If 3-4 results fall outside the proposed RI, what is recommended?

If 3-4 results fall outside the proposed RI, what is recommended?

• A. verify measurement accuracy and repeat testing
• B. expand the CI to include the outliers
• C. collect additional 20 samples and re-evaluate
• D. apply population-based RI to the dataset

Answer: (C)

When building a reference interval, a few results outside the current interval often reflect limited data rather than true abnormalities. The best course is to gather more data and re-evaluate the interval—in this case, collect additional samples (about twenty) and recalculate the RI. Increasing the sample size stabilizes the estimated lower and upper limits and helps determine whether those outliers are random variation or indicate a need to adjust the interval. Verifying measurement accuracy and repeating testing is useful as a preliminary quality check, but it doesn’t address the underlying precision of the RI estimate as effectively as adding data. Expanding the confidence interval to include the outliers would weaken the RI and reduce its clinical usefulness. Applying a population-based RI ignores the specific population and method used to generate the current data, which can misrepresent the local context.

When building a reference interval, a few results outside the current interval often reflect limited data rather than true abnormalities. The best course is to gather more data and re-evaluate the interval—in this case, collect additional samples (about twenty) and recalculate the RI. Increasing the sample size stabilizes the estimated lower and upper limits and helps determine whether those outliers are random variation or indicate a need to adjust the interval.

Verifying measurement accuracy and repeating testing is useful as a preliminary quality check, but it doesn’t address the underlying precision of the RI estimate as effectively as adding data. Expanding the confidence interval to include the outliers would weaken the RI and reduce its clinical usefulness. Applying a population-based RI ignores the specific population and method used to generate the current data, which can misrepresent the local context.