Bivariate familial correlation analysis of quantitative traits by use of estimating equations: Application to a familial analysis of the insulin resistance syndrome

✍ Scribed by David-Alexandre Trégouët; Bernard Herbeth; Irène Juhan-Vague; Gérard Siest; Pierre Ducimetière; Laurence Tiret

• Publisher: John Wiley and Sons
• Year: 1999
• Tongue: English
• Weight: 65 KB
• Volume: 16
• Category: Article
• ISSN: 0741-0395
• DOI: 10.1002/(sici)1098-2272(1999)16:1<69::aid-gepi6>3.0.co;2-h

✦ Synopsis

Familial correlation analysis involving two traits may give a better insight into the etiology of multifactorial syndromes than familial analysis focused on single traits. Significant cross-trait correlations between biological relatives but not between spouses suggest that the two traits share common transmissible factors whereas correlations between spouses additionally suggest the influence of shared lifestyle factors. We apply the Estimating Equations (EE) technique to the estimation of intra-trait and cross-trait familial correlations on two quantitative traits. Unlike maximum likelihood methods, the EE method does not require one to specify the joint distribution of the traits. Estimation of correlations and of their variance involves an iterative three-stage algorithm which converges rapidly. The generalized Wald test can be used to test any specific hypothesis of familial resemblance. This method has great flexibility for handling covariates and incomplete family data. A simulation study indicated that the EE technique performed well in large samples (100 families), both in terms of type I error and coverage probability . However, in small samples (50 families), an increase of the type I error and a decrease of the coverage probability was observed. As an illustration, we applied this technique to a family study of metabolic factors involved in the Insulin Resistance Syndrome (body mass index, insulin, triglycerides, HDL-cholesterol, and diastolic blood pressure). The study was carried out in a sample of 216 healthy nuclear families with > or =2 offspring. The results suggested the existence of a common transmissible (genetic or cultural) factor influencing both body mass index and insulin, whereas the weak clustering of triglycerides and HDL-cholesterol would be more compatible with the influence of shared lifestyle factors.