Evidence

Does losing sleep actually make you a worse operator?

The orthobro · 2026-08-12 · 6 min read

You came off a 24-hour shift, scrubbed into a case, and it went fine. So the whole thing is overblown — right?

Two pieces of evidence say two different things. Both are worth knowing, because most people quoting one have never read the other.

On a simulator, you do get worse

A group of orthopedic residents was tested on a shoulder arthroscopy simulator before and after 24-hour shifts, three times over six months. Ten residents, 120 exercises.

Their scores dropped on both tasks. On one exercise the overall performance score fell from 48.83 to 43.23. On the other it fell from 124.34 to 118.64. Both differences were statistically significant. Average sleep during those shifts was 3.3 hours.

This is about as close to your actual job as this research gets. It is orthopedic residents, on an arthroscopy simulator, after real call.

It is also ten people at one hospital. The authors say so themselves. Ten residents is not a law of nature.

Across whole programs, patients did the same either way

Now the other side.

A national trial randomized 117 general surgery residency programs to either standard duty-hour rules or flexible ones that dropped the shift-length caps. Then it looked at what happened to 138,691 patients.

Death or serious complications within 30 days: 9.1% under flexible rules, 9.0% under standard rules. No meaningful difference.

That is a very large study, and it did not find that letting residents work longer hurt patients.

Why both can be true

It is tempting to pick whichever result you like and stop reading. The more useful move is to notice that these two studies are not measuring the same thing.

The simulator study measures you, on a specific morning, doing a specific task, with your performance recorded to two decimal places. Nothing else changes.

The duty-hour trial measures a scheduling policy, across entire programs, where everything changes at once. Under flexible rules a tired resident might stay for a case rather than hand it over — and that trade runs both ways. Attendings supervise. Cases get reassigned. Someone else scrubs in. The system absorbs a lot.

So a measurable drop in your individual skill can be entirely real, and still not show up as more dead patients across a hundred programs. The system is designed to absorb exactly that.

What that trial did not test is whether you personally, post-call, operate as well as you do rested. The simulator study is the closer answer to that question, and its answer is no.

What this is not

It is not evidence that operating post-call is dangerous. Nobody has shown that cleanly, and the largest study we have points the other way.

It is also not permission. "The complication rates came out the same across 117 programs" is not the same claim as "I am fine right now."

The honest summary is narrower and less satisfying than either camp wants: your hands are measurably slower and less accurate after a night on call, the system usually catches the difference, and how much that matters on any given case depends on the case.

The better argument for sleeping

Everything above is framed around damage. That is the boring half of this, and it is the half everyone repeats at you.

The more interesting case for protecting your sleep has nothing to do with fear. It is that sleep is doing work you cannot do any other way.

Sleep is when a new skill actually sticks. In one study, 57 volunteers learned a finger-tapping sequence and then got faster overnight — between 17.5% and 28.9% faster — without touching the task again. The improvement came from sleeping, not from more practice.

The best part is which sequence improved most. The hardest one. The difficult two-handed pattern gained roughly 50% more than the easy patterns did. Sleep put its effort where the learning was hardest.

That was students on a keyboard, not surgeons on an arthroscope. Nobody has run that study on a knee, and I am not going to pretend they have. But the direction is worth knowing: the night after a case you found difficult is not dead time. It is the second half of learning it.

Sleep also cuts your error rate away from the OR. Interns working traditional extended shifts made 35.9% more serious medical errors than the same interns on a schedule with those shifts removed. Serious diagnostic errors were 5.6 times as common — 18.6 per 1000 patient-days against 3.3.

Note what that is and is not. Those are errors on the ward, not deaths on the table, which is why it sits comfortably beside the duty-hour trial above. Different outcome, different question, both real.

So

Some of your sleep is not yours to control. The rota is the rota, and no article fixes that.

But some of it is. The hour before bed. The nap between cases. The night after a hard case, specifically — that one is worth guarding, because that is the night your hands are consolidating what they just learned.

That is the honest pitch: sleep is not recovery time you are stealing from work. It is the cheapest training you will do all week, and it works hardest on exactly the things you found most difficult.

And for the mornings when you got none of it anyway — the fix is not heroics. It is having already decided. Knowing the steps, the positioning, the implant sizes and that attending's preferences before you walk in costs nothing at the moment you are most depleted, because you did that thinking when you were rested.

That last part is not a research finding. It is just the obvious response to one.