Comment by rayiner

> Surely there's a better way to filter out risky blood

It's simple Bayesian probability. Blood tests have a relatively high error rate. Hep-B tests have a 6-12% false negative rate early in the disease, and Hep-C is 3-6% even later in the course of the disease. That's considered a "very low" false-negative rate for a blood test.

In Bayesian terms, blood tests don’t “screen” for a disease. They reduce the odds ratio of contaminated blood by a factor of 10 or so. But the ultimate odds still depends heavily on the prior odds—the prevalence of the disease in the donor population: https://en.wikipedia.org/wiki/Prior_probability. Even with testing, you can reduce the risk of contaminated blood by drawing blood from a pool of donors that has lower prevalence rate of diseases.