Advanced Strategies for Longitudinal Causal Inference and Treatment Switching

In many late phase oncology randomized controlled trials (RCTs), control arm patients are permitted to take active treatment (1-way crossover), or patients in both control and active arms are permitted to take alternative treatments (2-way treatment switching) after disease progression due to ethical considerations. In both situations, the effect of active intervention on overall survival (OS) is no longer directly observable. The intent-to-treat (ITT) analysis of the observed data will reflect the trial outcome per the treatment policy strategy but may not be able to make causal inference for the active intervention effect on OS. The latter is important for the payer agency's evaluation and is helpful for regulatory decisions on drug applications. During the last decade, several complex statistical methods have been adapted and applied to RCTs to recover the causal OS effect of randomized active intervention under settings that allow for treatment switching. These methods included but not limited to Marginal Structural Model (MSM), Two-Stage Estimation (TSE), Inverse Probability of Censoring Weighting (IPCW), Rank Preserving Structure Failure Time Model (RPSFTM), etc. This talk will review these methods, regulatory guidance and strategies how to select appropriate adjusted analysis methods at the RCT design stage. Case studies will also be presented to illustrate the pros and cons along with practical issues when each method is applied under the RCT setting.  

The Average Treatment Effect on the Treated (ATT) is a common causal parameter defined as the average effect of a binary treatment among the subset of the population receiving treatment. We propose a novel family of parameters, Generalized ATTs (GATTs), that generalize the concept of the ATT to longitudinal data structures, multi-valued or continuous treatments, and conditioning on arbitrary treatment subsets. We provide a formal causal identification result that expresses the GATT in terms of sequential regressions, and derive the efficient influence function of the parameter, which defines its semi-parametric efficiency bound. Efficient semi-parametric inference of the GATT requires estimating the ratios of functions of conditional probabilities (or densities); we propose directly estimating these ratios via empirical loss minimization, drawing on the theory of Riesz representers. Simulations suggest that estimation of the density ratios using Riesz representation have better stability in finite samples. Lastly, we illustrate the use of our methods to evaluate the effect of chronic pain management strategies on the development of opioid use disorder among Medicare patients with chronic pain.  

A standard assumption for causal inference about the joint effects of time-varying treatment is that one has measured sufficient covariates to ensure that within covariate strata, subjects are exchangeable across observed treatment values, also known as 'sequential randomization assumption (SRA)'. SRA is often criticized as it requires one to accurately measure all confounders. Realistically, measured covariates can rarely capture all confounders with certainty. Often covariate measurements are at best proxies of confounders, thus invalidating inferences under SRA. In this paper, we extend the proximal causal inference (PCI) framework of Miao, Geng, et al. (2018. Identifying causal effects with proxy variables of an unmeasured confounder. Biometrika, 105(4), 987–993. https://doi.org/10.1093/biomet/asy038) to the longitudinal setting under a semiparametric marginal structural mean model (MSMM). PCI offers an opportunity to learn about joint causal effects in settings where SRA based on measured time-varying covariates fails, by formally accounting for the covariate measurements as imperfect proxies of underlying confounding mechanisms. We establish nonparametric identification with a pair of time-varying proxies and provide a corresponding characterization of regular and asymptotically linear estimators of the parameter indexing the MSMM, including a rich class of doubly robust estimators, and establish the corresponding semiparametric efficiency bound for the MSMM. Extensive simulation studies and a data application illustrate the finite sample behaviour of proposed methods. 

Data collected in the context of usual care present a rich source of longitudinal data for research, but often require analyses that simultaneously enable causal inferences in the presence of treatment switching, and handle irregular and informative assessment times. Assessment not at random occurs when outcome and assessment process remain dependent on conditioning upon observed variables; approaches to causal inference in this context are limited. This talk will show how a process known as multiple outputation can be used to simultaneously handle longitudinal treatment confounding and a special case of assessment not at random, where assessment and outcome processes are conditionally independent given past observed covariates and random effects.