In the last couple years there have been several articles in CAP TODAY about procalcitonin, which is most typically elevated in bacterial sepsis (here, here, here). It caught my attention that there was a favorable review of procalcitonin in Cochrane Reviews, whereas technology assessments usually beat the (x) out of clinical laboratory tests.
See short summary in Annals of Internal Medicine and full review by Shuetz at Cochrane. In a meta-analysis of 6708 patients in 26 trials, mortality was reduced from 10% to 8.6%, giving a number needed to treat of 65. That suggests, in simple math, circa 65x$20 or about $1200 per death avoided.