Turnaround, honestly

The average is the problem

A laboratory reporting a mean turnaround of six hours can be missing its promise on one report in five. The mean is dragged down by the hundreds of routine tests that ran perfectly and says nothing about the ones that did not.

Carelis reports the median, the 90th and 95th percentile, and the worst case. The mean is kept only so the difference between it and the 95th is visible.

The clock starts at registration

Not at the bench. Measuring from the moment a sample reaches the analyser excludes the queue at the counter and the journey in the van — which is to say, it excludes most of the delay a patient actually experiences.

Urgent and routine are never blended

When they are averaged together, an urgent flag that does nothing looks like an urgent flag that works. Separating them is how a laboratory finds out that its urgent samples are moving no faster than the rest — which is a common finding, and an invisible one until somebody looks.

Quality indicators

Ten indicators along ISO 15189 and NABL lines, each with a numerator, a denominator and a target recorded at the time. Signing off a missed target requires a real sentence about what was done. Raising a target later cannot rewrite an earlier miss into a pass.