Confidence intervals for genotype relative risks and allele frequencies from the case parent trio design for candidate-gene studies

Abstract

Objectives: Confidence intervals for genotype relative risks, for allele frequencies and for the attributable risk in the case parent trio design for candidate-gene studies are proposed which can be easily calculated from the observed familial genotype frequencies.

Methods: Likelihood theory and the delta method were used to derive point estimates and confidence internals. We used Monte Carlo simulations to show the validity of the formulae for a variety of given modes of inheritance and allele frequencies and illustrated their usefulness by applying them to real data.

Results: Generally these formulae were found to be valid for 'sufficiently large' sample sizes. For smaller sample sizes the estimators for genotype relative risks tended to be conservative whereas the estimator for attributable risk was found to be anti-conservative for moderate to high allele frequencies.

Conclusions: Since the proposed formulae provide quantitative information on the individual and epidemiological relevance of a genetic variant they might be a useful addition to the traditional statistical significance level of TDT results.