At 6:47 in the morning a chemistry analyser flags a dangerously high potassium result. The day shift arrives at seven. The night technologist has just started the handoff huddle. The result sits on a screen that, for the next hour, belongs to no one, and the call to the ward that should have followed within minutes instead follows at nearly eight. The patient, whose potassium had been climbing for entirely treatable reasons, does not wait that long.
Nothing in that sequence required a careless person. It required a gap, and the gap was in the process at its most fragile point, the moment of a shift change, where a result that flags in the seam belongs to whoever happens to notice, which on a busy morning is no one. A critical value is, by its nature, a piece of information the system has already worked hard to catch. When its notification fails, you have not merely missed a result. You have caught a danger and then dropped the catch, which is a detection failure squared, and the most consequential kind there is.
Laboratory medicine learned something years ago that the rest of the hospital is still absorbing. The analytic phase, the instrument actually measuring, is now so controlled that the risk has migrated outward, to the pre-analytic and post-analytic phases performed by hands outside the lab’s walls. The specimen mislabelled in a corridor. The sample that hemolysed and was rejected two hours later, the whole delay invisible to the ward that believes the test is running. And at the far end, the critical value that reaches a phone at an empty desk, or the ordering-provider field that still points at the hospitalist who rotated off yesterday.
The fixes that hold have one thing in common: they route around individuals. Notification goes to a role that is always staffed rather than to a named person who might be at lunch. Escalation timers fire automatically when acknowledgement does not come, so that a call unanswered does not simply end. And closure is defined as the ordering clinician acknowledging the result in the record, not as “someone answered the phone,” because the phone being answered is not the same as the danger being received.
The general lesson reaches well past the laboratory. Any time a process has already detected a problem, the communication of that detection is itself a step that can fail, and it fails most often in the seams between shifts and departments. Catching the danger is only half the job. The other half is making sure the catch cannot be dropped in the handoff, and that half is designed, or it is left to luck.
This article accompanies the book Before Harm Happens.