Build Trustworthy Data and Act on the Right Signal

Design the data before designing the form

Every collected field should serve a prespecified question, safety need, adjustment, or interpretation. Create a data dictionary that names the variable, operational definition, unit, allowed values, source, timing, instrument version, missing-value codes, and derivation. Define the numerator, denominator, eligible population, and observation period for each indicator. Use validated instruments in populations and languages for which their performance is adequate, and obtain permission when required. Pilot the workflow with realistic participants. A measurement can be reliable, producing consistent values, yet invalid because it consistently measures the wrong construct. It can be valid in one setting but perform poorly after translation, mode changes, or use in another population. Document calibration, staff training, interrater checks, and any protocol changes.