Assessing the commonly used assumptions in estimating the principal causal effect in clinical trials

TL;DR

This paper evaluates the validity of key assumptions used in estimating principal causal effects in clinical trials, using a cross-over study to compare assumptions and estimate effects without relying on the cross-over design.

Contribution

It is the first study to empirically assess the plausibility of assumptions for principal causal effect estimation using a cross-over trial.

Findings

Monotonicity and within-treatment principal ignorability assumptions did not hold well.

Cross-world principal ignorability and principal strata independence assumptions appeared reasonable.

Estimates based on plausible assumptions aligned with cross-over data results.

Abstract

In clinical trials, it is often of interest to understand the principal causal effect (PCE), the average treatment effect for a principal stratum (a subset of patients defined by the potential outcomes of one or more post-baseline variables). Commonly used assumptions include monotonicity, principal ignorability, and cross-world assumptions of principal ignorability and principal strata independence. In this article, we evaluate these assumptions through a 2$\times$2 cross-over study in which the potential outcomes under both treatments can be observed, provided there are no carry-over and study period effects. From this example, it seemed the monotonicity assumption and the within-treatment principal ignorability assumptions did not hold well. On the other hand, the assumptions of cross-world principal ignorability and cross-world principal stratum independence conditional on baseline covariates seemed reasonable. With the latter assumptions, we estimated the PCEs, defined by whether the blood glucose standard deviation increased in each treatment period, without relying on the cross-over feature, producing estimates close to the results when exploiting the cross-over feature. To the best of our knowledge, this article is the first attempt to evaluate the plausibility of commonly used assumptions for estimating PCEs using a cross-over trial.