HISTORICAL PERSPECTIVE

Delirium has been described by physicians for over 2000 years. Ancient Greeks described two states for what we now recognize as delirium. Phrenitis (Hippocrates) was characterized by cognitive and behavioral disturbances, sleep disruption, restlessness, and agitation. It was associated with fever. Lethargus was, on the contrary, described as a condition producing symptoms such as listlessness, sleepiness, inertia, memory impairment, and dulling of senses (Lipowski, 1990; Liptzin & Levkoff, 1992). Celsus, a Roman physician, coined, in the first century A.D., the term delirium (delirare: to go out of the furrow) to describe a state characterized by delusions and perceptual disturbances in association with fever. Galen, another Greek physician, practicing in Rome during the second century A.D., also observed that delirium was often associated with medical diseases.

Closer to us, in the 19th and 20th centuries, at least 30 different syndromes termed after what was considered the probable etiology of the disease (psychosis post-influenza, post-acute disease, post fever, etc.) would now be termed delirium (Camus, 2002).

Phenomenologically, two separate patterns of delirium distinguishable by contrasting levels of psychomotor activity have been recognized for over two centuries (Lipowski, 1990), that is, delirium and acute confusion, which have long been considered as two separate entities. Acute confusion or “Torpor” was defined as disorientation with a hypoactive presentation whereas hyperactive disoriented patients were termed delirious. In the 1890s, a French psychiatrist, Philippe Chaslin, gathered them under the unified entity “confusion mentale primitive” described as “an acute brain disorder, consecutive to an organic significant disease, with cognitive impairment associated with delusions, hallucinations, psychomotor agitation, or reciprocally, with psychomotor retardation and inertia” (Chaslin 1892, p. 225; Camus, 2002). Still, delirium and acute confusion have been treated as separate entities until recently and in particular in the neurological literature (Adams & Victor, 1977; Ross et al., 1991). Lipowski (1987) as early as 1983 suggested a similar fusion of those two entities, tailoring the two adjectives hypoactive and hyperactive to qualify the delirium. The formal definition of the two subtypes—a hypoalert–hypoactive type and a hyperalert–hyperactive type—based on alertness was coined by him in 1989 (Lipowski, 1989). In recognition of the fact that patients with transient cognitive impairment (Meagher et al., 2000) have varying motoric profiles, and that hypoactivity does not preclude other features of delirium of high cerebral activity (e.g., perceptual disturbances), he considered all of these patients to have delirium irrespective of their motoric presentations (Lipowski, 1989; Meagher et al., 1998).

THREE DELIRIUM SUBTYPES

Phenomenological findings of different motoric subtypes of delirium reveal that purely hypoactive/somnolent or hyperactive/agitated patients appear to be a minority of cases, with more than 50% of patients experiencing a mixed profile during the course of their illness (Koponen et al., 1989; Liptzin & Levkoff, 1992; Meagher et al., 1996). Lipowski, reviewing the emerging literature on delirium phenomenology, distinguished the three subtypes in 1990. Liptzin and Levkoff (1992), using a symptom checklist (Delirium Symptom Interview [DSI]), attempted to isolate specific symptom patterns for each subtype of delirium. Although the cluster of symptoms associated with one subtype or the other seems to be founded on common sense, with each symptom specifically characterizing hypo- or hyperactive delirium (Table 1), some of them, especially in the hyperactive symptoms' list, for example, uncooperativeness, irritability, distractibility, and nightmares, hardly fit with clinical observation. Nevertheless, although only 86% of the sample met the criteria for a DSM-III delirium, demented patients were not excluded, and cutoff scores were arbitrarily chosen, this study was one of the first to attempt to outline specific criteria based on motoric and psychological behaviors for classification into hypoactive, hyperactive, and mixed categories.

Liptzin and Levkoff's (1992) examples of hypoactive and hyperactive symptoms on the Delirium Symptom Interview (DSI)

PREVALENCE STUDIES

Table 2 summarizes the findings of the main studies evaluating the subtypes of delirium. The definition of three subtypes is used in most studies. Nonetheless, irrespective of the way delirium is diagnosed and the subtypes defined, the hypoactive subtype is the most prevalent one. It is estimated that approximately one half of deliria are of the hypoactive type; the combination of hypoactive and mixed subtypes reach a prevalence of approximately 80%. Hence, the prototypically agitated delirious patient most familiar to clinicians is actually a minority of the deliria that occur (Breitbart et al., 1995). Comparison of the mean values suggest that the mixed subtype is more likely to be diagnosed as hypoactive in those studies using the two subtypes' definitions (purely hyper- or hypoactive).

The findings of the main studies evaluating the subtypes of delirium

LITERATURE CONSENSUS OF THE PHENOMENOLOGICAL DIFFERENCES AND SIMILARITIES OF THE DELIRIUM SUBTYPES

There is a lack of consensus about the subtypes' definitions that limits the comparability of the different studies. Ross et al. (1991) classified patients as somnolent and activated, leaning on the arousal disturbances, using a Visual Analog Scale. O'Keeffe (1999) predefined delirium subtypes, using only two items defining the psychomotor activity (retarded and agitated) of the Delirium Assessment Scale (DAS), a tool based on the definitions of the DSM-III criteria for delirium. Liptzin and Levkoff arbitrarily attributed the 21 items of a Delirium Symptom Interview either to the hyper- or to the hypoactive subtype. As previously stated, the classification of delirium subtypes in the current literature is based either on motoric manifestation or on arousal disturbance. Liptzin and Levkoff (1992) focus his definition on arousal disturbances (hyperactive: hypervigilance; hypoactive: decreased alertness). And so does Ross (1991) (drowsiness vs. hyperalertness). Conversely, Meagher et al. (2000) defines subtypes by considering the motoric aspect as the core feature to be considered. As a result, the motoric aspects are not taken into account in the cluster analysis, but are considered as a matter of fact. For Marcantonio et al. (2002), the definition of the subtypes is based on the memorial delirium assessment scale (MDAS's) items quoting the motoric aspects of the presentation.

Two studies only (Trzepacz & Dew, 1995; Camus et al., 2000) tried to evaluate systematically all the symptoms and the abnormalities generally attributed to delirium, using factor analysis as a tool to isolate clusters of them related to each subtype. Trzepacz and Dew, discussing the factor structure of the first version of Delirium Rating Scale (DRS) isolated two clusters of symptoms including psychomotor and behavioral disturbances, one of them including delusions, psychomotor behavior, cognitive impairment, sleep–wake cycle disturbance, and mood lability, the other one comprising hallucinations, perceptual disturbances, abrupt temporal onset of symptoms, and fluctuation of symptoms. Those clusters, though, do not say much about the different clinical subtypes, as, for example, in the DRS the item “psychomotor disturbance” measures agitation as well as retardation. According to the authors, the second cluster of symptoms seems to be related to the severity of the delirium, suggesting that perceptual disturbances and not delusions are related to the more severe forms of the disorder. However the small size of the sample (n = 20) strongly limits the interpretation of the results.

Camus et al. (2000), on a larger sample (n = 154), isolated through a factor analysis procedure two clusters of symptoms correlated to the hypo- and hyperactive subtypes of delirium, thus supporting the evidence of clinical subtypes. In this study, the different subtypes are distinguished according to arousal disturbances. The hyperactive cluster of symptoms includes logorrhea, agitation, stereotyped activities, increased reactivity, delusions, expansive mood, perceptual disturbances and hallucinations, and mental slowing. The hypoactive cluster of symptoms includes facial inexpressiveness, motor retardation, speech retardation, decreased reactivity, lack of perceptual disturbance, of delusions, of mental acceleration, of decreased awareness. All patients were diagnosed with delirium according to the DSM-IIIR, and delirium symptomatology was evaluated with a delirium clinical symptom checklist rating 19 symptoms of different clinical dimension: perception of self and environment, mental and motor functioning, psychopathology, neurovegetative disturbances.

In those studies, clusters of symptoms have been identified for the hypo- and hyperactive subtypes of delirium, but not for the mixed one. There has been no study reporting on the stability of the motoric subtypes of delirium, although there exists a growing awareness of the relative instability of the clinical presentation. Overall, the definition of the hypoactive subtype of delirium, which is also defined with fewer symptoms, is more consensual. Lipowski (1990), Liptzin (1992), O'Keeffe (1999), Camus (1990, 1992, 2000) find a certain agreement in defining the hypoactive type as at least having symptoms such as motor retardation, speech slowing, and decreased arousal. In the clinical experience, patients clearly hypoaroused often show concomitant symptoms of agitation and restlessness. Would they be considered hypo- or hyperactive? Standing on the DSM-IV definition, we wonder if it would not be more sensible to diagnose those patients as hypoactive, basing the definition of subtypes on a key pathophysiologic abnormality such as arousal disturbance rather than motoric aspects.

PREVALENCE OF PSYCHOTIC SYMPTOMS IN THE HYPOACTIVE TYPE OF DELIRIUM

It is traditionally accepted that hallucinations and delusions are less frequent in the hypoactive type of delirium. Ross et al. (1991) revealed that hyperactive patients with delirium had significantly more hallucinations and delusions than the hypoactive type. Table 3 compares his results with a more recent observation: Breitbart et al. (in press) found Ross's observation in contradiction with their own. A sample of 154 patients diagnosed with a delirium according to the DSM-IV were administered the Memorial Delirium Assessment Scale in which items 7 and 8 investigate perceptual disturbances and delusions. The difference of both items remains statistically significant between the subtypes, but to a far lesser extent, suggesting that those psychotic features are more frequent, although harder to identify, than previously proposed in the hypoactive type of delirium. Those symptoms have then to be further investigated and treated even in somnolent patients because it has been demonstrated that perceptual disturbances and delusions are significantly associated with delirium-related patient distress (Breitbart et al., 2002).

Comparing Ross et al. (1991) and Breitbart results

ETIOLOGY

A number of studies have attempted to link the etiology of delirium to its phenomenological expression (Lipowski, 1989; Ross et al., 1991). Some of them indicated little relationship between symptoms pattern and etiology (Camus et al., 2000). However other authors have felt that there is some variation in phenomenology dependent upon the etiology. In particular, it has been suggested on the basis of clinical observation that delirium due to alcohol or sedative-hypnotic withdrawal is more likely to be of the hyperalert–hyperactive type, whereas metabolic encephalopathy is more likely to be of the hypoalert–hypoactive type. To further complicate this intricate issue it has to be mentioned that only a small percentage of delirious patients have a single etiologic factor; the condition seems to be essentially multifactorial. It has nevertheless to be mentioned that most of those etiologies are putative and that the etiologies of up to 75% of the cases remain unknown (Stiefel et al., 1992). For example, a recent study (Camus et al., 2000) showed that 16% of the sample had one etiologic factor, 27% had two, and 90% had up to four possible etiologies. The same study showed that the most frequent etiologies were drug toxicity (56%), acute cardiovascular disease (48%), and acute metabolic and endocrine disease (43%). However the study failed to correlate phenomenology of delirium and etiology. The most frequently observed etiologies of delirium are infection, drug intoxication and withdrawal, brain injury, low perfusion state, and metabolic disturbances. The hypoactive type of delirium is, in several studies, correlated with hypoxia (O'Keeffe, 1999), metabolic disturbances (Ross et al., 1991), and conditions triggering anticholinergic mechanisms (Meagher et al., 2000). The hyperactive type is correlated with alcohol and drug withdrawal (Ross et al., 1991; Meagher et al., 2000) and drug intoxication or side effect (O'Keeffe, 1999). O'Keeffe noted that there was no clear link between infection, the most common aetiological factor in his study, and a particular subtype, even though alcohol withdrawal and metabolic disturbances led, respectively, as expected, to hyper- and hypoactive deliria.

PATHOPHYSIOLOGY

There have been few attempts to investigate the relationship between the presentation of delirium and putative cerebral lesions, neuro-imaging, EEG, or neurotransmitters' activity. There is some evidence indicating that the more widespread cerebral lesions (i.e., mixed cortical/subcortical as compared to cortical), the higher the incidence of delirium (Robertsson et al., 1998). Hyperactive delirium has been linked with lesions of the cingulated gyrus, the orbitofrontal cortex, hippocampus, parahippocampus, fusiform cortex, and lingual gyri (Medina et al., 1974). Orbitofrontal lesions are known to cause restlessness, hyperactivity, and disinhibition. Lesions of the right cerebral hemisphere have been linked with agitated delirium whereas left cerebral lesions have been linked with somnolent delirium (Cutting, 1992). Mesial frontal/anterior cingulated lesions are known to cause abulia and hypoactivity (Meagher et al., 2000). More recently, advances in neuropsychiatry have revealed differences between brain regions, including the hemispheres, which may underlie the constellation of symptoms among different psychiatric disorders. For example, different neural pathways are involved in major depression and obsessive–compulsive disorder, including lateralization to one or the other hemisphere. Trzepacz (1999) has suggested in her recent review that “delirium, too, involves particular neural pathways and that lateralization to the right may be relevant. Structural and functional neuroimaging reports and recent neuropsychological studies support this lateralization. Prefrontal cortices, anterior and right thalamus, and right basilar mesial temporoparietal cortex may play a significant role in subserving delirium symptoms and may be the ‘final common pathway’ for delirium from a variety of etiologies. The final common pathway may be responsible for certain ‘core symptoms’ (disorientation, cognitive deficits, sleep–wake cycle disturbance, disorganized thinking, and language abnormalities), while other symptoms (delusions, hallucinations, illusions, and affective lability) may occur depending on the etiology causing delirium.”

There is little evidence of specific electroencephalographic patterns matching delirium subtypes. Those who have attempted such examination, first of all Engel and Romano (1959) seem to find a general slowing of the EEG, whatever the subtype of delirium (Ross et al., 1991; Breitbart, 2001). However, varying illnesses may account for some differences. The hypoactive delirium of hepatic encephalopathy shows a clear slowing of the EEG (Van der Rijt & Schalm, 1985). Nonetheless, it has been suggested that in alcohol or sedative-hypnotic withdrawal delirium, commonly observed as hyperactive, it is more common to find a low-voltage fast EEG (Pro & Wells, 1977). Neuroimaging has been used in a couple of studies to ascertain the cerebral metabolic activity in specific delirious states. Thus hyperactive alcohol-withdrawal deliria, even if patients have normal or even increased blood flow (Hemmingsen et al., 1988), also seem to have a specific single photon emission computed tomography presentation (Trzepacz, 1994). By contrast, patients with hepatic encephalopathy show a lower global cerebral metabolic activity (Posner & Plum, 1960).

The brain pathways involved in the maintenance of normal arousal and sleep–wake cycles are only partly known; those of altered states of arousal are less understood. Several neurotransmitters' imbalances have been implicated in delirium. Anticholinergic substances are known to cause delirium often with agitation and perceptive disturbances (Ross et al., 1991), whereas gabaergic mechanisms have been suggested to be involved in hypoactive delirium due to hepatic encephalopathy (Mullen et al., 1988). Table 4 summarizes the current hypotheses about the various neurotransmitters' activity and their possible implication in the phenomenology of the delirium subtypes, although there is little knowledge of the neurotransmitters' interrelationship. Some conditions commonly associated with the deregulation of a specific neurotransmitter's activity are mentioned as well. This may account for the absence of specific delirious symptoms, which are commonly described in other psychiatric disorders. It is interesting to note the age-related decrease in dopamine, which could be linked with the prevalence of hypoactive delirium among older patients. In fact age > 80 is the only predictive factor of hypoactive subtype of delirium, as will be stated in an article by Breibart et al. (in press). The more recent studies have suggested that an imbalance in the cholinergic and dopaminergic neurotransmitter systems is primarily implicated in causing delirium, and could both account for delirium symptoms and be consistent with the neuroanatomical pathways being implicated. Table 4 indicates the most frequently observed imbalances of some neurotransmitters. It shows as well some medical conditions usually correlated with specific neurotransmitter's dysregulation. Those medical conditions share part of their clinical presentation with either the hypoactive or the hyperactive type of delirium.

Adapted from Meagher et al. (2000)

TREATMENT RESPONSE

Delirium was, by definition, considered as a transient disorder most often of brief duration (Bucht et al., 1999). The validity of this statement, though, has now long been questioned. Literature suggests a poorer than previously thought outcome for a large proportion of patients. In a recent study in a geriatric population, mortality after an episode of delirium reached 30% at 1 year, and up to 32% had persistent courses notably with constant cognitive and functional impairment, irrespective of preexisting dementia (McCusker et al., 2003). However, little has been done to correlate the clinical presentation with a specific outcome. Research has been more prone to investigate predictors of clinical subtypes (as previously stated, age > 80 seems to be a good predictor for the hypoactive subtype of delirium), and general predictors of outcome—older age, longer duration of the episode, preexisting cognitive impairment, severe physical illness—are now generally accepted as predictors of poorer outcome.

Nonetheless, it is considered that the mixed subtype of delirium has the worst treatment response and thus the worst prognosis (Lipowski, 1983; Kobayashi et al., 1992; Liptzin & Levkoff, 1992). On the other hand, the hyperactive type shows definitely the best prognosis. Kobayashi (Kobayashi et al., 1992) had in that subgroup the highest rate of full recovery (54% vs. 40% for the other subtypes). The hypoactive type is more frequent in the elderly and more difficult to identify (Inouye, 1994), covered by or misdiagnosed as depression and, for those studies identifying pure hyper- or hypoactive deliria, is a predictor of poorer outcome (Breitbart et al., 2002). Patients with hypoactive delirium seem to have longer hospital stays (Liptzin & Levkoff, 1992; O'Keeffe, 1999). Immediate mortality rate seems to be more important for that group as well, even though this is not very well and constantly supported. It is noteworthy that there is a large variation in management practices among motoric subtypes and it is therefore difficult to dissociate improvements due to more active and earlier manipulations, medication, for example, from other factors such as underlying pathophysiology.

Different subtypes are associated with particular complications (O'Keeffe, 1999): Hypoactive patients are more likely to develop pressure sores and hospital-acquired infections. Those complications may be preventable, if anticipated with an early assessment of the condition and of its expression.

The treatment of delirium is a multistep or a multicomponent intervention. The underlying causes for delirium have to be identified if possible and treated. Several nonpharmacological interventions have proven effective (Inouye et al., 1999): fluid and electrolyte balance, nutrition, and measures to help reduce anxiety and disorientation (constant reorientation, correction of hearing and visual impairment, early mobilization). Physical restraint with one-to-one nursing observation may also be useful in reducing anxiety in agitated terminally ill patients (Breitbart, 2001). Specific interventions aimed to treat specific symptoms of hypoactive deliria have not been explored yet.

The pharmacological treatment is an important part of the multicomponent intervention. Haloperidol is still considered as the gold standard for the treatment of delirious patients. It can be used per os or parenterally with doses that seldom need to exceed 20 mg in a 24-h period. Olanzapine or risperidone have proven to be a valid alternative in those who have demonstrated intolerance to the extrapyramidal side effects of the classic neuroleptics. A switch to methotrimeprazine or chlorpromazine, two phenothiazine neuroleptics, is an alternative when agitation is not controlled satisfactorily (Oliver, 1985; Breitbart & Strout, 2000). In case of refractory agitation an augmentation of a haloperidol regimen with lorazepam is well documented as well (Akechi et al., 1996). Parenteral infusions of midazolam (de Sousa & Jepson, 1988; Bottomley & Hanks, 1990), a short-acting benzodiazepine, or of propofol (Mercadante et al., 1995; Moyle, 1995), a short acting anaesthetic agent, have been used to control agitation related to delirium in the terminally ill. The latter has an advantage over the former in that the level of sedation is more easily controlled and recovery is rapid upon decreasing the rate of infusion. Overall, in the literature on the treatment of delirium there is a consensus on how to treat the agitated delirious patients (Stiefel et al., 1992). The specificity of the management of hypoalert, somnolent patients in terms of pharmacological and nonpharmacological interventions has been taken less into account in the literature. Several studies however have been published on the use of psychostimulants in that population (Stiefel & Bruera, 1991; Morita et al., 2000; Rozans et al., 2002).