Year: 2020
Type: erratum
Abstract: In Section 3.2 of our paper 'Adjusted restricted mean survival times in observational studies', our notation for the pseudo-observation model of the restricted mean survival time (RMST) was incorrect.1 We incorrectly indexed the pseudo-observation on exposure group k, when we actually calculated the pseudo-observations in the overall sample. The correct notation for the pseudo-observation for individual i is μ τ , i = n · μ ^ τ − ( n − 1 ) · μ ^ τ ( − i ) , where μ ^ τ ( − i ) is the estimated RMST without individual i. Then, we model E ( μ τ , i | Z i , X i ) = α + β z Z i + β x ′ X i . The model coefficients are estimated by solving the estimating equation using the pseudo-observations, ∑ i ( ∂ ∂ β β ′ W i ) ′ V i − 1 ( μ τ , i − β ′ W i ) = 0 . The simulations and illustrative example in our paper are correct, and subsequent results are not affected by this error.
Source: Statistics in Medicine
DOI: 10.1002/sim.8742
Authors: Sarah C. Conner, Ludovic Trinquart
Institution: Boston University
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W3088083101