Today we discussed a case where the treating physician ordered a procalcitonin assay to decide whether or not the patient had a bacterial respiratory infection. The result was (very) negative. Nonetheless they continued antibiotics. Was that the right decision?
This Lancet ID meta-analysis looks at the operating characteristics of the test. There is a lot of heterogeneity of the pooled studies which is a caveat. But to me, I'm not sure that a pooled sensitivity of 77% and specificity of 79% is all that impressive.
This would yield a positive likelihood ratio of 3.6 and a negative likelihood ratio of 0.27
If the patient has a 20% pre-test probability of bacterial sepsis the test would perform as follows:
Positive predictive value = 47%
Negative predictive value = 6%
(FYI the numbers for 10% and 30% are 28%/3% and 60%/12%)
Does that help you in the care of that patient? Maybe if it is negative; however, it depends on the context and how ill the patient is and the consequences of being among the 6%.
FYI as of 2012 an estimate of the per unit cost was $10-15 for the test... Given the volume of tests which would be (maybe unnecessarily) ordered in the emergency department each day it is no wonder that not all hospitals cover the test (ours does not at time of posting). See technology assessment 2012 available online here.
