“Human error” is the most reassuring phrase in the whole language of safety, which is exactly what makes it dangerous. It closes the file. Once we have decided that a competent person simply erred, there is nothing left to examine except the person, and the person, being competent, will promise to try harder. Then the same thing happens to someone else.
Look closely at almost any error made by a good clinician and you find the same shape underneath. The action was easy to get wrong and hard to catch. Consider a nurse preparing two medications for two patients at the same counter, from vials that look nearly identical, under a workload that treats interruption as normal. When the vials are eventually swapped, we can write “nurse failed to verify” and feel that we have explained something. We have explained nothing. We have described the moment the accident became visible, and stopped there.
The productive question is the next one, and it is never about the person. What was it about this step that made the error easy to commit and unlikely to be caught? Side-by-side preparation of look-alike products was permitted. The final check happened in the same crowded visual field where the swap occurred. Interruptions during preparation were routine and unremarked. Each of those points at something you can change, and none of them is a character flaw. This is the quiet insight that decades of human factors work handed to healthcare: reliable people operating inside an unreliable process will produce unreliable results, and no amount of willpower closes that gap.
The reason this matters is practical, not philosophical. If you believe the cause was carelessness, your fix is re-education, and re-education is close to the weakest intervention available. You are asking people to compensate, through sustained vigilance, for a design that fights them. It works for a while, on day shift, when the unit is fully staffed and nobody is covering two assignments. It fails at three in the morning, which is when it matters.
There is a test I like for any corrective action. Imagine the most careful person you know, on their worst day, doing this task. If your fix depends on them being better than that, it is not a fix. Redesign the step so the wrong action becomes difficult or the right one becomes automatic, and you have protected every future person who will ever stand in that spot, including the ones who have not been hired yet.
None of this excuses genuine recklessness, which is rare and is a separate conversation. The point is narrower and more useful. When your good people keep making the same mistake, they are not the variable that changed. The process is telling you where it is weak, in the only language it has.
This article accompanies the book Before Harm Happens.