Chairs:
Haitao Chu, MD, PhD (Pfizer Inc.)
Xiang Zhang, PhD (CSL Behring)
Abstract: Comparative effectiveness research (CER) depends on rigorous analytical methods to evaluate the relative benefits and risks of treatments, especially when direct head-to-head evidence is lacking. As healthcare decision-makers increasingly rely on evidence synthesis for regulatory, reimbursement, and clinical policy decisions, advancing CER methodology is essential for ensuring transparency, rigor, and reliable inference. This session features four researchers presenting innovative statistical methods that strengthen the validity and practical impact of CER for informed healthcare decisions.
Dr. Hwanhee Hong (Duke University) presents Simulation-Based Methods for Power Calculation in Bayesian Network Meta-Analysis, introducing flexible simulation frameworks for estimating power under contrast-based and arm-based Bayesian NMA models. Through extensive simulations, the work clarifies how network structure, evidence balance, effect sizes, heterogeneity, and sample sizes collectively influence power, offering practical guidance for designing and interpreting Bayesian NMAs.
Dr. Moming Li (AbbVie) discusses Missing Data Handling in the Application of Matching-Adjusted Indirect Comparison, addressing a key gap in population-adjusted indirect comparisons used in HTA. The proposed weighting-based approaches integrate inverse probability weighting for missing outcomes and covariates directly into the MAIC framework, improving robustness when combining IPD and aggregate data.
Dr. Joseph G. Ibrahim (UNC–Chapel Hill) presents A Cure Rate Joint Model for Time-to-Event Data and Longitudinal Tumor Burden with Potential Change Points, proposing a flexible joint modeling strategy for capturing complex tumor burden trajectories in oncology trials. By identifying individualized change points and leveraging MCEM estimation, the model improves characterization of disease progression and demonstrates strong performance in a Phase 3 NSCLC trial.
Finally, Dr. Jiajia Zhang (University of South Carolina) introduces Learning Individualized Treatment Rules with Optimal Treatment Grouping for Survival Outcomes. This Cox-type model with supervised adaptive fusion clustering (CSCAF) simultaneously identifies treatment groupings and individualized treatment rules, showing strong performance in simulations and a large head and neck cancer dataset.
Together, these presentations advance the methodological foundations of CER, offering innovative solutions for power assessment, missing data, joint modeling, and individualized decision-making.
Speaker: Hwanhee Hong, PhD (Duke University)
Title: Simulation-Based Methods for Power Calculation in Bayesian Network Meta-Analysis
Abstract: Bayesian network meta-analysis (NMA) has become a popular tool for simultaneous comparisons of multiple treatments by synthesizing direct and indirect evidence from multiple clinical trials. However, methods for power calculation in Bayesian NMA remain underdeveloped, and the factors that influence power—such as evidence structure and effect size—are not well understood. In this study, we propose simulation-based methods for power calculation in Bayesian NMAs under both contrast-based and arm-based models. Using extensive simulations, we investigate how the network composition—particularly the balance of direct and indirect comparisons—affects power across varying network structures and heterogeneity levels. Results show that direct evidence consistently provides greater power than indirect evidence, with the strongest gains observed in closed-loop networks. Power also increases with larger effect sizes, more studies with larger sample sizes, and lower between-study heterogeneity. Indirect evidence contributes meaningfully to power only when anchored by adequate direct comparisons. We further illustrate the methods using two real-world case studies, demonstrating their practical application and interpretation. Our findings provide practical guidance for evaluating statistical power and enhancing the reliability of Bayesian NMAs.
Speaker: Morning Li, PhD (AbbVie Inc.)
Title: Missing data handling in the Application of Matching-Adjusted Indirect Comparison
Abstract: In support of Health Technology Assessment submission, we often need to conduct indirect treatment comparisons (ITC). One common type of ITC is population-adjusted indirect comparisons, in which individual patient data in one trial and aggregate data in the other trial are used to adjust for the difference in the distributions of covariates (prognostic factors or effect modifiers) that influence the outcome. The most popular PAIC method is the Matching- Adjusted Indirect Comparison (MAIC). However, the literature lacks guidance on how to handle missing data in the application of MAIC. In this talk, we propose some weighting-based methods to handle missing data in the outcome variable and/or covariates when applying MAIC. These weights can be expressed as products of the inverse probability of not missing the outcomes and weights that account for the difference in baseline characteristics. The proposed methods fit seamlessly into the original MAIC framework and obtain treatment effect estimates based on weighted difference between IPD and AgD.
Speaker: Joseph G. Ibrahim, PhD (UNC-Chapel Hill)
Title: Cure Rate Joint Model for Time-to-Event Data and Longitudinal Tumor Burden with Potential Change Points
Abstract: In non-small cell lung cancer (NSCLC) clinical trials, tumor burden (TB) is a key longitudinal biomarker for assessing treatment effects. Typically, standard-of-care (SOC) therapies and some novel interventions initially decrease TB; however, many patients subsequently experience an increase—indicating disease progression—while others show a continuous decline. In patients with an eventual TB increase, the change point marks the onset of progression and must occur before the time of the event. To capture these distinct dynamics, we propose a novel joint model that integrates time-to-event and longitudinal TB data, classifying patients into a change-point group or a stable group. For the changepoint group, our approach flexibly estimates an individualized change point by leveraging time-to-event information. We use a Monte Carlo Expectation-Maximization (MCEM) algorithm for efficient parameter estimation. Simulation studies demonstrate that our model outperforms traditional approaches by accurately capturing diverse disease progression patterns and handling censoring complexities, leading to robust marginal TB outcome estimates. When applied to a Phase 3 NSCLC trial comparing cemiplimab monotherapy to SOC, the treatment group shows prolonged TB reduction and consistently lower TB over time, highlighting the clinical utility of our approach. The implementation code is publicly available on https://github.com/quyixiang/JoCuR.
Speaker: Jiajia Zhang, PhD (University of South Carolina)
Title: Learning Individualized Treatment Rules with Optimal Treatment Grouping for Survival Outcomes
Abstract: Many methods for learning Individualized Treatment Rules (ITR) with survival outcomes have been proposed in the precision medicine literature. However, most existing methods focus on problems with a moderate number of treatment arms, such as binary and triple arms. When the treatment space becomes large and some arms have similar effects, the existing methods may provide suboptimal estimation due to small sample sizes for certain treatment arms. To fill the research gap, we develop a Cox-type model incorporating Supervised Clustering approach based on Adaptive Fusion (CSCAF) which can simultaneously learn the treatment grouping and estimate the optimal ITR maximizing the potential survival times. The optimization function is a penalized likelihood function and a computing algorithm utilizing group-lasso and accelerated proximal gradient descent is provided. Our simulations with varying data generation processes, sample sizes, and censoring rates demonstrate the model’s superior performance. Finally, we apply the CSCAF model to head and neck cancer (HNC) datasets, developing ITRs for the 1,625 HNC patients that underwent chemotherapy and identifying the optimal four treatment groups.
