In the Milan conference, the preferred specification is the effect of assay error on patient outcomes. This seems reasonable enough but consider the following two cases.
Case 1, a glucose meter reads 350 mg/dL, truth is 50 mg/dL; the clinician administers insulin resulting in severe harm to the patient.
Case 2, a glucose meter reads 350 mg/dL, truth is 50 mg/dL; the clinician questions the result and repeats the test. The second test is 50 mg/dL; the clinician administers sugar resulting in no harm to the patient.
One must realize that lab tests by themselves cannot cause harm to patients; only clinicians can cause harm by making an incorrect medical decision based in part on a lab test. The lab test in cases 1 and 2 has the potential (a high potential) to result in patient harm. Case 2 could also be considered a near miss. From a performance vs. specification standpoint, both cases should be treated equally in spite of different patient outcomes.
Thus, the original Milan statement should really be the effect of assay error on potential patient outcomes.