Hello everyone,

In order to compare two clinical methods, we usually use Passing & Bablok (PABA) regression. Most of the time, our samples are larger than n=50, but for the comparison I'm interested in today (method A vs method B), the samples are small (n = 10-15).

The PABA regression validates the equivalence between the two methods (method A vs method B). Indeed, the CI intercept crosses 0 and CI slope crosses 1 :

  • Intercept = -6 and confidence intervalle (CI) = [-56 ; 31]
  • Slope = 2. and confidence intervalle (CI) = [0,5 ; 4]

However, I have a few points of concern about these results because :

  • The Pearson coefficient is low (r = 0,63),
  • The size of the CI is very large,
  • The coefficient of variation (CV) between the two methods is high (CV > 20%).

Do you know of any criteria or rules that I could add to the analysis of PABA regression that would enable me to improve our validation method ?

Thanks in advance for your help ! :)

Similar questions and discussions