Combining Survival Trials Using Aggregate Data Based on Misspecified Models

TL;DR

This paper proposes a method to combine treatment effects from multiple clinical trials using aggregate data, addressing challenges posed by model misspecification in Cox proportional hazard models.

Contribution

It introduces an approach for unbiased aggregation of trial results under misspecified models, with efficient estimates and Wald tests based solely on aggregate data.

Findings

Provides a framework for robust data analysis with misspecified models.

Develops efficient estimators for combined hazard ratios.

Offers a Wald test for hypothesis testing using aggregate data.

Abstract

The treatment effects of the same therapy observed from multiple clinical trials can often be very different. Yet the patient characteristics accounting for these differences may not be identifiable in real world practice. There needs to be an unbiased way to combine the results from multiple trials and report the overall treatment effect for the general population during the development and validation of a new therapy. The non-linear structure of the maximum partial likelihood estimates for the (log) hazard ratio defined with a Cox proportional hazard model leads to challenges in the statistical analyses for combining such clinical trials. In this paper, we formulated the expected overall treatment effects using various modeling assumptions. Thus we are proposing efficient estimates and a version of Wald test for the combined hazard ratio using only aggregate data. Interpretation of the methods are provided in the framework of robust data analyses involving misspecified models.