Context

Coming face-to-face with a trainee who needs to be failed is a stern test for many supervisors. In response, supervisors have been encouraged to report evidence of failure through numerous assessment redesigns. And yet, there are lingering signs that some remain reluctant to engage in assessment processes that could alter a trainee’s progression in the program. Failure is highly consequential for all involved and, although rare, requires explicit study. Recent work identified a phase of disbelief that preceded identification of underperformance. What remains unknown is how supervisors come to recognise that a trainee needs to be failed.

Methods

Following constructivist grounded theory methodology, 42 physicians and surgeons in British Columbia, Canada shared their experiences supervising trainees who profoundly underperformed, required extensive remediation or were dismissed from the program. We identified recurring themes using an iterative, constant comparative process.

Results

The shift from disbelieving underperformance to recognising failure involves three patterns: accumulation of significant incidents; discovery of an egregious error after negligible deficits; or illumination of an overlooked deficit when pointed out by someone else. Recognising failure was accompanied by anger, certainty and a sense of duty to prevent harm.

Conclusion

Coming to the point of recognising that a trainee needs to fail is akin to the psychological process of a tipping point where people first realise that noise is signal and cross a threshold where the pattern is no longer an anomaly. The co-occurrence of anger raises the possibility for emotions to be a driver of, and not only a barrier to, recognising failure. This warrants caution because tipping points, and anger, can impede detection of improvement. Our findings point towards possibilities for supporting earlier identification of underperformance and overcoming reluctance to report failure along with countermeasures to compensate for difficulties in detecting improvement once failure has been verified.