| Without knowing the number of tests performed, you can both be correct. Theirs would be the false positive rate (FPR) or false negative rate (FNR), and yours would be an probability, which takes the FPR (or FNR) rate, the number of tests and the base incidence rate to arrive at 1/300, 1/5000. For example, in Canada, 1/750 of live born babies has Down's [2]. There are about 400,000 births every year in Canada [3]. So, there are 400,000 / 750 = 500 kids with Down's born every year. Assume the test has a FNR rate of 0, and a FPR of 5%. 400,000 * 0.05 + 500 = 20,500 children test positive Odds of positive test indicating Down's (5% FPR): 500/20,500 = 2.4% Not quite the 1/300, but as you said "soft marker", so maybe the FPR is higher for that, vs all the markers? To get to 1/300, the FPR rate on the marker would need to be 37.5%? The 1/5000 is (I believe) a change from a FPR rate to the FNR rate, since the second test is indicating that the disease _isn't_ present. 1/5000 would indicate a negative genetic test will miss one Down's diagnosis in a decade in Canada. Interestingly, it looks like the FRP increases with age, with it being 4% at age 30 (combined ultrasound/blood test)? [1] [1] https://pubmed.ncbi.nlm.nih.gov/11702835/ [2] https://www.canada.ca/en/public-health/services/publications... [3] https://www.statista.com/statistics/443051/number-of-births-... |