Statistical Analysis Guidelines

Paired Difference — Statistical Analysis

For each measurand level, the paired difference analysis compares the test group (maximum interferent concentration) to the control group (zero interferent):

Absolute Difference

Difference = Mean(Test) − Mean(Control)

The mean difference between test samples (containing the interferent) and control samples (without interferent) at the same measurand concentration level. A positive value indicates the interferent causes a positive bias; negative indicates a negative bias.

95% Confidence Interval

CI = Difference ± t(df, 0.975) × SE

The 95% confidence interval for the absolute difference, expressed as: Difference (lower, upper). The standard error accounts for variability in both the test and control groups. The CI indicates the range within which the true interference effect is expected to lie.

Pass/Fail (Paired Difference): The result passes when the magnitude of the absolute difference does not exceed the allowable difference. If the magnitude of the observed difference exceeds the allowable limit, the result is marked Fail — indicating that the interferent at the tested concentration causes a clinically significant bias.

Dose Response — Statistical Analysis

The dose-response analysis fits a regression model to the concentration-response data and uses it to locate the maximum interferent concentration within acceptable limits:

M0 (Baseline)

M0 = Mean of measurements at interferent concentration = 0

The baseline analyte measurement in the absence of interferent. M0 serves as the reference point for all bias calculations and is displayed as a reference line in the scatter plot. When the allowable difference is expressed as a percentage, it is converted to absolute units using M0 before comparison (e.g., 10% of M0 = 200 mg/dL → allowable difference = 20 mg/dL).

Regression Model

Y = β₀ + β₁X (linear) or Y = β₀ + β₁X + β₂X² (quadratic)

The best-fit regression model relating analyte measurement (Y) to interferent concentration (X). The module selects between linear and quadratic regression based on model fit. The quadratic equation and its coefficients are displayed above the scatter plot. The regression curve represents the expected analyte measurement at any given interferent concentration.

Interference Bias

Interference Bias = Predicted Mean − M0

The difference between the regression model's predicted analyte value at a given interferent concentration and the baseline (M0). This estimates the interference effect relative to the zero-interferent baseline at each concentration level.

% Difference

% Difference = ((Sample Mean − M0) / M0) × 100

The percentage deviation of the observed sample mean from the M0 baseline. Provides an intuitive sense of the relative magnitude of the interference effect at each concentration level.

95% Confidence Interval (Bias)

CI for Interference Bias at each concentration level

The 95% confidence interval for the interference bias, in analyte units. When the allowable difference is set as a percentage, the CI is displayed in converted analyte units (e.g., mg/dL) to align with the bias estimate. The upper bound of the CI is capped at the converted allowable limit — it will not be shown to exceed the user-defined threshold.