End of Life, Brain Death and Organ Donation

In depth

The one-page summary above is what the PDF prints. These sections go further.

Family outcomes and the communication evidence

At 90 days, 33.1% of 284 relatives had post-traumatic stress symptoms, more often after sharing end-of-life decisions (Azoulay 2005). In 51 audiotaped conferences, clinicians did 71% of the talking; families who spoke more were more satisfied (McDonagh 2004).

Lautrette 2007 randomized relatives of dying ICU patients to a proactive VALUE conference plus a bereavement brochure, or the customary conference: at 90 days the intervention group had fewer PTSD-related symptoms (45% vs 69%), anxiety (45% vs 67%) and depression (29% vs 56%). In PARTNER (White 2018; 1,420 patients), family support from the ICU team left 6-month anxiety, depression and PTSD scores unchanged but improved rated communication (69.1 vs 62.7) and shortened ICU stay (6.7 vs 7.4 days).

Running a goals-of-care meeting

1. Prepare: agree the facts, likely outcomes and a recommendation; pick one lead speaker; invite the bedside nurse; book a professional interpreter, not a relative.
2. Open: introductions, purpose, what the family already understands.
3. Inform: one headline in plain words, then stop; say died and dying (Truog 2008).
4. Respond to emotion before adding facts; allow silence.
5. Explore what mattered to the patient and what they would say now.
6. Recommend a plan that fits those values: what continues, what stops, and that the patient will not be abandoned or left to suffer (Truog 2008).
7. Close: summarize, set the next contact, document the decision, surrogate and code status.

Shared decision-making is the default for value-laden ICU decisions, but families differ in how much of the decision they want to carry; the ACCM/ATS statement (Kon 2016) asks clinicians to adapt, generally offering a recommendation and, if a family prefers, leading the decision while it stays free to disagree. Hoping for the best while planning for the worst keeps hope without leaving families unprepared (Truog 2008).

A time-limited trial uses agreed treatments for a set period, with markers of improvement or decline chosen in advance and a meeting to review them. After physicians in three public-hospital ICUs were trained to offer them to patients with advanced illness (Chang 2021; 209 patients), family meetings rose from 60% to 96% and moved from about day 5.5 to day 1, median ICU stay fell from 8.7 to 7.4 days, ventilation (85.8% vs 72.9%) and central lines (71.7% vs 40.0%) fell, and hospital mortality was unchanged (58.4% vs 58.3%).

Surrogates, conflict and moral distress

A surrogate decides as the patient would have (substituted judgment), using stated wishes and values; when those are unknown, the standard is the patient's best interests. A health care agent named by the patient comes first; otherwise state law sets a default order, commonly spouse or partner, adult children, parents, then siblings. Relatives outside that order have no authority.

Three principles frame US practice (Truog 2008): withholding and withdrawing are equivalent (the ANA 2024 adds legally); allowing death from disease differs from killing, because patients and surrogates may refuse any treatment (Quinlan); and double effect separates relieving symptoms from causing death. A nurse may never act with the intent of ending a patient's life (ANA 2024).

For requests the team believes should not be met, the 2015 multisociety statement (Bosslet) keeps futile for interventions that cannot reach their physiologic goal and calls the rest potentially inappropriate. Prevent conflict with early communication; if it persists, use a fair process: expert consultation, written notice, a second opinion, committee review, an offer of transfer, notice of external appeal, then carrying out the outcome, with high-quality palliative care if treatment stops. SCCM 2025 suggests every hospital have such a policy.

Moral distress, knowing the right action but feeling unable to take it, is common among ICU nurses when treatment seems only to prolong dying. Raise it at rounds, request an ethics consultation, and debrief after deaths (Truog 2008). A nurse may decline to take part on moral grounds but must hand off care safely, never abandon the patient.

Palliative care triggers, code status and POLST

The ICU team provides primary palliative care; specialists join when needs exceed it. In Norton 2007 (191 high-risk medical ICU patients), proactive consults for the five triggers on the one-pager shortened ICU stay from 16.3 to 9.0 days without changing mortality or discharge disposition. ESICM 2024 also suggests validated symptom-assessment tools for patients with advanced illness.

A DNR order says nothing about pressors, antibiotics, dialysis or ICU admission; each is decided on its merits. Turn off defibrillator shocks once comfort is the goal, and revisit code status whenever goals change.

Withdrawing life support step by step

Truog 2008 treats withdrawal as an ICU procedure with a plan, and SCCM 2025 suggests protocols that assess symptoms before, during and after extubation. Withdrawing treatment is never withdrawing care. Antibiotics, pressors, dialysis, blood, IV fluids and feeds add no comfort, and apart from the ventilator they can stop at once without distress. Explain why monitors come off, so the family watches the patient, not the screen.

1. Before: decision documented; OPO called (CMS requires it before withdrawal); family and chaplain offered time and rituals.
2. Before: paralytic off and worn off; ICD shocks off; labs, imaging and alarms stopped.
3. Before: comfort orders written (opioid and benzodiazepine bolus and infusion, suction); prescriber available.
4. During: opioid bolus first; suction, extubate or reduce support; treat distress at once; afterward, pronounce death per policy and give the family time.

Extubation or terminal wean? Truog 2008 finds no basis for a dogmatic choice: terminal weaning is not ethically required, and its only justification is time to titrate drugs to dyspnea. In ARREVE (Robert 2017; 458 patients, 43 French ICUs), immediate extubation brought more airway obstruction, gasping and higher Behavioral Pain Scale scores, but relatives' 3-month stress scores (IES-R 30.5 with extubation vs 31.9 with weaning), grief, anxiety, depression and time to death did not differ, and assistant nurses reported less job strain.

Of 822 terminally extubated patients (Fehnel 2020), 19% had a respiratory rate above 30 within 1 hour and 30% within 6 hours; no anticipatory opioid (adjusted OR 2.08), no opioid in the first hour, GCS above 8 and P/F below 200 predicted it, and anticipatory dosing did not change time to death. In Chan 2004 (75 patients), median time to death after ventilator withdrawal was 35 minutes (range 1 to 890).

Prepare the family for irregular breaths and pauses, noisy secretions, mottled skin and an unpredictable time course. Truog 2008 advises against the word agonal, which sounds like agony; with good palliation this breathing is part of dying, not suffering.

Symptom control and the ethics of comfort

Most dying ICU patients cannot report symptoms, so distress is read from behavior: tachypnea, tachycardia, a fearful face, accessory muscle use, paradoxical breathing and nasal flaring (Truog 2008), formalized in the Respiratory Distress Observation Scale; CPOT or BPS scores pain. Treat what you see, then reassess within minutes.

Opioids are the mainstay for pain and dyspnea. Truog 2008 names morphine the agent of choice, with fentanyl (by infusion, since a bolus wears off quickly) or hydromorphone as alternatives. Benzodiazepines add anxiolysis, sedation and amnesia but no analgesia; midazolam peaks in 5–10 minutes, lorazepam in 20–25. When distress recurs, bolus and raise the infusion together, since a rate change alone takes hours to reach steady state. Doses for opioid-tolerant patients scale to current use; the target is comfort, not a number.

Noisy secretions often trouble families more than the patient: turn the patient onto the side, suction the mouth gently, stop IV fluids, and consider glycopyrrolate or scopolamine, whose benefit is uncertain.

Double effect separates giving drugs to relieve suffering, with the foreseen but unintended possibility of hastening death, from giving them to cause death, a line US law also draws (Truog 2008). Because intent shows in practice, chart the sign of distress, the dose and the response, and raise doses in proportion to symptoms. Chan 2004 found no link between opioid or benzodiazepine dose and time to death after withdrawal; in the last 2 hours, each 1 mg/h more benzodiazepine was associated with 13 minutes longer survival.

Brain death: physiology and the 2023 standard

Rising ICP from a mass, hemorrhage or anoxic swelling stops cerebral blood flow and the brainstem herniates. A catecholamine surge gives way to vasoplegia; central diabetes insipidus affects 46–86% of donors; temperature control is lost (Meyfroidt 2019). The 2023 AAN/AAP/CNS/SCCM guideline (Greer), replacing those of 2010 and 2011, defines brain death as permanent loss of function of the entire brain, including the brainstem: coma, brainstem areflexia and apnea. It is legal death under the Uniform Determination of Death Act.

The examination confirms coma with no response beyond spinal reflexes, no pupillary light response, and absent corneal, oculocephalic (skipped if the neck is unsafe), oculovestibular, gag and cough reflexes. One examination by a credentialed clinician suffices in adults; children need two, by different clinicians at least 12 hours apart. No one involved in organ recovery may take part.

Apnea test: bring PaCO₂ to 35–45 mm Hg and pH to 7.35–7.45 (or a known chronic baseline); preoxygenate with 100% O₂ for at least 10 minutes toward a PaO₂ above 200; stop ventilator breaths and give oxygen by a catheter at the carina at 4–6 L/min (adults), CPAP, or a flow-inflating bag with a PEEP valve. No breathing with arterial PaCO₂ ≥60 and ≥20 above baseline and pH below 7.30 supports brain death. Stop for any breath, SBP <100 or MAP <75 despite titration, a progressive fall in SpO₂ below 85%, or an arrhythmia with hemodynamic instability. On ECMO, preoxygenate through the membrane lung too, then lower sweep gas to 0.2–1 L/min or add CO₂ to the circuit; on VA ECMO, sample the distal arterial line and post-oxygenator blood (Lewis 2023).

Ancillary tests are needed only when the examination or apnea test cannot be completed, or with chronic hypercapnia and an unknown baseline: catheter angiography, radionuclide flow scanning, or transcranial Doppler (adults only); EEG, CTA, MRA, MRI and evoked potentials do not qualify. Time of death is when the final apnea-test gas meets criteria, or when a clinician documents a confirming ancillary test. Consent is not needed unless state law or policy says so, but families are told testing is planned, offered reasonable accommodation, and given limited time before organ support stops.

Organ donation: referral, request and pathways

Hospitals must phone their OPO about every death and imminent death, ideally within 1 hour and before any life-sustaining therapy is withdrawn; batch lists are not allowed (CMS tag A-0886). Each hospital defines imminent death with its OPO, for example a ventilated patient with severe brain injury and a GCS at or below an agreed threshold, a planned brain death evaluation, or an order to withdraw support. Early referral lets the OPO judge suitability before any family is approached.

Only an OPO representative or a hospital designated requester trained by the OPO may ask the family, ideally together, and no family may be skipped because of assumptions about grief, race, religion or income (CMS tags A-0889 and A-0890). In Siminoff 2001, 56.7% of 420 families consented, more often when their pre-request profile, including knowing the patient's wishes, favored donation (adjusted OR 7.68) or they had more contact with OPO staff (OR 3.08). In most states the revised Uniform Anatomical Gift Act (2006) bars others from revoking a registered adult donor's own gift, so the family is informed rather than asked.

Care decisions must stand alone: the 2023 brain death guideline asks clinicians to weigh only their patient's interests. For DCD, the ATS/ISHLT/SCCM/AOPO/UNOS statement (Gries 2013) puts discussion of withdrawal before any mention of donation and gives transplant staff no role in the decision, the withdrawal or the declaration of death. OPTN rules effective November 18, 2026 add a 5-minute minimum between circulatory cessation and organ recovery, a formal pause families or staff can call over safety concerns, documented neurologic assessments, and specific consent for premortem DCD drugs or procedures (HRSA 2026).

Caring for the brain-dead donor

Meeting the UNOS Region 5 donor management goals (any 7 of 9) before recovery predicted more organs transplanted per donor (Malinoski 2012: 4 or more organs, OR 2.34).

The Canadian guideline (Ball 2020) suggests a MAP of at least 65, vasopressin as the first vasopressor and norepinephrine second, and suggests against dopamine at any dose (conditional, very low to low certainty). Diabetes insipidus shows as large volumes of dilute urine and a rising sodium: replace free water and give desmopressin 1–4 mcg IV, repeated as needed, or vasopressin when hypotension coexists, keeping sodium below 155 (Meyfroidt 2019).

For lung donors, Canada 2020 strongly recommends tidal volumes of 6–8 mL/kg, PEEP of at least 8 and recruitment after ventilator disconnections. In Mascia 2010 (118 donors), a protective strategy made 95% vs 54% of donors eligible to give lungs and doubled lungs harvested (54% vs 27%).

In Dhar 2023 (838 unstable heart donors), levothyroxine 30 mcg/h for at least 12 hours did not add hearts for transplant (54.9% vs 53.2% with saline) and caused more severe hypertension (26 vs 5) and tachycardia (16 vs 3); Canada 2020 suggests against routine thyroid hormone. Methylprednisolone (15 mg/kg or 1 g daily) is still used despite largely neutral trials, given after blood for HLA typing is drawn (Meyfroidt 2019).

Keep the core above 35 °C (Meyfroidt 2019). Mild hypothermia of 34–35 °C cut delayed kidney graft function from 39% to 28% (Niemann 2015), and Canada 2020 suggests it unless the kidneys will not be used, but it did worse than machine perfusion of the kidneys in a later trial (30% vs 19%, Malinoski 2023). Treat glucose above 180; a hemoglobin threshold of 7, platelets above 50,000 and INR below 1.5 before surgery are proposed goals, untested in donors (Meyfroidt 2019).

Exam pearls

• VALUE: value, acknowledge, listen, understand the person, elicit questions (Lautrette 2007).
• DNR withholds CPR only; POLST is a clinician-signed order, an advance directive the patient's own document.
• No paralytics at withdrawal; opioids for dyspnea, benzodiazepines for anxiety; bolus before extubation and with each rate increase.
• Double effect: intent to relieve suffering, doses matched to symptoms; titrated doses do not shorten dying.
• Brain death 2023: core ≥36 °C, SBP ≥100 and MAP ≥75, apnea PaCO₂ ≥60 and ≥20 above baseline with pH <7.30; EEG no longer counts.
• Refer every imminent death to the OPO, ideally within 1 hour and before withdrawal; only the OPO or a trained requester asks the family.
• DCD: withdrawal is decided first; the care team declares death after ≥2 minutes without circulation (OPTN 2026: 5 before recovery); transplant staff stay out.
• Donor care: vasopressin first for hypotension, desmopressin for DI, lung-protective ventilation, sodium below 155.