Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
Defining the estimand of interest in a clinical trial is crucial to align its planning, design, conduct, analysis, and interpretation. The need for more precise specifications of estimands is highlighted in the draft addendum ICH E9(R1) which was published for public consultation in August 2017. Although not explicitly mentioned in ICH E9(R1), the addendum brings causal reasoning – besides randomization and ITT – into our world of pharmaceutical statistics. In this webinar, we will discuss the link between the ICH E9(R1) and causal inference. Furthermore, per protocol analyses will be discussed from a causal inference perspective and a case study where a principal strata estimand was investigated will be presented.
For more information please click here.
Baldur Magnusson
Novartis Pharma AG
Using principal stratification to address post-randomization events: A case study
In a randomized controlled trial, occurrence of post-randomization events associated with treatment and the primary endpoint may complicate the interpretation of the overall treatment effect. In this presentation, we discuss how these events may be accounted for at the estimand and the estimator level in the context of a recent case study. We define a principal stratification estimand derived from the scientific question of interest. Consideration is given to identifying assumptions, model-based derivation of an estimator, handling of covariates and missing data. We also discuss the role of sensitivity analyses.
Please click here to view the slides.
Daniel Scharfstein
Professor of Biostatistics, Johns Hopkins Bloomberg School of Public Health
Estimands and Causal Inference
Recently, the ICH proposed an addendum to the E9 Guidance: Statistical Principles for Clinical Trials. This addendum is focused on estimands and sensitivity analysis for randomized trials with intercurrent events. In this webinar, I will discuss the potential outcomes framework for causal inference and use it to formally define estimands that address different types of intercurrent events. I will then discuss the assumptions required to identify these estimands from the observable data and discuss the important role of sensitivity analysis.
Please click here to view the slides.
Wanjie Sun
FDA/CDER/OB/DBVIII
Estimating Causal Effects in Clinical Endpoint Bioequivalence Studies in the Presence of Treatment Noncompliance and Missing Data
In clinical endpoint bioequivalence (BE) studies, the primary analysis for assessing equivalence between a generic and an innovator product is usually based on the observed per-protocol (PP) population (i.e., completers and compliers in general). The FDA Missing Data Working Group and the ICH E9 Revision 1 Working Group recommended using “causal estimands of primary interest.” The analysis based on the PP population, however, is not generally causal because PP is determined post-treatment, hence conditioning on it may introduce selection bias. To date, no causal inference has been proposed to assess to equivalence. In this paper, we propose a causal framework and co-primary causal estimands to test equivalence by applying Frangakis and Rubin (2002)’s principal stratification in causal inference. We identify three conditions when the current PP estimator is unbiased for one of the proposed co-primary causal estimands – the“Survivor Average Causal Effect” (SACE) estimand. Simulation was used to demonstrate the bias, type 1 error, and power associated with the PP estimator when these three conditions are not met. We also propose a tipping point sensitivity analysis to evaluate the robustness of the current PP estimator (primary analysis) in testing equivalence when the underlying sensitivity parameters vary across a clinically meaningful range. Data from a clinical endpoint BE study is used to illustrate the proposed co-primary causal estimands and sensitivity analysis method. Our work starts causal evaluation of equivalence assessment in clinical endpoint BE studies with non-compliance and missing data, and can be applied to clinical biosimilar and non-inferiority studies.
*The views expressed in this article represent the opinions of the authors, and do not represent the views and/or policies of the U.S. Food and Drug Administration.
To access the recording, please visit the Video-on-Demand Library.
