On confidence intervals for genotype relative risks and attributable risks from case parent trio designs for...

Area of Science:

• Genetics
• Statistical Genetics
• Epidemiology

Background:

• Investigates bias and coverage probabilities of genotype relative risks (GRRs) and attributable risk (AR) estimates in case parent trio designs.
• Examines scenarios where the marker single nucleotide polymorphism (SNP) locus is not identical to the disease locus.

Discussion:

• Bias in GRR and AR estimates is influenced by linkage disequilibrium (LD) and marker allele frequencies.
• Substantial bias occurs when marker allele frequencies differ from the functional SNP's frequencies or when LD is not maximal.

Key Insights:

• No bias is observed when the marker and disease loci coincide.
• Estimates are generally biased towards the null hypothesis, particularly with differing allele frequencies or suboptimal LD.
• Biases away from the null are rare and typically small.

Outlook:

• Emphasizes the need for caution when applying GRR and AR estimates in candidate gene studies.
• Highlights the impact of marker-disease locus mismatch on genetic association study results.
• Suggests careful consideration of LD and allele frequencies for accurate genetic risk assessment.