Ask a group to brainstorm everything that could go wrong with a process and you will get a lively hour and a partial list. Brainstorming is generous with the failures that are dramatic, recent, or personally remembered, and quietly blind to the rest. The failure that has never quite happened yet, the one nobody has a story about, does not raise its hand. And that is usually the one you needed.
There is a better way, and it is almost embarrassingly simple. For any step in a process, run it through the same six questions, every time, in order. Was the step not done at all? Was it done too late, or too early? Was it done incompletely, started but not finished? Was it done to the wrong object, the wrong patient or sample or medication? Was it done with the wrong input, the right action fed by the wrong information? Or was it done correctly and then communicated to no one, so the result died in transit? Six directions of wrongness, applied deliberately, and the quiet failures that brainstorming skips come out one after another.
Take a single respiratory step, resetting the ventilator alarms after a change in settings. Run the six. The alarms are not reset at all. They are reset late, after an unmonitored gap. They are reset incompletely, one limit left at the old value. They are reset on the wrong ventilator in a two-bed room. They are reset using the previous patient’s stored profile. Or they are reset correctly and the change is never mentioned to the nurse who will answer the alarm. Ninety seconds, six failure modes, each with different causes and, as it turns out, different fixes. A free-form discussion of the same step tends to find two of them.
The grammar does something else, quietly, that matters as much as the coverage. It keeps a team from writing down effects and calling them failure modes. “Patient aspirates” is not a failure mode of a food service process. It is the effect of several: the thickened liquid not prepared to the ordered consistency, the modified tray delivered to the wrong room, the diet upgraded before the swallow assessment cleared it. If a team fixes “aspiration” it fixes nothing, because there is no such lever. If it fixes the tray-to-room matching, it has changed something real. Separating the two is most of the difference between a worry list and an analysis.
There is a useful test hidden in all this. A good failure mode is specific enough that you could act it out. If you cannot mime it, it is still too vague, and vague failure modes produce vague fixes that reassure everyone and protect no one.
The pleasure of the six-question grammar is that it makes thoroughness cheap. You do not need inspiration or a particularly imaginative group. You need a process broken into steps and the discipline to ask all six of every one, especially the steps everyone is sure are fine.
In the field workbook that accompanies Before Harm Happens this grammar is a single worksheet, six columns wide, meant to be run down a process step by step.
This article accompanies the book Before Harm Happens.