Our selection criteria

Our risk score model, based on strict curation criteria, reflects high scientific rigor.

Navigenics has developed a set of criteria used to curate the scientific literature and evaluate genetic associations:

• We require that an association be consistently replicated within a given ancestral group. For candidate gene studies, we require at least two independent replications. (Since publication of such studies can be difficult when results are negative, publication bias is a major concern.) If the condition has been studied more than twice, the preponderance of evidence must show the same statistically significant result with the same risk allele.
• We take a close look at sample size and have a minimum requirement of 250 cases and 250 controls when the effect size is less than 1.5. However, most of the associations that we report on have been studied in multiple independent populations.
• We evaluate the study design to determine how the cases and controls were selected, phenotyped and genotyped.
• Analysis includes an evaluation and adjustment for population stratification and possibly other confounders that may lead to spurious results.

Navigenics carefully evaluates the overall quality of evidence for an association before including a SNP in our risk score model:

• Previously, genetic association studies had a poor reputation, with the available literature showing genetic associations that were observed once but could then never be replicated in subsequent studies.
• This poor prior performance has been attributed to multiple factors (Cardon and Bell. Nature Review Genetics 2:91, 2001), which include small sample size, subgroup analysis and multiple hypothesis testing, poorly selected control group, failure to attempt study replication, failure to detect linkage disequilibrium with adjacent loci, over-interpreting results, positive publication bias, inadequate attention to genotyping assay and algorithm quality, and failure to correct or avoid population substructure.
• More recently, genome-wide association studies (GWAS) have made noticeable improvements in addressing these failure points, although critical review is still required. For instance, although the great majority of recent studies has analyzed large sample sizes and often includes multiple replications in multiple population samples, it is still important to evaluate the overall study design, control selection and statistical correction for testing multiple hypotheses.
• With multiple independent studies being published, often using different genetic markers, subject matter experts need to carefully examine the overall evidence to come to a comprehensive conclusion.

Even with these guidelines for curation, interpretation of the genetic association literature is challenging. This is why Navigenics has developed a team of subject matter experts that carefully reviews each genetic risk factor that is included in our genetic testing services.

For more information on the outcome of our curation – the individual health conditions included in our genetic testing services – please see our Conditions we cover page.