Assessments That Have to Be Comparable

Part of Sub-Investigator (SUBI) Training: Clinical Assessment in a Trial.

About 2 minutes. Free, in your language, no account needed.

Your assessment is being pooled with strangers'

A severity grading you make is combined with gradings made by clinicians you will never meet. The protocol's definitions exist so those gradings mean the same thing.

Using your own clinical shorthand, however sound, makes your site's data different.

Use the specified instrument, as specified

If the protocol names a toxicity scale, a performance status or a rating instrument, use that one, with its own definitions.

Approximating from clinical impression is the commonest way a site drifts.

Consistency within a participant matters most

Where possible, the same assessor performs the same assessment across visits for a participant.

Inter-rater variation shows up as change that is not real, and for subjective scales it can be larger than the treatment effect.

Follow the order and the timing

Assessments taken before or after a dose, fasting or fed, before or after a result is disclosed. These are specified because they affect the value.

Doing them in a convenient order rather than the specified one is a deviation that produces a plausible, wrong number.

Record the value, not the interpretation

"Blood pressure 158/96, participant asymptomatic" is data. "Blood pressure a little up, not concerned" is an opinion that cannot be re-analyzed.

Give the number, then the judgment if you have one.

Do not round toward the expected

A tendency to record values nearer the expected range is well documented and unconscious. Read the instrument and write what it says.

Missed assessments are recorded as missed

Not estimated, not carried forward from the previous visit, not left blank.

A missing value handled honestly is manageable. An invented one is not detectable and corrupts the analysis.