A good data definition provides ALL the characteristics of forms (questionnaires, CRFs) and the associated variables. What is needed is good old "TLV".
| Stands For.. | Overview | |
|---|---|---|
T | Type of Variable | The most common types are dates, number, coded variables (e.g., 0 = No, 1 = Yes;). |
L | Layout and Logic | Indicate the look and feel of the variable, and when it should be displayed (e.g., only show if subject is a male). |
V | Validation | Indicate the rules that govern the acceptable variable values (acceptable range, units) |
What To Do
Use A Spreadsheet
Document your data definition using the spreadsheet here.
Leverage Standards
Learn About Data Definition Standards (see here)
Determine Forms
In cases where the forms are NOT predefined (e.g., a standardized depression measure), determine the variables on a form via a combination of Temporal and Conceptual considerations (see example image):
- Temporal - What variables are collected together at specific time points? The data definition should note the variables that may be exceptions (e.g., labs values that are rarely ordered)
- Conceptual - What variables belong to a given content domain (e.g., surgical vs. imaging variables)? As a rule of thumb, forms should be broken down no further than about 5 variables (e.g., separating "Surgical Characteristics" into "Intraoperative" vs. "Perioperative" Characteristics)
Determining Forms Example
Mark It UP
When sending us a form, mark it up with the data definition.
Example Marked Up Form
Keep It Simple
Forms should use the cleanest, simplest layout possible. Standard practice is to use one-row-per-variable (examples here).
Leverage Visual Validation
A form's spatial and visual layout can be it's most important validation (see examples here).
Indicate Timing
Document which forms are collected with each study event by completing the Data Collection Schedule (see Data Collection Schedule.doc).