Computation of a cardiac severity score with left-censored biomarkers for patients with heart failure

Abstract

Introduction: Biomarkers reported as being less than the limit of detection (LOD) are challenging to be included in analyses without being dichotomized. Substitutions by the LOD, LOD/2, LOD/√2, or zero have been attempted for left-censored values. We calculated a novel Modified Myocardial Injury Summary Score (MMISS) by incorporating four biomarkers (B-type natriuretic peptide, troponin, galectin-3, and suppression of tumorigenicity 2) to efficiently stratify heart failure (HF) patients. The objective of this study is to evaluate the differences in associations between MMISS, calculated by different methods of substitution, and therapeutic intensity index (TII), a composite pharmacologic score. Methods and results: This is a cross-sectional study including 39 HF patients aged ≥18 years who were treated in the Baylor Health Care System. MMISS was calculated with the left-censored biomarkers substituted by zero, LOD, LOD/2, and LOD/√2. Patients with biomarker values more than LOD (complete cases) were also considered separately. The computed TII was regressed separately on MMISS for each substitution method, while controlling for age and gender. All substitution methods yielded negative associations; however, statistical significance for the association was not achieved using substitution by zero or when considering only the complete cases. The association was quite comparable with the substitution of left-censored values by LOD, LOD/2, and LOD/√2. Conclusion: Substantial loss of information is inevitable if only the data with values above the LOD are considered for analysis or when the left-censored values are substituted with zero.

Plasma oxylipin levels associated with preterm birth in preterm labor✰

“…Oxylipins with a difference in concentrations between groups at a p-value <0.1 were further analyzed with logistic quantile regression in univariate and multivariate models. Here, only oxylipins with at least 40% of values above the limit of detection (LOD) were taken into consideration, and values below LOD were substituted with LOD/√2 [25]. Multivariate logistic quantile regression models were adjusted for usage of pneumatic tube systems for sample transport, as this may affect oxylipin levels [26].…”