Terminal Delirium

Over one quarter of Americans die in long-term care facilities,¹ and the prevalence of delirium in the last days or weeks of life is at least 80% to 90%.^2,3 Thus, the long-term care clinician will frequently encounter and need to manage terminal delirium, sometimes, but not always, in conjunction with a hospice program. It is fortunate that most patients die peacefully with decreasing levels of consciousness. Ferris⁴ has referred to this as the “usual road,” but at least 10% to 25% of patients develop varying degrees of agitated terminal delirium (the “difficult road”).⁵ Agitated terminal delirium is an urgent condition; it is profoundly distressing for the patient and is associated with falls, pulling out lines and catheters, and other dangers. It is arguably even more distressing for loved ones, and the image of an agitated, struggling death is highly traumatic and never forgotten.

Delirium is an acute condition characterized by deficits in attention and concentration and altered levels of consciousness; it is always indicative of a systemic problem. At the end of life, there are almost always multiple contributors to delirium.⁶ Hypoactive, lethargic delirium is the most common type at the end of life; hyperactive, agitated delirium is less common but far more urgent and more distressing. Some patients present with a mixed picture of alternating hypoactivity and hyperactivity. The cornerstone of delirium management is to determine and correct the underlying etiology, but at the end of life this is infrequently possible or even desirable. The decision to perform a work-up and intervene depends on many factors, with the goal being a mostly comfortable, peaceful death for all involved.

The cardinal features of any delirium are abrupt onset of impairment, a decrement in attention and focus, and alteration in level of consciousness.⁷ Other features, especially in hyperactive delirium, may include sleep-wake disturbance, illusions and hallucinations, vivid and frightening dreams, and carphologia (purposeless picking at clothing, bedcovers, or unseen stimuli).⁸ The presence of delirium is usually obvious in this setting, but there are numerous screening instruments available as well. Inouye’s Confusion Assessment Method (CAM),⁹ which has been used for over 30 years, is validated in palliative care populations¹⁰ and simply involves noting the presence of major features of delirium (ie, delirium is present if there is decreased attention, an acute onset with fluctuating course, and either disorganized thought or altered level of consciousness). To assess attention at the bedside, the patient can be asked to repeat digits or spell words, such as “world,” forward and then backward. Other quick and useful instruments are available that have been validated for general populations and show good sensitivity and specificity but have not been specifically validated in terminally ill patients. For example, Fick and colleagues¹¹ developed an ultra-brief delirium screen in which the patient is asked the current day of the week and to recite the months of the year backward; one error is indicative of delirium. Other quick screens include the 4 A’s Test (4AT),¹² the Delirium Diagnostic Tool-Provisional (DDT-Pro),¹³ the Nursing Delirium Screening Scale (Nu-DESC),¹⁴ and the Stanford Proxy Test for Delirium (S-PTD).¹⁵ Delirium may be particularly difficult to recognize in patients with end-stage dementia, as most of the typical features of delirium may already be present (eg, poor attention or unresponsiveness, mutism or disorganized speech, lethargy, sleep disturbance).¹⁶ In these patients, an abrupt change in condition may be the only clue.

In patients near the end of life, several conditions can mimic many of the features of delirium and also worsen delirium that is already present. These include pain or discomfort (including constipation and urinary retention), dyspnea, nausea, anxiety and fearfulness, and sleep disturbance. All of these features should be sought and managed aggressively. Conversely, delirium can be misinterpreted as any of these conditions, especially pain, and managed inappropriately.¹⁷

The differential diagnosis of delirium is broad and is almost always multifactorial at the end of life.⁶ Important contributors at this stage include medications (especially anticholinergic agents and benzodiazepines), opioid-induced neurotoxicity, dehydration, electrolyte abnormalities, major end organ failure (kidney, liver, lung, heart), and sepsis. Opioid-induced neurotoxicity is likely underdiagnosed; this is an idiosyncratic reaction to an opioid, often in the context of dehydration or renal impairment, with delirium and often agitation, hallucinations, and myoclonus.^18,19 It is more common with morphine, codeine, meperidine and tramadol, and less so with hydromorphone, oxycodone, methadone, and fentanyl.^20-22 Often, the contributors to terminal delirium are obvious; the decision to further work-up and intervene is a highly individualized one, based upon the anticipated remaining quantity and quality of life, the intrusiveness of the evaluation, the likelihood of finding an easily treatable etiology, the intrusiveness of that treatment and, above all, patient and family preference (Table 1). Morita and colleagues²³ have noted that overall, up to 50% of delirium in the last days of life is potentially reversible; whether it should be reversed is a more complicated question.

However, several contributors involve noninvasive evaluation and treatment (Table 2). These “low-hanging fruit” may yield improvements in alertness, interaction with family, quality of life, and sometimes in agitation. Ultimately, though, the decision to pursue these measures depends on the patient’s and loved ones’ goals of care. For example, some patients and families may see this as doing nothing more than inappropriately prolonging the dying process (although most patients and families will look to the clinician for guidance).

The following factors may be contributing to delirium symptoms:

• Medications. Consider reviewing and withdrawing potentially problematic medications (especially anticholinergics and sedatives), as well as any medications unlikely to provide comfort within the patient’s remaining time (eg, statins, antihypertensives).
• Opioid-induced neurotoxicity. Consider rotating opioids, especially in the presence of myoclonus or if there is no other obvious etiology of the delirium. Oxycodone and hydromorphone are probably less problematic than morphine, and methadone and fentanyl are rarely associated with neurotoxicity.^20-22
• Sedative/hypnotic withdrawal. Consider empirically restarting any recently discontinued sedative agent, which can often be continued until death ensues. If discontinuation is necessary, it can be withdrawn gradually or a better tolerated agent (often lorazepam) can be substituted and gradually weaned.
• Electrolyte abnormalities. Consider checking a basic metabolic panel, managing fluids, and giving electrolytes accordingly if the patient can tolerate the infusion. Consider withholding fluid for hyponatremia or offering hydration for hypercalcemia.
• Hypoglycemia. Consider checking finger stick glucose and then treating with oral or parenteral glucose or glucagon.
• Dehydration. Treatment is rarely warranted. See section below for more information.

Dehydration is common and often desirable at the end of life, and a trial of enteral or parenteral hydration is rarely warranted. In unusual cases, however, it may be reasonable to either check a basic metabolic panel or simply rely on physical signs of dehydration, and to consider a time-limited trial of enteral hydration if a gastrostomy or jejunostomy tube is already present, or intravenous or subcutaneous (ie, hypodermoclysis) hydration. Hypodermoclysis is an especially attractive option at the end of life, as the subcutaneous needle or cannula requires very little time and skill to insert, can be placed in difficult-to-reach places (eg, medial to the scapula) to discourage removal by a confused patient, can support an infusion rate as high as 75 mL/hr, and can deliver numerous medications, including most parenteral opioids, haloperidol, lorazepam, and furosemide.^24,25 Hydration should be discontinued if not yielding the desired outcome; it may sometimes improve alertness and interaction with loved ones, but risks include worsening dyspnea, edema, inability to tolerate the required hardware, or simply prolonging the dying process in a patient with little remaining quality of life.^26-28

In most instances, the goal for management of terminal delirium is simply to maintain comfort and safety and to facilitate a peaceful death.

Hypoactive Delirium
Nonpharmacologic measures (Table 3) are always appropriate and often all that is needed for patients with hypoactive delirium. These measures include avoiding invasive devices whenever possible, reducing the risk of falls and other safety concerns, maintaining a relatively quiet, understimulating but not sensory-depriving environment, minimizing alarms and other noise, and providing a gentle reality orientation. Delirious patients often misinterpret overheard conversations, especially those with potentially frightening content; this should be avoided. There is evolving evidence, mostly in general populations, that interventions to normalize the sleep-wake cycle may decrease the incidence of delirium²⁹; it is reasonable to try to maintain wakefulness during the day and to control lighting and noise. There is also preliminary evidence that certain hypnotics, including melatonin,^30,31 ramelteon,³⁰ and suvorexant,³² may decrease the incidence and severity of delirium. It is critical to remember that delirium, even hypoactive delirium, does not attenuate the experience of pain, dyspnea, or other unpleasant symptoms; opioids and other palliative medications should almost always be continued until death ensues. Acetylcholinesterase inhibitors are ineffective for preventing or treating delirium,^33,34 and neuroleptics have no role in the treatment of hypoactive terminal delirium.

Hyperactive Delirium
Agitated, hyperactive delirium is seen in a minority of patients at the end of life but is an urgent situation; it is highly distressing to patients and families and is associated with falls, pulling out lines, and other poor outcomes.⁴ Initially, nonpharmacologic measures (Table 3) should be implemented, adequate control of pain and other distressing physical symptoms should be ensured, and it may be reasonable to pursue a limited evaluation for easily treatable contributors (ie, the “low-hanging fruit”).

Pharmacologic Treatment
There are few high-quality studies of pharmacologic treatment of terminal delirium, and many of the following considerations are extrapolated from studies of delirium in general populations.

Neuroleptics remain the mainstay of symptomatic management of agitation in hyperactive terminal delirium. While evidence is weak overall^17,35 and there are a few dissenting reports,^36,37 most studies and expert opinion^38-42 support their use as first-line agents. All first- and second-generation neuroleptics seem equally effective, and the choice of agent is usually based on available routes of administration, side-effect profile, or previous exposure. Haloperidol is used most often in this population⁴³ and may be given orally, sublingually, subcutaneously, or intravenously. Typical adverse events, such as QTc prolongation and torsades de pointes, extrapyramidal effects (believed to be less problematic with intravenous administration⁴⁴) and metabolic syndrome, are not concerning in the last weeks of life. Alternatively, several second-generation neuroleptics can be given sublingually, and olanzapine can be given subcutaneously.⁴⁵ Neuroleptics should be titrated to relieve distressing symptoms but not cause undue sedation if possible. If ineffective, it is reasonable to switch to another neuroleptic or to augment with a low dose of lorazepam⁴⁶ or other benzodiazepine. Some practitioners have had success with chlorpromazine for refractory hyperactive terminal delirium^47,48; it is quite sedating and causes significant hypotension. It is also very irritating when given parenterally and must be given in a large vein (central line preferred)—never subcutaneously or intramuscularly. It can also be administered rectally or sublingually if feasible.

With the advent of the US Centers for Medicare & Medicaid Services’ National Partnership to Improve Dementia Care in Nursing Homes in 2012, the use of antipsychotics in skilled nursing facilities (other than for schizophrenia and a few other indications) has become a negative quality measure, open to public scrutiny. End of life care is not exempt from the standard, and there is evidence of reluctance to use antipsychotics and perhaps a greater reliance on other classes of medications in this population.⁴⁹ The role of benzodiazepines in terminal delirium is controversial and inadequately studied; they are commonly administered in the last days of life in hospice enrollees⁵⁰ despite numerous studies in general populations demonstrating worsening of delirium and worsened overall outcome.^51,52 The latter concern is of course irrelevant in patients with a short prognosis, and there is some evidence of their utility for agitation in terminal delirium, probably at the expense of level of alertness.^46,53 Lorazepam is most commonly used as it is readily available parenterally, has a short half-life and no active metabolites. There are also anecdotal reports of the use of midazolam, diazepam, and other benzodiazepines,⁵⁴ also of the barbiturate phenobarbital,^55,56 in the last days of life.

Dexmedetomidine, an alpha-2 agonist, has also shown some promise in terminal delirium.⁵⁷ It is often used in an intensive care setting as it does not appear to be deliriogenic at all and causes no respiratory depression, thus facilitating ventilator weaning. There are reports of benefits in terminal delirium and also for terminal weaning/extubation^58,59 and for palliative sedation.⁶⁰ Limitations include bradycardia and hypotension (again, probably less concerning in patients with short life expectancy), and expense. The oral alpha-2 agonists clonidine and guanfacine have also shown efficacy for agitation in hyperactive delirium in general populations, and they may be reasonable for the palliative care setting as well.^61,62 Of the two agents, guanfacine has the advantage of producing fewer cardiovascular adverse effects.⁶³ Palliative sedation is often considered the option of last resort for refractory and severely distressing agitated terminal delirium. Agitated delirium is the most common indication for palliative sedation^64,65; the most commonly used agents are midazolam and propofol, and mean survival is generally less than 24 hours.⁶⁶ With proper management, palliative sedation should not often be necessary.

The vast majority of dying patients develop a delirium in the last days of life, and an unfortunate few develop highly distressing and frightening hyperactivity and agitation. A systematic approach to evaluation and management of terminal delirium is an essential component in ensuring a peaceful death for the patient and their family.

Shixie Jiang, MD¹ • Jonathan T. Stewart, MD, DLFAPA, AGSF^1,2

Affiliations:
¹University of South Florida College of Medicine, Tampa, FL
²James A Haley VA Hospital, University of South Florida College of Medicine, Tampa, FL

Disclosures:
The authors report no relevant financial relationships.

Address correspondence to:
Shixie Jiang, MD
401 Quarry Road Stanford, CA 94304
Email: jiangs@stanford.edu