Definition
A case report form is the document in which the information required by the protocol is recorded, for each participant. Its electronic version, the eCRF, replaces the paper booklet — but most of its value does not come from being digital.
In the francophone hospital tradition the equivalent is often called a *fiche d'exploitation*: the document listing, before capture, the variables to be extracted from each record. The word differs, the object is the same, and a thesis collection form is an eCRF that does not call itself one.
The six properties of an eCRF
A web form collects answers. An eCRF collects study data, which requires more.
| Property | What it implies |
|---|---|
| Typed, constrained fields | Type, scale, unit, bounds, closed lists — the form refuses the impossible |
| Qualified absence | An empty field must be able to say why it is empty |
| Cross-field checks | A discharge date before admission is blocked at entry |
| Audit trail | Every change is recorded, with its author and reason |
| Discrepancy management | A doubtful value can be flagged, discussed and resolved without being overwritten |
| Locking | The form stops accepting changes when the study moves to analysis |
eCRF, collection form, dictionary: three words, two objects
Confusion between these terms is common and easily cleared up.
- The variable dictionary declares what each variable *is*: type, unit, bounds, categories. It is a specification. See the data dictionary.
- The eCRF (or collection form) is the interface through which those variables are *entered*: field order, groupings, visits.
- The second should be generated from the first. When they are maintained separately they drift — and it is almost always the dictionary that becomes wrong, because the form is what people actually use.
Designing a usable eCRF
Follow the clinical narrative, not the database structure
The operator reads a record in the order it was written. A form following the order of database tables forces back-and-forth navigation, which multiplies attention errors.
Cut to the strictly necessary
Every added field lengthens the time per record and lowers the completion rate of all the others. A variable appearing in no planned analysis should not appear in the eCRF.
Reuse existing definitions
CDASH supplies variable names and definitions designed for capture, and SDTM the downstream tabulation structure. Adopting them where they apply makes the dataset comparable; ignoring them forces a future reader to guess.
Plan for interruption
Data entry gets interrupted. An eCRF that loses the state of a partly filled record will be worked around with paper notes, and the workaround becomes the real source.
How AnzarSeha addresses it
- The collection form is generated from the dictionary, itself drawn from the pathology catalogue: staging, work-up, scores and markers are already there, typed.
- The primary endpoint is declared on that form, before capture, and the protocol lock refuses to engage until it is.
- Reasons for absence are specific to each variable, and each carries its analytical consequence.
- Every change passes through the audit trail, with its author and reason.
Frequent questions
Can an online form or survey tool serve as an eCRF?
For simple, one-off, single-operator capture it does the job. It carries no reasoned audit trail, no freeze, no discrepancy management and no qualified absence — so it cannot later demonstrate what was done. The choice depends on what the study will have to prove.
Is an eCRF needed for a retrospective study?
Yes, in an adapted form. The protocol always precedes capture, even when the data already exists: what changes is that the fields describe what will be sought in records rather than what will be measured on patients.
What is the difference between eCRF and EDC?
The eCRF is the form; EDC (electronic data capture) is the system hosting it, managing users, discrepancies, locks and exports. In common usage the two blur, but an eCRF without the surrounding system delivers only the first third of the properties listed above.