Nonparametric analysis of nonhomogeneous multi-state processes based on clustered observations

TL;DR

This paper develops nonparametric methods for analyzing complex multi-state event data with clustered observations, addressing dependence within clusters and providing valid inference tools for clinical trial data.

Contribution

It introduces nonparametric estimators and tests for multi-state models that handle dependence, informative cluster sizes, and both Markov and non-Markov processes, with theoretical guarantees.

Findings

Estimators are uniformly consistent and converge to Gaussian processes.

Proposed variance estimators and confidence bands are rigorously derived.

Ignoring within-cluster dependence can lead to invalid conclusions.

Abstract

Frequently, clinical trials and observational studies involve complex event history data with multiple events. When the observations are independent, the analysis of such studies can be based on standard methods for multi-state models. However, the independence assumption is often violated, such as in multicenter studies, which makes the use of standard methods improper. In this work we address the issue of nonparametric estimation and two-sample testing for the population-averaged transition and state occupation probabilities under general multi-state models based on right-censored, left-truncated, and clustered observations. The proposed methods do not impose assumptions regarding the within-cluster dependence, allow for informative cluster size, and are applicable to both Markov and non-Markov processes. Using empirical process theory, the estimators are shown to be uniformly consistent and to converge weakly to tight Gaussian processes. Closed-form variance estimators are derived, rigorous methodology for the calculation of simultaneous confidence bands is proposed, and the asymptotic properties of the nonparametric tests are established. Furthermore, we provide theoretical arguments for the validity of the nonparametric cluster bootstrap, which can be readily implemented in practice regardless of how complex the underlying multi-state model is. Simulation studies show that the performance of the proposed methods is good, and that methods that ignore the within-cluster dependence can lead to invalid inferences. Finally, the methods are applied to data from a multicenter randomized controlled trial.