The column called “status”
Does it mean test status, enrollment status, disease status, or record-processing status? A familiar word can hide several incompatible meanings.
A useful dictionary defines the concept rather than repeating the column name in a longer sentence.
| Field detail | Example of the needed information |
|---|---|
| Meaning | Whether a laboratory result is preliminary or final |
| Allowed values | Enumerated codes and their meanings |
| Missingness | Blank differs from unknown or not applicable |
| Timing | Status as of extraction, not necessarily at collection |
| Source | Originating system and transformation rule |
Preserve units and denominators
A field labeled “rate” needs a numerator, denominator, and time basis. A weight field needs units. A score needs its instrument and valid range. Without these, a clean table can still support incorrect calculations.
Version the definitions
If a field changes meaning halfway through a dataset, the dictionary should show when and how. Otherwise, a trend may reflect a software change rather than a health change.
Before analyzing, choose a few records and trace them back to their definitions and source. That small exercise often reveals ambiguity earlier than a complex model does. A dictionary is useful when it prevents a mistaken interpretation, not merely when every column has a description.
