DNR vs DNI

Every hospital patient has a code status, whether anyone has discussed it or not. If nothing is written, the status is full code. The National Institute on Aging puts it plainly: without a DNR order, medical staff "will attempt every effort to restore your breathing and the normal rhythm of your heart." That means chest compressions, electric shocks, IV drugs like epinephrine, and a breathing tube, for as long as the team thinks there is a chance.

Full code is not a choice someone made. It is the absence of a choice. That matters because the whole vocabulary of DNR and DNI is a set of subtractions from full code, and you cannot understand a subtraction without knowing the starting number.

The figure below is the version of this guide you can hold in your head. Each column is a code status, plus a comfort-measures-only order for contrast. Each row is a treatment. A filled dot means the treatment is still given. A dash means it is withheld. A dotted circle means it depends on the goals written in the plan: a breathing tube under a DNR alone, for instance. The column that needs that answer most is comfort measures only, which redirects the whole plan rather than one moment.

A DNR order, sometimes written DNAR (do not attempt resuscitation) or AND (allow natural death), tells the team not to attempt CPR if the heart stops or breathing stops. The American Medical Association's Code of Medical Ethics is specific about the boundary: DNAR orders "apply only to resuscitative interventions." Other treatment, it says, "such as antibiotics, dialysis, or appropriate symptom management will be provided or withheld in accordance with the patient's wishes."

So the moment a DNR applies is narrow. It is the moment the monitor goes flat or the breathing stops. Hospitals differ on whether a DNR alone covers the case where breathing stops but the heart has not, so if the order is DNR without DNI, ask that question directly. Up to that second, a person with a DNR can be in the ICU, on antibiotics for pneumonia, recovering from a hip operation, or getting chemotherapy. The order says nothing about any of that.

The name is the problem. "Resuscitate" is a word almost nobody uses outside a hospital, and in a hallway it sounds like "revive" or "treat." It means none of those. It means CPR. If the phrase "do not resuscitate" ever makes you picture a team walking away from a bed, replace it in your head with "no CPR," which is what the order actually says.

Some states write the comfort floor into law. Texas, for example, says an out-of-hospital DNR does not authorize withholding anything "considered necessary to provide comfort care or to alleviate pain or to provide water or nutrition." A DNR is not a reason for anyone to be left uncomfortable.

A DNI order says no breathing tube and no mechanical ventilator. The tube goes down the throat into the windpipe; the ventilator breathes for the person. Placing it usually requires sedation, and while the tube is in, the person cannot talk or eat by mouth.

One fact changes the whole decision. A breathing tube is almost never placed because the heart stopped. In the paper that hospital medicine still cites on this, Breu and Herzig (Journal of Hospital Medicine, 2014) note that cardiac arrest accounts for less than 2 percent of the reasons people are put on a ventilator. The usual reason is breathing failure from something treatable: pneumonia, a COPD flare, fluid on the lungs from heart failure. And the odds are different. In that paper's figures, mortality after in-hospital cardiac arrest is over 75 percent, while mortality for people ventilated for breathing failure alone is under 40 percent.

That is why bundling the two orders costs people something real. In a study the paper cites (Jesus and colleagues, Mayo Clinic Proceedings, 2013), 28 percent of hospitalized patients who already had a combined DNR/DNI order said they would in fact accept a trial of a ventilator for pneumonia. They had refused a treatment they wanted, because nobody separated the two questions.

The phrase to know is "time-limited trial." It means: put the tube in for a defined window, say three to seven days, with a plan to remove it if the lungs have not recovered. It is a middle option between "never" and "indefinitely," and it is what many people mean when they say they do not want to live on a machine. Ask for it by name.

One more thing a DNI does not rule out: a mask. BiPAP and CPAP push air through a tight face mask without a tube, and people with DNI orders get them routinely. A 2018 meta-analysis in Critical Care Medicine pooled 27 studies of 2,020 people with DNI orders treated with mask ventilation: 56 percent survived to leave the hospital, and survival was 68 percent for COPD flares and heart-failure fluid, 41 percent for pneumonia, and 37 percent for people with cancer (Wilson and colleagues). A DNI is not a decision to stop breathing support. It is a decision about one kind of it.

Because the two orders answer different questions, there are three real combinations. DNR and DNI together is the most common line on a hospital chart. DNR alone means no CPR after an arrest, but a breathing tube before one is still an option if it fits the goals. DNI alone means no tube, ever, but CPR would still be attempted.

DNI alone is the odd one, and some clinicians will tell you it does not make sense. Their point is that real CPR usually includes a breathing tube, so a DNI without a DNR asks for a version of CPR with a hand tied. The counterpoint is that a person can reasonably refuse a long stretch on a ventilator while still wanting a brief attempt if their heart stops suddenly. Both positions are defensible. What matters for you is that if someone proposes DNI alone, the team should say out loud what CPR would and would not include, so nobody is surprised.

New York documents the two as separate orders on its MOLST form. If a form in front of you has one checkbox for both, ask for them to be discussed one at a time anyway.

Charts sometimes call these mixes a "partial code" or "limited code," for example shocks and drugs but no compressions, or compressions but no tube. Most guidance discourages them, because a half-done resuscitation rarely works and the team is left guessing mid-code. If one is proposed, get the exact list of what will and will not be done written into the order.

On paper a DNR changes one thing. In practice, it changes more, and this is the part of the decision the forms never mention.

The 2020 survey of 268 clinicians at an academic hospital (Stevenson and colleagues, Journal of General Internal Medicine) gave them scenarios of a patient who was sick but had not arrested, and asked what they would do. When the patient had a DNR, clinicians were far less likely to offer treatments the order does not cover: the odds of withholding a BiPAP mask rose more than fourteenfold, intubation for breathing failure more than tenfold, and blood-pressure medication about sixfold. Nurses and surgical specialties were the most likely to pull back.

It shows up in outcomes, not just surveys. A 2011 study in the Archives of Surgery (Kazaure and colleagues) matched 8,256 surgical patients with DNR orders to similar patients without them. Complication rates were only modestly different. Death within 30 days was 23.1 percent with a DNR versus 8.4 percent without, and the adjusted odds of dying were about 2.2 times higher. The authors' reading, and the field's, is that when a complication happened, DNR patients were less likely to be rescued from it. This pattern has a name in the literature, "failure to rescue," and it is the cost of a word that sounds like "do not treat."

There is a fix, and it is a sentence. When a DNR is written, ask the team to write the rest of the plan next to it in words: "DNR. Full treatment otherwise," or "DNR/DNI, continue antibiotics, ICU transfer acceptable, BiPAP acceptable." The AMA guidance is that other interventions are provided "in accordance with the patient's wishes," which only works if those wishes are written down somewhere a night nurse can see them. A POLST form, covered in step 7, has a box for exactly this. A hospital chart often does not, so ask.

People decide about CPR based on what they have seen, and what they have seen is television. A study in the New England Journal of Medicine counted every CPR scene in episodes of ER, Chicago Hope, and Rescue 911: 75 percent of the patients survived the arrest, and 67 percent appeared to leave the hospital alive (Diem, Lantos, and Tulsky, 1996). Real numbers are not close to that, and they drop with age and illness.

For an adult whose heart stops in a US hospital, where a team is seconds away, survival to discharge is under one in four in recent years of the American Heart Association's Get With The Guidelines registry. Outside a hospital, it is far worse: 9.3 percent of people treated by paramedics for a cardiac arrest in 2022 survived to leave the hospital (AHA Heart Disease and Stroke Statistics, 2024).

The numbers fall with age and illness. For Medicare patients 65 and older who had CPR in the hospital, a study of 433,985 patients found 18.3 percent survived to discharge (Ehlenbach and colleagues, New England Journal of Medicine, 2009). For people with cancer, a meta-analysis of 42 studies found survival to discharge of 6.2 percent overall, 5.6 percent with metastatic disease, and 2.2 percent when the arrest happened in the ICU (Reisfield and colleagues, Resuscitation, 2006).

Survival also is not the only outcome. Effective compressions break bones. In an autopsy series of 2,148 people who had CPR, rib fractures were found in 77 percent of men and 85 percent of women, and sternum fractures in 59 and 79 percent (Kralj and colleagues, Resuscitation, 2015). Most survivors of a brief arrest recover well neurologically. Survivors of a long one often do not.

The gap between belief and reality is measurable. When 135 newly admitted hospital patients were asked to guess their own odds of surviving in-hospital CPR to discharge, the average guess was 60.4 percent (Kaldjian and colleagues, Journal of Medical Ethics, 2009). The real figure for Medicare patients is 18.3 percent. Most people making this decision are working from a number three times too high.

None of this argues for a DNR. For a healthy 50-year-old whose heart stops from a treatable rhythm, CPR is usually the right call and the odds are far better than the averages above. The numbers argue for deciding with the real odds for this person, with this illness, instead of with the television ones.

A DNR is a medical order. A physician writes it, and in a number of states a nurse practitioner or physician assistant can too; New York's nonhospital DNR form has a signature line for all three. The patient consents if they can. If they cannot, the AMA guidance is that the physician takes the decision "from the surrogate when the patient lacks capacity." Who the surrogate is depends on whether the patient named one, and otherwise on state law, which sets a default order of relatives, commonly starting with the spouse, and differs from state to state. When the family disagrees and nobody was named, the hospital's ethics committee is the usual next step, and it is slow. Naming a proxy in advance is the single thing that prevents that. (This guide is about adults; decisions for children run through parents and a different set of rules.)

This is where the other documents fit, and it helps to sort them by what kind of thing each one is. A living will is instructions: what the person wants if they cannot speak. A healthcare proxy (also called a medical power of attorney or healthcare agent) is a person: who speaks for them. Our advance directive vs living will vs medical power of attorney guide covers when each of those switches on and which one wins when they disagree. A DNR is an order: what the team will do, right now, in this facility. A POLST is also an order, a portable one, for people who are seriously ill.

The distinction that catches families is that instructions and a person are not an order. National POLST says it directly: advance directives, living wills, and healthcare powers of attorney "are not orders that EMS providers can follow." A living will that refuses CPR is valuable, but the NIA advises that "it is helpful to have a DNR order as part of your medical file if you go to a hospital," because the living will on its own does not stop a code from starting.

POLST (called MOLST, POST, or MOST in some states) is the bridge. It is a signed medical order that travels with the person, covers more than CPR (it has sections for intubation, hospital transfer, and artificial feeding), and is meant for people who are seriously ill or frail, not for healthy adults doing general planning. In most states the provider and the patient or surrogate both sign it. If the person you are caring for is sick enough that a DNR is being discussed, they are usually sick enough for a POLST, and a POLST is the document that writes down "no CPR, but full treatment otherwise" in a form a paramedic will honor.

About a third of American adults have any advance directive at all (Yadav and colleagues, Health Affairs, 2017, pooling 150 studies). If your family has a living will and a proxy, you are ahead of most. If you also have a POLST or out-of-hospital DNR on the refrigerator, you are ahead of almost everyone.

A DNR written in a hospital is an order in that hospital's chart. It generally does not follow the person home, to a nursing home, or into an ambulance. This surprises nearly every family, and it is the reason people with a hospital DNR get CPR in their own kitchen.

What paramedics can honor is a state-recognized out-of-hospital DNR form, a POLST, or in some states a DNR bracelet or necklace. National POLST describes the default when none of these is present: "The current standard of care during an emergency is to do everything possible in an attempt to save someone's life." EMS crews will start CPR unless they can see a valid order. A living will in a drawer does not count, and neither does a daughter on the phone saying "he has a DNR at Memorial."

So the practical checklist is short. Ask the discharging team whether the DNR is being converted into an out-of-hospital form or a POLST before the person leaves. A POLST is also the form a nursing home honors. Put the signed original where a responder will look: the refrigerator door is the convention, and New York's guidance is to "store it where emergency responders can easily find it." Carry a copy, or a bracelet where the state issues one. And tell every person who might be in the house when something happens, including the home health aide, where the paper is.

It fails in the other direction too. A study of advanced cancer patients found that admitting clinicians did not address code status in 53 percent of hospitalizations, which defaults to full code, and that about half of the patients later switched to DNR during the stay had wanted a DNR before they were admitted (El-Jawahri and colleagues, Cancer, 2017). An order from the last admission is not automatically carried into this one. At every admission, say it out loud on day one and ask to see it written.

A DNR can be revoked at any time, by the patient, for any reason. Texas law says it in one line: a person may revoke "at any time without regard to the declarant's mental state or competency," by destroying the form, removing the bracelet, or simply telling the responding crew or physician. Other states are similar in substance. To revoke in a hospital, tell the attending physician, who cancels the order in the chart; at home, destroy the form and let the people who knew about it know.

A surrogate's authority is narrower. A proxy decides when the patient cannot, and generally decides as the patient would have. A family member who disagrees with a DNR the patient signed while competent cannot simply undo it, and a hospital will usually involve ethics or legal staff if that is attempted.

Surgery is its own case. Anesthesia routinely does things that look like resuscitation, so for years many hospitals automatically suspended DNR orders in the operating room. The American Society of Anesthesiologists' guidelines, reaffirmed in 2023, say that automatic suspension "may not sufficiently address a patient's rights" and lay out three options to choose before the procedure: fully suspend the DNR during anesthesia and recovery, keep refusing specific procedures (for example chest compressions, but allow a breathing tube for the operation), or have the anesthesiologist use judgment guided by the patient's goals.

The AMA says the same: confirm before any procedure with arrest risk whether the DNR stays in effect. If someone with a DNR is heading to surgery, ask which of the three it will be, and get the answer written down.

Hospice does not require a DNR. Federal regulation (42 CFR 489.102) forbids hospices, hospitals, and home health agencies from conditioning care "on whether or not the individual has executed an advance directive." Most hospices will raise the subject, and it is usually a sensible conversation, because CPR rarely fits the goals of hospice. But "you have to sign a DNR to get hospice" is not true, and a hospice that says so is wrong. The wider comparison is in our guide on palliative care vs hospice. What hospice does require is covered in how to qualify for hospice.

The decision usually gets made badly because it gets made fast, by a resident, at admission, in the form of "if your heart stopped, would you want us to do everything?" Almost nobody says no to "everything." Here is a better version you can steer toward.

Ask the team to separate the questions: "Can we talk about CPR and the breathing tube as two different decisions?" Ask for the odds for this person: "For someone with her illness and age, what are the realistic chances CPR gets her home?" Ask about the middle option: "If she needed a ventilator for pneumonia, could we do a time-limited trial?" And close the gap from step 5 explicitly: "If we choose a DNR, can you write next to it that everything else continues, including ICU and antibiotics?" If the decision is being made overnight, ask whether the order is in the chart yet; a nurse can act on a written order, and a conversation that was never written down is a full code.

Then have three things ready before any of it is needed. The names: who the healthcare proxy is, in writing, and a backup. The paper: the living will, the proxy form, and, once it exists, the POLST or out-of-hospital DNR, with the original on the refrigerator and copies with the proxy. And the record of what was decided and when, kept somewhere the whole family can reach, because the order will be questioned at the next admission, by the next team, and the person who knows the answer may not be the one who picks up the phone.

If you want a printable list of questions for the broader meeting, our questions to ask your doctor guide covers how to ask about prognosis without asking for a countdown, and the questions to ask a hospice checklist covers what to ask when that conversation comes.