Pointing Fingers: Verbosity of Patient Safety Narratives Is Associated With Attribution of Blame

This explored how the length of the narrative in incident reports is connected to attributions of blame.

The authors hypothesised that longer written narratives in reports would lead to higher blame, due to more opportunities for things like opinions, assumptions and accusations [and presumably counterfactual & normative language].

All safety reports related to anaesthesia services over a four year period at a single centre were coded by three independent reviewers (263 reports; an anesthesiologist, a hospital chief risk officer and a medical student). Each reviewer evaluated attributions of blame & whether issues with communication were reported. Blame was defined as “evidence in the free-text of a judgment about a deficiency or fault by a person or people”.

Results

As shown in the table below, the degree of agreement between reviewers varied significantly about whether blame was present in incident narratives.

However, there was agreement between the reviewers that higher word counts and a communication issue were “more likely to be present in the safety reports that had attribution of blame”.

As shown below, a higher median word count was present in reports believed to indicate blame (avg 75 words) versus reports without a blame attribution (avg 36 words).

In discussing the findings, it’s said that reporting in health care traces its history back to aviation; but unlike aviation, it’s said many adverse events aren’t reported in health care; “leading health care workers to report only what they cannot conceal”.

As is well-known, a barrier to voluntary reporting is the “punishment meted out for making an error”.

Although there was weak inter-rater agreement on which report contained attributions of blame (but agreement on the length of narratives in reports and blame), it’s said that this indicates that a subjective filter exists for blame in all people and “perhaps the use of a common definition may not be able to override this”.

Further, the different perceptual filters reflect the different backgrounds of people (professions, training, experiences etc) reflects the reality of, really, any workplace. That is, even though having weak inter-rater agreement may be seen as a study limitation, it’s probably at least consistent with most workplace arrangements with incident investigations.

The presence of blame in reports has consequences. They cite one study which found that people who were given agentive descriptions of an incident (“he ripped the costume" versus “the costume ripped”) were more likely to blame people for the incident versus those without agentive language.

It's said that often the person reading (or even writing) an incident report didn’t personally witness the event, so will form impressions only from reading the report.

Furthermore, reports that found communication issues were also more likely to attribute blame.

I had a bit of trouble with practical takeaways with this; and I suspect the authors did too. They discussed some points, but not sure I really agree with them. You can read the report and form your own conclusions.

One challenge though, as the authors note, is that if you reduce the narrative to reduce the chance of blame, then you also lose the richness of information.

They also argue that the development of corrective actions is more important than the safety report itself, so perhaps more focus on actions rather than describing is warranted. In practice, I don’t entirely agree. This is because there’s research highlighting how corrective actions are strongly influenced by sociopolitical, perceptual, structural and resourcing factors such that people identify improvements that are easy to see, easy to fix, easy to close, easy to appease other stakeholders and not necessarily strongly linked to the elimination and/or mitigation of that class of issues and contributing factors.

There’s also the point that it’s difficult to really improve something without properly understanding the different narratives about work.

Link in comments.

Authors: Ackerman, R, Patel, S, Costache, M, Petitt, M, Mhaskar, R, Cohen, J, 2021, General Surgery News