Absolute bunk. Doctors are the last people in the world who should "insist they are not affected by sleep deprivation." Their whole job is to make decisions based on medical evidence. It doesn't affect you? Prove it. Show me a study looking for correlations between medical errors and sleep deprivation. Show me a study measuring alertness at the end of an 80-hour work week. Don't give me anecdotes. Give me proof. To do otherwise, as a doctor, is hypocritical.
Here's my anecdote: my wife is a resident, and she comes home from a 30-hour shift so tired, she's nervous about driving and sometimes feels like she has the flu.
Is she just weak? Should other doctors look down on her?
Not without a rock-solid study to prove their point. Instead, doctors seem to have a hazing mentality - "I survived this, so you should have to, too." But it's really a question of medical fact: how do medical work schedules affect patient outcomes? Any doctor who doesn't see that is being dangerously irrational.
A simple test, that worked very well for me, was testing my reading speed (I know it is only a weak indicator for mental acuity, but it is easy and fast to test).
Under perfect conditions (well rested, optimal nutrition, completely silent room, no visual distractions, after mediation and a good cup of green tea) I am able to read about 800 WPM (which is not to bad) of the kind of text I usually need to read with good retention. For me this declines rapidly to less than 200 WPM (which is quite bad for a "knowledge worker") with sleep deprivation and/or a noisy environment, even though I believe I am doing fine.
That being said, even though I would say that doctors are handicapped "mental performance" is handicapped by their working conditions, I fully agree that it is better for patient safety to not hand off a patient to a new doctor say every 8 or 12 hours due to the amount implicit knowledge that a doctor gathers during a >24h time period about his patients that can't be transferred in a handoff.
Software Developers, imagine what a screw up it would be if you'd have to hand off your code every 8h to another developer, who'd have to continue right were you left... add to this not having a version control system, spotty test coverage and highly ambiguous specifications
While this may be true, the process of the handoff could be vastly improved. My wife is a resident, and the quality of info she gets during a patient handoff varies drastically, from detailed notes in a computer, correlated to medical history, to scrawled notes on paper. Sometimes she has to start from square 1.
Maybe the reason handoffs are harmful is because they're done poorly. And maybe they're done poorly because the doctor who's handing off is freaking exhausted.
It's like a guy who's driving home after a 20 hour day that ended with a couple of drinks at the bar, calling home from the car to tell the wife that he's on his way and his driving is perfectly safe. In reality, he is so impaired by the tiredness, the alcohol, and the distraction of the call that he can no longer even understand how far off his game he must be, nor judge the horrific risk he must be taking by driving in that way.