Premium
Statistical consequences of a successful lung allocation system – recovering information and reducing bias in models for urgency
Author(s) -
Tayob Nabihah,
Murray Susan
Publication year - 2017
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.7283
Subject(s) - censoring (clinical trials) , computer science , ranking (information retrieval) , imputation (statistics) , statistics , survival analysis , medicine , econometrics , missing data , mathematics , machine learning
The lung allocation system has reduced the number of waitlist deaths by ranking transplant candidates on the basis of a lung allocation score that requires estimation of the current 1‐year restricted mean waitlist survival (urgency). Fewer waitlist deaths and the systematic removal of candidates from the waitlist for transplantation present statistical challenges that must be addressed when using recent waitlist data. Multiple overlapping 1‐year follow‐up windows are used in a restricted mean model that estimates patient urgency on the basis of updated risk factors at the start of the window. In simulation studies, our proposed multiple imputation procedure was able to produce unbiased parameter estimates with similar efficiency to those obtained if censoring had never occurred. The analysis of 10,740 lung transplant candidates revealed that for most risk factors incorporating additional follow‐up windows produced more efficient estimates. Copyright © 2017 John Wiley & Sons, Ltd.