Linearity of chromatographic systems in drug analysis part II: a Monte Carlo justification for the use of nonlinear regressions.

In Part I of this series of communications an alternative approach was outlined to quantify the deviation from linearity by estimation of the change in true sensitivity of several mathematically defined concentration-response relationships. In this paper, Monte Carlo techniques were used to compare the strengths and limitations of linear and nonlinear regressions in assay calibrations. Nonlinear regressions can define a calibration curve as efficiently as a linear regression. Provided that the nonlinear regression is matched to the actual nonlinear concentration-response relationship, the use of nonlinear regressions can improve the accuracy of the assay and increase the percentage of calibrations meeting acceptance criteria.