2023 Short Courses

1. Bayesian Approaches for Accelerating Drug and Medical Device Approvals

Instructors:

• Dr. Bruno Boulanger and Dr. Brad Carlin, PharmaLex

Abstract:

Thanks to the sudden emergence of Markov chain Monte Carlo (MCMC) computational methods in the 1990s, Bayesian methods now have a more than 25-year history of utility in statistical and biostatistical design and analysis.  However, their uptake in regulatory science has been much slower, due to the high premium this field places on Type I error control, and its historical reliance on p-values and other traditional frequentist statistical tools.  Fortunately, recent actions by regulators at FDA and elsewhere have indicated a new willingness to consider more innovative statistical methods, especially in settings where traditional methods are ill-suited or demonstrably inadequate.

In this half-day short course, after a very brief review of the Bayesian adaptive approach to clinical trial design and analysis, we will discuss a variety of areas in which Bayesian methods offer a better (and perhaps the only) path to regulatory approval.  Topics to be covered are expected to include:

• Leveraging historical controls and other auxiliary data (power/commensurate/robust mixture priors)
• Methods for rare and pediatric disease (including those utilizing patient natural history data)
• Bioequivalence and noninferiority studies
• Platform/umbrella/seamless multiphase studies
• Causal inference tools to incorporate real world data (RWD)/real world evidence (RWE), including synthetic controls

If time permits, we will also mention novel Bayesian methods for early-phase studies (BOIN and its “descendants”), as well as fully Bayesian approaches for benefit-risk tradeoff in economic decision making (e.g., Phase III go/no-go).  Our presentation will be illustrated with real-data examples and illustrations, as well as associated computer code that attendees can try at home.  The presentation will conclude with a brief review of existing FDA programs designed to encourage the use of Bayesian and other novel statistical approaches, as well as corresponding white papers and other documents from EMA and other non-North American regulatory agencies.

References:

1. Berry, S.M., Carlin, B.P., Lee, J.J., and Muller, P. (2011). Bayesian Adaptive Methods for Clinical Trials.  Boca Raton, FL:  Chapman and Hall/CRC Press.
2. Carlin, B.P. and Louis, T.A. (2009). Bayesian Methods for Data Analysis, 3rd ed. Boca Raton, FL:  Chapman and Hall/CRC Press.
3. Lesaffre, E., Baio, G., and Boulanger, B., eds. (2020). Bayesian Methods in Pharmaceutical Research.  Boca Raton, FL: Taylor and Francis/CRC Press.

2. Statistical Analysis of Zero-Inflated Continuous Data 

• Dr. Lei Liu, Washington University

Abstract:

Zero-inflated continuous (or semi-continuous) data arise frequently in medical, economical, and ecological studies. Such data are often characterized by the presence of a large portion of zero values, in addition to continuous non zero (i.e., positive) values that are often skewed to the right and heteroscedastic. Both features suggest that no simple parametric distribution is suitable for describing such ?zero-inflated continuous? data. Examples include, though certainly are not limited to, substance abuse, medical costs, medical care usage, microbiome abundance, and single cell expression levels. The topic will be of interest to a wide variety of researchers.

In this short course we will review statistical methods to analyze zero-inflated continuous data. We will start from the cross-sectional zero-inflated continuous data. Three approaches are presented to account for the point mass at zero: a two-part model; a sample selection approach (e.g., Tobit model); and a zero-inflated Tobit model. We will then introduce flexible models to characterize right skewness and heteroscedasticity in the positive values.

The second section involves modeling repeated measures zero-inflated continuous data. Random effects will be used to tackle the correlation on repeated measures of the same subject and that across different parts of the model. We will incorporate such random effects to the models introduced in Section 1. We will also present joint models of longitudinal zero-inflated continuous data and survival, to account for the possible dependent terminal event or informative dropout.

Finally, we will present applications to real datasets to illustrate our methods. We will use alcohol drinking data, medical costs, and microbiome abundance data as examples. Codes will be provided to facilitate the applications of these methods. Model comparison will also be conducted.

The lecturer has over 15 years of hands-on experience in the analysis of zero-inflated continuous data. This application oriented short course is of interest to researchers who would apply up-to-date statistical tools to zero-inflated continuous data.

Learning outcomes

This short course is designed to familiarize attendees with advanced statistical techniques on the analysis of zero-inflated continuous data. Upon completion of this course students should be able to:

• Know the essential features of zero-inflated continuous data. Describe situations in which the introduced methods are needed so they can recognize such type of data in practice.
• Choose the correct model for the data analysis, understand the model assumptions.
• Based on sample SAS codes provided in the class, they can use SAS to apply the statistical models to their data analysis and interpret the output, including regression parameter estimates, hypothesis tests, and model comparison.
• Develop new statistical methods tailored to the specific data of their research interest.

Syllabus

Section I: Cross sectional zero-inflated continuous data.

1. Features of zero-inflated continuous data
2. Accounting for point mass at zero
   1. Two-part model (2PM) -> Four-part model
   2. Tobit model
   3. Zero-inflated Tobit model
3. Characterizing positive values
   1. Log normal distribution
   2. Generalized linear model, e.g., Gamma regression model
   3. Generalized Gamma distribution
   4. Normal distribution after Box-Cox transformation
   5. Log skew normal distribution
   6. Heteroscedasticity
   7. Nonparametric regression model with flexible mean and variance functions

Section II: Correlated zero-inflated continuous data

1. Random effects two-part model
   1. Within-part correlation and cross-part correlation
   2. Different choices in Part II
   3. Multi-level random effects 2PM
   4. Joint model of longitudinal zero-inflated continuous data with survival
2. Random effects Tobit model
3. Random effects zero-inflated Tobit model

Section III: Examples

1. Longitudinal medical costs for heart failure cohort
2. Clustered pharmacy cost data
3. Longitudinal daily drinking data from alcohol treatment trials
4. Microbiome abundance data