Analysis and comparison of sigmoidal curves: application to dose-response data.

A number of physiological or pharmacological studies generate sigmoidal dose-response curves. Ideally, data analysis should provide numerical solutions for curve parameters. In addition, for curves obtained under different experimental conditions, testing for significant differences should be easily performed. We have reviewed the literature over the past 3 years in six journals publishing papers in the field of gastrointestinal physiology and established the curve analysis technique used in each. Using simulated experimental data of known error structure, we have compared these techniques with nonlinear regression analysis. In terms of their ability to provide accurate estimates of ED50 and maximal response, none approached the accuracy and precision of nonlinear regression. This technique is as easily performed as the classic methods and additionally provides an opportunity for rigorous statistical analysis of data. We present a method of determining the significance of differences found in the ED50 and maximal response under different experimental conditions. The method is versatile and applicable to a variety of different physiological and pharmacological dose-response curves.