A value that was never collected and a value nobody has got round to entering look identical in the export: both are empty. Declaring data as missing is how you tell the two apart, and it is the difference between a clean dataset and a statistician's email at the end of the study.
š Prerequisite: eCRF Records (Edit), the same permission as entering data. There is no separate missing-data permission, so anyone who can fill in the eCRF can declare a value missing.
Three levels, one gesture each
Declare | Where |
One question | The Ā·Ā·Ā· menu on that question |
A whole form | The bulk control at the top of the form |
A whole visit | The bulk control beside the visit name, in the left panel |
The visit level is the one worth knowing about. A subject lost to follow-up has an entire visit that will never be collected, and declaring it in one action beats working through every question in it.
Declare one data point as missing
š ļø Step by step guide
Open the subject's eCRF and go to the form.
On the question, select the Ā·Ā·Ā· menu.
Select Missing/Unknown.
Choose a standard reason and add a comment.
Select Submit.
Declare a form or a whole visit
š ļø Step by step guide
A whole form
Open the eCRF and go to the form.
Select the bulk missing data control at the top of the form.
Choose a standard reason and add a comment.
Select Submit.
A whole visit
In the left panel, find the visit.
Select the bulk control directly under the visit name.
Choose a standard reason and add a comment.
Select Submit.
š” Tip: the comment is the part your statistician will actually read. The code says the value is absent; the comment says why this subject, this visit. Subject lost to follow-up after D28 is worth writing. N/A is not.
What the codes mean in your export
Each reason maps to a code that appears in the exported data.
Reason | Code |
Not Done |
|
Not Applicable |
|
Unknown |
|
Not Asked |
|
Asked but Unknown |
|
š” Tip: an empty cell and a coded cell are not the same thing, and the difference matters at analysis. An empty cell means nobody said anything about that value. A code means somebody recorded why it does not exist. Every declaration also appears on its own sheet in the full export, with its reason and comment. See Export eCRF data.
Track declarations across the study
š ļø Step by step guide
Open the study and go to eCRF > Monitoring.
Open the Missing Data tab.
Use Add filter to narrow to a center, and Save it for next time.
Use Filters to narrow by subject, question, visit, form or randomization ID.
Select the eye icon on a row to jump to that data point.
The three figures at the top are the study's headline: how many values are declared missing, how many subjects they affect, and what percentage of the expected data that represents. The table below is one row per declaration, showing the variable, its question, the subject, the center and where it sits in the eCRF.
š” Tip: this tab has the most precise filters of the four dashboards. It is the only one that can narrow to a single form rather than a whole visit, which is what you want when you are checking whether one questionnaire is systematically going uncollected at one site.