Data dictionary

The eCRF is organised into nine sections. The section a field belongs to determines where its value is stored on the patient record, which import file it should appear in, and which completeness indicator it contributes to.

The two lymph-node sections account for most of the field count because the form asks the same short battery of questions — clinically positive, count of positive nodes, histologically confirmed, nodes removed, and so on — separately for every region and every numbered station. That is why the table below is searchable rather than printed in full: search for a station number, or filter to one section, rather than scrolling.

Each row of the table below gives you:

• Field label — the wording shown on the form, with the form's own help text beneath it where one exists, and the group the field sits in (for example the specific lymph-node station).
• Key — the identifier used everywhere else: the column header in the import template, the property name inside the record, and the name you will see in a validation flag or an export column.
• Type — how a value is interpreted on import. choice and choice + other match against a fixed vocabulary (the second also allows a free-text write-in); boolean is a tick box; number and range are numeric, often with a minimum and maximum; date is a date; text is free text; image choice is a diagram the investigator clicks.
• Required — whether the original form marked the field mandatory. Twenty-one fields are marked required. The importer does not enforce these. The mark is a property of the form definition shown here; the schema the importer validates against currently marks no field required, so an upload missing all twenty-one raises no flag. Treat the column as a statement about the form, not a guarantee about the data.
• Options / range — the allowed values, collapsed; expand to see the list. For numeric fields, the permitted minimum and maximum and the unit are shown instead.

Field values are stored per section rather than as one flat row, and most sections are encrypted at rest. Three details are worth knowing if you work with exports or the API:

• Adjuvant is a real section but shares storage. The adjuvant fields are a section of the form and are read and written as one, but their values live inside the pathology blob. This is deliberate: the record's completion percentage is calculated over a fixed set of eight sections, and adding a ninth would have silently dropped every existing record below 100 % and could have un-completed already verified records.
• Lymph-node regions are not encrypted. Region-level node data is stored in a plain column because it carries no identifying content; the numbered station data is encrypted like the clinical sections.
• Some fields are mirrored into plaintext columns. A small set of clinical values — tumour type, location and histology, stage components, surgical approach, node counts and ratio, complication grade and count, the mortality flags, resection margin, follow-up months, recurrence and vital status — are copied out of the encrypted sections into plaintext columns when a record is saved. Those columns are what the analytics read, and they are the columns you get in a non-PHI export. Everything else stays inside the encrypted section.

The importer maps to exactly these field keys, so a column headed with a key from this dictionary lands in the right section automatically. See Data import for how values are coerced and validated against the types and options listed here.

This dictionary is generated, not written by hand. It is derived from migration/drupal-reference/tiger-ecrf-elements.yaml — the element export of the study's original Drupal form — by scripts/gen-ecrf-dictionary.py, and the generated file is committed alongside the site so this page and the running platform cannot drift apart unnoticed.

The generator also cross-checks every field's section against the backend's import target schema and refuses to write the file if the two disagree, so the section shown here is the section the importer will actually use.