Subgroup analysis, covariate adjustment and baseline comparisons in clinical trial reporting: current practiceand problems

By Stuart J. Pocock, S. Assmann, L. Enos, L. Kasten. Statistics in Medicine, 2002

1,118 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/7a0e956afd6a310666b032deae670a0008bda805

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

Citations in the corpus

Bias-Reduced Doubly Robust Estimation (2015, background): “A first reason for this concern is the post hoc selection of covariates: the fact that covariate adjustment may prompt fishing expeditions of those covariates that yield the largest estimate or the smallest p-value for the treatment effect (Pocock et al. 2002).”
The performance of different propensity score methods for estimating marginal hazard ratios (2007, neutral): “, marginal) estimate [27].”
Misunderstandings between experimentalists and observationalists about causal inference (2008, result): “…of randomized experiments, where researchers have shown that even small (and statistically insignificant) differences in important (or what they call in this literature ‘prognostic’) covariates can result in large differences in the results of the experiment (Senn, 1994; Pocock et al., 2002).”
Subgroup analysis in clinical trials (2004, background): “au EBM: Trials on Trial whether there are subgroups of trial participants who a more (or less) likely to be helped (or harmed) by th intervention under investigation, and a recent survey trials published over 3 months in four leading journa found that 70% included subgroup analyses.(1,2) Furthe more”
Covariate adjustment for two-sample treatment comparisons in randomized clinical trials: A principled yet flexible approach (2008, background): “However, covariate adjustment has inspired considerable controversy among numerous authors [1, 7–9] and regulatory authorities [10, 11] because of the potential for biased estimation due to post hoc selection of covariates and, more ominously, the temptation for analysts to engage in a ‘fishing expe”
Estimation of Heterogeneous Treatment Effects from Randomized Experiments, with Application to the Optimal Planning of the Get-Out-the-Vote Campaign (2011, methodology): “We discuss our methodology in the context of getout-the-vote randomized field experiments and show how the proposed two-step framework can be applied in real-world settings.”
The International Journal of Biostatistics Empirical Efficiency Maximization : Improved Locally Efficient Covariate Adjustment in Randomized Experiments and Survival Analysis (2011, neutral): “= −E(O,U)∼L(O)×G0 [( I(T̃ ≥ U) G(3)0(u) )( ∆ψ(T̃ ,W )G0(U) G0(T̃ ) Q(U,W ) ] (20)”
Adjusting treatment effect estimates by post‐stratification in randomized experiments (2013, background): “However, in practice covariate selection is often done after-the-fact in part because, as is pointed out by Pocock et al. (2002), it is often quite difﬁcult to know which of a set of covariates are signiﬁcantly prognostic a priori .”
Modeling heterogeneous treatment effects in large-scale experiments using Bayesian Additive Regression Trees (2010, background): “Since measures of statistical uncertainty are usually not adjusted for this specification search, standard errors and confidence intervals will understate the uncertainty of the resulting CATE estimates (Pocock et al. 2002; Gabler et al. 2009).”
Subgroup Identification and Variable Selection from Randomized Clinical Trial Data. (2013, background): “A well-known risk in subgroup analysis is that of false positives [Yusuf et al., 1991, Peto et al., 1995, Assmann et al., 2000, Brookes et al., 2001, Cui et al., 2002, Pocock et al., 2002, Brookes et al., 2004, Rothwell, 2005, Lagakos, 2006, Wang et al., 2007].”
Towards causal benchmarking of bias in face analysis algorithms (2020, methodology): “Since any observable confounder may be annotated inDtransect, we can employ covariate-adjusted ATE estimators [63, 65, 78].”