Friday, November 8, 2013

Sleep deprivation, surgeons, operations, and outcomes


A new paper found that surgeons who performed elective laparoscopic cholecystectomies after having operated the night before had outcomes similar to those when they were presumably well-rested.

The retrospective study involved 331 surgeons who did 2078 cholecystectomies after operating the night before and 8,312 when not operating the night before. Outcomes both were matched for each surgeon.

Comparing outcomes after operating the night before to not found rates of conversion to open - 2.2% vs. 1.9%, risk of iatrogenic injuries - 0.7% vs. 0.9%, and death - 0.2% vs. 0.1%, respectively. None of those differences were significant.

The abstract concluded, "These findings do not support safety concerns related to surgeons operating the night before performing elective surgery."

This paper is the latest of several that show similar results.

So case closed, right?

As much as I hate to say this, the paper does not prove that sleep deprived surgeons don't have more complications than when they are well rested. What it does prove is that conversion rates, not complication rates, are the same whether the surgeon got adequate sleep the night before or not.

In the paper, which was published in JAMA, the authors said, "Although not always considered a complication, conversion to open cholecystectomy may serve as an aggregate end point for many complications."

I disagree. I know of no previous study confirming that conversion of a laparoscopic cholecystectomy to an open procedure is a marker for complications. Instead, I believe it is a sign of good judgment. The sooner a surgeon recognizes that he can't safely do the procedure laparoscopically, the better off the patient is. A surgeon should never be discouraged from converting a case to open.

The study probably included enough patients to support its conclusion that there is no difference in conversion rates, but it is underpowered to detect a difference in iatrogenic injury rates or mortality because those events are so infrequent. To conclude that there is no difference in iatrogenic injury or mortality rates is what is known in statistical circles as a "Type II error" or failure to reject a false null hypothesis. The two null hypotheses in this situation were that there is no difference in 1) iatrogenic injury or 2) mortality rates when surgeons are rested or not.

In other words, the rates of iatrogenic injuries and deaths may not really be different, but the lack of a difference could simply be due to the fact that there were not enough subjects in the study. Iatrogenic injuries and deaths occur so infrequently with laparoscopic cholecystectomy that a study would need a lot more patients in each group to conclude that sleep is not a factor.

Most media coverage of the paper did not question its findings. Even Atul Gawande was hooked. Yesterday he tweeted "New @jama study of daytime surgery by surgeons who operated during night before: found NO increased complications."

Better studies on the effects of sleep deprivation on surgeon performance are needed before the issue is settled.


25 comments:

Anonymous said...

Were there differences due to the ages of the surgeons? I know that as a 20-something resident I could get by with 2-3 hours sleep, work 10 hours next day, and then go out for a couple of beers. Now at age 50? Not so much.

(I am not a surgeon.)

Ray Collins said...

Thanks for this post. I am not a surgeon, so I found your viewpoint useful. I referenced it in my own post today. http://eraycollins.postach.io/conversion-vs-complication-rate-as-a-sleep-deprivation-metric

Surgeon said...

I am better in the OR if I am sleep deprived. I am worst just after a restful vacation.

Todd J. Scarbrough, M.D. said...

JAMA publishes negative retrospective studies that look at questionable surrogate measures which yield bold conclusions highly susceptible to errors of statistical inference? They do that now?

Well OK.

Skeptical Scalpel said...

I accidentally deleted a comment by Pamchenko (sorry) which said, "converting a lap to open gallbladder usually means its a really bad gallbladder. probably a male gallbladder." I agree some people think cholecystitis is worse in males. Of course, the study was about elective surgery and almost 75% of the cholecystectomies were performed on women.

Anon, the average age of the surgeons was 48. Other than that, there was no mention of a breakdown of cases by surgeon age.

Ray, thanks.



Skeptical Scalpel said...

Surgeon, I don't know about that. You're going to need a bigger series to convince people.

Todd, I am surprised that this appeared in JAMA. An editorial about this paper by two well-known surgeons did not raise any of the points I mentioned.

Todd J. Scarbrough, M.D. said...

me too. And one of the authors seemed to be a PhD biostatistician. My biostats profs taught me to be more measured in my conclusions, and what you were hinting at is that this needs to be an equivalence study (http://learnpkpd.com/2011/01/01/trial-designs-non-inferiority-vs-superiority-vs-equivalence/). "These results do not find a significant difference between group A and group B" would have been a far more statistically honest statement than what the conclusion said.

Skeptical Scalpel said...

Todd, I agree. there's nothing wrong with saying that conversion rates were similar but to equate them with complications? I don't think so.

Skeptical Scalpel said...

Drew Shirley, MD put this comment on NowTalk: I really agree with this post! I was a little shocked at how others on Twitter propped this up as definitive proof that sleep deprivation doesn't effect surgical outcomes. I think a fundamental problem with the study is that it used lap chole as the procedure to prove this when even the most inexperienced surgeons could do this instinctively and they have a fairly low complication rate baseline.

Good comments, Drew. The problem for the researchers is they needed a high volume procedure to study. But since the rates of iatrogenic injury and death are so low, they would need more than 20,000 patients in each group to avoid a Type II error.

Anonymous said...

SS,

Good point about the problem of using lap choles for this study. Most surgeons can do them in their sleep (oops).

I am puzzled about the statistical comments. I'll use Type 1 errors as an example. Say Drug A lowers blood pressure by 2 mmHg in a study of 1000 patients, and is deemed statistically insignificant. The researchers then studied 100,000 patients and got the same 2mmHg reduction, but the stats are now highly significant.

What is the clinical relevance?

Similarly, if you need to study 20,000 patients to avoid a Type 2 error, does that mean anything to a patient undergoing surgery?

Skeptical Scalpel said...

Anon, good questions. Regarding the difference between statistical and clinical significance--they are not necessarily equivalent. In your example, a 2 mm difference in BP might be statistically significant but not clinically important. I wrote about this 2 years ago. It's the most read post I've ever written. Here's the link http://skepticalscalpel.blogspot.com/2011/08/statistical-vs-clinical-significance.html.

Statistics don't mean much to an individual patient. The woman who suffers a common bile duct injury during a lap chole doesn't care that her odds of having that happen were less than 1%. But stats help us explain risks to patients and make policy decisions as a society.

Ray Collins said...

Surgeon, "worse after a holiday" has been studied at various times and places. The first I recall from past decades is the air force wondering why their jet fighter pilots crashed their planes more often on Monday.

Anonymous said...

Even if sleep deprivation does not cause more complications it doesn't mean it is OK. When I'm sleep deprived I'm unhappy, irritable, and unpleasant. Sleep deprivation affects my mood and family life. It is unhealthy also.

Skeptical Scalpel said...

Ray, I'm sure that you can find research to support both sides of the issue.

Anon, everyone reacts differently to lack of sleep. If sleep deprivation bother you that much, you probably shouldn't be a general surgeon.

Anonymous said...

Even if sleep deprivation worsens outcome, what is the solution?

If you are practicing in a small group in a moderately busy community hospital, you have a good chance of being wakened at night when you are on call. It is not feasible to cancel your next-day elective OR schedule or your office even if you had worked all night.

A growing number of larger hospitals now use surgical hospitalists, including orthopedic and obstetric hospitalists. Of course, since most of them cover a week or 2 at a time, they are still at risk of operating after little sleep.


Skeptical Scalpel said...

Anon, you are absolutely right. There aren't enough surgeons to cover especially in rural areas. and yes, surgical hospitalists can work several 24-hour shifts in a row. But when I was doing that, I was able to nap in the daytime after a bad night because I didn't have a long elective schedule or an office full of patients the next day.

medaholic said...

I was similarly surprised by my colleagues who interpreted this study as same conversion rates means same amount of complications. Complications and harm are already so hard to define and measure, that it's a gross oversimplification to say open chole's are complications.

The problem is the retrospective study design of elective (low risk) choles. I definitely would not want a sleep-deprived surgeon doing a big case

Anonymous said...

SS,

OT, but can you comment on your experiences as a hospitalist. It sounds like you can fit in anywhere, but I assume that surgeons who insist on such-and-such or I quit may not do as well.

Skeptical Scalpel said...

Medaholic, We are in complete agreement.

Anon, here's a post I wrote a while ago about surgical hospitalist work (http://skepticalscalpel.blogspot.com/2011/03/general-surgery-acute-care-surgery-and.html). Yes, it's not an ideal job for a fussy surgeon.

RuggerMD said...

I agree, conversion to open is not a complication, it is good judgement.
A complication is spending 8 hrs trying to do it laparoscopically or worse yet not converting and causing an injury.
IMO 99.9% of complications from lap chole are CAUSED by us.
The anatomy is usually straight forward with of course all the variations you can find in Netter's. If you don't know the anatomy then don't cut it.
If you can't see the anatomy from body habitus, inflammation, scarring or whatever, then open.
If you still can't find the cystic duct secondary to scarring then ligate at base of infundibulum and oversew it.
You don't always have to be right on the cystic duct. Some times as we say "the enemy of good is better".
But that is just what I do.

Skeptical Scalpel said...

Rugger, I agree with you. I also think it is still ok to do a cholecystostomy occasionally.

RuggerMD said...

Agree.
Just ordered a Perc Cholecystostomy on a 95 y/o with STEMI

Skeptical Scalpel said...

I rest my case. Thank you.

Doctor Which said...

Studies with small amounts of alcohol show that the person is aware of minor impairment and is more careful in risk taking. It has been shown to improve overall safety.
I believe all this has been hyped, something to do with junior doctors hours and the Geneva Convention. Nobody ever considers the cultural loss. The medical ethic of selflessness used to be indoctrinated as a part of absolute vocation. Oh, it was brainwashed into the subconscious. It was dehumanising....
Or was it superhumanising in terms of ethics.

We'll miss them as in the case of the scottish hospital. The errors are a lack of self discipline caused by putting their petty little selves before the patient

Skeptical Scalpel said...

Doctor Which, thanks for the comments. Good points about sleep and care. Yes the Scottish personnel who let the patient fall of the OR table weren't sleepy, just careless. That post is here http://skepticalscalpel.blogspot.com/2013/11/patient-falls-off-or-table-system-error.html.

Post a Comment

Note: Only a member of this blog may post a comment.