While I'm not in the medical profession or field and, quite honestly, the sight of blood or even the mere mention of it makes me queasy.....I can talk about nonconformances very easily!
1. Define a nonconformance at your site...the parameters for what makes a nonconformance a nonconformance
2. Define triggers for when you will take action...and to what level action will be taken.
3. Formalize process.
4. Record data.
5. Review nonconformances on a regular basis.
6. Adjust definitions for nonconformance and triggers accordingly.
Here's how we look at...
If the process has an unscheduled delay of let's say 5 seconds. There is no impact to the product. Machines are still okay. We don't sweat it. We do, however, record the fact that the delay occured, where it occured, when, etc.
If the delay was 20 minutes, that's more of an issue and we do all the above, but resolve the issue and get back to making product.
If the delay is 40 minutes, wow! That's caused a lot of problems and we'll do a full-blown corrective action on that nonconformance.
But we also have stats on our process that back us up on our defined triggers. And, as we start to see the larger problems decrease in number, we start to lower our triggers so that we can still focus on nonconformances.
This approach also helps in the demonstration of continual improvement.