Statistical Models for Toxicity and Safety Pharmacology Studies
- PDF / 589,610 Bytes
- 13 Pages / 504 x 719.759 pts Page_size
- 53 Downloads / 192 Views
Drug Infomarion J o u m l . Vol. 34. pp. 631-643. 2000 Printed in the USA. All rights reserved.
Copyright 0 2000 Drug Information Association Inc.
STATISTICAL MODELS FOR TOXICITY AND SAFETY PHARMACOLOGY STUDIES HELLEANDERSEN, MSc Novo Nordisk A/S, Bagsverd, Denmark, and Department of Mathematical Modelling, IMM, Technical University of Denmark, Lyngby, Denmark
HENRIKSPLIID, PHD, PROF STAT Department of Mathematical Modelling. IMM, Technical University of Denmark, Lyngby, Denmark
S0)REN
LARSEN,CAND STAT
Novo Nordisk A / S , Bagsvaerd, Denmark
Consistency in statistical analyses of preclinical studies is a request from authorities worldwide because itfacilitates the evaluation and comparison of similar studies. The goal of this paper is to describe mixed-effect analysis of variance models broadly applicable to the analysis of continuous responses from toxicio and safew pharmacology studies. Traditional models are discussed together with more complex models including the splitplot model and related models for repeated measurements. The model fitting process ranges from the choice of model to the final check of model assumptions which we exemplify using case studies analyzed by the S A P procedure. Key Words: Reclinical studies; Split-plot models; Repeated measurements; Analysis of variance; Model fitting process; Components of variance
THE MODEL FITTING PROCESS AND DIAGNOSTIC CHECKS: A BRIEF REVIEW
INTRODUCTION IN THIS PAPER we describe parametric models and methods that enable toxicologists to perform statistical analyses applicable to a broad class of preclinical studies. Typically, the design variables include gender, the number and types of treatments involved, and the number of times the response is measured. The responses of interest can be continuous, categoric (discrete), or, for example, ordinal categorical. In this paper we review mixed-effect analysis of variance models for the analysis of continuous responses, because most toxicity studies relate to continuous measurements.
The model fitting process is split into four stages: 1. Formulation: Choosing the general form of the model, 2. Estimation: Attaching numerical values to parameters, 3. Inference: Calculating confidence intervals andor testing the hypothesis about parameters of direct interest, and 4. Diagnostics: Checking that the model fits the data.
All general model assumptions are given in the model formulation. In the estimation Reprint address: Helle Andersen, MSc, Novo Nordisk stage numerical values are assigned to paN S , Novo Allt. DK-2880, Bagsvaerd, Denmark. E-mail: rameters and noninferential tests are [email protected]. 631
Downloaded from dij.sagepub.com at IOWA TESTING PROGRAMS on May 30, 2015
632
Helle Andersen, Henrik Spliid, and S)ren Lursen
scribed in order to simplify the model and its covariance structure. Depending on the complexity of the model, different estimation methods are recommended. This is followed by methods for inference, that is, calculation of confidence intervals and tests of hypothesis about paramete
Data Loading...
