Threshold Decision-Making

---

Example: diagnosing acute pancreatitis with a serum lipase level. Based on the patient's presenting features, you think that the pre-test probability for a specific patient having acute pancreatitis is only 20%. Literature review tells you that the lipase level is 94% sensitive and 96% specific. If the test is negative, this patient now has a 1% post-test probability of having acute pancreatitis. If positive, the patient has a 85% post-test probability. You weren't sure what to do with a 20% probability, but you feel comfortable treating the patient for pancreatitis with an 85% probability; and you feel comfortable ignoring the possibility of pancreatitis with only a 1% probability.