I sat recently in the conference room attached to my hospital’s intensive care unit, thinking about ethics and the end of life. Our ICU was remodeled last year, and the conference room has a clean, corporate feel; the wood intentionally unfinished, the chairs black nylon with hinges optimized for back support. My fellow medical residents—many of whom, having worked the night shift, were intermittently falling asleep—flanked me at the table. We wore blue scrubs.

With our help, the physician in charge compiled a list of the ethical precepts that guide end-of-life decision-making (in the US, in 2006) on the white erase board. There were four:

1. Autonomy
2. Non-malfeasance
3. Beneficence
4. Social justice

The order of the terms implied hierarchy. Autonomy, in fact, was written larger. Extra white space separated it from the other precepts. Ironically—since end-of-life decisions are often made by distraught relatives via a slippery concept called substituted judgment, whereby they divine what their loved one “would have wanted” were he or she of sound mind—self-determination rules the ICU.

Non-malfeasance, a fancy term for the promise we made in the Hippocratic oath to do no harm, came in a distant second. As with autonomy, which instructs us to defer to a patient’s wishes even when the patient couldn’t possibly have any, non-malfeasance, from the perspective of the physician as moral actor, is passive, avoidant—a cynic might have sniffed fear of litigation. It wasn’t until number three that we actually got to try to do some good.

I was paying close attention, and not only because I’d been able to sleep in my own bed the night before. This was a welcome departure from our usual stay-on-point talks about drugs, intravenous lines, and breathing machines. ICU doctors often find themselves delivering fantastically expensive and brutally invasive care to patients with little chance of survival. Such practices are based on complex assumptions about our society and what it values—but for doctors, the time to step back and think about those assumptions is a rare luxury.

“Is social justice ever invoked in ICU decisions?” I asked. “And if it is, could you give an example?”

“Um, not really,” replied my superior indifferently. He equivocated a bit, and the discussion was over.

While we learned, twenty patients under our care lay in solitary, glass-doored rooms, hooked up to breathing tubes, feeding tubes, hydrating tubes, urinating and defecating tubes, and virtually every antibiotic on God’s green earth. Yet many of them, inexorably, would succumb within days or weeks.

The problem with autonomy and end-of-life decisions is largely a problem of information—its transmission and, more fundamentally, its adequacy.

There is information asymmetry: doctors train for years to prepare to make decisions in the ICU, and the knowledge they acquire is scientifically complicated and shrouded in jargon. It takes a talented communicator to keep anxious, suffering patients up to speed, especially while alarms blare unremitting warnings of high or low vital signs. With notable exceptions, though, physicians are not known for their interpersonal skills. And while the emphasis placed on autonomy seems to demand that doctors play the role of disinterested newsbearers, in practice they more often attempt, consciously or otherwise, to nudge patients toward the decisions that they believe are correct—this is termed “framing bias” in decision-science literature.

More troublingly, many patients are incapacitated and unable to make decisions at all. One of the hardest things to get used to as a new ICU doctor is examining people who, at best, are sedated beyond a hint of purposeful movement. (At worst, patients are moaning in pain, unable to communicate but for barely distinguishable lip or eye movements that, no matter what the patient intends, always seem to mean help me.)

In an effort to obviate the problem of incapacity, we obtain “advanced directives,” whereby we ask patients well before an emergency what they want done should one arise. These directives are a major component of federal end-of-life policy—the Patient Self-Determination Act of 1990 requires hospitals and other medical institutions to inform patients about their state’s options for advanced directives and living wills. Unfortunately, as argued in a 2004 Hastings Center Report, the 100 million– plus dollars spent to comply with the PSDA have had little effect. Many patients balk at filling out a menu of hypothetical options for artificial resuscitation: If your heart were to stop, would you want chest compressions? With shocks or without? Even when patients do file directives, relatives and physicians find it very difficult to interpret them, and there is rarely meaningful influence on the care received. Every medical resident can tell stories of patients who were DNR/DNI—do not resuscitate or intubate—but wound up receiving full, aggressive care.

When a patient is incapable of deciding—because she is unconscious, because she has a breathing tube in, because she’s oblivious to person, place, and time—then a family member (provided one can be found) is called to stand in. But relatives are often no better than doctors at knowing what the patient would have wanted. And their decisions are tied inextricably to their own interests, whether an understandable wish to avoid the burden of “giving up on” their loved one, or messier concerns related to finances or family dynamics.

So the concept of autonomous, informed, rational decisions at the end of life is porous at best. Beyond this relative information gap, furthermore, lies a vast and growing information deficit. Our technologic capabilities—glorious as they are—have far outstripped our understanding of how and when they will work. Well-conducted studies of medical interventions test a medically homogeneous population by manipulating one aspect of their care. ICU patients are so sick, in such diverse and unique combinations of ways, that there is shockingly little sound information on whether or how much the interventions we can offer will help. And so a day seldom passes on rounds without us standing around scratching our heads about what to do next. Dialysis? Could work. Twelve-thousand-dollar-a-day drug that is “effective,” at least temporarily, in 10 percent of cases? Let’s give it a shot.

Put yourself in the position of an ICU doctor. You cannot tell a patient and his family with 100 percent certainty that all hope is gone—yet you’re damn sure that even the best-case scenario does not involve a comfortable, independent existence. You have an irrational fear of litigation and a more rational fear of withholding care in one of the rare cases when last-ditch efforts grant a patient extra months or years of quality life. You have little incentive to limit the interventions that you offer (scarce resources? whatever). In many settings, you are actually paid more money for performing more procedures. So what do you do?

In the name of preserving autonomy and choice, our system leads to little choice at all. Technology takes over. We simply do everything.

There is a real toll taken—on health-care workers, families, and patients alike—by persisting with highly aggressive medical care when meaningful recovery is impossible. The procedures that make up the daily ICU routine are exceedingly invasive and uncomfortable. Tubes are inserted through noses into stomachs and down throats into airways. Big needles are stuck into necks, chests, or thighs so that large IVs can be placed. The slang term we use for performing these procedures on people unlikely to benefit from them, “flogging,” is actually rather precise. Patients require continuous, heavy sedation to tolerate their ICU stays. They often become delirious and hysterical as a consequence.

Nurses describe taking hour-long showers to cleanse themselves of the metaphysical dirt that accumulates from providing invasive care in hopeless situations. But their hands will not come clean. I know the feeling.

My own most damning experience involved caring for a man in his sixties who was mentally disabled from a congenital disorder. The usual life expectancy for someone with this disorder is roughly forty years. This fellow, having experienced unusual longevity, had become both blind and deaf. Unfortunately, he grew critically ill after an arthritis-related surgery—his heart, like his eyes and ears, had ceased to function optimally, and he just didn’t tolerate the procedure. With no family around to allow him to pass naturally, as his body so clearly indicated he was ready to do, we aggressively resuscitated him. This meant, among other things, that he needed a feeding tube.

A feeding tube, even for a person who knows why he needs it, is one of the more unpleasant procedures people undergo in the name of a healthful recovery. I have grown accustomed to a particular look of betrayal in my patients’ eyes while I am placing a tube, having warned them beforehand that it might be “a little uncomfortable.” In this case, though, there was no way to communicate why the tube would help. It wasn’t even possible to give him fair warning before we began.

And so with the help of a colleague, I shoved the tube through his nose, down the back of his throat, and into his stomach while he thrashed wildly. It was terrible the first time we did it. Then, and amazingly, considering that his arms were tied to the bed and an attendant was constantly in the room, he managed to get the tube back out, requiring its replacement. This happened on three consecutive days. It felt as though I was torturing him. The dreams I had reflected this. So did the news, with reports of feeding tubes used as torture tools coming from Guantánamo Bay.

Despite all our efforts, and the pain we inflicted, the patient died a few weeks later. I felt angry and at fault. What could have been done differently? We might have worked harder to find a family member who might have allowed us to adjust the goals of his care—but when, amid the constant chaos of the ICU? And what if, long-distance from New Hampshire or California or wherever—far from the feeding tubes and needles—the indifferent relative we spent hours tracking down told us to forge ahead anyway?

And when no family exists? Who decides then? There’s no sound protocol to cover these cases in real time. Without patients’ wishes or certain prognoses to serve as guides, doctors are frequently trapped in a messy middle ground, where the path of least resistance for us is the path of most resistance for the patient. Keeping people alive at all costs becomes the protocol.

While our failure to make difficult choices on an individual level has painful consequences, on a collective level it amounts to a crisis.

The many failings of our health system have become cliché. We spend much more on health care than any other country—15 percent of our GDP in 2003, compared with an average of just under 9 percent for all developed countries. Yet our population is neither particularly healthy nor happy with its health care by comparison. Our major employers increasingly cut into the health-care coverage offered to employees because of cost, and the number of us who lack basic health insurance grows each year.

The reason that we are not as healthy as people in other countries, despite our excessive spending, is that overall population health is largely determined by factors other than health care. Modern medical interventions do prolong lives, but at such a high marginal cost that this expenditure does not affect average life expectancy nearly as much as the physical environment in which we live and work (clear air, clean water, safe streets, low stress); personal behaviors (Whoppers, Marlboros, trips to the gym); and genes.

So what do our extra dollars buy us? The short answer is: more. More tests, more procedures, more administrative costs, more technology. A few more machine-dependent days, months, or sometimes years at the end of life.

Prominent US health economists have argued that we have more because we want more. Who’s to say how much a country should be spending on its health care, anyway? Americans demand higher technology and more aggressive care, and are willing to bear the extra costs.

But economic theory itself predicts the endgame played out in our ICUs. While insurance in this country for cheap, effective preventative services is patchy at best, the coverage for high-cost, acute medical care is pretty close to a blanket. Hospitals are ethically and legally obliged to provide for patients who are acutely ill, regardless of insurance—care for the uninsured is paid out of a pool of funds from the insured, reimbursed at a higher rate and often bolstered by state funds. By the time a person is sick enough to wind up in the ICU, someone else is always footing the bill.

Our dependence on a poorly defined notion of autonomy leads us to treat patients as consumers; but in health care there is no functioning market to regulate their choices. Modern medical innovations have created an environment in which even seconds of life can have a considerable financial cost, but the cost at the end of life is always borne by others, and so frightened patients and rattled families have little incentive not to forge ahead, trying treatment after hopeless treatment. At present, these costly, compromised seconds are considered the right of every American. Meanwhile the modestly priced interventions that could add productive, pain-free years to millions of lives—for instance, better management of a chronic, epidemic disease like diabetes—are not.

The best study on the kinds of deaths people want, conducted by researchers from Duke and the University of Chicago, suggests that aggressive ICU care does not play a major role. In a large survey of patients living with chronic illnesses, families of the recently deceased, doctors, and other health-care workers, more than 90 percent of respondents agreed on the importance of such attributes as being kept clean, maintaining their dignity, trusting their physician, and being free of pain. Eighty-eight percent valued being able to help others, and 81 percent not being a burden to society. But only 48 percent of patients felt it was important to “use all available treatments, no matter what the chance of recovery.” Physicians and health-care workers were much more emphatic; only 7 percent and 5 percent, respectively, wanted all available treatments for themselves—testimony, perhaps, to the information asymmetry that exists between doctors and patients.

And so many patients end up dying deaths they do not want, deaths their caretakers certainly would not want for themselves. When families are involved, they agonize over decisions that often merely drag out the dying process by excruciating days or weeks. The information failures that plague health-care transactions throughout life only increase in magnitude and intensity as we approach the end.

A week out from my rotation in the ICU, I received an e-mail urging me to complete a dictation on another patient who had died during my tenure. He was a man with a painful terminal cancer whose death had been forestalled by “palliative” chemotherapy and aggressive ICU care.

The palliation was anything but successful, however; no amount of narcotics could dull his constant pain, while his life was artificially prolonged by a breathing tube, a feeding tube, kidney dialysis, three powerful antibiotics (vancomycin, meropenem, flagyl) and an antifungal (voriconazole), all delivered through a large IV that had been placed, at the risk of puncturing a lung, just under his collarbone. Each day, the extensive battery of blood tests, radiology tests, and scopes inserted into his airways and stomach documented his steady decline. The dictation needed to be done immediately, I was informed, so that the hospital could be paid the more than $250,000 it was owed.

I suppose it is only fitting that we should die as we live—surrounded by technology, consuming vast amounts of resources, too often alone—but that makes it no less distressing. There is such a thing as too much. But how will we say: “Enough”?