An Application of Covariate-Adjusted Partial Spearman's Rank Correlations with Probability-Scale Residuals

• Activity Number: 344 - Semiparametric Modeling
• Type: Contributed
• Date/Time: Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
• Sponsor: Biometrics Section
• Author(s): Cathy A Jenkins* and John R Koethe and Timothy R Sterling and Spyros A Kalams and Bryan E Shepherd
• Companies: Vanderbilt University Medical Center and Vanderbilt University School of Medicine and Vanderbilt University School of Medicine and Vanderbilt University School of Medicine and Vanderbilt University School of Medicine
• Keywords: Probability-scale residuals; Spearman's rank correlation; HIV

Abstract

When engaging in hypothesis generating research, it is important to summarize the relationships of interest in a meaningful way. Correlation coefficients are a common summary statistic that describe the strength and direction of relationships. Simple summary methods use unadjusted correlation coefficients such as Pearson or Spearman's; however, failing to adjust for covariates may result in misleading conclusions. Partial correlations allow for adjustment variables, but traditional approaches for measuring partial rank correlations lack a theoretical foundation. Liu et al (Biometrics, in press) proposed an alternative estimator for covariate-adjusted partial Spearman's rank correlations using probability-scale residuals (PSRs). We apply this method to explore the relationships between body composition and markers of innate and adaptive immune status in HIV+ subjects. Three measures of adiposity were considered - body mass index, fat mass index, and plasma leptin. Immunologic markers included soluble cytokines and other biomarkers, CD4+ and CD8+ T cell subsets.