**You Are Here:** U.S. Preventive Services Task Force > Topic Index > Genetic Risk Assessment and *BRCA* Mutation Testing for Breast and Ovarian Cancer Susceptibility > Systematic Evidence Review > Appendix 3

# Genetic Risk Assessment and *BRCA* Mutation Testing for Breast and Ovarian Cancer Susceptibility

## Appendix 3

Several estimates were used to develop the outcomes table. This appendix provides 2 examples of specification of sampling distributions for these estimates.

- Sampling Distribution for Penetrance
*P*(*D*^{+}|*G*):The estimate of

*P*(*D*^{+}|*G*) is obtained from our analysis. We assumed that logit(*P*(*D*^{+}|*G*)), denoted as logit(*P*) for concise notation, is approximately normally distributed and estimated logit(*P*) and its variance from data available in the literature. Estimate of*P*(*D*^{+}|*G*) and its CI is obtained by transforming logit(*P*) and its CI (Go to Appendix 2 for more information).In Monte Carlo simulation, random samples for the estimate of

*P*(*D*^{+}|*G*) are obtained as follows. First, random samples of logit(*P*) are drawn from the following normal distribution:

- 2. Sampling Distribution for Relative Risk
When developing the outcomes table, the estimates of relative risk (RR) are obtained from published studies. Usually, the point estimate and its 95% CI (RR

_{L}, RR_{U}) are reported.Since ln(RR) is usually assumed to be approximately normally distributed, we calculate

where Z

_{0.975}is the 97.5% quantile of the standard normal distribution. Random samples of RR are obtained by first drawing random samples of ln(RR) from and then transforming to RR by taking exponentiation.If we recalculate the 95% CI for RR by using

the resulting CI usually agrees very well with the reported (RR

_{L}, RR_{U}).