Updated: Mar 1, 2020
exploring how (and why) time pressure makes us worse at diagnosing things correctly.
The Quick Look
"Factors underlying suboptimal diagnostic performance in physicians under time pressure" ALQahtani, DA, et al. Medical Education, 2018. PMID: 30302783. Find it here.
Time pressure (the pressure we feel from things needing to happen right now) impacts our ability to think and perform optimally--knowing exactly how this works could help us figure out how to build training to mitigate and overcome these effects. In this study from King Saud University in Riyadh, Saudi Arabia, the investigators tried to explore whether time pressure leads to suboptimal diagnostic performance (choosing the right hypothesis from among many) via the emotional and physiologic effects of stress on our thinking, or because we simply run out of time and don't generate enough potential hypotheses in the first place.
This study suggests both are probably true, to different degrees. Basically, time pressure likely makes us generate fewer initial options as potential solutions, and also makes us worse at picking between them. As the authors put it: "Our findings suggest that making diagnoses under time pressure may result in distortions in hypothesis generation, the evaluation of available alternatives, and the production of an accurate final diagnosis." When we're making complex decisions under pressure, this study suggests therefore that there are two potential areas to improve on: (1) expanding on the initial options we generate to see if something is missed and (2) slowing down at the point of decision to improve our ultimate choice.
Prospective, randomized controlled experiment, with n=75 participants, all of whom were senior internal medicine residents at two teaching hospitals in Riyadh, Saudi Arabia. Residents in their 3rd or 4th year of internal medicine training (different than the American system) were tasked to identify the diagnosis for eight different standardized presentations, then afterward to rate their stress and list alternative diagnoses they had considered. The exposure group (n=40) had a set time limit to finish the cases (the time pressure), while the control group (n=35) did not. There were no demographic or training differences between the groups.
- The time pressure group really felt the pressure. They used significantly less time compared to the control group (85.5 +/- 33.8s per case compared to 181.8 +/- 96.6s) and were worse at making correct diagnoses (diagnostic score 0.44 +/- 0.18 vs 0.53 +/- 0.15, with a score of 1 being perfect).
- The time pressure felt significantly more stressed (5.8 +/- 2.2 compared to 4.7 +/- 2.2 on a scale of 1-9 with 9 being the most stressed), and also generated fewer good alternative diagnostic hypotheses (score of alternative diagnoses 0.37 +/- 0.22 vs 0.51 +/- 0.27, with a score of 1 being perfect).
- There did not appear to be a significant relationship between stress and the number of diagnostic alternatives considered after adjusting for the effects of time pressure, which the authors interpret as the two pathways (stress and hypothesis generation) being different from each other.
How exactly time pressure makes us worse at complex decision making is an interesting problem and a key step to figuring out better training for decision making under pressure. This study suggests that there are two separate pathways: (1) time pressure makes us generate fewer initial options as potential solutions to a problem, and (2) time pressure stresses us out, which makes us worse at analyzing and choosing between potential options. Both of these are reasonable pathways and probably contribute to the effects of time pressure on decision making. It is less clear how these pathways interact, what other pathways might be involved, and which among them are the most important and/or modifiable at the individual and system levels.
From a scientific perspective, the study was well executed, but small and likely underpowered, especially when it comes to detecting interactions between the potential pathways. Additionally, it's not immediately clear how other individuals with more training (experienced internal medicine doctors, for example, or specialists in emergency care) would behave under similar circumstances. Training for emergencies--if we do it correctly--will rewire the way that we think and perform under pressure, so it would be very interesting to see similar studies with trained responders.
- When under significant time pressure, consider looking harder for alternative initial hypotheses, as you're perhaps less likely to generate them "naturally" than usual.
- Keep training under pressure to (try to) mitigate the effects of pressure on complex decision making.
- Potential to develop similar work studying differences in pressurized decision making between novice and trained operators?
- More ideas? Comment below, we'd love to hear them.