Skip to main content
. 2021 Jan 13;11:1169. doi: 10.1038/s41598-020-80929-3

Figure 1.

Figure 1

Cross-sectional weighted associations of sphingomyelins and ceramides with physical function in ARIC visit 5. Multiple linear regression model was used to determine β-coefficients [95% confidence internal] between each metabolite (log-transformed concentration modeled in z score) and physical function. Model 1 covariates: age, sex, race-site, and batch effects. Model 2 covariates: model 1, plus body mass index, diabetes status, hypertension status, use of lipid lowering medication, total cholesterol, Digit Symbol Substitution test z-score, and prevalent stroke at the ARIC visit 5. Model 3 covariates: model 2, plus education, presence of an APOE4 allele, Mini-mental state examination (MMSE) score, sports index, previous heart failure, previous coronary heart disease, CES depression score, HDL cholesterol, triglyceride levels, SF bodily pain score, current smoking status, and current drinking status.