Despite progress in developing treatments that have improved life expectancies for patients with advanced-stage cancer, the American Cancer Society estimates that 606,520 Americans will die of cancer in 2020. People with cancer die under various circumstances. A report of the Dartmouth Atlas Project analyzed Medicare data from 2003 to 2007 for cancer patients older than 65 years who died within 1 year of diagnosis. Across the United States, 29% of patients died in a hospital, with 61.3% hospitalized at least once in the last month of life. In addition, 24% of patients were admitted to an intensive care unit at least once. Approximately 6% of patients nationwide received chemotherapy in the last month of life. Conversely, about 55% of patients who died used hospice service; however, the average length of stay was only 8.7 days, and 8.3% of patients were enrolled in the last 3 days of life. Significant regional variations in the descriptors of end-of-life (EOL) care remain unexplained.
Patient and family preferences may contribute to the observed patterns of care at the EOL. Patient recall of EOL discussions, spiritual care, or early palliative care, however, are associated with less-aggressive EOL treatment and/or increased utilization of hospice.[3,4,5,6,7] In addition, death in a hospital has been associated with poorer quality of life and increased risk of psychiatric illness among bereaved caregivers. (Refer to the PDQ summary on Planning the Transition to End-of-Life Care in Advanced Cancer for more information.) Thus, it is important to help patients and their families articulate their goals of care and preferences near the EOL.
This summary provides clinicians with information about anticipating the EOL; the common symptoms patients experience as life ends, including in the final hours to days; and treatment or care considerations. The decisions commonly made by patients, families, and clinicians are also highlighted, with suggested approaches. The goal of this summary is to provide essential information for high-quality EOL care.
In this summary, unless otherwise stated, evidence and practice issues as they relate to adults are discussed. The evidence and application to practice related to children may differ significantly from information related to adults. When specific information about the care of children is available, it is summarized under its own heading.
The Final Days to Weeks of Life
Providing excellent care toward the end of life (EOL) requires an ability to anticipate when to focus mainly on palliation of symptoms and quality of life instead of disease treatment. Prognostic information plays an important role for making treatment decisions and planning for the EOL. However, the available literature suggests that medical providers inaccurately predict how long patients will live and tend to overestimate survival times.[2,3] This appears to hold true even for providers who are experienced in treating patients who are terminally ill. In a multicenter cohort study of 230 hospitalized patients with advanced cancer, palliative care providers correctly predicted time to death for only 41% of patients. Survival time was overestimated in 85% of patients for whom medical providers gave inaccurate predictions, and providers were particularly likely to overestimate survival for black and Latino patients.
Providers attempting to make prognostic determinations may attend to symptoms that may herald the EOL (refer to the Impending Death section of this summary for more information), or they may observe trends in patients' functional status.
For example, one group of investigators  retrospectively analyzed nearly 71,000 Palliative Performance Scale (PPS) scores obtained from a cohort of 11,374 adult outpatients with cancer who were assessed by physicians or nurses at the time of clinic visits. The PPS is an 11-point scale describing a patient's level of ambulation, level of activity, evidence of disease, ability to perform self-care, nutritional intake, and level of consciousness.
The investigators assigned patients to one of four states:
Of the 4,806 patients who died during the study period, 49% were recorded as being in the transitional state, and 46% were recorded as being in the stable state. For a patient who was in the transitional state, the probability of dying within a month was 24.1%, which was less than that for a patient in the EOL state (73.5%). The results suggest that serial measurement of the PPS may aid patients and clinicians in identifying the approach of the EOL.
Impending death, or actively dying, refers to the process in which patients who are expected to die within 3 days exhibit a constellation of symptoms. In the final days of life, patients often experience progressive decline in their neurocognitive, cardiovascular, respiratory, gastrointestinal, genitourinary, and muscular function, which is characteristic of the dying process. A number of highly specific clinical signs can be used to help clinicians establish the diagnosis of impending death (i.e., death within days).[7,8]
Impending death is a diagnostic issue rather than a prognostic phenomenon because it is an irreversible physiologic process. The ability to diagnose impending death with confidence is of utmost importance to clinicians because it could affect their communication with patients and families, and inform complex health care decisions such as:[9,10]
One study examined five signs in cancer patients recognized as actively dying. Investigators reported that the median time to death from the onset of death rattle was 23 hours; from the onset of respiration with mandibular movement, 2.5 hours; from the onset of cyanosis in extremities, 1 hour; and from the onset of pulselessness on the radial artery, 2.6 hours.
The Investigating the Process of Dying study systematically examined physical signs in 357 consecutive cancer patients. Among the ten target physical signs, there were three early signs and seven late signs. The early signs had high frequency, occurred more than 1 week before death, and had moderate predictive value that a patient would die in 3 days. Early signs included the following:
The late signs occurred mostly in the last 3 days of life, had lower frequency, and were highly specific for impending death in 3 days. Late signs included the following:
In particular, the high positive likelihood ratios (LRs) of pulselessness on the radial artery (positive LR, 15.6), respiration with mandibular movement (positive LR, 10), decreased urine output (≤200 cc/d) (positive LR, 15.2), Cheyne-Stokes breathing (positive LR, 12.4), and death rattle (positive LR, 9) suggest that these physical signs can be useful for the diagnosis of impending death. Because of low sensitivity, the absence of these signs cannot rule out impending death. A prospective observational study that examined vital signs in the last 7 days of life reported that blood pressure and oxygen saturation decreased as death approached. However, a large proportion of patients had normal vital signs, even in the last 12 hours of life. Moderate changes in vital signs from baseline could not definitively rule in or rule out impending death in 3 days. Therefore, predicting death is difficult, even with careful and repeated observations.
Several other late signs that have been found to be useful for the diagnosis of impending death include the following:
In conclusion, bedside physical signs may be useful in helping clinicians diagnose impending death with greater confidence, which can, in turn, assist in clinical decision making and communication with families.
Care During the Final Hours of Life
The recognition of impending death is also an opportunity to encourage family members to notify individuals close to the patient who may want an opportunity to "say goodbye." In the final hours of life, care should be directed toward the patient and the patient's loved ones. In addition to continuing a careful and thoughtful approach to any symptoms a patient is experiencing (refer to the Symptoms During the Final Months, Weeks, and Days of Life section of this summary), preparing family and friends for a patient's death is critical. Preparations include the following:
Encouraging family members who desire to "do something" to participate in the care of the patient (e.g., moistening the mouth) may be helpful. In the final days to hours of life, patients often have limited, transitory moments of lucidity. Family members should be prepared for this and educated that this is a natural aspect of the dying process and not necessarily a result of medications being administered for symptoms or a sign that the patient is doing better than predicted. Despite their limited ability to interact, patients may be aware of the presence of others; thus, loved ones can be encouraged to speak to the patient as if he or she can hear them.
Educating family members about certain signs is critical. In the final hours of life, patients often experience a decreased desire to eat or drink, as evidenced by clenched teeth or turning from offered food and fluids. This behavior may be difficult for family members to accept because of the meaning of food in our society and the inference that the patient is "starving." Family members should be advised that forcing food or fluids can lead to aspiration. Reframing will include teaching the family to provide ice chips or a moistened oral applicator to keep a patient's mouth and lips moist. Finally, the death rattle is particularly distressing to family members. (Refer to the Death Rattle section in the Symptoms During the Final Months, Weeks, and Days of Life section of this summary for more information.) It is important to assure family members that death rattle is a natural phenomenon and to pay careful attention to repositioning the patient and explain why tracheal suctioning is not warranted.
For patients who die in the hospital, clinicians need to be prepared to inquire about the family's desire for an autopsy, offering reassurance that the body will be treated with respect and that open-casket services are still possible, if desired.
Health care professionals, preferably in consultation with a chaplain or religious leader designated by the patient and/or family, need to explore with families any fears associated with the time of death and any cultural or religious rituals that may be important to them. Such rituals might include placement of the body (e.g., the head of the bed facing Mecca for an Islamic patient) or having only same-sex caregivers or family members wash the body (as practiced in many orthodox religions). When death occurs, expressions of grief by those at the bedside vary greatly, dictated in part by culture and in part by their preparation for the death. Chaplains are to be consulted as early as possible if the family accepts this assistance. Health care providers can offer to assist families in contacting loved ones and making other arrangements, including contacting a funeral home. (Refer to the PDQ summary on Spirituality in Cancer Care for more information.)
The available evidence provides some general description of frequency of symptoms in the final months to weeks of the end of life (EOL). However, when the results of published studies of symptoms experienced by patients with advanced cancer are being interpreted or compared, the following methodological issues need to be considered:
Results of one of the larger and more comprehensive studies of symptoms in ambulatory patients with advanced cancer have been reported. Ambulatory patients with advanced cancer were included in the study if they had completed at least one Edmonton Symptom Assessment System (ESAS) in the 6 months before death. The ESAS is a patient-completed measure of the severity of the following nine symptoms:
Analysis of the changes in the mean symptom intensity of 10,752 patients (and involving 56,759 assessments) over time revealed two patterns:
In terms of symptoms closer to the EOL, a prospective study documented the symptom profile in the last week of life among 203 cancer patients who died in acute palliative care units.[Level of evidence: II] The proportion of patients able to communicate decreased from 80% to 39% over the last 7 days of life. ESAS anorexia, drowsiness, fatigue, poor well-being, and dyspnea increased in intensity closer to death. In contrast, ESAS depression decreased over time. Dysphagia of solids and liquids and urinary incontinence were also present in an increasing proportion of patients in the last few days of life. Less common but equally troubling symptoms that may occur in the final hours include death rattle and hemorrhage.
The following sections summarize some of the common symptoms and potential approaches to ameliorating those symptoms, based on available evidence. A final note of caution is warranted. Symptoms often cluster, and the presence of a symptom should prompt consideration of other symptoms to avoid inadvertently worsening other symptoms in the cluster. For example, a systematic review of observational studies concluded that there were four common clusters of symptoms (anxiety-depression, nausea-vomiting, nausea-appetite loss, and fatigue-dyspnea-drowsiness-pain).
Terminal delirium occurs before death in 50% to 90% of patients. Most patients have hypoactive delirium, with a decreased level of consciousness. Agitation, hallucinations, and restlessness may occur in a small proportion of patients with hyperactive and/or mixed delirium. Delirium is associated with shorter survival and complicates symptom assessment, communication, and decision making. Furthermore, it can be extremely distressing to caregivers and health professionals.[6,7,8] Risk factors associated with terminal delirium include the following:
Safety measures include protecting patients from accidents or self-injury while they are restless or agitated. The use of restraints should be minimized. Reorientation strategies are of little use during the final hours of life. Education and support for families witnessing a loved one's delirium are warranted.
There are few randomized controlled trials on the management of delirium in patients with terminal or irreversible delirium. (Refer to the PDQ summary on Delirium for a complete review.) Care of the patient with delirium can include stopping unnecessary medications, reversing metabolic abnormalities (if consistent with the goals of care), treating the symptoms of delirium, and providing a safe environment. Agents known to cause delirium include:
In a small, open-label, prospective trial of 20 cancer patients who developed delirium while being treated with morphine, rotation to fentanyl reduced delirium and improved pain control in 18 patients.[Level of evidence: II]
The onset of effect and non-oral modes of delivery are considered when an agent is being selected to treat delirium at the EOL. Agents that can be used to manage delirium include haloperidol, 1 mg to 4 mg orally, intravenously (IV), or subcutaneously. The dose is usually repeated every 4 to 6 hours but in severe cases can be administered every hour. Chlorpromazine can be used, but IV administration can lead to severe hypotension; therefore, it should be used cautiously. Other agents that may be effective include olanzapine, 2.5 mg to 20 mg orally at night (available in an orally disintegrating tablet for patients who cannot swallow);[Level of evidence: II] quetiapine; and risperidone (0.5–2 mg). While no randomized clinical trial demonstrates superiority of any agent over haloperidol, small (underpowered) studies suggest that olanzapine may be comparable to haloperidol.
In addition, a small, double-blind, randomized trial at the University of Texas MD Anderson Cancer Center compared the relative sedating effects of scheduled haloperidol, chlorpromazine, and a combination of the two for advanced-cancer patients with agitated delirium. The investigators screened 998 patients from the palliative and supportive care unit and randomly assigned 68 patients who met the inclusion criteria for having agitated delirium refractory to scheduled haloperidol 1 to 8 mg/day to three intervention groups: haloperidol 2 mg every 4 hours, chlorpromazine 25 mg every 4 hours, or haloperidol 1 mg combined with chlorpromazine 12.5 mg every 4 hours. Rescue doses equivalent to the standing dose were allowed every 1 hour as needed and once at protocol initiation, with the goal of producing sedation with a Richmond Agitation-Sedation Scale (RASS) score of 0 to –2. Dose escalations and rescue doses were allowed for persistent symptoms.
Of the 68 randomized patients, 45 patients were treated and monitored until death or discharge. Patients in all three groups demonstrated clinically significant decreases in RASS scores within 30 minutes and remained sedated at 24 hours. No statistically significant difference in sedation levels was observed between the three protocols. However, the chlorpromazine group was less likely to develop breakthrough restlessness requiring rescue doses or baseline dosing increases. The most common adverse event was hypotension, which was seen in 40% of patients in the haloperidol group, 31% of those in the chlorpromazine group, and 21% of those in the combination group. One patient in the combination group discontinued therapy because of akathisia. The study was limited by a small sample size and the lack of a placebo group. Although all three interventions were effective at controlling agitation, it is worth noting that they controlled agitation via significant sedation, which may not be desired by all patients and/or their families.
Although benzodiazepines (such as lorazepam) or antidopaminergic medications could exacerbate delirium, they may be useful for the treatment of hyperactive delirium that is not controlled by other supportive measures. In intractable cases of delirium, palliative sedation may be warranted. A randomized controlled trial compared the effect of lorazepam versus placebo as an adjunctive to haloperidol on the intensity of agitation in 58 patients with delirium in a palliative care unit. Patients were eligible for the study if they had a diagnosis of delirium with a history of agitation (hyperactive delirium subtype). Once enrolled, patients began a regimen of haloperidol 2 mg IV every 4 hours, with 2 mg IV hourly as needed for agitation. The RASS score was monitored every 2 hours until the score was 2 or higher. At this threshold, the patient received lorazepam 3 mg or matching placebo with one additional dose of haloperidol 2 mg. The primary outcome of RASS score reduction was measured 8 hours after administration of the study drug. Patients in the lorazepam group experienced a statistically significant reduction in RASS score (increased sedation) at 8 hours (–4.1 points for lorazepam/haloperidol vs. –2.3 points for placebo/haloperidol; mean difference, –1.9 points [95% confidence interval, –2.8 to –0.9]; P < .001). Lorazepam-treated patients also required significantly lower doses of rescue neuroleptics and, after receiving the study medication, were perceived to be in greater comfort by caregivers and nurses. No differences in mortality were noted between the treatment arms.
Caution should be exercised in the use of this protocol because of the increased risk of significant sedation. The reduction in agitation is directly proportional to increased sedation to the point of patients being "minimally responsive to verbal stimulus" or conversion to hypoactive delirium during the remaining hours of life. Communication with patients and surrogates to determine goal-concordant care in the setting of terminal or hyperactive delirium is imperative to ensure that sedation is an intended outcome of this protocol in which symptom reduction is the primary intention of the intervention. Intensive evaluation of RASS scores may be challenging for the bedside nurse. Finally, this study examined a single dose of lorazepam 3 mg; repeat doses were not studied and may accumulate in patients with liver and/or renal dysfunction.
In dying patients, a poorly understood phenomenon that appears to be distinct from delirium is the experience of auditory and/or visual hallucinations that include loved ones who have already died (also known as EOL experience). Although patients may sometimes find these hallucinations comforting, fear of being labeled confused may prevent patients from sharing their experiences with health care professionals. Family members at the bedside may find these hallucinations disconcerting and will require support and reassurance. Consultation with the patient's or family's religious or spiritual advisor or the hospital chaplain is often beneficial.
Fatigue is one of the most common symptoms at the EOL and often increases in prevalence and intensity as patients approach the final days of life. Fatigue at the EOL is multidimensional, and its underlying pathophysiology is poorly understood. It may be associated with drowsiness, weakness, and sleep disturbance. Scores on the Palliative Performance Scale also decrease rapidly during the last 7 days of life. No clinical trials have been conducted in patients with only days of life expectancy. Methylphenidate may be useful in selected patients with weeks of life expectancy. (Refer to the PDQ summary on Fatigue for more information.)
Treatment options for dyspnea, defined as difficult, painful breathing or shortness of breath, include opioids, nasal cannula oxygen, fans, raising the head of the bed, noninvasive ventilation, and adjunctive agents.
Opioids are often considered the preferred first-line treatment option for dyspnea. However, the evidence supporting this standard is controversial, according to a 2016 Cochrane review that found only "low quality evidence" to support the use of opioids to treat breathlessness. The review's authors suggest that larger, more rigorous studies are needed to conclusively determine whether opioids are effective for treating dyspnea, and whether they have an impact on quality of life for patients suffering from breathlessness.
Nasal cannula oxygen
The routine use of nasal cannula oxygen for patients without documented hypoxemia is not supported by the available data. One group of investigators reported a double-blind randomized controlled trial comparing the severity of morning and evening breathlessness as reported by patients who received either supplemental oxygen or room air via nasal cannula. No differences in the primary outcome of symptomatic relief for refractory dyspnea were found in the 239 subjects enrolled in the trial. There were no significant differences in secondary outcomes such as extreme drowsiness or nasal irritation. Potential criticisms of the study include the trial period being only 7 days and a single numerical scale perhaps inadequately reflecting the palliative benefit of oxygen.
A small pilot trial randomly assigned 30 Chinese patients with advanced cancer with unresolved breathlessness to either usual care or fan therapy. The outcome measures included a self-report measure of breathlessness, respiratory rate, and measured oxygen saturation. The measurements were performed before and after fan therapy for the intervention group. There was a significant improvement in the self-reported scores of the patients in the fan group but not in the scores of controls. There were no changes in respiratory rates or oxygen saturations in either group.
Raising the head of the bed
Because dyspnea may be related to position-dependent changes in ventilation and perfusion, it may be worthwhile to try to determine whether a change in the patient's positioning in bed alleviates air hunger.
A randomized trial compared noninvasive ventilation (with tight-fitting masks and positive pressure) with supplemental oxygen in a group of advanced-cancer patients in respiratory failure who had chosen to forgo all life support and were receiving palliative care. Patients had to have significant oxygen needs as measured by the ratio of the inhaled oxygen to the measured partial pressure of oxygen in the blood. Two hundred patients were randomly assigned to treatment. Patients in the noninvasive-ventilation group reported more-rapid improvement in dyspnea and used less palliative morphine in the 48 hours after enrollment. Eleven patients in the noninvasive-ventilation group withdrew because of mask discomfort.
It should be noted that all patients were given subcutaneous morphine titrated to relief of dyspnea. Whether patients with less severe respiratory status would benefit is unknown. The generalizability of the intervention is limited by the availability of equipment for noninvasive ventilation. In addition, while noninvasive ventilation is less intrusive than endotracheal intubation, a clear understanding of the goals of the intervention and whether it will be electively discontinued should be established.
Bronchodilators, corticosteroids, and antibiotics may be considered in select situations, provided the use of these agents are consistent with the patient's goals of care. Bronchodilators may help patients with evidence of bronchoconstriction on clinical examination. Their use carries a small but definite risk of anxiousness and/or tachycardia. Corticosteroids may also be of benefit but carry a risk of anxiety, insomnia, and hyperglycemia. The treatment of potential respiratory infections with antibiotics likewise calls for a consideration of side effects and risks.
The prevalence of pain is between 30% and 75% in the last days of life.[21,29] The assessment of pain may be complicated by delirium. Many patients fear uncontrolled pain during the final days of life, but experience suggests that most patients can obtain pain relief and that very high doses of opioids are rarely indicated. Indeed, the average intensity of pain often decreases as patients approach the final days.
Some patients, family members, and health care professionals express concern that opioid use may hasten death. Several studies refute the fear of hastened death associated with opioid use. In several surveys of high-dose opioid use in hospice and palliative care settings, no relationship between opioid dose and survival was found.[30,31,32,33]
The principles of pain management remain similar to those for patients earlier in the disease trajectory, with opioids being the standard option. (Refer to the PDQ summary on Cancer Pain for a more complete review of parenteral administration of opioids and opioid rotation.) Because consciousness may diminish during this time and swallowing becomes difficult, practitioners need to anticipate alternatives to the oral route. In one study, as patients approached death, the use of intermittent subcutaneous injections and IV or subcutaneous infusions increased. Both IV and subcutaneous routes are effective in delivering opioids and other agents in the inpatient or home setting. For patients who do not have a preexisting access port or catheter, intermittent or continuous subcutaneous administration provides a painless and effective route of delivery.
Cough is a relatively common symptom in patients with advanced cancer near the EOL. In one small study, 33% of patients with advanced cancer who were enrolled in hospice and who completed the Memorial Symptom Assessment Scale reported cough as a troubling symptom. This compares to a prevalence of lack of energy (68%), pain (63%), and dyspnea (60%). The treatment of troublesome coughing in patients in the final weeks to days of life is largely empiric, although diagnostic imaging or evaluation may occasionally be of value. In another study of patients with advanced cancer admitted to acute palliative care units, the prevalence of cough ranged from 10% to 30% in the last week of life. (Refer to the PDQ summary on Cardiopulmonary Syndromes for more information about common causes of cough for which evaluation and targeted intervention may be indicated.) However, simple investigations such as reviewing medications or eliciting a history of symptoms compatible with gastroesophageal reflux disease are warranted because some drugs (e.g., angiotensin-converting enzyme inhibitors) cause cough, or a prescription for antacids may provide relief. In addition, patients may have comorbid conditions that contribute to coughing.
The cough reflex protects the lungs from noxious materials and clears excess secretions. The reflex is initiated by stimulation of peripheral cough receptors, which are transmitted to the brainstem by the vagus nerve. Activation of the central "cough center" mechanism causes a deep inspiration, followed by expiration against a closed glottis; then the glottis opens, allowing expulsion of the air. The empiric approach to cough may be organized as follows:
As discussed in the Dyspnea section of this summary, the use of bronchodilators, corticosteroids, or inhaled steroids is limited to specific indications, given the potential risks and the lack of evidence of benefit outside of specific indications.
The prevalence of constipation ranges from 30% to 50% in the last days of life.[3,29] The use of laxatives for patients who are imminently dying may provide limited benefit. Such patients often have dysphagia and very poor oral intake. Treatment of constipation in patients with only days of expected survival is guided by symptoms. If indicated, laxatives may be given rectally (e.g., bisacodyl or enemas).
Functional dysphagia and structural dysphagia occur in a large proportion of cancer patients in the last days of life. Specifically, patients often experience difficulty swallowing both liquids and solids, which is often associated with anorexia and cachexia. In one study of cancer patients, the oral route of opioid administration was continued in 62% of patients at 4 weeks before death, in 43% at 1 week before death, and in 20% at 24 hours before death. The clinical implication is that essential medications may need to be administered through other routes, such as IV, subcutaneous, rectal, and transdermal. Nonessential medications are discontinued.
Artificial nutrition is of no known benefit at the EOL and may increase the risk of aspiration and/or infections. Instead of tube-feeding or ordering nothing by mouth, providing a small amount of food for enjoyment may be reasonable if a patient expresses a desire to eat. (Refer to the Artificial Hydration and Artificial Nutrition sections of this summary for more information.)
Death rattle, also referred to as excessive secretions, occurs when saliva and other fluids accumulate in the oropharynx and upper airways in a patient who is too weak to clear the throat. Rattle does not appear to be distressing for the patient; however, family members may perceive death rattle as indicating the presence of untreated dyspnea. Thus, the family will benefit from learning about the nature of this symptom and that death rattle is not associated with dyspnea.
Rattle is an indicator of impending death, with an incidence of approximately 50% to 60% in the last days of life and a median onset of 16 to 57 hours before death.[23,40,41] Two types of rattle have been identified:[42,43]
In one retrospective chart review, rattle was relieved in more than 90% of patients with salivary secretions, while patients with secretions of pulmonary origin were much less likely to respond to treatment.
The results of clinical trials examining various pharmacologic agents for the treatment of death rattle have so far been negative. A small, double-blind, randomized, controlled trial that compared scopolamine to normal saline found no statistical significance. Another randomized study revealed no difference between atropine and placebo. Results of other randomized controlled studies that examined octreotide, glycopyrrolate, and hyoscine butylbromide  versus scopolamine were also negative.
Despite the lack of clear evidence, pharmacologic therapies are used frequently in clinical practice.[50,51] Among the options described above, glycopyrrolate may be preferred because it is less likely to penetrate the central nervous system, and fewer adverse effects are reported with glycopyrrolate than with other antimuscarinic agents, which could worsen delirium. Glycopyrrolate is available parenterally and in oral tablet form. Doses typically range from 1 mg to 2 mg orally or 0.1 mg to 0.2 mg IV or subcutaneously every 4 hours, or by continuous IV infusion at a rate of 0.4 mg to 1.2 mg per day. Repositioning may be helpful. Suctioning of excessive secretions may be considered for some patients.
Opioid-Induced Neurotoxicity and Myoclonus
Health care professionals need to monitor patients for opioid-induced neurotoxicity, which can cause symptoms such as myoclonus, hallucinations, hyperalgesia, seizures, and confusion, and which may mimic terminal delirium. One retrospective study of 390 advanced-cancer patients at the University of Texas MD Anderson Cancer Center who had been taking opioids for 24 hours or longer and who received palliative care consultations suggested that 15% of these patients developed at least one symptom of opioid-induced neurotoxicity, the most common of which was delirium (47%). When opioid-induced neurotoxicity is suspected, opioid rotation may be considered.
Myoclonus is defined as sudden and involuntary movements caused by focal or generalized muscle contractions. The duration of contractions is brief and may be described as shocklike. There are many potential causes of myoclonus, most of which probably stem from the metabolic derangements anticipated as life ends. Medications, particularly opioids, are another potential etiology. The reported prevalence of opioid-induced myoclonus ranges greatly, from 2.7% to 87%.
When opioids are implicated in the development of myoclonus, rotation to a different opioid is the primary treatment. In patients with rapidly impending death, the health care provider may choose to treat the myoclonus rather than make changes in opioids during the final hours. Benzodiazepines, including clonazepam, diazepam, and midazolam, have been recommended.[53,54,55] The anticonvulsant gabapentin has been reported to be effective in relieving opioid-induced myoclonus, although other reports implicate gabapentin as a cause of myoclonus.[57,58][Level of evidence: III] In one small randomized study, hydration was found to reduce myoclonus.[Level of evidence: I]
There are no reliable data on the frequency of fever. A prospective study of 232 adults with terminal cancer admitted to a hospice and palliative care unit in Taiwan indicated that fever was uncommon and of moderate severity (mean score, 0.37 on a scale of 1–3). There was no increase in fever in the 2 days immediately preceding death. In addition to considering diagnostic evaluation and therapeutic intervention, the clinician needs to carefully assess whether the patient is distressed or negatively affected by the fever.
There are no data showing that fever materially affects the quality of the experience of the dying person. While infection may cause a fever, other etiologies such as medications or the underlying cancer are to be strongly considered. While the main objective in the decision to use antimicrobials is to treat clinically suspected infections in patients who are receiving palliative or hospice care,[61,62,63][Level of evidence: II] subsequent information suggests that the risks of using empiric antibiotics do not appear justified by the possible benefits for people near death.
Hemorrhage is an uncommon (6%–14%) yet extremely distressing event, especially when it is sudden and catastrophic. Patients with bone marrow failure or liver failure are susceptible to bleeding caused by lack of adequate platelets or coagulation factors; patients with advanced cancer, especially head and neck cancers, experience bleeding caused by fungating wounds or damage to vascular structures from tumor growth, surgery, or radiation. Patients may also experience gastrointestinal bleeding from ulcers, progressive tumor growth, or chemotherapy-induced mucositis.
The management of catastrophic bleeding may include identification of patients who are at risk of catastrophic bleeding and careful communication about risk and potential management strategies. However, two qualitative interview studies of clinicians whose patients experienced catastrophic bleeding at the EOL suggest that it is often impossible to anticipate bleeding and that a proactive approach may cause patients and families undue distress.[66,67] Furthermore, the lack of evidence that catastrophic bleeding can be prevented with medical interventions such as transfusions needs to be taken into account in discussions with patients about the risks of bleeding.
Another strategy is to prepare to administer anxiolytics or sedatives to patients who experience catastrophic bleeding, between the start of the bleeding and death. However, there is little evidence supporting the effectiveness of this approach;[65,67] the experience of clinicians is often that patients become unconscious before the drugs can be administered, and the focus on medications may distract from providing patients and families with reassurance that suffering is unlikely. Nevertheless, the availability of benzodiazepines for rapid sedation of patients who experience catastrophic bleeding may provide some reassurance for family caregivers.
After the death of a patient from a catastrophic hemorrhage, team members are encouraged to verbalize their emotions regarding the experience, and their questions need to be answered.
Palliative Sedation to Treat EOL Symptoms
In rare situations, EOL symptoms may be refractory to all of the treatments described above. In such cases, palliative sedation may be indicated, using benzodiazepines, barbiturates, or neuroleptics. Refractory dyspnea is the second most common indication for palliative sedation, after agitated delirium. (Refer to the Palliative Sedation section of this summary for more information.)
The Decision to Discontinue Disease-Directed Therapies
A significant proportion (approximately 30%) of patients with advanced cancer continue to receive chemotherapy toward the end of life (EOL), including a small number (2%–5%) who receive their last dose of chemotherapy within 14 days of death.[1,2,3,4] Chemotherapy administered until the EOL is associated with significant adverse effects, resulting in prolonged hospitalization or increased likelihood of dying in an intensive care unit (ICU).[5,6] Thus, the lack of definite or meaningful improvement in survival leads many clinicians to advise patients to discontinue chemotherapy on the basis of an increasingly unfavorable ratio of benefit to risk.
Ultimately, the decision to initiate, continue, or forgo chemotherapy should be made collaboratively and is ideally consistent with the expected risks and benefits of treatment within the context of the patient's goals of care. However, the following reasons independent of the risks and benefits may lead a patient to prefer chemotherapy and are potentially worth exploring:
In the era of personalized medicine, genomic tumor testing is indicated for multiple tumor types, and finding actionable mutations for targeted therapy holds promise for many patients with metastatic cancers. Similarly, immune checkpoint inhibitors have revolutionized the standard of care for multiple cancers. These drugs are increasingly used in older patients and those with poorer performance status for whom traditional chemotherapy may no longer be appropriate, though they may still be associated with unwanted side effects. Predictive factors for whether any given patient will have a significant response to these newer agents are often unclear, making prognostication challenging.[10,11] This uncertainty may lead to questions about when systemic treatment should be stopped and when supportive care only and/or hospice care should begin.
Data on immune checkpoint inhibitor use at the EOL are limited, but a single-institution retrospective study showed that immunotherapy use in the last 30 days of life is associated with lower rates of hospice enrollment and a higher risk of dying in the hospital.
The Decision to Enroll in Hospice
Patients with advanced cancer who receive hospice care appear to experience better psychological adjustment, fewer burdensome symptoms, increased satisfaction, improved communication, and better deaths without hastening death.[12,13,14,15] The rate of hospice enrollment for people with cancer has increased in recent years; however, this increase is tempered by a reduction in the average length of hospice stay. Because of the association of longer hospice stays with caregivers' perceptions of improved quality of care and increased satisfaction with care, the latter finding is especially concerning.[16,17]
Multiple patient demographic factors (e.g., younger age, married status, female gender, white race, greater affluence, and geographic region) are associated with increased hospice enrollment. In one study, however, physician characteristics were more important than patient characteristics in determining hospice enrollment. This may reflect the observation that patients concede more control to oncologists over time, especially if treatment decisions involve noncurative chemotherapy for metastatic cancer.
There are many potential barriers to timely hospice enrollment. Forgoing disease-directed therapy is one of the barriers cited by patients, caregivers, physicians, and hospice services. The Medicare Care Choices Model, a novel Centers for Medicare & Medicaid Services (CMS) pilot program, is evaluating a new supportive care model that allows beneficiaries to receive supportive care from selected hospice providers, alongside therapy directed toward their terminal condition. This 5-year project enrolled its first cohort of patients in January 2016 and the second cohort in January 2018. CMS will evaluate whether providing these supportive services can improve patient quality of life and care, improve patient and family satisfaction, and inform a new payment system for the Medicare and Medicaid programs.
The purpose of this section is to provide the oncology clinician with insights into the decision to enroll in hospice, and to encourage a full discussion of hospice as an important EOL option for patients with advanced cancer. The related study  provides potential strategies to address some of the patient-level barriers.
Barriers are summarized in the following subsections on the basis of whether they arise predominantly from the perspective of the patient, caregiver, physician, or hospice, including eligibility criteria for enrollment.
Patients often express a sense that it would be premature to enroll in hospice, that enrolling in hospice means giving up, or that enrolling in hospice would disrupt their relationship with their oncologist. The difficulty in recognizing when to enroll in hospice may explain the observations that the trend in increasing hospice utilization has not led to a reduction in intensive treatment, including admission to ICUs at the EOL.[21,22]
Patients may agree to enroll in hospice in the final days of life only after aggressive medical treatments have clearly failed. Given the likely benefit of longer times in hospice care, patient-level predictors of short hospice stays may be particularly relevant. One group of investigators conducted a retrospective cohort study of 64,264 adults with cancer admitted to hospice. Sixteen percent stayed 3 days or fewer, with a range of 11.4% to 24.5% among the 12 participating hospices. The following factors (and odds ratios [ORs]) were independently associated with short hospice stays in multivariable analysis:
A diagnosis of depression may also affect how likely patients are to enroll in hospice. A 2018 retrospective cohort study of 13,827 patients with non-small cell lung cancer (NSCLC) drew data from the Surveillance, Epidemiology, and End Results (SEER)–Medicare database to examine the association between depression and hospice utilization. The authors found that NSCLC patients with precancer depression (depression recorded during the 3–24 months before cancer diagnosis) and patients with diagnosis-time depression (depression recorded between 3 months before and 30 days after cancer diagnosis) were more likely to enroll in hospice than were NSCLC patients with no recorded depression diagnosis (subhazard ratio [SHR], 1.19 and 1.16, respectively).
Patients with precancer depression were also more likely to spend extended periods (≥90 days) in hospice care (adjusted OR, 1.29). In contrast, patients with postdiagnosis depression (diagnosed >30 days after NSCLC diagnosis) were less likely to enroll in hospice (SHR, 0.80) than were NSCLC patients without depression.
The authors hypothesized that patients with precancer depression may be more likely to receive early hospice referrals, especially given previously established links between depression and high symptom burden in patients with advanced cancer. They also suggested that enhanced screening for depression in patients with cancer may impact hospice enrollment and quality of care provided at the EOL.
In a survey of 53 caregivers of patients who died of lung cancer while in hospice, 35% of caregivers felt that patients should have received hospice care sooner. The highest rates of agreement with potential reasons for deferring hospice enrollment were for the following three survey items:
Only 22% of caregivers agreed that the family member delayed enrollment because enrolling in hospice meant giving up hope.
In a survey of 273 physicians, 65% agreed that a barrier to hospice enrollment was the patient preference for simultaneous anticancer treatment and hospice care. Almost one-half of physicians believed (incorrectly) that patients must have do-not-resuscitate and do-not-intubate orders in place to qualify for hospice. The lack of timely discussions with oncologists or other physicians about hospice care and its benefits remains a potentially remediable barrier to the timing of referral to hospice.[26,27,28]
The Medicare hospice benefit requires that physicians certify patients' life expectancies that are shorter than 6 months and that patients forgo curative treatments. Patients who are enrolled in hospice receive all care related to their terminal illnesses through hospice, although most hospice reimbursement comes through a fixed per diem. Thus, hospices may have additional enrollment criteria.
One group of investigators conducted a national survey of 591 hospices that revealed 78% of hospices had at least one policy that could restrict enrollment. Sixty-one percent of patients could not be receiving chemotherapy, 55% could not be receiving total parenteral nutrition, and 40% could not be receiving transfusions. Only 8% restricted enrollment of patients receiving tube feedings. The percentage of hospices without restrictive enrollment practices varied by geographic region, from a low of 14% in the East/West South Central region to a high of 33% in the South Atlantic region.
A further challenge related to hospice enrollment is that the willingness to forgo chemotherapy does not identify patients who have a high perceived need for hospice care. Investigators conducted conjoint interviews of 300 patients with cancer and 171 family caregivers to determine the perceived need for five core hospice services (visiting nurse, chaplain, counselor, home health aide, and respite care). Patients willing to forgo chemotherapy did not have different levels of perceived needs. In a multivariable model, the following patient factors predicted a greater perceived need for hospice services:
The following family factors predicted a greater perceived need for hospice services:
Choosing the Desired Place of Death
Many patients with advanced-stage cancer express a desire to die at home, but many will die in a hospital or other facility. Patients who die at home, however, appear to have a better quality of life than do patients who die in a hospital or ICU, and their bereaved caregivers experience less difficulty adjusting. Dying at home is also associated with better symptom control and preparedness for death and with caregivers' perceptions of a higher-quality death.
Enrollment in hospice increases the likelihood of dying at home, but careful attention needs to be paid to caregiver support and symptom control. One group of investigators analyzed a cohort of 5,837 hospice patients with terminal cancer for whom the patient's preference for dying at home was determined. Of the 5,837 patients, 4,336 (79%) preferred to die at home. Patients who preferred to die at home were more likely to do so (56% vs. 37%; OR, 2.21). In multivariable analysis, the following factors (with percentages and ORs) were correlated with a greater likelihood of dying at home:
Conversely, patients were less likely to die at home (OR, <1) if there was:
However, not all patients prefer to die at home, e.g., patients who are unmarried, nonwhite, and older. Thus, the oncology clinician strives to facilitate a discussion about preferred place of death and a plan to overcome potential barriers to dying at the patient's preferred site.
Approximately one-third to one-half of pediatric patients who die of cancer die in a hospital.[34,35] Dying in an inpatient setting has been associated with more intensive and invasive interventions in the last month of life for pediatric cancer patients and adverse psychosocial outcomes for caregivers. For example, parents of children who die in the hospital experience more depression, anxiety, and complicated grief than do parents of children who die outside of the hospital.
A retrospective analysis of 321 pediatric cancer patients who died while enrolled on the palliative care service at St. Jude Children's Research Hospital suggests that the following factors (with ORs) were associated with a higher likelihood of dying in the pediatric ICU:
Pediatric care providers may want to consider the factors listed above to identify patients at higher risk of dying in an intensive inpatient setting, and to initiate early conversations about goals of care and preferred place of death.
A necessary goal of high-quality end-of-life (EOL) care is the alleviation of distressing symptoms that can lead to suffering. An important strategy to achieve that goal is to avoid or reduce medical interventions of limited effectiveness and high burden to the patients. There is, however, a great deal of confusion, anxiety, and miscommunication around the question of whether to utilize potentially life-sustaining treatments (LSTs) such as mechanical ventilation, total parenteral nutrition, and dialysis in the final weeks or days of life. To ensure that the best interests of the patient—as communicated by the patient, family, or surrogate decision maker—determine the decisions about LSTs, discussions can be organized around the following questions:
Medicine is a moral enterprise. The decisions clinicians make are often highly subjective and value laden but seem less so because, equally often, there is a shared sense of benefit, harm, and what is most highly valued. Occasionally, disagreements arise or a provider is uncertain about what is ethically permissible. For example, an oncologist may favor the discontinuation or avoidance of LST, given the lack of evidence of benefit or the possibility of harm—including increasing the suffering of the dying person by prolonging the dying process—or based on concerns that LST interferes with the patient accepting that life is ending and finding peace in the final days. A provider also may be uncertain about whether withdrawing treatment is equivalent to causing the patient's death. Conversely, the patient may continue to request LST on the basis of personal beliefs and a preference for potential prolonged life, independent of the oncologist's clinical risk-benefit analysis.
Individual values inform the moral landscape of the practice of medicine. The potential conflicts described above are opportunities to refine clinicians' understanding of their beliefs and values and to communicate their moral reasoning to each other as a sign of integrity and respect. As clinicians struggle to communicate their reasons for recommendations or actions, the following three questions may serve as a framework:
In the event of conflict, an ethics consult may be necessary to identify the sources of disagreement and potential solutions, although frameworks have been proposed to guide the clinician. The following paragraphs summarize information relevant to the first two questions.
Respect for patient autonomy is an essential element of the relationship between oncology clinician and patient. Respect for autonomy encourages clinicians to elicit patients' values, goals of care, and preferences and then seek to provide treatment or care recommendations consistent with patient preferences. Autonomy is primarily a negative right to be free from the interference of others or, in health care, to refuse a recommended treatment or intervention. On the basis of potential harm to others or deliberate harm to themselves, there are limits to what patients can expect in terms of their requests. More controversial limits are imposed when oncology clinicians feel they are asked to violate their ethical integrity or when the medical effectiveness of a treatment does not justify the burden.
Perceived conflicts about the issue of patient autonomy may be avoided by recalling that promoting patient autonomy is not only about treatments administered but also about discussions with the patient. It is acceptable for oncology clinicians to share the basis for their recommendations, including concerns such as clinician-perceived futility.[6,7]
The following criteria to consider forgoing a potential LST are not absolute and remain a topic of discussion and debate; however, they offer a frame of reference for deliberation:
Religious and Spiritual Beliefs
Awareness of the importance of religious beliefs and spiritual concerns within medical care has increased substantially over the last decade. (Refer to the PDQ summary on Spirituality in Cancer Care for a review.) National consensus guidelines were published in 2009 and recommended the following:
An interprofessional approach is recommended: medical personnel, including physicians, nurses, and other professionals such as social workers and psychologists, are trained to address these issues and link with chaplains, as available, to evaluate and engage patients. A survey of nurses and physicians revealed that most nurses (74%) and physicians (60%) desire to provide spiritual care, which was defined as "care that supports a patient's spiritual health." The more commonly cited barriers associated with the estimated amount of spiritual care provided to patients included inadequate training and the belief that providing spiritual care is not part of the medical professional's role. Most nurses (79%) desired training in spiritual care; fewer physicians (51%) did.
Evidence strongly supports that most cancer patients desire dialogue about these issues with their physicians, other staff as appropriate, and hospital chaplains, if indicated. About one-half of patients acknowledge that they are not receiving such support from a religious community, either because they are not involved in one or because they do not perceive their community as supportive. Regardless of such support, patients may report substantial spiritual distress at the EOL, ranging from as few as 10% or 15% of patients to as many as 60%. (Refer to the PDQ summary on Spirituality in Cancer Care for more discussion about spiritual distress.) Distress may range from anger at God, to a feeling of unworthiness, to lack of meaning. Such distress, if not addressed, may complicate EOL decisions and increase depression.
Finally, it has been shown that addressing religious and spiritual concerns earlier in the terminal-care process substantially decreases the likelihood that patients will request aggressive EOL measures. In contrast, patients who have received strong support from their own religious communities alone are less likely to enter hospice and more likely to seek aggressive EOL care. It has also been shown that providing more comprehensive palliative care increases spiritual well-being as the EOL approaches.
The potential indications for artificial hydration in the final weeks or days of life may be broadly defined by the underlying goal of either temporarily reversing or halting clinical deterioration or improving the comfort of the dying patient. Although artificial hydration may be provided through enteral routes (e.g., nasogastric tubes or percutaneous gastrostomy tubes), the more common route is parenteral, either IV by catheter or subcutaneously through a needle (hypodermoclysis).
Although uncontrolled experience suggested several advantages to artificial hydration in patients with advanced cancer, a well-designed, randomized trial of 129 patients enrolled in home hospice demonstrated no benefit in parenteral hydration (1 L of normal saline infused subcutaneously over 4 hours) compared with placebo (100 mL of normal saline infused subcutaneously over 4 hours). There were no differences in survival, symptoms, quality of life, or delirium.
A prospective evaluation of the outcomes of 161 patients with advanced-stage abdominal cancers who received parenteral hydration in accordance with Japanese national guidelines near the EOL suggests there is little harm or benefit in hydration. The median survival of the cohort was 20 days (range, 1–84 days); the mean volume of parenteral hydration was 912 ± 495 mL/day. There were no significant trends in global quality of life, discomfort, or physical symptoms for ill or good; signs of fluid retention were common but not exacerbated. However, patients expressed a high level of satisfaction with hydration and felt it was beneficial. The guidelines specify that patients with signs of volume overload should receive less than 1 L of hydration per day. When the investigators stratified patients into two groups—those who received at least 1 L of parenteral hydration per day and those who received less than 1 L/day—the prevalence of bronchial secretions was higher and hyperactive delirium was lower in the patients who received less than 1 L.
Any discussion about the risks or benefits of artificial hydration must include a consideration of patient and family perspectives. Requests for artificial hydration or the desire for discussions about the role of artificial hydration seem to be driven by quality-of-life considerations as much as considerations for life prolongation. Families may be helped with this decision when clinicians explain that use of artificial hydration in patients with cancer at the EOL has not been shown to help patients live longer or improve quality of life. Additionally, families can be educated about good mouth care and provision of sips of water to alleviate thirst. The oncology clinician needs to approach these conversations with an open mind, recognizing that the harm caused by artificial hydration may be minimal relative to the perceived benefit, which includes reducing fatigue and increasing alertness.
The following discussion excludes patients for whom artificial nutrition may facilitate further anticancer treatment or for whom bowel obstruction is the main manifestation of their advanced cancer and for whom enteral or total parenteral nutrition may be of value. Furthermore, artificial nutrition as a supplement may benefit the patient with advanced cancer who has a good performance status, a supportive home environment, and an anticipated survival longer than 3 months.[26,27]
The decisions about whether to provide artificial nutrition to the dying patient are similar to the decisions regarding artificial hydration. The American Academy of Hospice and Palliative Medicine (AAHPM) recommends that individual clinical situations be assessed using clinical judgment and skill to determine when artificial nutrition is appropriate. Recognizing that the primary intention of nutrition is to benefit the patient, AAHPM concludes that withholding artificial nutrition near the EOL may be appropriate medical care if the risks outweigh the possible benefit to the patient.
Food should be offered to patients consistent with their desires and ability to swallow. The benefit of providing artificial nutrition in the final days to weeks of life, however, is less clear. One study has concluded that artificial nutrition—specifically, parenteral nutrition—neither influenced the outcome nor improved the quality of life in terminally ill patients.
Providing artificial nutrition to patients at the EOL is a medical intervention and requires establishing enteral or parenteral access. Considerations of financial cost, burden to patient and family of additional hospitalizations and medical procedures, and all potential complications must be weighed against any potential benefit derived from artificial nutrition support.
In discussions with patients, the oncology clinician needs to recognize that the patient perception of benefit is worth exploring; as a compromise or acknowledgment of respect for the patient's perspective, a time-limited trial may be warranted.
The administration of anti-infectives, primarily antibiotics, in the last days of life is common, with antibiotic use reported in patients in the last week of life at rates ranging from 27% to 78%.[31,32,33] Because of wide heterogeneity in the measurement of antibiotic use, assessment of symptom response, and lack of comparisons between those receiving antimicrobials with those not receiving them, the benefit of antimicrobials is currently hard to define. The infections most likely to respond are urinary tract, skin, or eye infections; those less likely to be managed by antimicrobials, such as pneumonia, may be better managed with symptomatic measures.[34,35,36] One additional setting in which antimicrobial use may be warranted is that of contagious public health risks such as tuberculosis. There is also concern that the continued use of antimicrobials in the last week of life leads to increased risk of developing drug-resistant organisms.
In general, most practitioners agree with the overall focus on patient comfort in the last days of life rather than providing curative therapies with unknown or marginal benefit, despite their ability to provide the therapy.[31,34,36,37,38]
Anemia is common in patients with advanced cancer; thrombocytopenia is less common and typically occurs in patients with progressive hematological malignancies. The decision to transfuse either packed red cells or platelets is based on a careful consideration of the overall goals of care, the imminence of death, and the likely benefit and risks of transfusions. The decision to use blood products is further complicated by the potential scarcity of the resource and the typical need for the patient to receive transfusions in a specialized unit rather than at home.
There are no randomized or controlled prospective trials of the indications, safety, or efficacy of transfused products. It should be recognized, however, that many patients will have received transfusions during active disease treatment or periods of supportive care. Such patients may have notions of the importance of transfusions related to how they feel and their life expectancies.
Several considerations may be relevant to the decision to transfuse red blood cells:
Broadly defined, resuscitation includes all interventions that provide cardiovascular, respiratory, and metabolic support necessary to maintain and sustain the life of a dying patient. It is important for patients, families, and proxies to understand that choices may be made to specify which supportive measures, if any, are given preceding death and at the time of death. People often believe that there is plenty of time to discuss resuscitation and the surrounding issues; however, many dying patients do not make choices in advance or have not communicated their decisions to their families, proxies, and the health care team. If these issues are unresolved at the time of EOL events, undesired support and resuscitation may result. Studies suggest that this aggressive care is associated with worse patient quality of life and worse adjustment to bereavement for loved ones.[42,43]
Narrowly defined, a do-not-resuscitate (DNR) order instructs health care providers that, in the event of cardiopulmonary arrest, cardiopulmonary resuscitation (CPR, including chest compressions and/or ventilations) should not be performed and that natural death be allowed to proceed. DNR orders must be made before cardiac arrest and may be recommended by physicians when CPR is considered medically futile or would be ineffective in returning a patient to life. A DNR order may also be made at the instruction of the patient (or family or proxy) when CPR is not consistent with the goals of care.
It is advisable for a patient who has clear thoughts about these issues to initiate conversations with the health care team (or appointed health care agents in the outpatient setting) and to have forms completed as early as possible (i.e., before hospital admission), before the capacity to make such decisions is lost. Although patients with end-stage disease and their families are often uncomfortable bringing up the issues surrounding DNR orders, physicians and nurses can tactfully and respectfully address these issues appropriately and in a timely fashion. Lack of standardization in many institutions may contribute to ineffective and unclear discussions around DNR orders.
No Further Escalation of Care
Patients with advanced cancer are often unprepared for a decline in health status near the end of life (EOL) and, as a consequence, they are admitted to the hospital for more aggressive treatments. This complicates EOL decision making because the treatments may prolong life, or at least are perceived as accomplishing that goal. At that point, patients or families may express ambivalence or be reluctant to withdraw treatments rather than withhold them.
One strategy to explore is preventing further escalation of care. The goal of this strategy is to provide a bridge between full life-sustaining treatment (LST) and comfort care, in which the goal is a good death. It has been suggested that clinicians may encourage no escalation of care because of concerns that the intensive medical treatments will prevent death, and therefore the patient will have missed the "opportunity to die." One study  described the care of 310 patients who died in the intensive care unit (ICU) (not all of whom had cancer). For 95 patients (30%), there was a decision not to escalate care. The average time from ICU admission to deciding not to escalate care was 6 days (range, 0–37), and the average time to death was 0.8 days (range, 0–5). The interventions most likely to be withheld were dialysis, vasopressors, and blood transfusions. Fifty-five percent of the patients eventually had all life support withdrawn. The transition to comfort care did not occur before death for the other decedents for the following reasons: waiting for family to arrive, change of family opinion, or waiting for an ethics consultation.
Conversely, some situations may warrant exploring with the patient and/or family a time-limited trial of intensive medical treatments. The patient or surrogates may choose to withdraw all LST if there is no improvement during the limited trial. From an ethical standpoint, withdrawing treatment is equivalent to withholding such treatment. Both actions are justified for unwarranted or unwanted intensive care.
Withdrawal of Ventilatory Support
Two methods of withdrawal have been described: immediate extubation and terminal weaning.
Immediate extubation. Immediate extubation includes providing parenteral opioids for analgesia and sedating agents such as midazolam, suctioning to remove excess secretions, setting the ventilator to "no assist" and turning off all alarms, and deflating the cuff and removing the endotracheal tube. Gentle suctioning of the oral cavity may be necessary, but aggressive and deep suctioning should be avoided. In some cases, patients may appear to be in significant distress. Analgesics and sedatives may be provided, even if the patient is comatose. Family members and others who are present should be warned that some movements may occur after extubation, even in patients who have no brain activity. Such movements are probably caused by hypoxia and may include gasping, moving extremities, or sitting up in bed. Immediate extubation is generally chosen when a patient has lost brain function, when a patient is comatose and unlikely to experience any suffering, or when a patient prefers a more rapid procedure.
Terminal weaning. Terminal weaning entails a more gradual process. Ventilator rate, oxygen levels, and positive end-expiratory pressure are decreased gradually over a period of 30 minutes to a few hours. A patient who survives may be placed on a T-piece; this may be left in place, or extubation may proceed. There is some evidence that the gradual process in a patient who may experience distress allows clinicians to assess pain and dyspnea and to modify the sedative and analgesic regimen accordingly. In a study of 31 patients undergoing terminal weaning, most patients remained comfortable, as assessed by a variety of physiologic measures, when low doses of opioids and benzodiazepines were administered. The average time to death in this study was 24 hours, although two patients survived to be discharged to hospice.
Paralytic agents have no analgesic or sedative effects, and they can mask patient discomfort. These neuromuscular blockers need to be discontinued before extubation. Guidelines suggest that these agents should never be introduced when the ventilator is being withdrawn; in general, when patients have been receiving paralytic agents, these agents need to be withdrawn before extubation. The advantage of withdrawal of the neuromuscular blocker is the resultant ability of the health care provider to better assess the patient's comfort level and to allow possible interaction between the patient and loved ones. One notable exception to withdrawal of the paralytic agent is when death is expected to be rapid after the removal of the ventilator and when waiting for the drug to reverse might place an unreasonable burden on the patient and family.
Regardless of the technique employed, the patient and setting must be prepared. Monitors and alarms are turned off, and life-prolonging interventions such as antibiotics and transfusions need to be discontinued. Family members should be given sufficient time to make preparations, including making arrangements for the presence of all loved ones who wish to be in attendance. They need to be given information about what to expect during the process; some may elect to remain out of the room during extubation. Chaplains or social workers may be called to provide support to the family.
Suffering at the End of Life (EOL)
Published in 2013, a prospective observational study of 64 patients who died of cancer serially assessed symptoms, symptom intensity, and whether symptoms were unbearable. Weakness was the most prevalent symptom (93% of patients). Fifty-one percent of patients rated their weakness as high intensity; of these, 84% rated their suffering as unbearable. Pain, loss of control over one's life, and fear of future suffering were unbearable when symptom intensity was high.
Informal caregiver perspectives
Caregiver suffering is a complex construct that refers to severe distress in caregivers' physical, psychosocial, and spiritual well-being. A number of studies have reported strong associations between patients' and caregivers' emotional states. In a systematic review of 19 descriptive studies of caregivers during the palliative, hospice, and bereavement phases, analysis of patient-caregiver dyads found mutuality between the patient's condition and the caregiver's response. Caregivers were found to be at increased risk of physical and psychological burden across studies, with caregiver distress sometimes exceeding that of the patient.
Several studies have categorized caregiver suffering with the use of dyadic analysis. In a qualitative study involving 22 dyadic semistructured interviews, caregivers dealing with advanced medical illness, including cancer, reported both unique and shared forms of suffering. Suffering was characterized as powerlessness, threat to the caregiver's identity, and demands exceeding resources. Because caregiver suffering can affect patient well-being and result in complicated bereavement, early identification and support of caregiver suffering are optimal. (Refer to the PDQ summary on Informal Caregivers in Cancer: Roles, Burden, and Support for more information.)
Palliative sedation may be defined as the deliberate pharmacologic lowering of the level of consciousness, with the goal of relieving symptoms that are unacceptably distressing to the patient and refractory to optimal palliative care interventions. In considering a patient's request for palliative sedation, clinicians need to identify any personal biases that may adversely affect their ability to respond effectively to such requests. Several points need to be borne in mind:
The following questions may serve to organize discussions about the appropriateness of palliative sedation within health care teams and between clinicians, patients, and families:
Indications for palliative sedation
The two broad indications for palliative sedation are refractory physical symptoms and refractory existential or psychological distress. A retrospective study at the MD Anderson Cancer Center in Houston included 1,207 patients admitted to the palliative care unit. Palliative sedation was used in 15% of admissions. The most common indications were delirium (82%) and dyspnea (6%).
The use of palliative sedation for refractory existential or psychological symptoms is highly controversial. In a survey of the attitudes and experiences of more than 1,000 U.S. physicians toward intentional sedation to unconsciousness until death revealed that 68% of respondents opposed palliative sedation for existential distress. Of note, only 10% of physician respondents had prescribed palliative sedation in the preceding 12 months. A previous survey conducted by the same research group reported that only 18% of surveyed physicians objected to "sedation to unconsciousness in dying patients" without a specified indication.
One potential objection or concern related to palliative sedation for refractory existential or psychological distress is unrecognized but potentially remediable depression. (Refer to the Requests for Hastened Death section of this summary for more information.)
Further objections or concerns include (1) whether the principle of double effect, an ethical basis for the use of palliative sedation for refractory physical distress, is adequate justification; and (2) cultural expectations about psychological or existential suffering at the EOL. The principle of double effect is based on the concept of proportionality. When applied to palliative sedation, this principle supports the idea that the intended effect of palliative sedation (i.e., relief of suffering) may justify a foreseeable-but-unintended consequence (such as possibly shortening life expectancy—though this is not supported by data, as mentioned above —or eliminating the opportunity to interact with loved ones) if the intended (positive) outcome is of greater value than the unintended (negative) outcome. Thus, in the case of palliative sedation for refractory psychological or existential distress, the perception that palliative sedation is not justified may reflect a devaluation of the distress associated with such suffering or that other means with fewer negative consequences have not been fully explored.
Intended level of sedation
Another decision to be made is whether the intended level of sedation is unconsciousness or a level associated with relief of the distress attributed to physical or psychological symptoms. In a survey of U.S. physicians, two-thirds of respondents felt that unconsciousness was an acceptable unintended consequence of palliative sedation, but deliberate unconsciousness was unacceptable. This finding may relate to the sense of proportionality. A qualitative study of 54 physicians who had administered palliative sedation indicated that physicians who were more concerned with ensuring that suffering was relieved were more likely to administer palliative sedation to unconsciousness. Physicians who chose mild sedation were guided more by their assessment of the patient's condition.
Intended duration of sedation
Palliative sedation may be provided either intermittently or continuously until death. If a clinician anticipates that a distressing symptom will improve with time, then that clinician should discuss with the patient any recommendations about a deliberate reduction in the depth of sedation to assess whether the symptoms persist. Furthermore, deliberate reductions in the depth of sedation may be appropriate if there is a desire for communication with loved ones.
For example, a single-center observational study monitored 89 (mostly male) hospice patients with cancer who received either intermittent or continuous palliative sedation with midazolam, propofol, and/or phenobarbital for delirium (61%), dyspnea (20%), or pain (15%). Mental status changes in the 37 patients who received intermittent palliative sedation for delirium were as follows, after sedation was lightened: 43.2% unchanged, 40.6% improved, and 16.2% worsened. Unsurprisingly, mental status remained the same or worsened for all patients who received continuous palliative sedation for delirium. Notably, median survival time was only 1 day for patients who received continuous sedation, compared to 6 days for the intermittent palliative sedation group, though the authors hypothesize that this difference may be attributed to a poorer baseline clinical condition in the patients who received continuous sedation rather than to a direct effect of continuous sedation.
Role of potentially LSTs during palliative sedation
The intent of palliative sedation is to relieve suffering; it is not to shorten life. The decision to discontinue or maintain treatments such as artificial hydration or nutrition requires a review of the patient's goals of care and the potential for benefit or harm. In general, the absence of evidence for benefit seems to justify recommendations to forgo LSTs in the context of palliative sedation. On the other hand, open lines of communication and a respectful and responsive awareness of a patient's preferences are important to maintain during the dying process, so the clinician should not overstate the potential risks of hydration or nutrition. There is consensus that decisions about LSTs are distinct from the decision to administer palliative sedation.
Requests for Hastened Death
Potential reasons for requests
Requests for hastened death or statements that express a desire to die vary from expression of a temporary or passive wish to a sustained interest in interventions to end life or a statement of intent to plan or commit suicide. Reliable data on the frequency of requests for hastened death are not available.
Requests for hastened death provide the oncology clinician with an opportunity to explore and respond to the dying patient's experience in an attentive and compassionate manner. From the patient's perspective, the reasons for requests for hastened death are multiple and complex and include the following:
The cited studies summarize the patient's perspectives. Surveys of health care providers demonstrate similar findings and reasons.
Responding to requests
Recommendations are based on principles of counseling and expert opinion. Specific studies are not available. The first and most important consideration is for health care providers to maintain awareness of their personal reactions to requests or statements. It is imperative that the oncology clinician expresses a supportive and accepting attitude. Providers who are too uncomfortable to engage in a discussion need to explain to a patient the need for a referral to another provider for assistance. Other prudent advice includes the following:
Family members are likely to experience loss at the death of their loved one. If left unattended, loss, grief, and bereavement can become complicated, leading to prolonged and significant distress for either family members or clinicians. Furthermore, clinicians are at risk of experiencing significant grief from the cumulative effects of many losses through the deaths of their patients. Burnout has also been associated with unresolved grief in health care professionals. (Refer to the PDQ summary on Grief, Bereavement, and Coping With Loss for more information.)
Oncologists and nurses caring for terminally ill cancer patients are at risk of suffering personally, owing to the clinical intensity and chronic loss inherent in their work. Lack of training in advance care planning and communication can leave oncologists vulnerable to burnout, depression, and professional dissatisfaction. One group of investigators studied oncologists' grief related to patient death and found strong impact in both the personal and professional realms. Negative effects included a sense of distraction and withdrawal from patients.
One study made an important conceptual distinction, explaining that while grief is healthy for oncologists, stress and burnout can be counterproductive. Other terms used to describe professional suffering are moral distress, emotional exhaustion, and depersonalization. Moral distress was measured in a descriptive pilot study involving 29 physicians and 196 nurses caring for dying patients in intensive care units. Both groups of professionals experienced moral distress related to pressure to continue aggressive treatment they considered futile. Nurses experienced more moral distress than did physicians, and perceived less collaboration than did their physician colleagues.
Several strategies have been recommended to help professionals manage the emotional toll of working with advanced-cancer patients and terminally ill cancer patients, including self-care, teamwork, professional mentorship, reflective writing, mindfulness techniques, and working through the grief process.
The PDQ cancer information summaries are reviewed regularly and updated as new information becomes available. This section describes the latest changes made to this summary as of the date above.
Symptoms During the Final Months, Weeks, and Days of Life
Added text about a small, double-blind, randomized trial at the University of Texas MD Anderson Cancer Center that compared the relative sedating effects of scheduled haloperidol, chlorpromazine, and a combination of the two for advanced-cancer patients with agitated delirium (cited Hui et al. as reference 17).
The Opioid-Induced Neurotoxicity and Myoclonus subsection was renamed from Myoclonus.
Revised text to state that health care professionals need to monitor patients for opioid-induced neurotoxicity, which can cause symptoms such as myoclonus, hallucinations, hyperalgesia, seizures, and confusion, and which may mimic terminal delirium. Also added text to state that one retrospective study of 390 advanced-cancer patients at the University of Texas MD Anderson Cancer Center who had been taking opioids for 24 hours or longer and who received palliative care consultations suggested that 15% of these patients developed at least one symptom of opioid-induced neurotoxicity, the most common of which was delirium (cited Lim et al. as reference 52).
Care Decisions in the Final Weeks, Days, and Hours of Life
Added Fang et al. as reference 4.
This summary is written and maintained by the PDQ Supportive and Palliative Care Editorial Board, which is editorially independent of NCI. The summary reflects an independent review of the literature and does not represent a policy statement of NCI or NIH. More information about summary policies and the role of the PDQ Editorial Boards in maintaining the PDQ summaries can be found on the About This PDQ Summary and PDQ® - NCI's Comprehensive Cancer Database pages.
Purpose of This Summary
This PDQ cancer information summary for health professionals provides comprehensive, peer-reviewed, evidence-based information about patient care during the last days to last hours of life. It is intended as a resource to inform and assist clinicians who care for cancer patients. It does not provide formal guidelines or recommendations for making health care decisions.
Reviewers and Updates
This summary is reviewed regularly and updated as necessary by the PDQ Supportive and Palliative Care Editorial Board, which is editorially independent of the National Cancer Institute (NCI). The summary reflects an independent review of the literature and does not represent a policy statement of NCI or the National Institutes of Health (NIH).
Board members review recently published articles each month to determine whether an article should:
Changes to the summaries are made through a consensus process in which Board members evaluate the strength of the evidence in the published articles and determine how the article should be included in the summary.
The lead reviewers for Last Days of Life are:
Any comments or questions about the summary content should be submitted to Cancer.gov through the NCI website's Email Us. Do not contact the individual Board Members with questions or comments about the summaries. Board members will not respond to individual inquiries.
Levels of Evidence
Some of the reference citations in this summary are accompanied by a level-of-evidence designation. These designations are intended to help readers assess the strength of the evidence supporting the use of specific interventions or approaches. The PDQ Supportive and Palliative Care Editorial Board uses a formal evidence ranking system in developing its level-of-evidence designations.
Permission to Use This Summary
PDQ is a registered trademark. Although the content of PDQ documents can be used freely as text, it cannot be identified as an NCI PDQ cancer information summary unless it is presented in its entirety and is regularly updated. However, an author would be permitted to write a sentence such as "NCI's PDQ cancer information summary about breast cancer prevention states the risks succinctly: [include excerpt from the summary]."
The preferred citation for this PDQ summary is:
PDQ® Supportive and Palliative Care Editorial Board. PDQ Last Days of Life. Bethesda, MD: National Cancer Institute. Updated <MM/DD/YYYY>. Available at: https://www.cancer.gov/about-cancer/advanced-cancer/caregivers/planning/last-days-hp-pdq. Accessed <MM/DD/YYYY>. [PMID: 26389307]
Images in this summary are used with permission of the author(s), artist, and/or publisher for use within the PDQ summaries only. Permission to use images outside the context of PDQ information must be obtained from the owner(s) and cannot be granted by the National Cancer Institute. Information about using the illustrations in this summary, along with many other cancer-related images, is available in Visuals Online, a collection of over 2,000 scientific images.
The information in these summaries should not be used as a basis for insurance reimbursement determinations. More information on insurance coverage is available on Cancer.gov on the Managing Cancer Care page.
More information about contacting us or receiving help with the Cancer.gov website can be found on our Contact Us for Help page. Questions can also be submitted to Cancer.gov through the website's Email Us.
Last Revised: 2020-10-28
Healthwise, Healthwise for every health decision, and the Healthwise logo are trademarks of Healthwise, Incorporated.