Effects of a continuous work-flow on residents in the intensive care unit
Posted Mar 11 2010 6:54am
It is well known that the average physician in training will be expected to work more than a few 24 hour shifts during their training. It is also well known that sleep deprivation affects performance (how much? Now that’s the real question, but I digress).
I read a paper, evaluated by Faculty of 1000 member Samuel Ajizian, in which a team of scientists led by R Sharpe at the University of British Columbia have studied the effects of a continuous work-flow on residents in the intensive care unit.
Although monitoring performance of sleep deprived workers is a popular research subject (see, here , here and here ), there are a couple of nice aspects to this study. Firstly, the use of a realistic total patient simulation is novel to this type of study; much of the previous body of research has focussed on more general cognitive testing and surgical simulators. The participants also assessed their own performance at various stages of the study allowing the researchers to see if the physicians could notice a drop in performance for themselves.
The simulation included performing advanced cardiac life support scenarios and management of a simulated critically ill patient.
I could spot a potential problem with this paper however. And I was not surprised to read the following statement
For the advanced cardiac life support scenarios, the mean number of major errors committed… decreased during the study period.
We have what looks like (and is subsequently judged to be) a significant ‘learning effect’, and I am surprised the team couldn’t have found a way to assuage this, perhaps by demonstrating the simulation and training physicians before observation?
Results from the patient management scenario followed expectation
The mean number of errors went up from 0.92 +/- 0.90 in the first session to 1.58 +/- 0.79 in the fourth session (p = 0.9).
Another interesting aspect of this study was that Sharpe and his team asked the physicians to conduct subjective self assessment at various stages. Compare the following results of personally assessed global score (again from patient management) with the number of errors displayed above
mean global score decreased from 56.8 ± 14.6 to 49.6 ± 12.6 (p = .02).
A discrepancy between the mean number of errors and the mean global score can be seen here, strongly suggesting that our ability to judge personal performance is not nearly as accurate at hour 20 as it is at hour 1.
Although I wouldn’t like to make any serious judgements from this paper alone, I think there is significant room for further research in this area, especially using realistic patient simulation.
Ajizian also noted in his evaluation that it would be interesting to [repeat this study and allow the residents unlimited access to their preferred caffeine source].
Doing so might provide us with some better data, especially as we tend to work under the influence of caffeine even when we aren’t particularly tired. I wonder if performance levels might correlate to the intensity or the type of stimulant used. Perhaps coffee offers a greater mental boost for physicians than tea? Any anecdotal evidence is welcome below.
Sharpe R, Koval V, Ronco JJ, Dodek P, Wong H, Shepherd J, Fitzgerald JM, Ayas NT. The impact of prolonged continuous wakefulness on resident clinical performance in the intensive care unit: a patient simulator study. Crit Care Med 2010 38:766-70