Nonparametric estimation of broad sense agreement between ordinal and censored continuous outcomes. (23rd March 2020)

• Record Type: Journal Article
• Authors: Dai, Tian; Guo, Ying; Peng, Limin; Manatunga, Amita
• Is Part Of: Statistics in medicine. Volume 39:Number 14(2020)
• Page Start: 1952
• Page End: 1964
• Publication Date: 2020-03-23
• Subjects: agreement -- censored observations -- multiple imputation -- nonparametric estimation; Medical statistics -- Periodicals; Statistique médicale -- Périodiques; Statistiques médicales -- Périodiques; 610.727
• DOI: 10.1002/sim.8523
• Languages: English
• ISSNs: 0277-6715

Abstract

Abstract : The concept of broad sense agreement (BSA) has recently been proposed for studying the relationship between a continuous measurement and an ordinal measurement. They developed a nonparametric procedure for estimating the BSA index, which is only applicable to completely observed data. In this work, we consider the problem of evaluating BSA index when the continuous measurement is subject to censoring. We propose a nonparametric estimation method built upon a derivation of a new functional representation of the BSA index, which allows for accommodating censoring by plugging in the nonparametric survival function estimators. We establish the consistency and asymptotic normality for the proposed BSA estimator. We also investigate an alternative approach based on the strategy of multiple imputation, which is shown to have better empirical performance with small sample sizes than the plug‐in method. Extensive simulation studies are conducted to evaluate our proposals. We illustrate our methods via an application to a Surgical Intensive Care Unit study.