Precision of ED50 Estimate for Four-Parameter Logistic Model with Different Slopes and Testing Plans

• Activity Number: 420
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Quality and Productivity Section
• Author(s): Jianfang Hu*+
• Companies: Merck
• Keywords: Precision ; ED50 ; Four parameter Logistic

Abstract

Objective: To compare the precision of ED50 estimate for four parameter logistic curves with different slopes and testing plans with different dilution factors. Method: Log normal data were simulated using four parameter logistic equation. Upper and lower asymptotes, and ED50 were set to be fixed values. Slope varied from 0.5 to 2. Bend points refer to the starting and end of the linear dose-response region from the four parameter logistic curve. Number of points on the linear portion or asymptotes varied by changing dilution factors with same starting concentration. Relative Standard Deviation of ED50 estimates were compared from above scenarios. Results: Precision was greatly improved with at least two points beyond the bend points on each asymptote. Precision improved as slope increased. As dilution factor increased from 2 to 4, precision only improved slightly if there are enough points beyond bend points on each side for each testing plan. If there is a limit on the total number of points that can be studied on the curve (especially for low slopes), better precision can be achieved by allocating more points on asymptotes and skipping points on the linear portion.