Admixture analysis of high density lipoprotein cholesterol distribution in a Jerusalem population sample.

Abstract

The distribution of high density lipoprotein cholesterol values (HDL-C) in 2003 nuclear families (including both parents and one child aged 17) participating in the Jerusalem Lipid Research Clinic survey, were analyzed by a maximum-likelihood procedure for evidence of bimodality. HDL-C was age-sex adjusted by the mean-variance method to 17 year old male levels. After covariance adjustment for Quetelet's index, season, education and social class, when using untransformed data a mixture of two distributions fitted the data significantly better than one distribution in both sexes. The results indicated that 4.7% of males and 14.1% of females came from an upper distribution with mean values of 2.15-3.15 standard deviations above the major mode. We next applied the MacLean et al. (1976) method of transformation and used a maximum-likelihood method in which the skewness parameter was estimated jointly with the other parameters of the model. When the data were transformed the mixture of two distributions fitted the data significantly better than one for males only. We tested a mixture of three normal distributions and estimated that 6% of both males and females belonged to a higher distribution, while 1% came from a lower distribution. This model did not provide a significantly better fit to the data than a mixture of two distributions. Our findings imply that in the Israeli population there is evidence for an admixture in the distribution of HDL-cholesterol. Segregation analysis is necessary in order to determine whether a major gene or some environmental factors are responsible for the commingling detected.