One clinician decides not to offer CPR to a patient due to an extremely poor prognosis. Another clinician caring for the exact same patient might see things differently, instead asking the family to decide for themselves whether they want CPR offered.

“That could potentially result in two very different outcomes for a patient and their family,” says Ann (Long) Jennerich, MD, MS, an assistant professor of medicine in the division of pulmonary, critical care, and sleep medicine at the University of Washington. “We know there is a lot of variability in end-of-life care throughout the world. There is debate about what drives this variability.”

It is unclear whether the variability is because of differences among clinicians, hospital culture, or the region where they practice. “We thought it would be interesting to use an international sample of clinicians to explore the primary contributor to variability related to agreement with consensus statements about end-of-life care,” Jennerich reports.

Researchers examined views of 1,068 providers from 178 hospitals and 31 countries.1 Of declining to offer CPR when it was not clinically indicated: 1% strongly disagreed, 7% disagreed, 11% were neutral, 44% agreed, and 36% strongly agreed. There were similar mixed results regarding withdrawing or withholding life-sustaining interventions that were believed inappropriate. “We were surprised at the degree to which variability was explained at the level of the clinician,” Jennerich says.

The authors expected most variability would exist at the clinician level. They did not expect so little variability would be explained at the level of the hospital or country. “Variability in and of itself is not a bad thing,” Jennerich offers. “Variability in end-of-life care is acceptable — when the source of variability is the patient.”

A patient’s views about end-of-life care might be different than those of their loved ones, and vice versa. The problem arises when these views differ depending on the provider and the particular values the provider espouses.

“The care someone receives at the end of their life should not differ dramatically based on who the attending physician is on any given day,” Jennerich says.

There are few clues to help ethicists know when the variability is occurring. “It’s not as if someone makes an announcement when their approach to end-of-life care differs from their colleague’s approach,” Jennerich says.

Efforts to prevent variability need to occur “upstream, through conversation and development of a common understanding related to end-of-life care,” Jennerich adds.

“For the most part, consensus statements emphasize giving patients accurate information, a voice, and allowing patients to make decisions that reflect the patient’s own values and priorities,” says Rosamond Rhodes, PhD, director of bioethics education at Icahn School of Medicine at Mount Sinai in New York City.

Imagine a patient who is dying of a disease that might include a genetic component. Perhaps this patient is a potential participant in an investigation of a new drug. Despite facing possible unpleasant side effects, all for no benefit, this theoretical patient might want to try the new medication anyway because it could help his or her children or grandchildren.

Even if the new medication would not benefit the theoretical patient or loved ones, he or she still may carry a strong desire to contribute to society. In other words, this patient may be willing to do whatever it takes to prolong his or her life. But the provider may wish to spare the patient enduring unpleasant medication side effects for no clinical benefit.

The Australian Commission on Safety and Quality in Health Care National Consensus Statement maintains the importance of not harming patients at the end of life by providing burdensome investigations of no benefit.2 “The traditional medical thinking is to try things on people with the least to lose and the most to gain,” Rhodes says.

The Australian recommendations state that “unless required by law, doctors are not obligated to initiate or continue treatment that will not offer a reasonable hope of benefit.” A physician might take the view that a 1% chance of survival is not reasonable, but patients might strongly disagree. “If we want to respect autonomy and give patients the role of decision-maker, then patients are the ones who have to decide what is a ‘reasonable’ hope of benefit,” says Rhodes. Further, while some may use “intervention” and “treatment” interchangeably, Rhodes argues that if an intervention offers no benefit, it should not be considered a treatment.

Most interventions involve some risks and burdens. “Such harms can only be justified by the counterbalancing benefits. When there are no benefits, the infliction of harms cannot be justified,” Rhodes says. “Also, research risks are justified by potential benefits for others.”

Typically, recommendations emphasize palliative care, prioritize avoiding pain, and promote comfort. That might not be what the patient prioritizes. “Such policies often prioritize avoiding anything that might be construed as burdens. But in our lives, we sometimes accept burdens for the sake of something else,” Rhodes notes.

Some patients will accept pain and discomfort to maintain consciousness and the ability to communicate with people. Those choices belong to patients, and their wishes should be honored, Rhodes argues.

Certain physicians or institutions might reject parts of the guidance offered in such documents. “[Physicians] might refuse to honor patient decisions to refuse life-prolonging care because [physicians] regard withholding such interventions as a violation of their own personal, moral commitments,” Rhodes explains.

For example, patients may refuse a nasogastric tube. Physicians might persist in providing the tube anyway over the patient’s objection because those clinicians believe they are killing the patient by not prolonging life. “That is the patient’s choice,” Rhodes says.

Other clinicians might not honor patients’ advance directives when those choices are not consistent with the clinician’s own values. “We have to be very careful about imposing medical staff priorities over patients’ priorities. It’s dangerous to make presumptions about what is in the patients’ interest,” Rhodes cautions.

There might be physicians who object to this part of the Australian statement, which calls on clinicians to “provide timely and accurate information regarding the patient’s clinical condition and its severity or stage, the expected disease trajectory, the available treatments, and the likelihood of response to such treatment.” Clinicians may feel obligated to mislead patients by giving them false hope. “The commitment to truth-telling, to recognize when the patient is at the end of their life and being truthful about that is a critical medical responsibility. There are a lot of very well-meaning doctors who don’t live by that commitment,” Rhodes says.

Physicians may make a well-meaning statement such as, “You’ll go to rehab. When you can walk again, you’ll come back to us, and we’ll start another round of chemotherapy.” But in reality, the clinician is quite sure the patient is dying. “Medicine is entirely based on trust of patients and society. That trust requires truthful reporting,” Rhodes stresses.

Certain physicians might think misleading a patient about prognosis actually serves the interest of the patient. “But that does not justify the deception,” Rhodes argues.

Whatever the problem, ethicists can guide these conversations and decisions by explaining the provider’s responsibility to offer accurate information. Help clinicians identify the right words and how to use them. Even after this education, some medical professionals are just not prepared for this conversation. Ethicists can help identify somebody else on the team who can share the conversational burden. For instance, someone from the palliative care team can describe the prognosis while the physician is present and nods in agreement. “Even if the doctor isn’t the one who says the words, and is just witnessing, it says ‘We’re all on the same team,’” Rhodes adds.


  1. Long AC, Brumback LC, Curtis JR, et al. Agreement with consensus statements on end-of-life care: A description of variability at the level of the provider, hospital, and country. Crit Care Med 2019;47:1396-1401.
  2. Australian Commission on Safety and Quality in Health Care. National consensus statement: Essential elements for safe and high-quality end-of-life care. 2015.