The Slow Code

The Times ran a story a few days ago that hospitals respond too slowly when patients go into cardiac arrest. Many of those patients have cardiac abnormalities—ventricular fibrillation or pulseless ventricular tachycardia—that are “shockable”; quick defibrillation can restart their hearts. Nonetheless, a recent study found that in 30 percent of cases, it took more than the recommended two minutes from arrest to defibrillation.


Many of the more specific findings are unsurprising. It takes longer on nights and weekends (when units are more thinly staffed), when patients entered the hospital with other illnesses (so the staff are less primed to expect the need for rapid defibrillation), and in small hospitals and ones without heart monitors (greater resources typically help the quality of care). In a telling indictment of our health system, the study also found:

Being black also increased the odds of a delay, but the researchers said this finding probably reflected the quality of hospitals in areas where most blacks live and are treated, rather than a decision by medical workers to drag their feet because of a patient’s race.

(Emphasis added.)

The study, the accompanying editorial, and the media coverage go into some of the policy implications. Perhaps counterintuitively, the automated external defibrillation kits now common in offices and public places are in some ways more sophisticated than the crash carts used by hospitals. Because the AEDs are designed for anyone to use, anyone can use them; hospitals often need to wait for doctors or other trained specialists to arrive. In addition to technological improvements, there’s also room for the kind of process improvements that Atul Gawande has harped on. Best practices from hospitals with quick responses could, and should, be adopted at other hospitals, to institutionalize this comparatively simple but effective procedure.

But the mechanics of quality care are only half the story. Some doctors and nurses don’t want to be faster. Or at least not always.


Meet the “slow code.” Sometimes a patient with severe dementia or a terminal disease goes into cardiac arrest. The doctors think that resuscitation would be futile; an aggressive intervention would provide only a temporary reprieve or more suffering. But the patient or her representative has refused to provide a do-not-resuscitate order. The doctors and staff will call a code, but they’ll respond slowly. A slow code. Or:

Hospital slang like “slow code” (to suggest a leisurely walk to the bedside), “Hollywood code” (in deference to TV hospital programs) or “light blue code” (an allusion to code blue, the term for a cardiac arrest resuscitation) became public.

The ethics are complicated, but dubious. Medicine subscribes to—and often adheres to—the principle that patient autonomy is paramount. This comforting bright line rule, however, produces many uncomfortable cases. The futile resuscitation is one. Rather than confront the problem head-on, the slow code is a bit of subterfuge. Doctors can leave a grieving family with the belief that all appropriate measures were tried. At the same time, they can spare the patient a few last indignities and agonies. And as for themselves, they can say, with a kind of hair-splitting logic, that they neither ignored a patient’s wishes nor contributed to the continuation of suffering.

It’s not a completely absurd position. DNR ethics are hard, contested, and messy. I could believe that sometimes misleading a patient about your plans has justifiable consequences. For some, death may be less psychologically terrible if it comes without the wrenching decision to sign a DNR order. Other patients are no longer even physically capable of making that decision.

Still, this kind of reasoning gets you only so far. In the end, the code blue is a lie. It’s a humanely motivated lie, but the deception is profound. It’s a lie to the patient: you say you’ll do everything you can to bring them back, and then you let them die. And it’s a lie to yourself. You go through the motions, you really do provide defibrillation, but you pull your punches and pretend that something other than your own willful act is responsible for the predictable consequences. It sits ill with our image of doctors and other medical professionals to think that there are situations where they’ll neither do what we ask them to do nor take a stance for what they actually think is right. The slow code lacks the courage of its convictions, and that essential cowardice may be more telling than any sophisticated moral reasoning about means and ends.


When I heard about slow codes, they struck me as a terrible idea for another, perhaps less idealistic reason. It’s hard for me to imagine a medical practice that would sound worse to a jury.

There’s a whole history here, one that’s in large part a story of growing patient autonomy. Slow codes have developed in parallel with state laws giving patients and families greater rights to determine end-of-life care: particularly, to sign or to withhold DNR orders. Previously, doctors would simply enter the orders themselves. No code, not slow code. The slow code is a physician’s response, a way of getting around the law.

But “getting around the law” is a dangerous business. Especially on such a fraught subject, subverting the clear intent of a patient-autonomy rule looks and smells awful. I’m as far as you can get from a health law expert, but I suspect the potential liability involved is genuinely awful. Not only is a slow code something of a deliberate decision, it’s a decision on which the legislature has specifically said that the physician’s medical judgment is irrelevant. I would not like to be a doctor sued for malpractice over a slow code.

True, there are proof problems. Slow codes are the sort of thing it sounds like medical professionals will talk about amongst themselves, and on anonymous surveys, but never with outsiders. Call it the blue-scrubs wall of silence. And indeed the point of going through the motions in almost good faith is to put on a genuinely convincing show, one that hopefully produces no untoward evidence.

Still, I wonder whether the very mystery of the slow code is not in fact a looming litigation disaster for the profession. A sharp plaintiff’s attorney could invoke the specter of the slow code in almost any end-of-life-care suit. Whether there was a slow code or not, the mere possibility that there was one puts defendant doctors in a poor position from the get-go.

It strikes me a perfect deposition subject. Establish that the defendant has heard of slow codes. Establish that he’s heard that they might be carried out in cases that share some of the characteristics of the one at bar. Establish the the defendant counseled the family that resuscitation in this case would be futile. Congratulations. You’ve now put defense counsel on notice that at trial you can present a thin but plausible argument that your client died as the result of a deliberate slow code. Of course, you’ll need to question the defendant on this theory, which means that the defendant’s knowledge of slow codes will be a relevant subject for cross-examination. That can’t possibly look good to the jury. You’ve just dealt yourself a few excellent cards for the forthcoming settlement negotiations.

I always love getting comments, but I’d particularly welcome informed thoughts about any of the three angles to this post.

I’m not certain that patient autonomy is necessarily the crucial issue here: certainly in a case where the patient had refused to sign a DNR, or had asked to be resuscitated no matter what it would be. But that isn’t generally the case, and from what I know of it (ie, only so much) the laws generally require a code when someone’s heart stops unless they’ve specifically requested otherwise. So there are a rather large number of cases where no specific DNR has been signed (for any reason) and in which running a code really would be futile - the patient would die the next day if resuscitated, and would be unconscious/suffering extensively for that extra day, say. And in these cases there’s a very good case to be made against running a code and very little to be said in favor of it.

It’s also worth noting that while patient autonomy is important in medicine it’s far from an overriding principle. And futile medical treatments are often cases in which the principle has to be very carefully weighed against other factors. Ethically, at least, this strikes me as more similar to cases of euthanasia of terminal patients by doctors (of the slightly less terrifying variety, though - the ‘injected the morphine a little too fast’/’gave the patient enough to ease their suffering cough-cough’ sort). That is, it’s the sort of compromise you tend to see when legal requirements or general ethical principles simply give what is evidently the wrong answer for a particular case and so some sort of workaround has to be developed. The ethics might be dubious in certain cases, but I think this is more plausibly the sort of thing that results from a poorly designed ethical system to begin with (see also: doctrine of double effect).

I agree with James Grimmelham fully. Many physicians are afraid to talk about death and dying and by extension about the realities of DNR. Many physicians like to have the decision taken out of their hands. Do you have any thoughts regarding unilateral overriding of patient’s demand for “futile” treatments?