Blockwise Passing Bablok Regression  for method comparison studies with repeated measurements

Passing-Bablok regression (PBR) is a non-parametric method widely used in clinical biochemistry and laboratory medicine to compare two measurement methods. It estimates the slope β and intercept α of the linear relationship y = α + βx by taking the median of all pairwise slopes between measurements.

When the data contain repeated measurements (multiple observations from the same sample or group), the within-group slopes have an expected value of 0 rather than β, and including them biases the slope estimate toward zero. Block-PBR corrects this by excluding within-group slopes from the median calculation and provides a corresponding asymptotic confidence interval that accounts for group sizes and inter-group overlap.

An example script on how to use our method is available at  github