Paper: Those Found Responsible Have Been Sacked
In light of recent news, I decided to revisit Richard Cook's paper on the concept of "error" titled "Those found responsible have been sacked": Some observations on the usefulness of error.
The paper takes a stance that's consistent with what has been advocated by people in "The new view" (or "Safety-II" approach) of safety science of the last few decades, particularly those that state that error as a broad category is not very useful as a concept: it has no stable definition, leads to dead-ends in reasoning, and is generally to be avoided. However, the paper makes a point that error as a concept survives in organizations because it has a use that is both socially and organizationally productive, mostly in terms of social control.
In short, error is useful, but not in the way you'd expect. To illustrate that point, a case study from co-author Christopher P. Nemeth is used, where "a patient was being prepared for spinal surgery and members of the surgical team placed the anaesthetized patient onto a modular table. One member noticed that there was a slight tilt to the table and began to correct the table’s position. The table swung loose, and the patient fell from the table to the floor but sustained no injury."
Many factors were identified by their investigation: the table design is such that it can rotate in a way an anaesthetized patient can fall off, no individual was assigned as responsible to operate the table, the controls of the table were unclear even while following the labelling due to their complexity and ambiguity, and it requires substantial force to manage to lock the table, more than operators were willing or able to exert.
The hospital's risk manager ran a "root cause analysis" where these elements were presented, along with FDA reports showing this type of even wasn't unseen. When it came to finding solutions, the surgical team said they still wanted to use the table despite its safety issues since no alternative could do better. The hospital was left with warning signs or labels, training and restrictions, checklists, or buddy systems:
In the end, each of the influences that led to the adverse event remained in place, including the prospect of clinicians being assigned blame for failing to operate the dangerous table correctly. The unit was not removed or modified. The warning was not used. Care providers were cautioned, setting the stage for further "errors".
Finding the sources of adverse outcomes is challenging, and organizations can lack the freedom or resources to effect true change. Hollnagel is cited as making the point that "it is meaningless to talk about mechanisms that produce errors. Instead, we must be concerned with the mechanisms that are behind normal action." The decision-making that leads to success and failure is generally similar in intent, and context and outcome may define whether something was an error or a failure.
The "first stories" (overly simplified accounts biased by apparent causes and known outcomes) are often attractive explanations for failure, but they lead to basic responses that limit both learning and improvement:
Those who are closest to the sharp (operator) end of the healthcare organization understand the difficulty and uncertainty that underlies their daily activities. Those who are closest to the blunt (management) end are most remote from sharp end operations and are concerned with maintaining the organization, and threats to the organization are minimized by casting adverse events as anomalies. Identification and removal of the event’s proximate cause gives the appearance of restoring the organization to "normal" conditions.
[...]
From the management viewpoint, operational failures that erode productivity and reputation can impair an organization’s financial performance . The question a manager asks of herself or himself when adversity strikes is ‘Which facets of the situation can I influence, no matter how impossible the situation may seem?’ It’s not about controlling everything, and everyone, to get what you want. It’s about being able to influence something in the situation to make it better
So getting back to the original question, amidst all this complexity, Cook suggests that what may be interesting about the concept of "error" is not what it is, but in how people use it. Four uses are proposed:
1. A defense against entanglement with accidents
An organization blaming human error for an incident lowers its own liabilities. That type of error is safe for the organization because it curtails investigations and directs all attention onto isolated humans, away from systemic pressures moving practice toward risk.
As an organizational defense, human (operator) error serves as a kind of lightning rod that conducts the potentially harmful consequences produced by an accident along an (organizationally) safe pathway.
2. The illusion of control
By blaming people for incidents, we create the prospect of being able to curb their behavior and of preventing future issues by restraining or containing them. If error is not individual and instead is systemic, then substantial changes may need to be made to create safety, which could impact productivity for example.
By asserting control over the circumstances, organizations can keep a public image of reliability, to promise that "this will never happen again" (even if it will).
3. A means for distancing
It just feels safer if the error is attributable to the character or moral failing of a person. If incidents happen from environmental sources, then colleagues, despite any level of competence, have to increase the sense of hazard and uncertainty.
Attributing error to human sources therefore can feel safer for people in the system.
4. A marker for failed investigations
This is what Richard Cook calls "the most important value" of "human error": it provides an acceptable point to stop an investigation into incidents:
investigation halts most often when the traceback process encounters a human with apparent freedom of action. This appearance forms a "cognitive barrier" beyond which investigators do not make much progress, mainly because it is so difficult to work through the psychology and behavior of human agents. The cognitive wall is not impenetrable.
However, not managing to go past that cognitive wall is often a sign of a flawed technical investigation. As stated in the paper, operator or user error is often a catchall term for events that aren't obviously mechanical failures.
So in seeing "human error", researchers now get a strong signal that an investigation was incomplete or failed. As Cook puts it, "this use for error may ultimately be the greatest contribution of human error to the creation of high reliability systems."
The authors conclude by stating that treating the presence of "human error" as noise to be ignored or as a technical inadequacy diminishes the value of "error"; noting its use can indeed provide critical information about the dynamics of incidents, from genesis to the response to their happening.