1st Edition

Medical Product Safety Evaluation
Biological Models and Statistical Methods

ISBN 9781466508088
Published September 10, 2018 by Chapman and Hall/CRC
354 Pages 50 B/W Illustrations

USD $125.00

Prices & shipping based on shipping country


Book Description

Medical Product Safety Evaluation: Biological Models and Statistical Methods presents cutting-edge biological models and statistical methods that are tailored to specific objectives and data types for safety analysis and benefit-risk assessment. Some frequently encountered issues and challenges in the design and analysis of safety studies are discussed with illustrative applications and examples.

Medical Product Safety Evaluation: Biological Models and Statistical Methods presents cutting-edge biological models and statistical methods that are tailored to specific objectives and data types for safety analysis and benefit-risk assessment. Some frequently encountered issues and challenges in the design and analysis of safety studies are discussed with illustrative applications and examples.

The book is designed not only for biopharmaceutical professionals, such as statisticians, safety specialists, pharmacovigilance experts, and pharmacoepidemiologists, who can use the book as self-learning materials or in short courses or training programs, but also for graduate students in statistics and biomedical data science for a one-semester course. Each chapter provides supplements and problems as more readings and exercises.


Table of Contents

List of Figures

List of Tables


1. Introduction

Expecting the unexpected

A brief history of medical product regulation

Science of safety

Differences and similarities between efficacy and safety endpoints

Regulatory guidelines and drug withdrawals

Medical product safety, adverse events and adverse drug reactions

Medical product safety

Adverse events versus adverse drug reactions

Safety data coding

Drug dictionaries

WHO Drug Dictionary

Anatomical Therapeutic Chemical (ATC) classification

NCI Drug Dictionary

Adverse event dictionaries

Medical Dictionary for Regulatory Activities (Med-DRA)

Common Terminology Criteria for Adverse Events (CTCAE)

WHO’s Adverse Reaction Terminology (WHO-ART)


Serious adverse events and safety signals

Statistical strategies for safety evaluation and a road map for readers

Safety data collection and analysis

Safety databases and sequential surveillance in pharmacovigilance

An interdisciplinary approach and how the book can be read

Supplements and problems

2. Biological Models and Associated Statistical Methods

Quantitative structure-activity relationship (QSAR)

Toxicity endpoints

Molecular descriptors

Statistical models in QSAR/QSTR

Model validation

Pharmacokinetic-pharmacodynamic models

Analysis of preclinical safety data


Reproductive and developmental toxicity

Correlated binary and trinary outcomes within litters

Dose response

Predictive cardiotoxicity

The Comprehensive in vitro Proarrythmia Assay (CiPA)


Ion channels, in silico models and stem-cell

derived cardiomyocyte assays

Phase I ECG studies

Concentration-QTc modeling

Toxicogenomics in predictive toxicology

Components of TGx

TGx biomarkers

Regulatory framework in predictive toxicology

Regulatory guidelines

Safety biomarker qualification

In silico models in predictive toxicology

Supplements and problems

3. Benefit-Risk

Some examples of B-R assessment




Critical ingredients for B-R evaluation

Planning process

Qualitative and quantitative evaluations

Benefit-risk formulations

A multidisciplinary approach incorporating multiple perspectives

Multi-criteria statistical decision theory

Multi-criteria decision analysis

Stochastic multi-criteria acceptability analysis

Stochastic multi-criteria discriminatory method

B-R methods using clinical trial data

Quality-adjusted benefit-risk assessment methods


Quality-adjusted survival analysis

Testing QAL differences between treatment and control

Additional statistical methods

Number needed to treat(NNT)

Incremental net benefits (INB)

Weighting schemes, uncertainty, models, supplemental data and patient-level data

Bayesian methods

Endpoint selection and other considerations

Other statistical considerations

Supplements and problems

4. Design and Analysis of Clinical Trials with Safety Endpoints

Dose escalation in phase I clinical trials

Rule-based designs

Model-based designs: CRM EWOC, Bayesian threshold designs

Individual versus collective ethics and approximate dynamic programming

Extensions to combination therapies

Modifications for cytostatic cancer therapies

Safety considerations for the design of phase II and III studies

Challenges of safety evaluation in phase II andphase III trials

Conditioning on rare adverse events and the RESTexample

A sequential conditioning method and an efficient sequential GLR test

Designs for both efficacy and safety endpoints

Summary of clinical trial safety data

Integrated summary of safety (ISS)

Development safety update

Clinical safety endpoints

Laboratory test results

Vital signs


Graphic display of safety data

Graphic display for proportions and counts

Graphic displays for continuous data

Statistical methods for the analysis of clinical safety data

Incidence rates and confidence intervals

Confidence intervals based on Wald’s approximation and moment

Confidence intervals based on variance estimate recovery

Confidence intervals based on parameter constraint

Confidence intervals with stratification

Regression models

Poisson regression

Negative binomial models

Rare event analysis

Zero-inflated regression models

Generalized extreme value regression

Time-to-event analysis and competing risks

Recurrent events

Mean cumulative function

Regression models

Supplements and problems

5. Multiplicity in the Evaluation of Clinical Safety Data

An illustrative example

A three-tier adverse event categorization system

The MMRV combination vaccine trial

Multiplicity issues in efficacy and safety evaluations

P-values, FDR and some variants

Double false discovery rate and its control

FDR control for discrete data

Bayesian methods for safety evaluation

Berry and Berry’s hierarchical mixture model

Gould’s Bayesian screening model

Compound decisions and an empirical Bayes approach

Supplements and Problems

6. Causal Inference from Post-Marketing Data

Post-marketing data collection

Clinical trials with safety endpoints

Observational pharmacoepidemiologic studies using registries

Prospective cohort observational studies

Retrospective observational studies

Potential outcomes and counterfactuals

Causes of effects in attributions for serious adverse health outcomes

Counterfactuals, potential outcomes, and Rubin’s causal model

Frequentist, Bayesian, and missing data approaches

Causal inference from observational studies

Matching, sub classification, and standardization

Propensity score: Theory and implementation

Control for confounding via estimated propensityscore

Inverse probability weighting

Unmeasured confounding

Instrumental variables

Econometrics background, instrumental variable tests and GMM extensions

Trend-in-trend research design of observational studies

Supplements and problems

7. Safety Databases: Statistical Analysis and Pharmacovigilance

Safety databases

Preclinical data

Clinical trial data

FDA Adverse Event Reporting System (FAERS)

Vaccine Adverse Event Reporting System and Vaccine

Safety Link


Medicare, Medicaid Database, and health insurance claims databases

Adverse event reporting database for medical devices

Statistical issues in analysis of spontaneous AE databases

Statistical methods for the analysis of safety database

Reporting ratios and disproportionality analysis

Empirical Bayes shrinkage estimation of log ratios

Combining results from multiple safety studies by meta-analysis

Fixed effects model for combining studies

Random effect model for combining studies

Meta-analysis of rare events

Meta-analysis using individual subject data

Bayesian meta-analysis

Pharmacoepidemiologic approaches

Information content differences among different safety databases and from web-based epidemiologic studies

Case-control and self-controlled case series (SCCS) approaches

OMOP and systematic pharmacovigilance

Postmarketing pharmacoepidemiologic studies: Examples from biologic therapies

Quantitative signal detection and machine learning

Likelihood ratio tests

Supplements and problems

8. Sequential Methods for Safety Surveillance

Overview of sequential change detection and diagnosis

Combining sequential testing and detection for pharmacovigilance

MaxSPRT, CMaxSPRT and sequential GLR tests

Implementation of CMaxSPRT

Adjustment for confounding

Supplement and problems



View More



Jie Chen is a distinguished scientist at Merck Research Laboratories. He has more than 20 years of experience in biopharmaceutical R&D with research interest in the areas of innovative trial design, data analysis, Bayesian methods, multiregional clinical trials, data mining and machining learning methods, and medical product safety evaluation.

Joseph F. Heyse is a Scientific Assistant Vice President at Merck Research Laboratories, Fellow of the ASA and AAAS, and founding editor of Statistics in Biopharmaceutical Research. He has more than 40 years of experience in pharmaceutical R&D with research interest in safety evaluation and health economics and has more than 70 publications in peer reviewed journals. He is an editor of Statistical Methods in Medical Research.

Tze Leung Lai is the Ray Lyman Wilbur Professor of Statistics, and by courtesy, of Biomedical Data Science and Computational & Mathematical Engineering, and Co-director of the Center for Innovative Study Design at Stanford University. He is a Fellow of the IMS and ASA. His research interest includes sequential experimentation, adaptive design and control, change-point detection, survival analysis, time series and forecasting, multivariate analysis and machine learning, safety evaluation and monitoring. He has published 12 books and 300 articles in peer reviewed journals, and has supervised over 70 PhD theses at Columbia and Stanford Universities.


"This book provides comprehensive coverage of the statistical methods for evaluating medical product safety in different stages of development life-cycle: from pre-clinical to clinical, and to post marketing studies. As the evaluation of safety of medical products including drugs, vaccines, devices are becoming increasingly important, more and more novel and complex statistical methods have recently been proposed and used. This book gives a very detailed account of the framework for safety evaluation as well as in-depth descriptions of many advanced statistical methods. As far as I am aware, it is the only book that covers such a broad arrays of topics in safety evaluation. Therefore, this book should be appealing to a very large audience, including graduate students and professional statisticians in industry, government, and academia. I think it can be used as a textbook (as many parts of the materials have been used for short courses or part of graduate degree course) or a reference book for practicing statisticians... Overall, I found the book to be a very important contribution to the scientific community."
~Ivan Chan, AbbVie