How does simple binomial validation work and what is the number used?

How does simple binomial validation work and what is the number used?

• A. Compare 20 healthy reference individuals from the lab's population to proposed RI; Screen for outliers; replace if found.
• B. Compare 10 healthy references; perform t-tests; adjust RI.
• C. Use 30 individuals; assess distribution.
• D. Use 15 individuals; compute percent differences.

Answer: (A)

Simple binomial validation evaluates a proposed reference interval by applying a binomial rule to a small set of healthy individuals from the lab population. You take 20 healthy reference individuals, measure the analyte, and see how many results fall outside the proposed interval. You also screen for obvious outliers and replace them if identified, then re-evaluate. If zero or one of the 20 values lies outside the proposed RI, the interval is considered validated for that population; if two or more lie outside, the RI should be revised. The number 20 is used because it provides a practical sample size that yields enough power to detect a mis-specified interval while remaining feasible; with a correctly specified 95% RI you’d expect about 1 outside on average, so 0–1 outside supports validity. Other choices mentioning larger sample sizes, t-tests, or distribution testing describe different methods and are not the standard simple binomial validation approach.

Simple binomial validation evaluates a proposed reference interval by applying a binomial rule to a small set of healthy individuals from the lab population. You take 20 healthy reference individuals, measure the analyte, and see how many results fall outside the proposed interval. You also screen for obvious outliers and replace them if identified, then re-evaluate. If zero or one of the 20 values lies outside the proposed RI, the interval is considered validated for that population; if two or more lie outside, the RI should be revised. The number 20 is used because it provides a practical sample size that yields enough power to detect a mis-specified interval while remaining feasible; with a correctly specified 95% RI you’d expect about 1 outside on average, so 0–1 outside supports validity. Other choices mentioning larger sample sizes, t-tests, or distribution testing describe different methods and are not the standard simple binomial validation approach.