Summary & Reference Interval Guidelines

Summary Statistics

After computing the NPM and INR values for all subjects, the module calculates the following statistics for both PT and INR independently:

Geometric Mean (Normal Patient Mean)

NPM = exp(mean of ln(values))

For PT: the NPM itself. For INR: the geometric mean of all computed INR values (which, by definition, will be very close to 1.0 for a normal population, since each INR is anchored to the NPM).

2SD Range

x̄ ± 2s (arithmetic mean ± 2 sample SDs)

The arithmetic mean plus and minus two standard deviations. For PT, this characterizes the spread of the observed PT distribution. For INR, it characterizes the spread of the computed INR values. Note: this is the observed 2SD range of the study sample, not the reference interval — these are different statistics.

% Out of Limits

(count of values outside [Lower Limit, Upper Limit]) / n × 100

The percentage of normal subjects whose PT (or INR) falls outside the configured reference interval limits. Values exactly at the limits are counted as in-range. This is the key metric used for pass/fail determination.

Pass/Fail: A study passes when ≤10% of normal subjects fall outside the configured reference interval limits — for both PT and INR independently. If more than 10% of subjects are out of limits, the reference interval is too narrow for the local population and must be revised before the reagent lot is placed into service.

INR Reference Interval Derivation (Compute INR from PT)

When the "Compute INR from PT" option is enabled, the INR reference limits are derived from the PT reference limits using the same INR formula:

INR Lower Limit (derived)

INR_lower = (PT_lower / NPM)^ISI

The INR value corresponding to the lower PT reference limit. Because the INR formula is monotonically increasing with PT (for positive ISI), the PT lower limit maps directly to the INR lower limit.

INR Upper Limit (derived)

INR_upper = (PT_upper / NPM)^ISI

The INR value corresponding to the upper PT reference limit. These derived limits are rounded to 2 decimal places for display and comparison.