After 10 hours on the job, a truck driver must pull off the road. After 16 hours, an airline pilot can no longer legally fly a plane. But after 24 hours or more on the job, with perhaps an hour nap somewhere along the line, a first-year medical resident can perform a surgical procedure, write a prescription, or insert a chest tube.

This introduction to a 2004 article in Focus Online, the newsletter from Harvard Medical, Dental, and Public Health Schools, says it all: we expect doctors to be superhuman.

When you stop and think about it, the expectation that medical residents, especially first-year interns, can, or should, perform incredibly complex procedures with minimal sleep is crazy. Medical researches agree—which is why hospital shifts of doctors-in-training have come under much scrutiny in recent years.

In 2003, the Accreditation Council for Graduate Medical Education (ACGME) created standards to restructure residents’ hours in order to combat fatigue, and the dangerous mistakes that accompany exhaustion. The council’s reforms limited residents’ hours to:

• No more than 80 hours a week
• No more than 6 work days a week, averaged over 4 weeks
• No more than 24 continuous hours of duty, except for another 6 hours of education or transfer of care In-house call no more often than every third night
• No less than 10 hours of rest between duty periods

These changes were meant to benefit not only residents, but also patients: an exhausted doctor is a careless doctor. But for all of the Council’s good intentions, studies show that the ACGME reforms don’t go far enough—residents need more of a break if they are to maximize their effectiveness and ensure the safety of their patients.

How do we know?  Last year, two studies from Kevin G. Volpp, MD, PhD, from the Center for Health Equity Research and Promotion, VA Hospital in Philadelphia, Pennsylvania looked at whether the first four years of the ACGME guidelines had resulted in a meaningful decline in patient mortality—in other words, if the guidelines had saved lines by keeping doctors alert and rested.

Unfortunately, after looking at 318,636 VA patients from June 2000 to
June 2005, and a whopping 8,529,595 Medicare beneficiaries over the
same period, Volpp found that “the overall impression remains that duty
hour restrictions have had little effect on patient mortality.” For the
VA patients, there was a slight improvement in mortality after the
introduction of the ACGME guidelines, but not for surgical patients.
Volpp’s Medicare study showed that “the ACGME duty hour reform was not
associated with either significant worsening or improvement in
mortality.” According to Volpp, these mixed results mean that there is
still “controversy about whether the current duty hour rules are
sufficient for regulating work hours."

This last sentence is key: the problem with ACGME reforms isn’t that
reducing resident hours is a bad idea, but rather that the guidelines
don’t go far enough in restructuring schedules. Again, the question
must be asked: how do we know? 

A 2004 study by a Harvard Medical School (HMS) team led by Charles
Czeisler, the Frank Baldino Jr., PhD professor of sleep medicine at
Brigham and Women’s Hospital, compared the ACGME paradigm to an even
more flexible one they designed—and found that, when it comes to
resident shifts, less is more.

Czeisler’s team followed “20 interns [who] completed two three-week
rotations in the intensive care unit,” and divided the interns into two
groups.  One group followed the ACGME guideline of 80 hours a week
(although in practice, “nearly all interns” worked at least 80 hours),
and the other kept to a revised schedule which allowed a maximum of 16
continuous hours of work at a time, for a total of around 65 hours per
week. (This revised schedule was called the “intervention schedule” in
the study).

Here’s a run-down of the study’s methodology and results, courtesy of
Focus Online: “The interns logged the hours they spent sleeping and
were monitored at work and at home with electrodes to record eye
movements and brain waves. On the intervention schedule, interns
averaged 5.8 hours more sleep per week and experienced less than half
the rate of ‘attentional failures’ at night, slow-rolling eye movements
caused by profound sleepiness during waking activities at work.”

As unimpressive as the absence of “slow-rolling eye movements” may
sound at first, in reality it’s critical that residents be well-rested
(as we’ll soon see). In a follow-up study, Christopher Landrigan, an
HMS assistant professor of pediatrics, led a team that looked at the
Czeisler’s intervention schedule in order to determine whether it
reduced serious medical errors. Rested interns are all well and good,
but what really matters is whether or not rested interns screw up less.
Once again, Focus Online summarizes the methodology and findings:

“A team of observing physicians (working in eight-hour shifts)
shadowed the interns at all times over the course of their rotation
through the intensive care unit. The observers recorded and categorized
every error that the interns made, and the results were verified by two
physicians who were unaware of the schedule the interns were following.
The study found that interns made 35.9 percent more medical errors
considered serious during the traditional [ACGME] schedule, including
20.8 percent more medication errors  and more than five times as many
serious diagnostic errors].”

This is scary stuff, and the major problem seems to be that the ACGME
shift framework is still too darn demanding—especially because it
allows for 30 hour continuous shifts.

In a more recent study,
Harvard’s Czleisar was part of an American and Canadian team of medical
researchers who tracked 2,737 interns via web surveys where they “gave
information about their working hours, hours of sleep, and number of
extended-duration shifts [i.e. at least 24 hours continuously at
work]…and completed questions about medical errors in the past month.
Then, for each intern in the study, researchers compared month by month
the number of medical errors and the number of extended-duration shifts
that had been worked.”

The researchers found that months heavy on extended-duration shifts
were also heavy on medical errors. When “compared to months in which no
extended-duration shifts were worked,” it turned out that in  months
where an intern completed between one and four extended-duration
shifts  researchers were 3.5 times as likely to see “at least one
fatigue-related significant medical error.” And in months when an
intern worked five or more extended-duration shifts, this probability
shot up 7.5 times—meaning there was a 750 percent greater chance of
major screw-ups in the five-times-or-more extended-duration months than
in those without extended hours.

Equally shocking are the studies’ numbers on “fatigue-related
preventable adverse events.”  In months where a resident worked
extended-duration shifts, the probability of these events increased
anywhere from 700 to 850 percent! Most strikingly, “interns working
five or more extended-duration shifts per month…reported 300 percent
more fatigue-related preventable adverse events resulting in a
fatality.” Thus, quite literally, medical shifts place patients’ lives on the
line—and if a resident is too exhausted, patients can lose out
big-time. Lo and behold, doctors are human—indeed, all too human.

Given all of this data, there seems an obvious conclusion: if a tired
resident is a dangerous resident, why not just put strict limits on
their work hours?

There are many reasons why we need to be careful about cutting medical
shifts. For one, medicine will always be a high-endurance profession,
no matter how resident shifts are structured. Some surgeries can take
thirteen or fourteen hours to complete, and the unexpected is routine
in hospitals and emergency rooms, meaning that stamina is of the utmost
importance. Pampering residents with neatly bounded shift schedules
won’t provide adequate preparation for the reality of being in the
trenches.

Another reason we need to be careful in chopping up shifts is that the
doctor-patient relationship is both important and sensitive. A “punch
the clock” system of medicine, where providers check in and out
regardless of what’s going on with their patient, is no more desirable
than a marathon framework. Do you really want your doctor checking his
watch during surgery?

The last concern is that the more switches there are, the more often
patients are handed-off, i.e. transitioned from one provider to
another. Hand-offs are dangerous moments, where miscommunication can
lead to serious problems for patients—something which I’ve discussed in the past.

But there’s a huge middle ground between an 80 hour work week and too many punch-the-clock hand-offs.

The Internet might be of help here—if virtual medicine is in fact all
its cracked up to be, shifts could become less labor intensive.
Perhaps, some day, some of what goes on in hospitals today can be done
online. But that’s a ways off, and until we enter that wired utopia,
the best we can do is go through a process of trial-and-error in
structuring medical shifts.

The two main variables to be adjusted are the total hours worked per
week and the length of a continuous shift. In both cases, there’s a lot
of room to experiment. In 2004, for example, Brigham and Women’s
Hospital, (BWH) where the HMS studies were centered, instituted 12-hour shifts—less than half as long as the common 30 hours stints—and 12 hours off between shifts.

These changes brought BWH more in line with Europe, where junior
doctors don’t work more than 13 hours. The European Working Time
Directive also limits junior docs to a total of 56 work hours per week,
with that number to be reduced to 48 hours by next year.

It’s hard to say where the sweet spot is in terms of resident
scheduling. But the data on current shift guidelines makes one thing
crystal clear: we’re not quite there yet.