Most rotating components on an engine have some sort of life limit (often expressed in cycles rather than hours - sometimes both) and are designed to never fail within that life limit.
However sometimes $hit happens and one lets go due to a manufacturing defect or improper maintenance.
While the regulations basically say you need to design the aircraft in such a way to minimize the hazard, the associated regulatory guidance generally points to a "1 in 20" analysis. Basically, the airframer analyzes all the potential disc burst scenarios, and makes sure that 5% (or less) will have a catastrophic outcome. I've had secondary involvement in 1-20 analysis, but the prime 1-20 task is almost black magic. For small, lightweight debris such as compressor/turbine blades, you can take credit for shielding - for large pieces of disc shielding is deemed to be impractical (Where does a third of a fan disc go? Anywhere it wants to...) so location of flight critical components becomes critical.
Uncontained engine failures are considered to be a 10-8/hr. event - so if 1 in 20 is catastrophic the probability of a catastrophic uncontained engine failure is less than 10-9/hr. which in the regulatory arena means 'it'll never happen' (yea, I know, but you have to draw the line someplace and it was agreed that 1/billion was good enough).