LoB Calculation Guidelines
LoB Calculation Methods
Two methods are available. The nonparametric method is the default and requires no assumptions about the shape of the blank measurement distribution. The parametric method assumes normally distributed blank measurements and yields a formula-based estimate.
Nonparametric Method (Default)
Rank Position
Rank Position = 0.5 + B × 0.95
Where B = total number of blank measurements for the lot. All blank values are sorted in ascending order and the LoB is read from this rank position. The 0.95 factor corresponds to the 95th percentile — the LoB is defined such that 95% of blank sample results fall at or below it.
LoB (Nonparametric)
LoB = X(floor(rank)) + fractional part × (X(floor(rank)+1) − X(floor(rank)))
Linear interpolation between the two values bracketing the rank position. For example, if the rank position is 57.5 with 60 blank measurements, LoB = X(57) + 0.5 × (X(58) − X(57)), where X(57) and X(58) are the 57th and 58th sorted blank values.
Parametric Method
LoB (Parametric)
LoB = Mean_blank + cp × SD_blank
Where Mean_blank and SD_blank are the mean and standard deviation of all blank measurements for the lot, and cp is a correction factor that accounts for the sample size and number of unique blank samples.
Correction Factor (cp)
cp = 1.645 × (1 + 1 / (4(B − K)))
Where B = total blank measurements and K = number of unique blank samples. The base factor 1.645 corresponds to the 95th percentile of the standard normal distribution (one-sided, α = 0.05). The correction term 1/(4(B−K)) adjusts for the degrees of freedom in the SD estimate, ensuring the LoB is not underestimated in small studies.
Maximum LoB (Final Reported Value)
Maximum LoB
Maximum LoB = max(LoB_lot1, LoB_lot2, …, LoB_lotN)
When 2–3 reagent lots are tested, the LoB is calculated separately for each lot and the maximum value is reported as the final LoB. This conservative approach ensures the reported LoB represents worst-case performance across the reagent lots evaluated.