Step-by-step guide to performing a method comparison study under CLIA, including sample requirements, statistical analysis, and acceptable bias thresholds.
Frequently Asked Questions
When is a method comparison study required?
CLIA requires a laboratory running the same test on different instruments or methods, or at multiple sites, to evaluate the relationship between them twice a year (42 CFR §493.1281(a)). A comparison is also a common way to verify accuracy when a new analyzer or method goes into service (§493.1253(b)(1)(i)(A)).
How many specimens does it need?
CLIA and TJC set no minimum; the medical director or designee decides. CLSI EP09 suggests about 40 patient specimens spanning the full analytical measurement range, deliberately collected at low, mid, and high concentrations.
Why patient specimens instead of QC material?
Fresh patient samples reflect the actual matrix the methods encounter in routine testing; QC materials do not.
What statistics does it use?
Slope, y-intercept, Pearson correlation coefficient, and bias at medical decision points. Deming or Passing-Bablok regression is preferred over ordinary least squares because it accounts for error in both methods.
How do you set pass or fail?
The laboratory or manufacturer sets acceptable limits for slope, intercept, and bias, defined before running the study, with bias at decision points within total allowable error. CLSI EP09 is the reference protocol.