SafetyInsights.org

This study, with Amy Edmondson as one of many co-authors, explored how different types of near miss and incident event types lead to different learning outcomes.

The focus is radiation oncology, but the categories may still be applicable elsewhere.

Providing background:

·        Prior work suggests that both near miss events and actual incidents are rare given the total amount of work; e.g. an average of 466 incidents, including near miss events, over a year in a department. Of that, just 53 events make it through to the patient

·        Another study associated 14 near miss events per actual incident

·        Nevertheless, only a fraction of actual incidents are reported

·        Six event types were differentiated, based on their attributes of latent error and enabling conditions: hit, potential hit, almost happened, fortuitous catch, could have happened, and process-based catch

·        Near miss events that “could have happened” are differentiated from those that “almost happened”, based on how the former elicit cues of system resilience (that is, the event is perceived as a success), and the latter elicits cues of system vulnerability (the event is perceived as a failure)

·        They reason that the difference may lie in the varying presences of latent error/variability in performance, and the enabling conditions that predispose the patient to harm

·        I.e. “could have happened” events feature only the underlying latent error of behavior or process. On the other hand, “almost happened” events feature the latent error as well as the enabling conditioning signaling predisposition to harm” (p257)

They provide the example of a patient with a pacemaker needing radiation. Staff are required to check whether the patient has a pacemaker, and if so, seek authorisation to administer radiation. If the patient has a pacemaker and gets radiation without the staff knowing/nor getting authorisation, it is possible for immediate harm to manifest or harm later on (with failure of the pacemaker).

For definitions of incidents:

·        Hit: adverse events resulting in patient harm

·        Potential hit: events that reach the patient, but do not result in immediate harm (but may manifest in harm later on)

·        Almost happened: near misses that reach the patient, but no harm results. They note “In retrospect, these events contain prior signaling of potential harm (eg, the presence of a pacemaker), that is, the enabling condition for harm, as well as a latent error by staff (eg, not checking the pacemaker status)” (p258)

·        Fortuitous catch: Also called a good catch, these are events that don’t reach the patient and are averted by chance from human intervention.

·        Could have happened: near misses that don’t reach the patient and don’t contain prior signals of the enabling conditions. E.g. the pacemaker status wasn’t checked prior to radiation, but the patient didn’t have a pacemaker anyway. These types are said to contain only active conditions (rather than latent).

·        Process-based catch: near misses that are averted by a systems-based process or check for the enabling conditions. E.g. the process requires staff to check for pacemaker status and the staff correspondingly check for the pacemaker.

An overview of the types are below:

Results

·        Perceived success scores of the events and willingness to report the events differed by event/near miss types

·        “Could have” happened events were viewed as ‘less successful’ (e.g. closer to actual unmitigated harm/failure), and were more likely to be reported by staff compared to “almost happened” events

·        Even though fortuitous catch events had the presence of enabling conditions (the pacemaker was present), people perceived these types as more indicative of success compared to “could have” events

·        Nevertheless, more people were likely to report fortuitous events because they were deemed more successful [* and I suppose therefore, people were less worried to report these events]

They say that given enabling conditions “cognitively signal greater proximity to the negative outcome than without such conditions”, these events, having a higher disposition to actual harm, can “elicit greater counterfactual learning, possibly due to negative outcome bias” (p260).

Staff in this sample perceived the almost happened events (featuring the enabling condition) as closer to failure and being more report-worthy for organisational learning.

They state that the “counterfactual scenario of the patient (almost) having an arrhythmia due to the missed pacemaker check may be more cognitively accessible and result in greater perceived risk due to dominant associative information-processing” (p261); therefore, events that can be more readily associated with these events (the enabling conditions) are more likely to result in concern and disposition to learning.

In contrast, events that only feature the latent error and not the enabling condition, “could have happened” events, may not result in cognitively assessable counterfactual scenarios. Thus, these events are less likely to elicit counterfactual learning.

People were less likely to view them as failures. “Could have happened” events were “more likely to be viewed as a case of system resilience, and in the absence of salient information about the enabling condition that signals the predisposition to harm, these events have been shown to increase risk-taking behavior, that is, the normalization of deviance” (p261).

They argue that current definitions of potential harm in healthcare “anchor on the enabling condition and less so on the more pervasive but subtle latent errors in behavior and processes” (p261).

Moreover, fortuitous catch events were seen as more successful than failure orientated—irrespective of the enabling condition signalling predisposition to harm. They suggest that, perhaps, staff may “take responsibility and credit for a “good catch” (p261).

They also note that there could be false confidence in relying on safety processes if those processes are inefficient or not adhered to [* no surprises here of course.] Hence, organisations must better engage from workers and learn from normal situated work to remove “deadweight” processes and decentralised problem-solving mechanisms.

Ref: Kundu, P., Jung, O. S., Valle, L. F., Edmondson, A. C., Agazaryan, N., Hegde, J., … & Raldow, A. (2021). Practical Radiation Oncology, 11(3), e256-e262.

Study link: https://doi.org/10.1016/j.prro.2020.09.007

LinkedIn post: https://www.linkedin.com/pulse/missing-near-miss-recognizing-valuable-learning-ben-hutchinson