10/17/2024: 10:00 AM - 11:30 AM EDT
Panel
Room: Grand Ballroom
Missing data is a common occurrence in medical research. Inappropriate handling of missing data can compromise the validity of study findings. Understand the underlying mechanisms of missing data and handling methods is essential to make informed decision about the most appropriate approach for specific cases. In this panel, we will explore various statistical methods that can be used to handle different types of missing data. Dr. Yun Wang from U.S. Food and Drug Administration will introduce five statistical methods to impute missing data occurring after intercurrent events in clinical trials to align with treatment policy strategy. Dr. Qianqian Wang from Eli Lilly and Company will discuss various scenarios of missing data occurrence in clinical trials and elucidate the rationale behind the selection of handling method along with illustrative case examples. Dr. Ge Zhao from Portland State University will present his research on a nested semiparametric model to analyze a case-control study where genuine case status is missing for some individuals. Dr. Lucy D'Agostino McGowan from Wake Forest University will discuss why including the outcome in an imputation model is not 'double dipping' or 'peeking' at the outcome in a way that can negatively impact the Type 1 error in studies along with practical advice.
Organizer
Qianqian(Jessie) Wang, Eli Lilly and Company
Chair
Qianqian(Jessie) Wang
Target Audience
Mid-Level
Tracks
Knowledge
Women in Statistics and Data Science 2024
Presentations
We propose a nested semiparametric model to analyze a case-control study with missingness on the genuine cases. The concept of non-case is introduced to allow imputing the missing genuine cases. The odds ratio parameter of the genuine cases is of interest. The imputation predicts the probability of genuine case over non-case semiparametrically in a dimension reduction fashion. This procedure is flexible, and vastly generalizes the existing methods. We establish the root-n asymptotic normality of the odds ratio parameter estimator. Our method yields a stable odds ratio parameter estimation owing to the application of an efficient semiparametric sufficient dimension reduction estimator. We conduct finite sample numerical simulations to illustrate the performance of our approach, and apply it to a dilated cardiomyopathy study.
Speaker
Ge ZHAO
Missing data with ignorable missingness or non-ignorable missingness often occur in clinical trials, which can arise from various sources. In this talk, we will first discuss various scenarios of missing data occurrence in clinical trials and elucidate the rationale behind the selection of handling method along with illustrative case examples, and then delve into a detailed discussion of Pattern mixture models (PMMs), which provide a flexible approach by considering different patterns of missingness and are commonly used to assess treatment effects for clinical trials with non-ignorable missing data.
Speaker
Qianqian(Jessie) Wang, Eli Lilly and Company
An often repeated question is whether including the outcome in an imputation model is 'double dipping' or 'peeking' at the outcome in a way that can negatively impact the Type 1 error in studies. This talk will dive into this myth and help dispel these concerns. We mathematically demonstrate that including the outcome variable in imputation models when using stochastic methods is required to avoid biased results. A discussion of these results along with practical advice will follow.
Speaker
Lucy D'Agostino McGowan, Wake Forest University
The primary objective of a clinical study is usually to assess a product's effectiveness and safety based on the planned treatment regimen instead of the actual treatment received. The estimand using the treatment policy strategy, which collects and analyzes data regardless of the occurrence of intercurrent events, is usually utilized to align with this study objective. In this presentation, the speaker will explain how missing data can be handled using the treatment policy strategy in connection with antihyperglycemic product development programs. Five statistical methods (retrieved dropouts, return-to baseline, placebo wash out, jump to reference, copy reference) to impute missing data occurring after intercurrent events will be discussed and compared via Markov Chain Monte Carlo simulations. Case studies will be presented to show how three of these five methods have been applied to estimate the treatment effects published in the labels for three antihyperglycemic agents currently on the market.
Speaker
Yun Wang, Food and Drug Administration