PERMUTATION ANALYSES OF FAMILIAL ASSOCIATION ARRAYS FOR LIPOPROTEIN CONCENTRATIONS IN FAMILIES OF THE STANFORD-5 CITY PROJECT AMERICAN JOURNAL OF EPIDEMIOLOGY Williams, P. T., Karlin, S., Farquhar, J. W. 1987; 126 (6): 1126-1140

Abstract

Permutation models are introduced as a formal method for assigning significance to association matrices that assess the correlation of spouse, parent-offspring, and sibling similarity over an entire class of data transformations (usually, the class of all increasing functions). Analysis of 218 nuclear families who participated in the Stanford Five City Project revealed that parent and offspring triglyceride concentrations correlated more strongly when data transformations emphasized contrasts among low to moderate levels, and that high density lipoprotein (HDL) cholesterol correlated more strongly between family members with relatively higher HDL cholesterol concentrations. Application of family weights to the association matrices revealed a tendency for greater correlation among sibling triglyceride concentrations in larger families. Parent-child, mother-child, father-child, parent-daughter, and sibling total cholesterol concentrations correlated significantly for all monotonically increasing transformations (designated strong association), and father-daughter and parent-son cholesterol concentrations correlated significantly for most increasing transformations of the data (moderate association). There were fewer significant associations for plasma triglyceride concentrations: parent-child and sibling (both strong), parent-daughter and mother-daughter (both moderate), and mother-child (weak). HDL cholesterol showed no strong or moderate familial associations and was weakly associated only among siblings. Thus, concordance in familial lipoprotein levels appears to be restricted to a narrower range of values for triglycerides and HDL cholesterol than total cholesterol levels, possibly reflecting in part the influences of diet or other environmental factors on specific regions of the HDL cholesterol or triglyceride distributions in casual blood samples.

View details for Web of Science ID A1987L240400013

View details for PubMedID 3687922