On adjustment for auxiliary covariates in additive hazard models for the analysis of randomized experiments

TY - JOUR

T1 - On adjustment for auxiliary covariates in additive hazard models for the analysis of randomized experiments

AU - Vansteelandt, S.

AU - Martinussen, Torben

AU - Tchetgen, E. J Tchetgen

PY - 2014/3

Y1 - 2014/3

N2 - We consider additive hazard models (Aalen, 1989) for the effect of a randomized treatment on a survival outcome, adjusting for auxiliary baseline covariates. We demonstrate that the Aalen least-squares estimator of the treatment effect parameter is asymptotically unbiased, even when the hazard's dependence on time or on the auxiliary covariates is misspecified, and even away from the null hypothesis of no treatment effect. We furthermore show that adjustment for auxiliary baseline covariates does not change the asymptotic variance of the estimator of the effect of a randomized treatment. We conclude that, in view of its robustness against model misspecification, Aalen least-squares estimation is attractive for evaluating treatment effects on a survival outcome in randomized experiments, and the primary reasons to consider baseline covariate adjustment in such settings could be interest in subgroup effects or the need to adjust for informative censoring or baseline imbalances. Our results also shed light on the robustness of Aalen least-squares estimators against model misspecification in observational studies.

AB - We consider additive hazard models (Aalen, 1989) for the effect of a randomized treatment on a survival outcome, adjusting for auxiliary baseline covariates. We demonstrate that the Aalen least-squares estimator of the treatment effect parameter is asymptotically unbiased, even when the hazard's dependence on time or on the auxiliary covariates is misspecified, and even away from the null hypothesis of no treatment effect. We furthermore show that adjustment for auxiliary baseline covariates does not change the asymptotic variance of the estimator of the effect of a randomized treatment. We conclude that, in view of its robustness against model misspecification, Aalen least-squares estimation is attractive for evaluating treatment effects on a survival outcome in randomized experiments, and the primary reasons to consider baseline covariate adjustment in such settings could be interest in subgroup effects or the need to adjust for informative censoring or baseline imbalances. Our results also shed light on the robustness of Aalen least-squares estimators against model misspecification in observational studies.

KW - Additive hazard model

KW - Model misspecification

KW - Randomized experiment

KW - Robustness

KW - Survival time

UR - http://www.scopus.com/inward/record.url?scp=84897745953&partnerID=8YFLogxK

U2 - 10.1093/biomet/ast045

DO - 10.1093/biomet/ast045

M3 - Journal article

AN - SCOPUS:84897745953

SN - 0006-3444

VL - 101

SP - 237

EP - 244

JO - Biometrika

JF - Biometrika

IS - 1

ER -