Design and outcome-model estimability in doubly robust survival analysis: a breast cancer reanalysis
- 1Lakehead University
Abstract
Doubly robust survival estimators require nuisance models that the available information can support. We examine design and estimator dependence in TCGA breast cancer data. Treatment indicators disagree for 337 of 1076 patients; 67.4% have unobserved five-year binary outcomes; and receptor imputation expands the triple-negative subgroup from 156 directly measured patients to 226. We compare chemotherapy initiation by day 90 with non-initiation, using adjusted restricted mean survival time (RMST) over the next 1095 days. Weighting-only estimates are 55 days (95% bootstrap interval: -24, 133) in triple-negative disease (16 deaths) and 38 days (12, 72) in the whole cohort (72 deaths). A Cox outcome-model benchmark yields 108 and 35 days, respectively, demonstrating material model dependence in the smaller population. A parameter-to-information ratio summarizes the burden on complete-outcome regressions within cross-fitting. Its simulation-calibrated warning level is 0.31 (0.28-0.35) for the original regime; additional simulations examine covariate-dependent censoring, delayed benefit and nonlinear misspecification. The ratio does not certify estimator validity or replace weight diagnostics. The contribution is a reproducible workflow that separates design, nuisance-model support and estimator stability before interpreting observational survival contrasts.
Related
- Baseline Tumor Transcriptional Features and Event-Free Survival After Anti-CD19 CAR T-Cell Therapy with Axicabtagene Ciloleucel in Large B-Cell Lymphoma: An Exploratory Reanalysis of the Randomized ZUMA-7 Trial
- Prognostic Modeling of Lysosomal Cell Death Signatures in Lung Adenocarcinoma: A Machine Learning Framework for Survival Outcome Prediction
- Therapies for Alcohol Use Disorder in Alcohol-associated Hepatitis Clinical Trials: Integrated Trial Design with a Composite Clinical Outcome as the Primary Endpoint
- Post-90-Day Survival After Stroke According to 90-Day Functional Outcome in the 2013 and 2020 Cohorts: A Swedish Stroke Register Study
- Establishing a core outcome set for use in trials of severe acute paediatric asthma exacerbation (2–17 years): study protocol for an international co-design Delphi survey – a Paediatric Emergency Research Network study initiative
BibTeX
@misc{fahmy2026design,
title = {{Design and outcome-model estimability in doubly robust survival analysis: a breast cancer reanalysis}},
author = {W. Fahmy and E. Krikun and A. Alkhateeb},
howpublished = {medRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.29.26364358},
url = {https://www.medrxiv.org/content/10.64898/2026.09.29.26364358v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Design and outcome-model estimability in doubly robust survival analysis: a breast cancer reanalysis
AU - W. Fahmy
AU - E. Krikun
AU - A. Alkhateeb
PY - 2026
DA - 2026/09/29
PB - medRxiv
DO - 10.64898/2026.09.29.26364358
UR - https://www.medrxiv.org/content/10.64898/2026.09.29.26364358v1
AB - Doubly robust survival estimators require nuisance models that the available information can support. We examine design and estimator dependence in TCGA breast cancer data. Treatment indicators disagree for 337 of 1076 patients; 67.4% have unobserved five-year binary outcomes; and receptor imputation expands the triple-negative subgroup from 156 directly measured patients to 226. We compare chemotherapy initiation by day 90 with non-initiation, using adjusted restricted mean survival time (RMST) over the next 1095 days. Weighting-only estimates are 55 days (95% bootstrap interval: -24, 133) in triple-negative disease (16 deaths) and 38 days (12, 72) in the whole cohort (72 deaths). A Cox outcome-model benchmark yields 108 and 35 days, respectively, demonstrating material model dependence in the smaller population. A parameter-to-information ratio summarizes the burden on complete-outcome regressions within cross-fitting. Its simulation-calibrated warning level is 0.31 (0.28-0.35) for the original regime; additional simulations examine covariate-dependent censoring, delayed benefit and nonlinear misspecification. The ratio does not certify estimator validity or replace weight diagnostics. The contribution is a reproducible workflow that separates design, nuisance-model support and estimator stability before interpreting observational survival contrasts.
ER -