betaFeedback

WorksMathematics

Limited Follow Up Overlap in Causal Survival Analysis: Censoring-Adjusted Weighting for Restricted Mean Survival Time

Abstract

Overlap weighting addresses limited treatment overlap by targeting individuals for whom treatment assignment is uncertain. With right censored outcomes, however, adequate treatment overlap does not ensure adequate follow up. We show that the overlap weighted restricted mean survival time difference can have an infinite semiparametric efficiency bound even under randomized treatment and pointwise censoring positivity. We propose censoring adjusted overlap (CO) weighting, that accounts jointly for treatment assignment and the probability of remaining under observation through the restriction horizon. CO approaches ordinary overlap weighting as the strength of the follow up adjustment decreases and progressively downweights covariate profiles with weak follow up. We derive the efficient influence function and show that it is uniformly bounded without requiring censoring survival probabilities to be uniformly bounded away from zero. A cross fitted one step estimator is asymptotically normal with consistently estimated variance under follow up weighted nuisance rate conditions. Simulations show little difference from overlap weighting when follow up is adequate, but substantially greater stability as follow up deteriorates. An analysis of the Rotterdam breast cancer cohort illustrates both the protection against weak follow up and the potential cost in precision.

Paper details

vlnc:W_q626pujq
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
arXiv nonexclusive license
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…