Estimating restricted mean treatment effects with stacked survival models

  title={Estimating restricted mean treatment effects with stacked survival models},
  author={Andrew Wey and David M. Vock and John E. Connett and K.D. Rudser},
  journal={Statistics in Medicine},
  pages={3319 - 3332}
  • A. Wey, D. Vock, K. Rudser
  • Published 15 October 2014
  • Mathematics, Environmental Science
  • Statistics in Medicine
The difference in restricted mean survival times between two groups is a clinically relevant summary measure. With observational data, there may be imbalances in confounding variables between the two groups. One approach to account for such imbalances is estimating a covariate‐adjusted restricted mean difference by modeling the covariate‐adjusted survival distribution and then marginalizing over the covariate distribution. Because the estimator for the restricted mean difference is defined by… 

Non-parametric individual treatment effect estimation for survival data with random forests

The results indicate that the proposed method has a very good and stable performance to estimate the individual treatment effects and can detect a treatment effect in a sub-population even when the overall effect is small or nonexistent.

Stacked survival models for residual lifetime data

A stacking procedure to overcome the non-robustness of model misspecification is proposed; the proposed estimator comprises a linear combination of individual nonparametric/parametric survival function estimators, with optimal stacking weights obtained by minimizing a Brier Score loss function.

Summarizing causal differences in survival curves in the presence of unmeasured confounding

A semi-parametric procedure for adjusting survival curves for measured and unmeasured confounders, which augments the novel instrumental variable estimation method for survival time data in the presence of unme measured confounding with a procedure for mapping estimates onto the survival probability and the expected survival time scales.

The relationship between the C‐statistic and the accuracy of program‐specific evaluations

  • A. WeyN. Salkowski J. Snyder
  • Psychology
    American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons
  • 2019
A simulation study investigated the association of the C‐statistic with several metrics of program evaluation accuracy, including categorizing programs into the 5‐tier system and identifying programs for regulatory review, finding no information on the accuracy of program evaluations.

Association of pretransplant and posttransplant program ratings with candidate mortality after listing

  • A. WeyS. Gustafson J. Snyder
  • Medicine
    American journal of transplantation : official journal of the American Society of Transplantation and the American Society of Transplant Surgeons
  • 2019
For lung transplant patients, mortality after listing was highest at Programs with above‐ and below‐average transplant rates and lowest at programs with average transplant rates, suggesting that aggressive acceptance behavior may not always provide a survival benefit.

“Open mesh” or “strictly selected population” recruitment? The experience of the randomized controlled MeMeMe trial

The experience suggests that participants do not change their mind depending on the allocation group (intervention or control), and there is no single answer to sort out the short blanket syndrome.



Causal Inference on the Difference of the Restricted Mean Lifetime Between Two Groups

Estimators for the difference of the restricted mean lifetime between two groups that account for treatment imbalances in prognostic factors assuming a proportional hazards relationship are proposed and large-sample properties of these estimators based on martingale theory for counting processes are derived.

Distribution‐free inference on contrasts of arbitrary summary measures of survival

This work as a whole supports the use of survival summaries appropriate to a given medical application, whether that be, for example, the median or 75th percentile in some settings or perhaps a restricted mean in others.

Restricted mean life with adjustment for covariates

Double‐Robust Semiparametric Estimator for Differences in Restricted Mean Lifetimes in Observational Studies

This report uses semiparametric theory to develop an estimator for differences in restricted mean lifetimes although accounting for confounding factors and shows that the proposed estimator possesses the double robust property; i.e., when either the time-to-event or coarsening process is modeled correctly, the estimator is consistent and asymptotically normal.

Combining parametric, semi-parametric, and non-parametric survival models with stacked survival models.

An extensive simulation study demonstrates that stacked survival models consistently perform well across a wide range of scenarios by adaptively balancing the strengths and weaknesses of individual candidate survival models.

Censored quantile regression with recursive partitioning-based weights.

A new weighting approach that uses recursive partitioning, e.g. survival trees, that offers greater flexibility in handling covariate-dependent censoring in moderately high dimensions and can incorporate both continuous and discrete covariates is proposed.

Double Inverse‐Weighted Estimation of Cumulative Treatment Effects Under Nonproportional Hazards and Dependent Censoring

Three measures are proposed, including the ratio of cumulative hazards, relative risk, and difference in restricted mean lifetime, and a double inverse-weighted estimator, constructed by using inverse probability of treatment weighting to balance the treatment-specific covariate distributions.

Regression Analysis of Restricted Mean Survival Time Based on Pseudo-Observations

It is concluded that while existing methods may be superior for analysis of the mean, pseudo-observations seem well suited when the restricted mean is studied.

Model averaged double robust estimation

The model averaged double robust (MA-DR) estimators are introduced, which account for model uncertainty in both the propensity score and outcome model through the use of model averaging and can reduce mean squared error substantially, particularly when the set of potential confounders is large relative to the sample size.

A Proportional Hazards Model for the Subdistribution of a Competing Risk

This article proposes methods for combining estimates of the cause-specific hazard functions under the proportional hazards formulation, but these methods do not allow the analyst to directly assess the effect of a covariate on the marginal probability function.