How It Works
Study Design
A method comparison study collects paired measurements — the same patient sample tested on both the reference method (X) and the test method (Y). CLSI EP09C specifies the following minimum design requirements:
| Parameter | CLSI Recommendation |
|---|---|
| Patient samples | Minimum 40; preferably 100 or more for tighter bias estimates |
| Measurement range | Samples should cover the full clinically relevant measurement range across multiple concentration levels |
| Sample freshness | Test samples as fresh as possible; handle identically on both methods to avoid pre-analytical bias |
| Days | One or more days recommended; multiple days help assess consistency of method agreement over time |
| Replicates per day | One or more replicates per sample per day; additional replicates improve precision of the comparison |
| Pairing | Each sample must be tested on both methods — results are analyzed as pairs, not populations |
Why 40 samples minimum? Fewer than 40 paired observations provide insufficient statistical power to reliably detect clinically important bias. Studies with 100 or more samples yield substantially narrower confidence intervals for the bias estimate, making pass/fail determinations more defensible.