05/04/2011 08:29 am ET | Updated Jul 04, 2011

When Medical Residents Don't Get Enough Sleep

As someone with a large supply of frequent flyer miles, I wince along with many others at every story of a somnolent or inattentive air traffic controller. There, but for a timely nap, go I! But I -- and you -- might have cause to wince at the remedy, as well.

The only way to ensure we never again hear of an ATC officer's ill-timed siesta is to take steps that will place one or more well-rested, fully-alert, suitably trained and unfailingly diligent controllers in every tower of every airport, all the time. Airlines pay overflight fees that cover the salaries of ATC, and of course pass the costs along to us. So if we want this security, we will pay for it.

Presumably we would be willing to pay some amount for peace of mind. But there is also a line over which incremental gains in security are too small to justify a deeper reach into our pockets. The question is: where is that line?

A similar issue arises all the time in medicine, where decisions about how best to proceed are often encumbered by trade-offs. The trade-offs are concisely expressed in terms of sensitivity and specificity.

Sensitivity is the reliability with which we find a condition that's there. Specificity is the reliability with which we exclude a condition that isn't. The trouble is: it's hard to have both.

The more faithful we are to sensitivity, the harder it is to be faithful to specificity -- and vice versa. If we never want to miss what's there, we risk getting carried away, and reacting to what isn't. If we never want to react to what isn't there, we risk an inadequate response to what is.

This same trade-off governs the adequacy, and potential excesses, of air traffic control. To ensure that every plane is overseen by a well-rested, fully alert ATC officer, we more or less guarantee ourselves many salaried hours of ATC oversight of empty skies.

For those of us concerned with medical education, a parallel dilemma related to the working hours of medical residents is playing out at lower altitudes. There has long been attention to, and controversy over, the notoriously long shifts of doctors in training. Over the years, increasing restrictions have been imposed by the relevant authority, the Accreditation Council for Graduate Medical Education, or ACGME for short.

Residents are currently restricted to 80 work-hours per week, averaged over any given four-week span. We routinely quip in medicine about the "days of the giants," referring to our own training, when we worked much harder. But what constitutes a giant is relative! I recall my own training with call every third night, and weekly work routinely exceeding 100 hours. But my father, a cardiologist, scoffs. Back in his day, call was routinely every other night, and work weeks of 120 hours were far from rare. In fact, some weeks he simply didn't leave the hospital at all.

There is currently pressure on the ACGME to restrict resident work hours further, resulting from public attention to medical error and its consequences. But this is only a remedy if it fixes the problem, and that is far from clear. We have no reliable data. It may be that sleep-deprived residents are relevant to this concern; it may be they are a handy scapegoat.

And this issue -- like air traffic control and all matters medical -- is subject to trade-offs.

The obvious trade-off is between time and money. Residents are salaried by the work week, not work hours. In fact, I recall another reason for wincing -- when we lined up to receive our weekly paychecks that insultingly stated "40 hours." Enduring the "injury" of 100-hour-work-weeks was bad enough, but the insult of getting paid for only 40 was almost too much to bear! But the fact is that salary is what it is, however many hours residents actually work. So shorter work weeks mean more residents, which in turn mean more resident salaries, and as with the ATC officers -- someone has to pay.

There is an alternative, of course: less resident coverage. But that in turn means fewer doctors to attend to patients in teaching hospitals. This could well take us from the frying pan into the fire. After all, it would take pretty severe sleep deprivation for the physician running a resuscitative effort to be as inadequate as no physician at all!

There is another trade-off as well. Given how much medical care is out-patient, hospitalized patients these days tend to be pretty darn sick. The sicker and more complicated a patient, the more time and involvement with their case is required to really be on top of all the details. The sooner and more often the primary resident involved in the care of such a patient hits the wall of work-day restrictions, the more often he or she needs to "sign out" that patient to a colleague. No matter the diligence, each such passing of the baton risks dropping it in some fashion.

I was not just a resident myself; I was at one time a residency director as well. And in both roles, I acutely felt the trade-off between fatigue and continuity of care. In both roles, I felt there was some irreducible minimum in work hours and patient contact required. Those long work hours kept the physician who knew the patient best on hand for as long as possible. When it was me, and I was exhausted and wanting to go home, I still felt the tug of staying on the scene where my patient needed me. Leaving a scene of crisis because your shift is over is not an easy call.

Those long hours also served as the incubator for the sense of what it meant to be a doctor: that lives were at stake, and you needed to stay the course for the sake of your patients, however painful that might be at times. Personally, I can say that profound sense of responsibility, duty and privilege served me well over the years.

I am not sure about the optimal work hours for medical residents, nor the optimal number and distribution of ATC officers, because I'm not aware of any good data. Studies can and should inform such decisions. Different models could be tested, in both cases, and an evidence-based case built for the optimal trade-off between costs, and benefits. Guessing -- at the risk of guessing wrong in either direction -- is unacceptable when money, mission and lives are potentially at stake. In both of these scenarios, they clearly are.

Whether the control of nocturnal air traffic, or the oversight of residents who race along hospital corridors in wee hours to crises in the ICU, we can -- and should -- do better than fly-by-night decision making.

Dr. David L. Katz;