Fall 2008 Meeting
Held on October 16,
2008 at the Radisson Hotel Northbrook.
The Program consisted of three presentations:
Trial Design Questions and Answers
Peter A. Lachenbruch,
Oregon State University
Clinical Trials with Dropout: Longitudinal Assessment of Chronic Pain
Ronald A. Thisted, Ph.D., University
Distributed Lag Models: Estimating Effects of Particulate Matter Air
Pollution on Daily Mortality
Leah Welty, Ph.D., Northwestern
Some Clinical Trial Design Questions
and Answers by Peter A. Lachenbruch, Ph.D.,
Oregon State University
Dr. Peter Lachenbruch received his Ph. D. from
UCLA in Biostatistics. He has held positions on the faculties of the
University of North Carolina (1965-1976), the University of Iowa
(1976-1985), and UCLA (1985-1984). He was employed by the FDA/CBER from 1994
to 2005 and recently retired as the Director of the Division of
Biostatistics. He is currently Professor of Public Health at Oregon State
University (2006– present). He is a Fellow of the American Statistical
Association and a former elected member of the International Statistical
Institute. He has held many professional offices and is the President of the
American Statistical Association for 2008.
He has statistical interests in Discriminant
Analysis, Two-part Models, Model-Independent Inference, Statistical
Computing, and Data Analysis. He has application interests in Rheumatology,
Psychiatry, Pediatrics, Gerontology and Accident Epidemiology. He has more
than 180 publications in these fields.
Dr. Lachenbruch serves on the Editorial Boards of Statistics in Medicine,
Methods of Information in Medicine, Journal of Biopharmaceutical
Statistics, and Statistical Methods in Medical Research. He
served on advisory panels to the George Mason University Department of
Statistics, the Ohio State University Department of Statistics, Cytel. He
serves on the DSMB to a VA clinical trial, an NIH clinical trial and on the
OSMB for the Women’s Health Initiative.
Many questions arise in clinical trials by
those who are just beginning, and, of course, some gaps are in each of our
knowledge bases. This talk is in response to questions I’ve fielded over
several years. Some of the answers represent my opinions and some the
experience I’ve had at the FDA and elsewhere. If you are preparing a
submission for the FDA, you should rely on their insights. Here are the
• What pitfalls do the FDA see when
information from preclinical data (or early clinical data on a similar
investigational product) is formulated into a Phase I protocol?
• What are the various types of study designs available when there are more
than two comparators?
• What controls are appropriate when there cannot be any blinding in the
• What are the problems seen by the FDA with randomization in trials?
• What are the steps in designing a dose-ranging/dose-escalation study?
• How does one deal with multiple variables that will affect the outcome
measure (with an understanding of fixed randomization schemes and
adaptive/dynamic randomization schemes)?
• What factors are used to estimate sample size?
• How does the investigator choose the margin of equivalence or
non-inferiority (delta or
comparative clinical trials?
• How does the investigator deal with missing data?
• How does an investigator determine what data should and should not be
included in analyses (especially in the cases of protocol violations,
withdrawals and drop-outs)?
• What types of analyses should be used
when there are multiple time points
(multiple observations) of data collection (and there is no dichotomous
outcome / endpoint)?
• In analyzing data from clinical trials involving multiple sites, should
site be treated as a fixed or random effect?
• When should a planned interim analysis (for safety and/or efficacy and/or
sample size reestimation) be appropriate? What are the pros and cons? What
are the methods used?
Clinical Trials with
Dropout: Longitudinal Assessment of Chronic Pain by
Ronald A. Thisted, Ph.D., University of Chicago
Ronald Thisted received his Ph.D. in Statistics
from Stanford in 1977. He has been on the faculty in the Department of
Statistics at the University of Chicago since 1976. Since 1999 he has also
been Chairman of the Department of Health Studies, which covers the areas of
biostatistics, epidemiology, and health services research. He is active in
the design and analysis of clinical trials, with an emphasis on Phase III
studies. Since 1979 he has consulted with the pharmaceutical industry on
study design and analysis methods, and he has served as an expert witness in
pharmaceutical patent litigation. In addition to methodological issues
related to clinical trials, his research also includes computational methods
for statistics, statistical studies of authorship, and observational
Some patients in long-term treatment
trials withdraw from the study before their participation is scheduled to
end, so that outcome data for a subset of patients is incomplete. Methods
for handling missing data include analysis only of complete cases, ad hoc
approaches to imputation such as carrying forward baseline or most recent
observations, and model-based methods such as mixed effects models. In the
context of evaluating treatments for chronic pain, regulators have suggested
that replacing missing outcome data with patients' baseline values—the
baseline carried- forward (BCF) algorithm—leads to a conservative estimate
of treatment effect. Many factors can lead to dropout, including side
effects, disease remission or exacerbation, inability to comply with study
requirements, unrelated illness, or death. We examine how the validity of
estimates and tests for treatment effects for several missing-data methods
are affected by underlying models for pain response and by the factors that
lead to dropout. We examine the BCF algorithm in some detail, focusing on
validity, conservativeness, sample size implications, pitfalls, and
Bayesian Distributed Lag
Models: Estimating Effects of Particulate Matter Air Pollution on Daily
Mortality by Leah J. Welty, Ph.D.,
Assistant Professor, Department of Preventive
Medicine, Northwestern University. Postdoc, Dept of Biostatistics, Johns
Hopkins University. PhD from the University of Chicago, 2003
A distributed lag model (DLM) is a
regression model that includes lagged exposure variables as covariates; its
corresponding distributed lag (DL) function describes the relationship
between the lag and the coefficient of the lagged exposure variable. DLMs
have recently been used in environmental epidemiology for quantifying the
cumulative effects of weather and air pollution on mortality and morbidity.
Standard methods for formulating DLMs include unconstrained, polynomial, and
penalized spline DLMs. These methods may fail to take full advantage of
prior information about the shape of the DL function for environmental
exposures, or for any other exposure with effects that are believed to
smoothly approach zero as lag increases, and are therefore at risk of
producing sub-optimal estimates. We propose a Bayesian DLM (BDLM) that
incorporates prior knowledge about the shape of the DL function and also
allows the degree of smoothness of the DL function to be estimated from the
data. In a simulation study, we compare our Bayesian approach with
alternative methods that use unconstrained, polynomial and penalized spline
DLMs. We also show that BDLMs encompass penalized spline DLMs: under certain
assumptions, imposing a prior on the DL coefficients is analogous to
smoothing the DL coefficients with a penalty specified by the prior. We
apply our BDLM to data from the National Morbidity, Mortality, and Air
Pollution Study (NMMAPS) to estimate the short term health effects of
particulate matter air pollution on mortality from 1987-2000 for Chicago,