Clinical Data Validation Rules

Clinical Data Encoding Rules

Identifier fields

The data dictionary contains certain data elements regarded as "identifiers". These fields have an descriptor in the data dictionary and include:

• Donor: submitter_donor_id
• Specimen: submitter_specimen_id
• Sample: submitter_sample_id
• Primary Diagnosis: submitter_primary_diagnosis_id
• Treatment: submitter_treatment_id
• Follow Up: submitter_follow_up_id

These fields must be coded specifically for ICGC ARGO purposes using the following rules:

• These identifiers should not be derived from biobank or hospital identifiers or any other personal identifying information. These IDs are to be coded in such a way that they cannot be tracked back to the individual donors, except by the submitting program. Only the program will keep the key that will permit the data to be linked back to the individual donors. This key must not be communicated to the data users.
• Identifiers are assigned by each submitting program and must be unique within all the data submitted by that program (no duplicate IDs allowed).
• Identifiers referring to the same entity should be consistent across separate program submissions and should not be re-used for different entities. For example, the same donor should not be assigned different identifiers in different files or subsequent data submissions.
• Identifiers are case-sensitive.

Primary Diagnosis, Treatment and Follow Up Identifiers:

These identifier fields allow for linking across the different clinical events and should be coded using the following rules:

• Each primary diagnosis should be assigned a unique submitter_primary_diagnosis_id, so in the case where a donor has multiple primary diagnoses, each primary diagnosis should have a different submitter_primary_diagnosis_id. You will be required to submit the submitter_primary_diagnosis_id in the Specimen file - this provides information about which primary diagnosis the specimen is linked to. The submitter_primary_diagnosis_id is also required in the Treatment file, so it is understood which primary diagnosis the treatment is being administered for.
• Each treatment regimen in the Treatment file should be assigned a unique submitter_treatment_id. If the treatment regimen consists of chemotherapy, hormone therapy or radiation therapy, then you will use the same submitter_treatment_id in the appropriate Chemotherapy, Radiation or Hormone Therapy files. For example, a treatment regimen consisting of Chemotherapy and Radiation therapy is assigned cr01 as the submitter_treatment_id in the Treatment file. You would then submit the relevant clinical treatment information in the Chemotherapy and Radiation files using the same submitter_treatment_id (cr01) in those files. This allows the information in the two files to be linked together so it is understood that the two therapies were combined.
• Each follow up should be assigned a unique submitter_follow_up_id. Optionally, if a follow up is linked to a specific treatment, you may include the submitter_treatment_id for that follow up.

Time Intervals

To prevent potential identification of donors, actual calendar dates are not permitted. The timing of different clinical events are collected in days counted from the date of primary diagnosis. The date of primary diagnosis should be based on the earliest diagnosis of cancer. Validation checks are in place to ensure the values submitted for the different time interval fields make sense according to the following assumptions:

• The age_at_primary_diagnosis is used as the reference time point.
• The day the patient dies is the clinical endpoint (survival_time).

Examples of time interval validation checks:

• If a patient's vital_status is Deceased, all time intervals must be less than or equal to survival_time.
• The relapse_interval must be less than the interval_of_followup in the follow up entry that the relapse was recorded.
• If a follow up is associated with a particular treatment (via the submitter_treatment_id), the interval_of_followup must be greater than the treatment_start_interval.

Donors Older than 90 years old

Since the occurrence of individuals over the age of 90 is rare, it is therefore considered a potentially identifiable value. Thus, the allowed value for the age_at_diagnosis field is capped at 90.

Submitting Missing Values for Extended Clinical Fields

If reporting missing values is required for extended fields, data submitters must use the appropriate term, as defined below:

Term	Definition	Example
Unknown	A value that would be meaningful for analysis if observed, but is not available.	The ER status is relevant for breast cancer, but the value cannot be found in the patient's medical record. (er_status: Unknown)
Not applicable	The determination of the value is not relevant in the current context. (Reference NCIt C48660)	Clinical data regarding tobacco smoking status is not relevant for pediatric cancers. (ie. tobacco_smoking_status: Not applicable)
Cannot be assessed/determined	Aspects of the context prevent the evaluation needed to determine a value. (Reference NCIt C48657)	Lymph nodes cannot be assessed for metastases because the lymph nodes were previously removed, or surgery is not possible because the patient is too frail. (lymph_nodes_examined_status: Cannot be determined)

Cross Field Validations

A number of cross-field consistency checks within files are implemented to ensure quality control and data correctness. This requires the value of another field to validate the current field. The cross-field validation checks are implemented using Javascript. For the advanced user, you will be able to see the actual cross-field validation scripts in the Dictionary Viewer by clicking on the View Script buttons in the notes column. Examples include:

• Criteria for staging fields are dependent on the selected clinical_staging_system.
• Submitted tumour_grade is checked against selected tumour_grading_system.
• Valid values for specimen_type are cross-checked with the tumour_normal_designation field.
• The requirement for fields related to relapse/recurrence are dependent on the disease_status_at_followup field.
• The requirement for survival_time is dependent on the vital_status field.

Cross File Validations

Relationships between different clinical fields across files are validated to ensure data integrity and correctness. This requires checking the existence and relationships of different identifiers in different files, and checking the value of a field in another file to validate the current field or enforce supplemental file requirements. Examples include:

• A submitter_sample_id must belong to only one submitter_specimen_id. A submitter_specimen_id must belong to only one submitter_donor_id.
• A submitter_donor_id or submitter_specimen_id submitted in any of the clinical submission files must have been submitted in the Sample Registration file.
• A submitter_specimen_id in the Specimen file must belong to a registered submitter_donor_id in the Sample Registration file.
• A submitter_primary_diagnosis_id in the Treatment, Specimen or Follow Up file must have been submitted using the Primary Diagnosis file.
• A submitter_treatment_id in the Follow Up file must have been submitted using the Treatment file.
• The value of a specimen's tumour_normal_designation field in the Sample Registration file is checked to determine whether fields in the Specimen file are required.
• If survival_time is submitted in the Donor file, all time interval fields are validated to ensure they are less than or equal to the survival_time.
• Depending on the treatment_type selected in the Treatment file, additional treatment details may be required to be submitted. For example, if treatment_type includes Chemotherapy, the supplemental Chemotherapy treatment file is required.

Clinical Data Completion

Once all core clinical fields and files have been submitted for a donor, the donor is considered "clinically complete".

A donor must be clinically complete before any of their molecular analysis files are released to the program members for download.

How is clinical data completion calculated?

Complete clinical data means that a donor has a valid value submitted for all fields labelled "core" in the data dictionary, for a minimum set of clinical files. In more detail:

• A donor must have a donor file submitted with all core fields provided.
• A donor must have at least one primary diagnosis with all core fields provided.
• A donor must have at least one tumour and one normal specimen submitted.
• For each registered specimen, a donor must have all specimen core fields provided.
• A donor must have at least one treatment and a corresponding treatment detail file (if applicable, e.g. for chemotherapy, hormonal therapy, radiation, or immunotherapy) with all core fields provided.
• A donor must have at least one follow up with all core fields provided.