OVERVIEW

Delirium has likely replaced syphilis as “the great imitator,” as its varied presentations have led to misdiagnoses among almost every major category of mental illness. A syndrome, caused by an underlying physiological disturbance (and marked by a fluctuating course with impairments in consciousness, attention, and perception), delirium is often mistaken for depression (with a “withdrawn” or “flat” affect), for mania (with agitation and confusion), for psychosis (with hallucinations and paranoia), for anxiety (with restlessness and hypervigilance), for dementia (with cognitive impairments), and for substance abuse (with impairment in consciousness). With so diverse an array of symptoms, delirium assumes a position of diagnostic privilege in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), in that almost no other diagnosis can be made in its presence.

Perhaps even more noteworthy is that delirium is a signifier of (often serious) somatic illness.¹ Delirium has been associated with increased length of stay in hospitals² and with an increased cost of care.^3,4 Among intensive care unit (ICU) patients prospective studies have noted that delirium occurs in 31% of admissions⁵; when intubation and mechanical ventilation are required the incidence soars to 81.7%.¹

Sometimes, delirium is referred to as an acute confusional state, a toxic-metabolic encephalopathy, or acute brain failure; unquestionably, it is the most frequent cause of agitation in the general hospital. Delirium ranks second only to depression on the list of all psychiatric consultation requests. Given its prevalence and its importance (morbidity and mortality), the American Psychiatric Association issued practice guidelines for the treatment of delirium in 1999.⁶

Placed in this context, the consequences of misdiagnosis of delirium can be severe; prompt and accurate recognition of this syndrome is paramount for all clinicians.

DIAGNOSIS

The essential feature of delirium, according to the DSM-IV, is a disturbance of consciousness that is accompanied by cognitive deficits that cannot be accounted for by past or evolving dementia (Table 18-1).⁷ The ICD-10 includes disturbances in psychomotor activity, sleep, and emotion in its diagnostic guidelines (Table 18-2).⁸

Table 18-1 DSM-IV Diagnostic Criteria for Delirium

Table 18-2 ICD-10 Diagnostic Guidelines for Delirium

Disturbance of the sleep-wake cycle is also common, sometimes with nocturnal worsening (“sundowning”) or even by a complete reversal of the night-day cycle, though it should be emphasized that (despite previous postulation) sleep disturbance alone does not cause delirium.⁹ Similarly, the term “ICU psychosis” has entered the medical lexicon; this is an unfortunate misnomer because it is predicated on the beliefs that the environment of the ICU alone is capable of inducing delirium and that the symptomatology of delirium is limited to psychosis.⁹

Despite wide variation in the presentation of the delirious patient, the hallmarks of delirium (although perhaps less immediately apparent) remain quite consistent from case to case. Both Chedru and Geschwind¹⁰ and Mesulam and co-workers¹¹ regard impaired attention as the main deficit of delirium. This inattention (along with an acute onset, waxing and waning course, and overall disturbance of consciousness) forms the core of delirium, while other related symptoms (e.g., withdrawn affect, agitation, hallucinations, and paranoia) serve as a “frame” that can sometimes be so prominent as to detract from the picture itself.

Psychotic symptoms (such as visual or auditory hallucinations and delusions) are common among patients with delirium.¹² Sometimes the psychiatric symptoms are so bizarre or so offensive (e.g., an enraged and paranoid patient shouts that pornographic movies are being made in the ICU) that diagnostic efforts are distracted. The hypoglycemia of a man with diabetes can be missed in the emergency department (ED) if the accompanying behavior is threatening, uncooperative, and resembling that of an intoxicated person.

While agitation may distract practitioners from making an accurate diagnosis of delirium, disruptive behavior alone will almost certainly garner some attention. The “hypoactive” presentation of delirium is more insidious, since the patient is often thought to be depressed or anxious because of his or her medical illness. Studies of quietly delirious patients show the experience to be equally as disturbing as the agitated variant¹³; quiet delirium is still a harbinger of serious medical pathology.^14,15

The core similarities found in cases of delirium have led to postulation of a final common neurological pathway for its symptoms. Current understanding of the neurophysiological basis of delirium is one of hyperdopaminergia and hypocholinergia.¹⁶ The ascending reticular activating system (RAS) and its bilateral thalamic projections regulate alertness, with neocortical and limbic inputs to this system controlling attention. Since acetylcholine is the primary neurotransmitter of the RAS, medications with anticholinergic activity can interfere with its function, resulting in the deficits in alertness and attention that are the heralds of delirium. Similarly, it is thought that loss of cholinergic neuronal activity in the elderly (e.g., due to microvascular disease or atrophy) is the basis for their heightened risk of delirium. Release of endogenous dopamine due to oxidative stress is thought to be responsible for the perceptual disturbances and paranoia that so often lead to the delirious patient being mislabeled as “psychotic.” As we shall discuss later, both cholinergic agents (e.g., physostigmine) and dopamine blockers (e.g., haloperidol) have proven efficacious in the management of delirium.

DIFFERENTIAL DIAGNOSIS

Treatment relies on a careful diagnostic evaluation; there is no substitute for a systematic search for the specific cause of delirium. The temporal relationship to clinical events often gives the best clues to potential causes. For example, a patient who extubated himself was almost certainly in trouble before self-extubation. When did his mental state actually change? Nursing notes should be studied to help discern the first indication of an abnormality (e.g., restlessness, mild confusion, or anxiety). If a time of onset can be established as a marker, other events can be examined for a possible causal relationship to the change in mental state. Initiation or discontinuation of a drug, the onset of fever or hypotension, or the acute worsening of renal function, if in proximity to the time of mental status changes, become more likely culprits.

Without a convincing temporal connection, the cause of delirium may be discovered by its likelihood in the unique clinical situation of the patient. In critical care settings, as in EDs, there are several (life-threatening) states that the clinician can consider routinely. These are states in which intervention needs to be especially prompt because failure to make the diagnosis may result in permanent central nervous system (CNS) damage. These conditions are (1) Wernicke’s disease; (2) hypoxia; (3) hypoglycemia; (4) hypertensive encephalopathy; (5) hyperthermia or hypothermia; (6) intracerebral hemorrhage; (7) meningitis/encephalitis; (8) poisoning (whether exogenous or iatrogenic); and (9) status epilepticus. These conditions are usefully recalled by the mnemonic device WHHHHIMPS (Table 18-3). Other, less urgent but still acute conditions that require interven-tion include subdural hematoma, septicemia, subacute bacterial endocarditis, hepatic or renal failure, thyrotoxicosis/myxedema, delirium tremens, anticholinergic psychosis, and complex partial seizures. If not already ruled out, when present, these conditions are easy to verify. A broad review of conditions frequently associated with delirium is provided by the mnemonic I WATCH DEATH (Table 18-4).

Table 18-3 Life-Threatening Causes of Delirium as Recalled by Use of the Mnemonic WHHHHIMPS

Table 18-4 Conditions Frequently Associated with Delirium, Recalled by the Mnemonic I WATCH DEATH

Bacteremia commonly clouds an individual’s mental state. In prospectively studied seriously ill hospitalized patients, delirium was commonly correlated with bacteremia.¹⁷ In that study, the mortality rate of septic patients with delirium was higher than in septic patients with a normal mental status. In the elderly, regardless of the setting, the onset of confusion should trigger concern about infection. Urinary tract infections (UTIs) and pneumonias are among the most common infections in older patients, and when bacteremia is associated with a UTI, confusion is the presenting feature nearly one-third (30%) of the time.¹⁸ Once a consultant has eliminated these basic conditions as possible causes of a patient’s disturbed brain function, there is time enough for a more systematic approach to the differential diagnosis. A comprehensive list of differential diagnoses, similar to the one compiled by Ludwig¹⁹ (slightly expanded in Table 18-5), is recommended. A quick review of this list is warranted even when the consultant is relatively sure of the diagnosis.

Table 18-5 Differential Diagnoses of Delirium

Modified from Ludwig AM: Principles of clinical psychiatry, New York, 1980, Free Press.

To understand the acute reaction of the individual patient, one should begin by completely reviewing the medical record. Vital signs may reveal periods of hypotension or fever. The highest temperature recorded will also be key. Operative procedures and the use of anesthetics may also induce a sustained period of hypotension or reveal unusually large blood loss and its replacement. Laboratory values should be scanned for abnormalities that could be related to an encephalopathic state.

The old chart, no matter how thick, cannot be overlooked without risk. Some patients have had psychiatric consultations for similar difficulties on prior admissions. Others, in the absence of psychiatric consultations, have caused considerable trouble for their caregivers. Similar to a patient’s psychiatric history, the family psychiatric history can help make a diagnosis, especially if a major mood or anxiety disorder, alcoholism, schizophrenia, or epilepsy is present.

Examination of current and past medications is essential because pharmacological agents (in therapeutic doses, in overdose, or with withdrawal) can produce psychiatric symptoms. Moreover, these considerations must be routinely reviewed, especially in patients whose drugs have been stopped because of surgery or hospitalization or whose drug orders have not been transmitted during transfer between services. Of all causes of an altered mental status, use of drugs and withdrawal from drugs are probably the most common. Some, such as lidocaine, are quite predictable in their ability to cause encephalopathy; the relationship is dose related. Other agents, such as antibiotics, usually cause delirium only in someone whose brain is already vulnerable, as in a patient with a low seizure threshold.²⁰ Table 18-6 lists some drugs used in clinical practice that have been associated with delirium.

Table 18-6 Drugs Used in Clinical Practice That Have Been Associated with Delirium

ACTH, Adrenocorticotropic hormone; GABA, γ-aminobutyric acid.

Adapted from Cassem NH, Lake CR, Boyer WF: Psychopharmacology in the ICU. In Chernow B, editor: The pharmacologic approach to the critically ill patient, Baltimore, 1995, Williams & Wilkins; and from Drugs that may cause psychiatric symptoms, Med Lett Drugs Ther 44:59-62, 2002.

The number of drugs that can be involved either directly or indirectly (e.g., because of drug interactions) is numerous. Fortunately, certain sources provide regular reviews of published summaries and drug updates.²¹ Although physicians are usually aware of these hazards, a common drug, such as meperidine, when used in doses above 300 mg/day for several days, causes CNS symptoms because of the accumulation of its excitatory metabolite, normeperidine, which has a half-life of 30 hours and causes myoclonus (the best clue of normeperidine toxicity), anxiety, and ultimately seizures.²² The usual treatment is to stop the offending drug or to reduce the dose; however, at times this is not possible. Elderly patients and those with mental retardation or a history of significant head injury are more susceptible to the toxic actions of many of these drugs.

Psychiatric symptoms in medical illness can have other causes. Besides the abnormalities that may arise from the effect of the patient’s medical illness (or its treatment) on the CNS (e.g., the abnormalities produced by systemic lupus erythematosus or use of high-dose steroids), the disturbance may be the effect of the medical illness on the patient’s mind (the subjective CNS), as in the patient who thinks he is “washed up” after a myocardial infarction, quits, and withdraws into hopelessness. The disturbance may also arise from the mind, as a conversion symptom or as malingering about pain to get more narcotics. Finally, the abnormality may be the result of interactions between the sick patient and his or her environment or family (e.g., the patient who is without complaints until his family arrives, at which time he promptly looks acutely distressed and begins to whimper continuously). Nurses are commonly aware of these sorts of abnormalities, although they may go undocumented in the medical record.

EXAMINATION OF THE PATIENT

Appearance, level of consciousness, thought, speech, orientation, memory, mood, judgment, and behavior should all be assessed. In the formal mental status examination (MSE), one begins with the examination of consciousness. If the patient does not speak, a handy commonsense test is to ask oneself, “Do the eyes look back at me?” One could formally rate consciousness by using the Glasgow Coma Scale (see Chapter 81, Table 81-3), a measure that is readily understood by consultees in other specialties.²³

If the patient can cooperate with an examination, attention should be examined first because if this is disturbed, other parts of the examination may be invalid. One can ask the patient to repeat those letters of the alphabet that rhyme with “tree.” (If the patient is intubated, ask that a hand or finger be raised whenever the letter of the recited alphabet rhymes with tree.) Then the rest of the MSE can be performed. The Folstein Mini-Mental State Examination (MMSE),²⁴ which is presented in Chapter 2, Table 2-8, is usually included. Specific defects are more important than is the total score. Other functions (such as writing, which Chedru and Geschwind¹⁰ considered to be one of the most sensitive indicators of impairment of consciousness) are often abnormal in delirium. Perhaps the most dramatic (though difficult to score objectively) test of cognition is the clock-drawing test, which can provide a broad survey of the patient’s cognitive state (Figure 18-1).²⁵

Figure 18-1 The clock-drawing test. The patient is provided with a circular outline and asked to draw the numbers as they appear on the face of a clock. Once that is complete, the patient is asked to set the hands to a particular time (often “ten past” the hour to test if the patient can suppress the impulse to include the number 10). Example A demonstrates good planning and use of space. Example B features some impulsiveness as the numbers are drawn without regard for actual location, and the time “ten past four” is represented by hands pointing to the digits 10 and 4. Note the perseveration indicated by the extra loops on the digits 3 and 6. Impulsiveness and perseveration are indicative of frontal lobe dysfunction. Example C demonstrates gross disorganization, although the patient took several minutes to draw the clock, and believed it to be a good representation.

The patient’s problem may involve serious neurological syndromes as well; however, the clinical presentation of the patient should direct the examination. In general, the less responsive and more impaired the patient is, the more one should look for “hard” signs. A directed search for an abnormality of the eyes and pupils, nuchal rigidity, hyperreflexia (withdrawal), “hung-up” reflexes (myxedema), one-sided weakness or asymmetry, gait (normal pressure hydrocephalus), Babinski’s reflexes, tetany, absent vibratory and position senses, hyperventilation (acidosis, hypoxia, or pontine disease), or other specific clues can help to verify or refute hypotheses about causality that are stimulated by the abnormalities in the examination.

Frontal lobe function deserves specific attention. Grasp, snout, palmomental, suck, and glabellar responses are helpful when present. Hand movements thought to be related to the premotor area (Brodmann’s area 8) can identify subtle deficiencies. The patient is asked to imitate, with each hand separately, specific movements. The hand is held upright, a circle formed by thumb and first finger (“okay” sign), then the fist is closed and lowered to the surface on which the elbow rests (Figure 18-2). In the Luria sequence (Figure 18-3), one hand is brought down on a surface (a table or one’s own leg) in three successive positions: extended with all five digits parallel (“cut”), then as a fist, and then flat on the surface (“slap”). Finally, both hands are placed on a flat surface in front of the patient, one flat on the surface, the other resting as a fist. Then the positions are alternated between right and left hands, and the patient is instructed to do likewise.

Figure 18-2 Testing of the premotor area.

Figure 18-3 The Luria sequence.

For verbally responsive patients, their response to the “Frank Jones story” can be gauged (i.e., “I have a friend, Frank Jones, whose feet are so big he has to put his pants on over his head. How does that strike you?”). Three general responses are given. Type 1 is normal: The patient sees the incongruity and smiles (a limbic response) and can explain (a neocortical function) why it cannot be done. Type 2 is abnormal: The patient smiles at the incongruity (a limbic connection), but cannot explain why it cannot be done. Type 3 is abnormal: The patient neither gets the incongruity nor can explain its impossibility.

Laboratory studies should be carefully reviewed, with special attention paid to indicators of infection or metabolic disturbance. Toxicological screens are also frequently helpful in allowing the inclusion or exclusion of substance intoxication or withdrawal from the differential. Neuroimaging may prove useful in the detection of intracranial processes that can result in altered mental status. Of all the diagnostic studies available, the electroencephalogram (EEG) may be the most useful tool in the diagnosis of delirium. Engel and Romano²⁶ reported in 1959 their (now classic) findings on the EEG in delirium, namely, generalized slowing to the theta-delta range in the delirious patient, the consistency of this finding despite wide-ranging underlying conditions, and resolution of this slowing with effective treatment of the delirium. EEG findings may even clarify the etiology of a delirium, since delirium tremens is associated with low-voltage fast activity superimposed on slow waves, sedative-hypnotic toxicity produces fast beta activity (>12 Hz), and hepatic encephalopathy is classically associated with triphasic waves.²⁷

SPECIFIC MANAGEMENT STRATEGIES FOR DELIRIUM

Thoughts about (and treatment of) delirium have changed dramatically in the past three decades. Thirty years ago, atropine was routinely administered to newly admitted coronary care unit (CCU) patients with bradycardia. Some patients, particularly older ones with preexisting organic brain disease, developed delirium. For such patients, parenteral propantheline bromide (Pro-Banthine), a quaternary ammonium compound that does not cross the blood-brain barrier and is equally effective in treating bradycardia, was substituted. This approach may still be taken, but problems are seldom so simple. Often the drugs that cause delirium (such as lidocaine or prednisone) cannot be changed without causing harm to the patient. Alternatively, pain may cause agitation in a delirious patient. Morphine sulfate can relieve pain but may unfortunately lead to decreases in blood pressure and respiratory rate.

Psychosocial or environmental measures are rarely effective in the treatment of a bona fide delirium of uncertain or unknown cause. Nevertheless, it is commendable to have hospital rooms with windows, calendars, clocks, and a few mementos from home on the walls²⁸; soft and low lighting at night helps “sundowners”; and, most of all, a loving family in attendance reassures and reorients the patient. The psychiatric consultant is often summoned because psychosocial measures have failed to prevent or to treat the patient’s delirium. Restraints (e.g., Poseys, geriatric chairs, “vests,” helmets, and locked leather restraints for application to one or more extremities, chest, and [even] head) are also available and quite useful to protect patients from inflicting harm on themselves or staff. One or several of these is often in place when the consultant arrives. One hoped-for outcome of the consultation is that the use of these devices can be reduced or eliminated. The unfortunate misnomer chemical restraint is frequently applied to the most helpful class of drugs for delirium (i.e., neuroleptics). However, physicians do not use chemical restraints (i.e., tear gas, pepper spray, mace, or nerve gas) in the treatment of agitated patients.

When the cause of the delirium seems straightforward, the treatment revolves around resolution or reversal of the underlying cause. A discovered deficiency can be replaced (e.g., of blood, oxygen, thiamine, vitamin B₁₂, synthroid, or glucose). Pathological conditions can be treated (e.g., volume replacement for hypotension, diuretics for pulmonary edema, antibiotics for infection, calcium administration for hypocalcemia, or dialysis for acute lithium toxicity). Implicated drugs (such as meperidine and cimetidine) can be stopped or reduced.

Specific antidotes can reverse the delirium caused by some drugs. Flumazenil and naloxone reverse the effects of benzodiazepines and opioid analgesics, respectively. However, caution is required because use of flumazenil may precipi-tate seizures in a benzodiazepine-dependent patient, and naloxone can also precipitate narcotic withdrawal in a narcotic-dependent patient.

Anticholinergic delirium can be reversed by intravenous (IV) physostigmine (in doses starting at 0.5 to 2 mg). Caution is essential with use of this agent as the autonomic nervous system of the medically ill is generally less stable than it is in a healthy patient who has developed an anticholinergic delirium as a result of a voluntary or accidental overdose. Moreover, if there is a reasonably high amount of an anticholinergic drug on board that is clearing from the system slowly, the therapeutic effect of physostigmine, although sometimes quite dramatic, is usually short-lived. The cholinergic reaction to intravenously administered physostigmine can cause profound bradycardia and hypotension, thereby multiplying the complications.^29,30 It should also be noted that a continuous IV infusion of physostigmine has been successfully used to manage a case of anticholinergic poisoning.³¹ Because of the diagnostic value of physostigmine, one may wish to use it even though its effects will be short-lived. If one uses an IV injection of 1 mg of physostigmine, protection against excessive cholinergic reaction can be provided by preceding this injection with an IV injection of 0.2 mg of glycopyrrolate. This anticholinergic agent does not cross the blood-brain barrier and should protect the patient from the peripheral cholinergic actions of physostigmine.

DRUG TREATMENT

All too often, the cause of delirium is not readily identified or treated. These situations call for management of the symptoms of delirium while a more specific and effective treatment can be initiated. Opioids, benzodiazepines, neuroleptics, barbiturates, neuromuscular blocking agents, inhalant anesthetics, and assorted other agents (such as propofol, ketamine, isoflurane, chloral hydrate, and clonidine) are available (alone or in creative combinations).

Benzodiazepines (e.g., 2.5 mg of IV diazepam or 0.5 to 1 mg of midazolam) are often effective in mild agitation in the setting of withdrawal from drugs that work at the alcohol, benzodiazepine, and barbiturate receptor. Morphine is also used frequently as it calms agitation and is easily reversed if hypotension or respiratory depression ensues. Especially in higher doses, these agents can cause or exacerbate confusion in older patients. This occurs much less often with neuroleptics, unless thioridazine (which has potent anticholinergic properties) is used.

Neuroleptics are the agent of choice for delirium. Haloperidol is probably the antipsychotic most commonly used to treat agitated delirium in the critical care setting; its effects on blood pressure, pulmonary artery pressure, heart rate, and respiration are milder than those of the benzodiazepines, making it an excellent agent for delirious patients with impaired cardiorespiratory status.³²

Although haloperidol can be administered orally or parenterally, acute delirium with extreme agitation typically requires use of parenteral medication. IV administration is preferable to intramuscular (IM) administration because drug absorption may be poor in distal muscles if delirium is associated with circulatory compromise or with borderline shock. The deltoid is probably a better IM injection site than the gluteus muscle, but neither is as reliable as is the IV route. Second, as the agitated patient is commonly paranoid, repeated painful IM injections may increase the patient’s sense of being attacked by enemy forces. Third, IM injections can complicate interpretations of muscle enzyme studies if enzyme fractionation is not readily available. Fourth, and most important, haloperidol is less likely to produce extrapyramidal symptoms (EPS) when given IV than when given IM or by mouth (PO), at least for patients without a prior serious psychiatric disorder.³³

In contrast to the immediately observable sedation produced by IV benzodiazepines, IV haloperidol has a mean distribution time of 11 minutes in normal volunteers³⁴; this may be even longer in critically ill patients. The mean half-life of IV haloperidol’s subsequent, slower phase is 14 hours. This is still a more rapid metabolic rate than the overall mean half-lives of 21 and 24 hours for IM and PO doses. The PO dose has about half the potency of the parenteral dose, so 10 mg of PO haloperidol corresponds to 5 mg given IV or IM.

Haloperidol has not been approved by the Food and Drug Administration (FDA) for IV administration. However, any approved drug can be used for a nonapproved indication or route if justified as “innovative therapy.” For critical care units desirous of using IV haloperidol, one approach is to present this to the hospital’s institutional review board (IRB) or human studies committee with a request to use the drug with careful monitoring of results based on the fact that it is the drug of choice for the patient’s welfare, it is the safest drug available for this purpose, and it is justifiable as innovative therapy. After a period of monitoring, the committee can choose to make the use of the drug routine in that particular hospital.

In Europe, IV haloperidol has been used to treat delirium tremens and acute psychosis and to premedicate patients scheduled for electroconvulsive therapy (ECT). Its use has been associated with few adverse effects on blood pressure, heart rate, respiratory rate, or urinary output and has been linked with few EPS. The reason for the latter is not known. Studies of the use of IV haloperidol in psychiatric patients have not shown that these side effects were fewer. The reason for their rare appearance after IV administration in medically ill patients may be due to the fact that many of the medically ill patients have other medications in their system, especially benzodiazepines (that are protective against EPS) or that patients with psychiatric disorders are more susceptible to EPS.³³

Before administration of IV haloperidol, the IV line should be flushed with 2 ml of normal saline. Phenytoin precipitates with haloperidol, and mixing the two in the same line must be avoided. Occasionally, haloperidol may also precipitate with heparin, and because many lines in critical care units are heparinized, the 2 ml flush is advised. The initial bolus dose of haloperidol usually varies from 0.5 to 20 mg; usually 0.5 mg (for an elderly person) to 2 mg is used for mild agitation, 5 mg is used for moderate agitation, and 10 mg for severe agitation. The only time when a higher initial dose should be used is when the patient has already been unsuccessfully treated with reasonable doses of haloperidol. To adjust for haloperidol’s lag time, doses are usually staggered by at least a 30-minute interval. If one dose (e.g., a 5 mg dose) fails to calm an agitated patient after 30 minutes, the next higher dose, 10 mg, should be administered. Calm is the desired outcome. Partial control of agitation is usually inadequate, and settling for this only prolongs the delirium or guarantees that even higher total doses of haloperidol will be used over the prolonged course.

Haloperidol can be combined every 30 minutes with simultaneous parenteral lorazepam doses (starting with 1 to 2 mg). Because the effects of lorazepam are noticeable within 5 to 10 minutes, each dose can precede the haloperidol dose, observed for its impact on agitation, and increased if it is more effective. Some believe that the combination leads to a lower overall dose of each drug.³⁵

After calm is achieved, agitation should be the sign for a repeat dose. Ideally, the total dose of haloperidol on the second day should be a fraction of that used on day 1. After complete lucidity has been achieved, the patient needs to be protected from delirium only at night, by small doses of haloperidol (1 to 3 mg), which can be given PO. As in the treatment of delirium tremens, the consultant is advised to stop the agitation quickly and completely at the outset rather than barely keep up with it over several days. The maximum total dose of IV haloperidol to be used as an upper limit has not been established, although IV administration of single bolus doses of 200 mg have been used, and up to 1,600 mg has been used in a 24-hour period.³⁶ The highest requirements have been seen with delirious patients on the intraaortic balloon pump.³⁷ A continuous infusion of haloperidol has also been used to treat severe, refractory delirium.³⁸

When delirium does not respond and agitation is unabated, one may wonder if the neuroleptic (e.g., haloperidol) is producing akathisia. The best indication as to whether the treatment is causing agitation is the patient’s description of an irresistible urge to move—usually the limbs, lower more often than upper. If dialogue is possible, even nodding yes or no (provided that the patient understands the question) can confirm or exclude this symptom. If the patient cannot communicate, limited options remain: to decrease the dose or to increase it and judge by the response. In our experience, it is far more common for the patient to receive more haloperidol and to improve.

Hypotensive episodes following the administration of IV haloperidol are rare and almost invariably result from hypovolemia. Ordinarily, this is easily checked in ICU patients who have indwelling pulmonary artery catheters, but because agitation is likely to return, volume replacement is necessary before one administers further doses. Local caustic effects on veins do not arise. IV haloperidol is generally safe for epileptics and for patients with head trauma, unless psychotropic drugs are contraindicated because the patient needs careful neurological monitoring. Although IV haloperidol may be used without mishap in patients receiving epinephrine drips, after large doses of haloperidol a pressor other than epinephrine (e.g., norepinephrine) should be used to avoid unopposed β-adrenergic activity. IV haloperidol does not block a dopamine-mediated increase in renal blood flow. It also appears to be the safest agent for patients with chronic obstructive pulmonary disease.

Several centers have discovered that use of IV haloperidol is associated with the development of torsades de pointes (TDP).^39–43 The reasons for this are unclear, although particular caution is urged when levels of potassium and magnesium are low, when a prolonged QT interval is noted, when hepatic compromise is present, or when a specific cardiac abnormality (e.g., mitral valve prolapse or a dilated ventricle) exists. Progressive QT widening after haloperidol administration should alert one to the danger, however infrequent it may be in practice (4 of 1,100 cases in one unit).⁴⁰ Delirious patients who are candidates for IV haloperidol require care-ful screening. Serum potassium and magnesium should be within normal range and a baseline electrocardiogram (ECG) checked for the pretreatment QT interval corrected for heart rate (QTc) interval. QT interval prolongation occurs in some patients with alcoholic liver disease; this finding is associated with adverse outcomes (e.g., sudden cardiac death).⁴⁴

In the past, IV droperidol was recommended as an alternative agent to haloperidol in part because it was approved for IV administration by the FDA. Unfortunately, its label was revised to include a black box warning for QT prolongation leading to TDP and death.⁴⁵ This has limited the frequency of its use.

Other available parenteral first-generation neuroleptics for treatment of agitation are perphenazine, thiothixene, trifluoperazine, fluphenazine, and chlorpromazine. Perphenazine is approved for IV use as an antiemetic. Chlorpromazine is extremely effective, but its potent α-blocking properties can be dangerous for critically ill patients. When administered IV or IM, it can abruptly decrease total peripheral resistance and cause a precipitous fall in cardiac output. Nevertheless, used IV in small doses (10 mg) it can be both safe and effective in the treatment of delirium.

The availability of injectable formulations of both olanzapine and ziprasidone has prompted a growing interest in the use of the second-generation antipsychotics in the management of delirium.⁴⁶ Risperidone has the most available data supporting its use, with multiple studies showing it to be efficacious and safe for the treatment of delirium,^47–49 and one small randomized double-blind comparative study finding no significant difference in efficacy when compared with haloperidol.⁵⁰ The other members of this class (olanzapine, ziprasidone, quetiapine, clozapine, and aripiprazole) have far less supporting data, though some small studies seem to indicate some promise for management of delirium.^51–55 Agranulocytosis associated with clozapine and the resultant regulation of its use effectively eliminates any routine application of it in the management of delirium. All drugs in this class feature an FDA black box warning indicating an increased risk of death when used to treat behavioral problems in elderly patients with dementia. Similar warnings regarding a potential increased risk of cerebrovascular events are reported for risperidone, olanzapine, and aripiprazole. With decades of clinical experience in the use of haloperidol, and a dearth of available data on these newer agents, haloperidol remains the agent of choice for the treatment of delirium.

To date, there are few data to support pharmacological prophylaxis of delirium for the critically ill, although one recent randomized, double-blind, placebo-controlled study examining the preoperative use of haloperidol in elderly patients undergoing hip surgery indicated decreases in the severity and duration of delirium, as well as the length of hospital stay, but no statistically significant decrease in the actual incidence of delirium.⁵⁶ There are also some limited data suggesting that the procholinergic action of cholinesterase inhibitors provides some protection against the development of delirium.^57,58

DELIRIUM IN SPECIFIC DISEASES

Critically ill patients with human immunodeficiency virus (HIV) infection may be more susceptible to the EPS of haloperidol, and to neuroleptic malignant syndrome (NMS),^59–61 leading an experienced group to recommend use of molindone.⁶² The latter is associated with fewer of such effects; while it is available only as an oral agent, it can be prescribed from 5 to 25 mg at appropriate intervals or, in a more acute situation, 25 mg every hour until calm is achieved. Risperidone (0.5 to 1 mg per dose) is another recommended PO agent. If parenteral medication is required, 10 mg of chlorpromazine has been effective. Perphenazine is readily available for parenteral use as well, and 2 mg doses can be used effectively.

Patients with Parkinson’s disease pose a special problem because dopamine blockade aggravates their condition. If PO treatment of delirium or psychosis is possible, clozapine, starting with a small dose of 6.25 or 12.5 mg, is probably the most effective agent available that does not exacerbate the disease. With the risk of agranulocytosis attendant to the use of clozapine, quetiapine may play a valuable role in this population, since its very low affinity for dopamine receptors is less likely to exacerbate this disorder.⁶³

IV benzodiazepines (particularly diazepam, chlordiazepoxide, and lorazepam) are routinely used to treat agitated states, particularly delirium tremens, and alcohol withdrawal.⁶⁴ Neuroleptics have also been used successfully, and both have been combined with clonidine. IV alcohol is also extremely effective in the treatment of alcohol withdrawal states, particularly if the patient does not seem to respond as rapidly as expected to higher doses of benzodiazepines. The inherent disadvantage is that alcohol is toxic to both liver and brain, although its use can be quite safe if these organs do not show already extensive damage (and sometimes quite safe even when they do). Nonetheless, use of IV alcohol should be reserved for extreme cases of alcohol withdrawal when other, less toxic measures have failed. A 5% solution of alcohol mixed with 5% dextrose in water run at 1 ml per minute often achieves calm quickly.

Propofol, a relatively new agent for the sedation of critically ill patients, can also be extremely effective in managing agitation.^65–67 It has both moderate respiratory depressant and vasodilator effects, although hypotension can be minimized by avoidance of boluses of the drug. Impaired hepatic function does not slow metabolic clearance, but clearance does decline with age and its half-life is significantly longer in the elderly. When compared with midazolam as a sedative for postoperative coronary artery bypass graft patients, both drugs, without affecting cardiac output, provided safe and effective sedation and lowered arterial blood pressure and heart rates. The group treated with propofol, however, had significantly lower heart rates during the first 2 hours and significantly lower blood pressures 5 to 10 minutes after the initial dose (mean loading doses for propofol and midazolam were 0.24 mg/kg and 0.012 mg/kg, and maintenance doses 0.76 mg/kg/hr and 0.018 mg/kg/hr). In the propofol group, there was a reduced need for sodium nitroprusside and supplemental opioids were less often required.⁶⁸

In a prospective, randomized, crossover, placebo-controlled trial, propofol’s effect on critically ill patients’ response to chest physiotherapy was studied. Propofol, in an IV bolus dose of 0.75 mg/kg, significantly reduced the hemodynamic and metabolic stresses caused by chest physiotherapy.⁶⁹ This drug’s rapid onset and short duration make it especially useful for treatment of short periods of stress. When rapid return to alertness from sedation for an uncompromised neurological examination is indicated, propofol is a nearly ideal agent⁷⁰; however, its use in the treatment of a prolonged delirious state has specific disadvantages.⁷¹ Delivered as a fat emulsion containing 0.1 g of fat per milliliter, propofol requires a dedicated IV line, and drug accumulation can lead to a fat-overload syndrome that has been associated with overfeeding and with significant CO₂ production, hypertriglyceridemia, ketoacidosis, seizure activity 6 days after discontinuation, and even fatal respiratory failure.^71–73 Obese patients provide a high volume of distribution, and their doses should be calculated using estimated lean, rather than actual, body mass. If the patient is receiving fat by parenteral feeding, this must be accounted for or eliminated and adequate glucose infusion provided to prevent ketoacidosis. Although no clear association has been demonstrated with addiction, tolerance, or withdrawal, doses seem to require escalation after 4 to 7 days’ infusion. Seizures seen after withdrawal and muscular rigidity during administration are poorly understood. The drug is costly when prolonged infusions are used because propofol costs about three to four times as much as equivalent quantities of midazolam.

Dexmedetomidine is a selective α₂-adrenergic agonist used for sedation and analgesia in the ICU setting. Its relative lack of amnestic effect may limit its use in the treatment of the delirious patient due to an increased likelihood of distressing recollections persisting from the period of sedation.⁷⁴

Drug infusions may be more effective and efficient than intermittent bolus dosing because the latter may intensify side effects (such as hypotension), waste time of critical care personnel, and permit more individual error. The contents of the infusion can address simultaneously multiple aspects of a patient’s difficulties in uniquely appropriate ways. The report of the sufentanil, midazolam, and atracurium admixture for a patient who required a temporary biventricular assist device is an excellent example of multiple-drug infusion and creative problem-solving.^75,76

CONCLUSIONS

Of all psychiatric diagnoses, delirium demands the most immediate attention since delay in identification and treatment might allow the progression of serious and irreversible pathophysiological changes. Unfortunately, delirium is all too often underemphasized, misdiagnosed, or altogether missed in the general hospital setting.^77–79 Indeed, it was not until their most recent editions that major medical and surgical texts corrected chapters indicating that delirium was the result of anxiety or depression, rather than an underlying somatic cause that required prompt investigation. In the face of this tradition of misinformation, it often falls to the psychiatric consultant to identify and manage delirium, while alerting and educating others to its significance.