There is a particular kind of safety document that looks flawless and describes a hospital that does not exist. Its process map is clean. Every step follows the last in good order, nothing is interrupted, no one is covering two assignments, and the night shift behaves exactly like the policy says it should. It is a beautiful artefact, and it is worse than useless, because it certifies as safe a process that only runs that way in the manual.
Safety science has a plain name for the gap this document falls into. On one side is work-as-imagined, the process as it was designed, written down, and described in meetings. On the other is work-as-done, the process as real people actually perform it under real conditions. The distance between the two is not a scandal and not a sign of bad staff. It is a law of nature. No procedure survives contact with a short-staffed Saturday intact, and the people at the sharp end are constantly, skilfully adapting: bridging missing information, working around broken equipment, resequencing steps to absorb the interruption that just walked in. Most of those adaptations are the reason the system works at all.
The problem is that some of them quietly remove a barrier the designers were relying on, and nobody upstream knows. The double check that has become a second signature on a form. The barcode scan performed from a printed sheet of stickers because the scanner drops its signal in the far room. The specimen labelled in the corridor because the printer in the room jams. Each of these is invisible from the conference room, and each is where the next event is waiting.
Which is why a process cannot be analysed from a chair. It has to be watched. Somebody has to follow the real thing, shadowing a medication pass, walking a specimen from the tray to the analyser, riding along on a transfer to imaging, and writing down what happens rather than what is supposed to. The rules for doing this well are few and strict. Announce that you are studying the process and not the person. Record the deviation as a discovery, never as a violation. And talk to the hours and the people the conference room cannot represent, which means, for a process that runs around the clock, actually speaking with the night shift rather than about them.
There is a rough test worth trusting. A map that discovered nothing during the walking has not been walked. Real processes are messy, and a map that came back tidy came back early. The messiness is not noise to be cleaned up before the analysis begins. It is the analysis, arriving ahead of schedule.
The method in Before Harm Happens treats this walk as non-negotiable, because everything the later steps produce is only ever as honest as the map they start from.
This article accompanies the book Before Harm Happens.