Confidence interval approach for evaluating bias in laboratory methods

Area of Science:

• Laboratory science
• Biostatistics

Background:

• Laboratory quantitation methods are crucial for accurate measurements.
• Statistical significance of differences between methods is often interpreted as bias.
• Minute differences may not represent practical concerns in method comparison.

Purpose of the Study:

• To propose an alternative statistical approach for evaluating laboratory method agreement.
• To determine if two quantitation methods are practically interchangeable, not just statistically different.
• To utilize confidence intervals for assessing method comparability without additional statistical tests.

Main Methods:

• Statistical analysis of mean values from two laboratory quantitation methods.
• Calculation of the confidence interval for the mean difference between methods.
• Interpretation of the confidence interval to assess practical equivalence.

Main Results:

• A statistically significant difference does not always imply practical concern.
• Confidence intervals of the mean difference provide a direct measure of method agreement.
• No additional statistical tests are required beyond confidence interval calculation.

Conclusions:

• Focusing solely on statistical significance can misrepresent practical method differences.
• Confidence intervals effectively evaluate if laboratory methods are sufficiently close for practical use.
• This approach simplifies method comparison by avoiding complex statistical tests.