Marcus A. Björnsson (1,2), Lena E. Friberg (2), Ulrika S.H. Simonsson (2)
(1) Clinical Pharmacology & Pharmacometrics, AstraZeneca R&D, Södertälje, Sweden; (2) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: The objective of this simulation study was to compare bias in EC50 estimates when using interval censored dropout and data where the exact time of dropout is known, in the case of informative dropout.
Methods: An efficacy variable was simulated using an inhibitory Emax drug effect model combined with an exponential placebo effect model. Simultaneously, dropout was simulated using a hazard function in which the hazard was exponentially related to the individual predicted efficacy variable, adapted from Björnsson and Simonsson [1]. In the simulations, the dropout was either recorded at the actual time of dropout, or interval censored, i.e. dropout occurred somewhere between two pre-defined assessment times. The impact of number of pre-defined assessment times, and hence the interval between assessments, as well as the impact of extent of dropout was evaluated. The dataset consisted of a placebo group and three active treatment groups, each with 45 subjects, and with a dropout rate between 25 and 55% in the base scenario. The simulated efficacy and dropout data were simultaneously analysed using non-linear mixed effects modelling. The Laplacian estimation method in NONMEM 7 and PsN [2] were used for the stochastic simulations and estimations.
Results: In all simulations bias in EC50 was low, less than 10%. When simulations were performed without recording the exact time of dropout (interval censoring), bias in EC50 was higher than when the exact time of dropout was recorded. This was especially evident when there were few measurements of the effect variable. When there were only three measurements of the effect variable, bias in EC50 was almost five times higher for interval censoring, compared to when the exact time of dropout was known. Bias in EC50 also increased with increasing dropout rate, regardless of whether exact times or interval censoring is used.
Conclusions: When a dropout model is used, bias in EC50 is low, regardless of whether the exact dropout times or interval censoring is used. Bias is, however, lower if the exact dropout times are used, especially if the interval between observations is long or the extent of dropout is large.
References:
[1] Björnsson MA, Simonsson USH. Modelling of pain intensity and informative dropout in a dental pain model after naproxcinod, naproxen and placebo administration. Br J Clin Pharmacol 2011; 71:899-906
[2] Lindbom L, Pihlgren P, Jonsson EN. PsN-Toolkit–a collection of computer intensive statistical methods for non-linear mixed effect modeling using NONMEM. Comput Methods Programs Biomed 2005; 79:241-257
Reference: PAGE 22 (2013) Abstr 2759 [www.page-meeting.org/?abstract=2759]
Poster: Other Modelling Applications