Proposed guidance for carryover studies, based on elementary equivalence testing techniques.

Area of Science:

• Clinical Chemistry
• Laboratory Diagnostics
• Analytical Chemistry

Background:

• Carryover experiments are crucial for validating clinical and immunochemistry analyzers.
• Current statistical methods for analyzing carryover effects are inadequate, especially for demonstrating absence of carryover.
• Existing reporting in parts per million (ppm) lacks uncertainty, hindering accurate assessment.

Purpose of the Study:

• To propose a step-by-step guidance for the statistical analysis of carryover studies.
• To introduce a novel statistical design based on equivalence testing for carryover data analysis.
• To provide a practical, sample-based tutorial for implementing the proposed methodology.

Main Methods:

• Employed a one-sided equivalence testing approach, a form of non-superiority testing.
• Compared the observed carryover difference against a predefined limit.
• Demonstrated the methodology using total beta-hCG measurements on a UniCel DxI 880 analyzer.

Main Results:

• Developed a new statistical approach for analyzing carryover study data using equivalence testing.
• Determined that 8 (11) cycles of high/low concentration samples provide 80% (90%) power to validate absence of carryover at alpha=0.05.
• Proposed defining acceptance criteria based on the imprecision of unaffected low-concentration samples.

Conclusions:

• Appropriate statistical methods are essential for method validation, particularly for demonstrating absence of effect, non-inferiority, or equivalence.
• One-sided equivalence testing is the correct statistical model for carryover studies.
• The proposed methodology is applicable to various experimental approaches, including method comparison, commutability, and robustness studies.