Covariate adjustment for two-sample treatment comparisons in randomized clinical trials: A principled yet flexible approach

By Anastasios A. Tsiatis, Marie Davidian, Xiaomin Lu. 2008

394 citations recorded by the source corpus.

There is considerable debate regarding whether and how covariate‐adjusted analyses should be used in the comparison of treatments in randomized clinical trials. Substantial baseline covariate information is routinely collected in such trials, and one goal of adjustment is to exploit covariates associated with outcome to increase precision of estimation of the treatment effect. However, concerns are routinely raised over the potential for bias when the covariates used are selected post hoc and the potential for adjustment based on a model of the relationship between outcome, covariates, and treatment to invite a ‘fishing expedition’ for that leading to the most dramatic effect estimate. By appealing to the theory of semiparametrics, we are led naturally to a characterization of all treatment effect estimators and to principled, practically feasible methods for covariate adjustment that yield the desired gains in efficiency and that allow covariate relationships to be identified and exploited while circumventing the usual concerns. The methods and strategies for their implementation in practice are presented. Simulation studies and an application to data from an HIV clinical trial demonstrate the performance of the techniques relative to the existing methods. Copyright © 2007 John Wiley & Sons, Ltd.

Subfields: Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Statistics and Probability

1. Peng Ding
2. Xinran Li
3. Anqi Zhao
4. C. Fogarty
5. Karel Vermeulen
6. W. Lin
7. Stijn Vansteelandt
8. Kevin Guo

Citations in the corpus

• Covariate-adjusted Fisher randomization tests for the average treatment effect (2020, background): “It is also conventional to view the experimental units as random samples from a super population (Tsiatis et al. 2008; Berk et al. 2013; Negi and Wooldridge 2020; Ye et al. 2020).”
• Regression‐assisted inference for the average treatment effect in paired experiments (2016, methodology): “A natural and important extension of this work would be to consider other forms of regression adjustment in paired experiments by leveraging semiparametric theory, aligning with the approach taken for completely randomized designs by Zhang et al. (2008) and Tsiatis et al. (2008).”
• Rerandomization and regression adjustment (2019, methodology): “As a side note, Lin (2013)’s estimator also appeared in the semiparametric efficiency theory for the average treatment effect under independent sampling from a superpopulation (Koch et al. 1998; Yang and Tsiatis 2001; Leon et al. 2003; Tsiatis et al. 2008; Rubin and van der Laan 2011).”
• The International Journal of Biostatistics Empirical Efficiency Maximization : Improved Locally Efficient Covariate Adjustment in Randomized Experiments and Survival Analysis (2011, neutral): “Under the local efficiency framework of Robins and Rotnitzky (1992) and subsequent papers, adjustment in experiments is also considered by Tsiatis et al. (2007) and Moore and van der Laan (2007).”
• Agnostic notes on regression adjustments to experimental data: Reexamining Freedman's critique (2012, methodology): “(ii) With two treatment groups of equal size, the conventional OLS variance estimator for ÂTEadj is also consistent or asymptotically conservative [Freedman (2008a)]. (iii) Freedman (2008a) shows analogous results for variance estimators for the difference in means; the issue there is whether to ass”
• On Optimal Treatment Regimes Selection for Mean Survival Time (2014, neutral): “As in [28] and [29], besides the treatment indicators, we considered 12 covariates, including five continuous variables: age (years), weight (kg), Karnofsky score (scale of 0–100), CD4 count (cells/mm3) at baseline and CD8 count (cells/mm3) at baseline, and seven binary covariates: hemophilia (0 = n”
• Bias-Reduced Doubly Robust Estimation (2015, methodology): “Efficient consistent and asymptotically normal estimator ν̂n(Heff) The Theorem below shows that efficient estimation of the MPI under modelMindep relies on the CPI P(Y Y ∗|A = 0,A∗ = 1,X ,X ∗), just like efficient estimation of the additive treatment effect demands the conditional expectations E(Y |”