Hello Emily, During the Phoenix hands-on training last October, you mentioned Linear Up/Log Down is a better method than Linear Trapezoidal (Linear Interpolation). We have been using the latter, because it’s the default setting. Would you please provide reference/literature on the Linear Up/Log Down method? Thanks! Best, Rosa
Rosa, there has been quite some discussion on this on various resources like PharmPK and BEBAC forums if you want to go into more details but some initial references would bee; 1. Yeh, KC and KC Kwan A Comparison of Numerical Integrationg Algorithms by Trapezoidal, Lagrange, and Spline Approximation J Pharmacokin Biopharm 6, 79-98 (1978) 2. RD Purves Optimum Numerical Integration Methods fo Estimation of Area-Under-the-Curve (AUC) and Area-under-the-Moment-Curve (AUMC) J Pharmacokin Biopharm 20/3, 211-226 (1992) For some nice illustrations of potential bias using Linear trapezoids, please look at slides 31 and 32 of this presentation. http://bebac.at/lectures/Best%20Design%20of%20BE%20Studies%20(4).pdf Simon.
Thank you very much Simon! I will look up those references and search for the discussion on pharmPK.
Dear Rosa, see also this presentation (slide 22). That’s an example from the ‘real world’. The concentration at 48 hours was measured with 3.805 µg/mL. Next I removed this data point in order to mimick a missing sample. PHX/WNL’s default would interpolate linear between 36 and 72 hours resulting in a positive bias of AUC[sub]84[/sub] of 3.49%. If lin/log is used the bias would be only 0.14%. You may find following threads of the BEBA-Forum interesting: #3720, #4182, #5209.