id No integer A foreign key to a Standard Concept ID of Measurement Units in the Standardized Vocabularies that belong to the ‘Unit’ domain.
range_low No float The lower limit of the normal range of the Measurement result. The lower range is assumed to be of the same unit of measure as the Measurement value.
range_high No float The upper limit of the normal range of the Measurement. The upper range is assumed to be of the same unit of measure as the Measurement value.
provider_id No integer A foreign key to the provider in the PROVIDER table who was responsible for initiating or obtaining the measurement.
visit_occurrence_id No integer A foreign key to the Visit in the VISIT_OCCURRENCE table during which the Measurement was recorded.
visit_detail_id No integer A foreign key to the Visit Detail in the VISIT_DETAIL table during which the Measurement was recorded.
measurement_source_value No varchar(50) The Measurement name as it appears in the source data. This code is mapped to a Standard Concept in the Standardized Vocabularies and the original code is stored here for reference.
measurement_source_concept_id Yes integer A foreign key to a Concept in the Standard Vocabularies that refers to the code used in the source.
unit_source_value No varchar(50) The source code for the unit as it appears in the source data. This code is mapped to a standard unit concept in the Standardized Vocabularies and the original code is stored here for reference.
value_source_value No varchar(50) The source value associated with the content of the value_as_number or value_as_concept_id as stored in the source data.
modifier_of_event_id No bigint A foreign key identifier to the event (e.g. condition, procedure, episode) record for which the modifier is recorded.
modifier_of_field_concept_id No integer The concept representing the table field concept that contains the value of the event id for which the modifier is recorded (e.g. CONDITION_OCCURRENCE.condition_occurre nce_id).


EPISODE

The EPISODE table aggregates lower-level clinical events (VISIT_OCCURRENCE, DRUG_EXPOSURE, PROCEDURE_OCCURRENCE, DEVICE_EXPOSURE) into a higher-level abstraction representing clinically and analytically relevant disease phases/outcomes and treatments. The EPISODE_EVENT table connects qualifying clinical events (VISIT_OCCURRENCE, DRUG_EXPOSURE, PROCEDURE_OCCURRENCE, DEVICE_EXPOSURE) to the appropriate EPISODE entry.

Field Name Required Datatype Field Description
episode_id Yes bigint A unique identifier for each Episode event.
person_id Yes bigint A foreign key identifier to the Person who is experiencing the episode. The demographic details of that Person are stored in the PERSON table.
episode_concept_id Yes integer A foreign key that refers to a Standard Concept identifier in the Standardized Vocabularies belonging to the ‘Episode’ domain.
episode_start_datetime Yes datetime The date and time when the Episode begins.
episode_end_datetime No datetime The date when the instance of the Episode is considered to have ended.
episode_parent_id No bigint A foreign key that refers to a parent Episode entry representing an entire episode if the episode spans multiple cycles.
episode_number No integer An ordinal count for an Episode that spans multiple times.
episode_object_concept_id Yes integer A foreign key that refers to a concept identifier in the Standardized Vocabularies describing the disease, treatment, or other abstraction that the episode describes. Episode entries from the ‘Disease Episode’ concept class should have an episode_object_concept_id that comes from the Condition domain. Episode entries from the ‘Treatment Episode’ concept class should have an episode_object_concept_id that comes from the ‘Procedure’ or ‘Regimen’ domain.
episode_type_concept_id Yes integer A foreign key to the predefined Concept identifier in the Standardized Vocabularies reflecting the source data from which the Episode was recorded, the level of standardization, and the type of occurrence. These belong to the ‘Episode Type’ vocabulary.
episode_source_value No varchar(50) The source code for the Episdoe as it appears in the source data. This code is mapped to a Standard Condition Concept in the Standardized Vocabularies and the original code is stored here for reference.

EPISODE_EVENT

The EPISODE_EVENT table connects qualifying clinical events (VISIT_OCCURRENCE, DRUG_EXPOSURE, PROCEDURE_OCCURRENCE, DEVICE_EXPOSURE) to the appropriate EPISODE entry. The EPISODE_EVENT table supports the linkage of an EPISODE abstraction to the low-level clinical events that implement the EPISODE abstraction.

Field Required Type Description
episode_id Yes bigint A foreign key identifier to the Episode that the Episode Event belongs to.
event_id Yes bigint A foreign key identifier to the underlying event (condition, procedure, measurement, etc.) record in a respective table for which an episode is recorded.
episode_event_field_concept_id Yes int A foreign key identifier to the standardized concept corresponding to the table primary key column (condition_occurrence.condition_occurrence_id, procedure_occurrence.procedure_occurrence_id, measurment.measurment_id etc.) where the underlying event is stored.

CONCEPT_NUMERIC

CONCEPT_NUMERIC is an extension of the OMOP CDM and vocabulary that supports formal representation of concepts containing numeric values or ranges. This proposal has not yet been ratified by a larger CDM Workgroup. However, it plays a critical role in supporting ETL from tumor registries.

Background * NAACCR vocabulary includes concepts representing numeric values or numeric ranges. Often, these concepts also contain measurement units. For example, “Described as less than 1 centimeter (cm)”. In OMOP CDM, these concepts are normally used in Measurement and Observation tables to store value_as_concept_id. Analysis of these data is currently possible only if the user knows exactly which concepts are used to represent range or value, including their respective units. It is not possible to perform analysis on numeric values of these data, nor is it possible to differentiate numeric values by units. For example, it is not impossible to find tumors which size is less than 2 cm by issuing the following simple query:

    SELECT * 
    FROM measurement
    WHERE measurement_concept_id = 4139794  -- Tumor size
    AND ( (value_as_number <= 2 and unit_concept_id = 8582) -- Centimeter
    OR (value_as_number <= 20 and unit_concept_id = 8588)) --Millimeter
    AND COALESCE(operator_concept_id, 4171754) = 4171754 -- Less than or equal

One has to know exactly the concepts that cover this range.

Proposal 1

Add a new table, CONCEPT_NUMERIC

In this table, numeric values, units and math operators indicating range limits (less than) corresponding to “numeric” concepts will be stored.

Field Required Type Description
concept_id Yes integer A foreign key that refers to a respective concept in the Standardized Vocabularies.
value_as_number Yes float A value of the concept expressed as a numeric value.
unit_concept_id No integer A foreign key to a Standard Concept ID of the concept units in the Standardized Vocabularies that belong to the ‘Unit’ domain.
operator_concept_id yes float A foreign key identifier to the predefined Concept in the Standardized Vocabularies reflecting the mathematical operator that is applied to the value_as_number. Operators are <, <=, =, >=, > and these concepts belong to the ‘Meas Value Operator’ domain.

Proposal 2.

Conventions for representing “numeric” concepts in CONCEPT_NUMERIC

Numeric ranges that have lower and upper limits will be represented by two records in this table. Numeric ranges containing only lower or upper limit will be represented by one record.

Below are examples of representing the following “numeric” concepts in CONCEPT_NUMERIC:

concept_id concept_name domain_id vocabulary_id concept_class_id standard_concept concept_code
45883535 Mild Meas Value LOINC Answer S LA6752-5
35920053 Described as “less than 1 centimeter (cm)” Meas Value NAACCR Cancer Identification S @991
35919956 Described as “less than 2 cm,” or “greater than 1 cm,” or “between 1 cm and 2 cm” Meas Value NAACCR Cancer Identification S @992


CONCEPT_NUMERIC

concept_id value_as_number unit_concept_id operator_concept_id Description
45883535 1 1 is numeric value of ‘Mild’
35920053 1 8582 4171756 ‘Less than 1 cm’
35919956 1 8582 4171755 Lower limit of ‘between 1 cm and 2 cm’
35919956 2 8582 4171754 Upper limit of ‘between 1 cm and 2 cm’

Proposal 3.

Population of Measurement and Observation tables with the data represented by “numeric” concepts

When data are represented by “numeric” concepts, we recommend populating value_as_number, unit_id and operator_id (if available) along with value_as_concept_id. These additional values can be easily extracted from Concept_Numeric for the respective concept. Below is an example of Measurement record representing ‘Mild’ (45883535) ‘Pain intensity rating scale’ (4137083):

measurement_id measurement_concept_id value_as_concept_id value_as_number
1111111 4137083 45883535 1

In case, when a numeric concept represents a range with both lower and upper limits (e.g. ‘between 1 and 2 cm’), we recommend creating two records representing respectively lower and upper limit of the range. Values for value_as_number, unit_id and operator_id will be extracted from Concept_Numeric. Below is an example of Measurement record representing ‘Tumor size’ (4139794) ‘Described as “less than 2 cm,” or “greater than 1 cm,” or “between 1 cm and 2 cm”’ (35919956):

measurement_id measurement_concept_id value_as_concept_id value_as_number unit_concept_id operator_concept_id
2222222 4139794 35919956 1 8582 4171755
3333333 4139794 35919956 2 8582 4171754

This representation was also suggested in this Forum post: https://forums.ohdsi.org/t/how-to-implement-a-lab-test-range-result-in-measurement-table/5221

Proposal 4.

Queries to analyze data represented by “numeric” concepts When data represented by “numeric” concepts is stored in the Measurement table as described in the previous section, queries based on numeric values become possible against this data.

For example:

SELECT * 
FROM measurement
WHERE measurement_concept_id = 4139794  -- Tumor size
AND ( (value_as_number <= 2 and unit_concept_id = 8582) -- Centimeter
OR (value_as_number <= 20 and unit_concept_id = 8588)) --Millimeter
AND COALESCE(operator_concept_id, 4171754) = 4171754 -- Less than or equal

This query will identify records containing either regular numeric values (e.g. value_as_number = 0.5) or “numeric” concepts.


Vocabularies

  • Cancer Modifier (OMOP)
  • ICD-0
  • NAACCR
  • HemOnc
  • CAP
  • Genomic Vocabulary (OMOP)
  • Nebraska Lexicon