International Psychogeriatrics,Vol. 3. No.2,1991
0 1991 Springer Publishing Company
Delirium: Phenomenologic and Etiologic Subtypes Christopher A. Ross, M.D., Ph.D., Carol E. Peyser, M.D., Ira Shapiro, M.D., and Marshal E Folstein, M.D. Department of Psychiatry The Johns Hopkins University School of Medicine Baltimore, Maryland, U.S.A. ABSTRACT. While all delirious patients have clouding of consciousness (alteration of attention) and cognitive dysfunction, the level of alertness of different patients may range from stuporous to hyperalert. We, therefore, developed an analog scale to rate the alertness of delirious patients, and a separate scale to rate the severity of their clouding of consciousness. Based on these scales, patients were categorized overall as relatively “activated” (relatively alert despite clouding of consciousness), or “somnolent” (relatively stuporous along with clouding of consciousness). Cognitive function was estimated using the Mini-Mental Status Exam. Separate ratings were made of hallucinations, delusions, illusions, and agitated behavior. Activated and somnolent patients had similar ages, overall severity of delirium, and MiniMental Status Exam scores. Activated patients, however, were more likely to have hallucinations, delusions, and illusions than somnolent patients, and were more likely to have agitated behavior. Patients with hepatic encephalopathy were more likely to have somnolent delirium, while patients with alcohol withdrawal appeared more likely to have activated delirium. Thcse data indicate that phenomenologic subtypes of delirium can be defined on the basis of level of alertness. These subtypes arc validated in part by their differing associations with symptoms unrelated to alertness. These subtypes may havc different pathophysiology, and thus, potentially different treatments.
INTRODUCTION Delirium is a common complication of medical illness (Engel & Romano, 1959; Lipowski, 1983, 1989; Trzepacz, Teague, Lipowski, 1985; Rabins & Folstein, 1982).It has been defined as global cognitive dysfunction in the context of clouding of consciousness. The patient is unable to focus sustained attention o r to shift it appropriately in response to a changing environment. This results in incoherent 135
136
C.A . Ross et al.
thinking and speech. In addition to clouding of consciousness, alteration of alertness is manifested by drowsiness or hyperalertness. The sleep-wake cycle becomes irregular, with nighttime insomnia and daytime sleepiness. There is often increased or decreased psychomotor activity, as well. Delirious patients may differ in their symptoms. Lipowski (1989) distinguished two subtypes based on alertness-a hypoalert-hypoactive type and a hyperalerthyperactive type. Similar distinctions have also been made in the neurological literature, with the hyperalert-hyperactive type of alertness being called delirium, while the hypoalert-hypoactive type of alertness is ofien labeled an acute confusional state (e.g., Adams & Victor, 1977). Accompanying the alterations in alertness and attention are other symptoms such as hallucinations, illusions, and delusions (Wolff & Curran, 1935). Hallucinations are often visual and delusions are often paranoid. Delirious patients may be agitated, which may result in injury or reksal to cooperate with medical personnel for treatment (Lipowski, 1983). Patients with the hyperalert-hyperactive type of delirium are reported to be agitated (Lipowski, 1989), though this has not been studied quantitatively. The relationship between alertness and symptoms such as hallucinations, illusions, and delusions is unknown. Numerous etiological agents may lead to delirium. It has long been uncertain whether there is any relationship between the symptoms of delirium in a given patient and the etiology (see Bonhoeffer, reprinted 1974).Wolff and Curran (1935) stated that “no evidence was found that there was any specific relationship between a particular noxious agent and the form and content of the accompanying psychobiologic disturbances”; 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, while metabolic encephalopathy (such as hepatic or renal enccphalopathy) is more likely to be of the hypoalert-hypoactivetype (e.g., Engel & Romano, 1959; Lipowski, 1989; Pro & Wells, 1977). For the current study, we developed two scales to rate separately the severity of clouding of consciousness and of altered alertness. Thus, patients with a given severity of clouding of consciousness might be relatively somnolent or relatively alert (or even hyper-alert). The former type we would categorize overall as
TABLE 1. Characteristics of Delirious Patients Number Age (mean rt SD) Education (years) MMSE Digit span Clouding of consciousness score Somnolence score
58 52.8 10.4 14.3 4.4 5.6 4.2
rt rt rt rt k
+
18.2 3.7 7.5 2.4 2.4 2.6
Delirium: Phemmnologic and Etiologic Subtypes
137
“somnolent” and the latter type as “activated.” corresponding to Lipowski’s terms of hypoalert-hypoactive and hyperalert-hyperactive, respectively. We investigated whether positive symptoms such as hallucinations, illusions, and delusions were more common in one or the other subtype. Finally, we attempted to determine whether these phenomenologic subtypes, based on an assessment of alertness, were related to etiology.
METHODS Fifty-eight adult general medical and surgical inpatients (see Table 1) were ascertained in one of three ways. Investigators checked several medical floors each day, asking the nursing staff if there were any patients who appeared confused or disoriented. Other patients were referred by the psychiatry consultation team. Finally, some patients with chronic liver disease were referred from the gastroenterology service. Most patients (52/58) were interviewed by two psychiatrists (CAR and CEP, or CAR and IS). In the initial portion of the interview, efforts were made to rouse the patient as much as possible, so that patients who were simply sleepy would be fully alert. The patient was asked about his subjective level of alertness and clarity of thinking, and his level of alertness was observed. The patient was also asked questions about his present illness, with particular attention to the time course of symptoms in order to gauge his understanding of his current circumstances, causal relationships, and time sequences. He was then asked about major past life events and his educational level in order to assess his remote memory and to determine thc extent of his education. While the general outline of questioning was similar for each patient, the specific questions were determined by the responses to previous questions. A structured interview was not used. The patient’s ability to focus on questions and answer appropriately without distraction from other stimuli was carefully assessed. His cognitive status was evaluated by administration of the Mini-Mental Status Examination (MMSE) (Folstein, Folstein, & McHugh, 1975) and forward digit span. At this point, the two investigators independently completed analog scales (Folstein & Luria, 1973) of(a) severity ofclouding of consciousness and (b) dcgrcc of diminished alertness. In the first scale clear consciousness was defined as 0, and maximally clouded consciousness, i.e., maximally impaired ability to focus and shift attention, was defined as 10, with intermediate degrees of impairment indicated by marking a point on a lOmm line (“visual analog scale,” Folstein & Luria, 1973). Raters considered the degree of difficulty in engaging the paticnl’s attention and the degree of impairment of the patient’s ability to focus his attention on the examiner’s questions without distraction by extraneous stimuli. A similar scalcof“Globa1 Accessibi1ity”has been shown to be useful in the rating of delirium (Anthony, LeResche, Von Korff, Niaz, & Folstein, 1985). We therefore also rated patients on the Global Accessibility scale. Somnolence was rated purely on thc
138
C. A . Ross et al.
basis of how alert or somnolent the patient appeared to be. Full alertness was defined as 0, and stupor as 10. Patients were rated on several other analog 10-point scales: hyperactivity, agitation, roving eye movements, mood lability, rapid speech, and incoherence. Patients were also scored independently by each investigator as either somnolent or activated overall. The investigators then asked about hallucinations, delusions, and illusions. Patients were included whom both investigators agreed had an alteration of attention, and who were cooperative in the interview. All patients gave informed consent, and the protocol was approved by The Johns Hopkins Joint Committee on Clinical Investigation. A screening neurological exam was performed (CAR). Patients with focal deficits (hemiplegia, hemisensory deficits, or cranial nerve deficits) were excluded. Finally, pupillary diameter (in mm) was estimated, and blood pressure, pulse, and temperature were recorded. After the patient interview, medical charts were reviewed. Etiology of delirium was determined from this chart review and consultation with house staff. Patients with a previous history of dementia, as determined from chart review and interviews with family members, were excluded. Patients with history of focal CNS damage were also excluded, as were patients with focal deficits (hemiplegia, hemisensory deficits, or cranial nerve deficits) on a screening neurological exam. In addition, presence or absence of abnormal behaviors within the prcvious two days was recorded after reviewing the patient’s chart and consulting with the nursing staff. Agitated behavior was recorded when patients behaved in an excitcd, disturbed fashion, shouting out, or actively resisting staffdircctions for no apparcnt reason. “Hallucinatory behavior” was recorded if paticnts had bccn noted to appcar to respond to stimuli not apparent to the observer, such as seeming to carry on a conversation when no one else was present. “Delusional behavior” was rccordcd if patients were noted to behave in accordance with consistent false beliefs, such as a belief that nursing staff were trying to poison them. Eighteen control patients were also interviewed in a similar fashion in order to obtain scores on the subscales of the MMSE in this population. Control paticnts were matched on age, sex, race, and by floor. They were excluded if they appeared confused or disoriented. Other exclusion criteria were the same as for delirious patients.
RESULTS Characteristics of the patients are presented in Table 1 . The mean MMSE scorc of 14.3 F 7.5 (mean k SD) and digit span of 4.4 k 2.4 indicate that paticnts wcrc moderately cognitively impaired. Control patients (n = 18) matched on age, racc, and sex who were not delirious had MMSE scores of 29.6 f l . 9 (mean t- SD) and digit spans of 6.8 k 1.3. MMSE subscale scores for controls, mildly delirious (MMSE 2&30, mean 21.8), moderately delirious (MMSE 10-20, mean 15.8), and severely delirious
Delirium: Phemmemlogic and Etiologic Subtypes
139
(MMSE 0-10, mean 4.2) patients are shown in Figure 1 . Subscales relatively most affected in the mildly impaired patients were calculation, orientation, and recall. Subscales relatively preserved in the severely impaired patients were naming and registration. There were no differences in the subscales between patients with
MMSE SUBSCALE
1
MMSE SUBSCALE
MMSE SUBSCALE
Figure 1. MMSE J S A f ~~ ~ ~ ~ ~ ~ r n ~ ,tomtrols. a n Note that Orientation to t h e calculation and dare mo5t affected in mild delirium. Registration, naming,and following a command are m t preservedfor severedelirium.
d
~
v
C . A. Ross et 01.
140
activated or somnolent delirium, or between patients with or without hallucinations or delirium. Agreements between the two raters for clouding of consciousness and somnolence ratings are indicated in Table 2. The kappa for the two raters characterizing patients overall as activated or somnolent was 0.87. There was a high correlation between severity of clouding of consciousness and degree of abnormality on the MMSE ( r = -0.77, n = 58, p < 0.001). There was also a high correlation between scores of cloudingof consciousness and global accessibility (Anthony et al., 1985) ( r = 0.88, n = 58, p < 0.001). Almost one-third of all delirious patients had psychosis (defined as hallucinations or delusions or both) at the time of examination (TabIe 3). Patients with hallucinatory or delusional behavior or both weremore likely to have been recorded on the chart as being agitated during these two days than patients without (Table 4). In addition,these agitated patients were more likely to have a psychosis at the time of exam (Table 4). Patients with activated or somnolent delirium had similar ages, mean MMSE scores, digit spans, and observer ratings of clouding of consciousncss {Table 5). MMSE subscale scores also did not differ between the two groups (data not shown). There were significant differences, however, in the presence of psychotic symptoms. The activated patients had more hallucinations, delusions, and illusions on examination than the somnolent patients, and had more hallucinatory and delusional behavior in the two days prior to examination (Table 5). Activated patients were also more likely to have been agitated in the two days prior to examination.
TABLE 2. Reliability of DeIirium Rating Scales Clouding of consciousness (Rater I vs. rater 2) Somnolence (Rater 1 vs. rater 2)
P
r 0.78
n 52
0 1991 Springer Publishing Company
Delirium: Phenomenologic and Etiologic Subtypes Christopher A. Ross, M.D., Ph.D., Carol E. Peyser, M.D., Ira Shapiro, M.D., and Marshal E Folstein, M.D. Department of Psychiatry The Johns Hopkins University School of Medicine Baltimore, Maryland, U.S.A. ABSTRACT. While all delirious patients have clouding of consciousness (alteration of attention) and cognitive dysfunction, the level of alertness of different patients may range from stuporous to hyperalert. We, therefore, developed an analog scale to rate the alertness of delirious patients, and a separate scale to rate the severity of their clouding of consciousness. Based on these scales, patients were categorized overall as relatively “activated” (relatively alert despite clouding of consciousness), or “somnolent” (relatively stuporous along with clouding of consciousness). Cognitive function was estimated using the Mini-Mental Status Exam. Separate ratings were made of hallucinations, delusions, illusions, and agitated behavior. Activated and somnolent patients had similar ages, overall severity of delirium, and MiniMental Status Exam scores. Activated patients, however, were more likely to have hallucinations, delusions, and illusions than somnolent patients, and were more likely to have agitated behavior. Patients with hepatic encephalopathy were more likely to have somnolent delirium, while patients with alcohol withdrawal appeared more likely to have activated delirium. Thcse data indicate that phenomenologic subtypes of delirium can be defined on the basis of level of alertness. These subtypes arc validated in part by their differing associations with symptoms unrelated to alertness. These subtypes may havc different pathophysiology, and thus, potentially different treatments.
INTRODUCTION Delirium is a common complication of medical illness (Engel & Romano, 1959; Lipowski, 1983, 1989; Trzepacz, Teague, Lipowski, 1985; Rabins & Folstein, 1982).It has been defined as global cognitive dysfunction in the context of clouding of consciousness. The patient is unable to focus sustained attention o r to shift it appropriately in response to a changing environment. This results in incoherent 135
136
C.A . Ross et al.
thinking and speech. In addition to clouding of consciousness, alteration of alertness is manifested by drowsiness or hyperalertness. The sleep-wake cycle becomes irregular, with nighttime insomnia and daytime sleepiness. There is often increased or decreased psychomotor activity, as well. Delirious patients may differ in their symptoms. Lipowski (1989) distinguished two subtypes based on alertness-a hypoalert-hypoactive type and a hyperalerthyperactive type. Similar distinctions have also been made in the neurological literature, with the hyperalert-hyperactive type of alertness being called delirium, while the hypoalert-hypoactive type of alertness is ofien labeled an acute confusional state (e.g., Adams & Victor, 1977). Accompanying the alterations in alertness and attention are other symptoms such as hallucinations, illusions, and delusions (Wolff & Curran, 1935). Hallucinations are often visual and delusions are often paranoid. Delirious patients may be agitated, which may result in injury or reksal to cooperate with medical personnel for treatment (Lipowski, 1983). Patients with the hyperalert-hyperactive type of delirium are reported to be agitated (Lipowski, 1989), though this has not been studied quantitatively. The relationship between alertness and symptoms such as hallucinations, illusions, and delusions is unknown. Numerous etiological agents may lead to delirium. It has long been uncertain whether there is any relationship between the symptoms of delirium in a given patient and the etiology (see Bonhoeffer, reprinted 1974).Wolff and Curran (1935) stated that “no evidence was found that there was any specific relationship between a particular noxious agent and the form and content of the accompanying psychobiologic disturbances”; 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, while metabolic encephalopathy (such as hepatic or renal enccphalopathy) is more likely to be of the hypoalert-hypoactivetype (e.g., Engel & Romano, 1959; Lipowski, 1989; Pro & Wells, 1977). For the current study, we developed two scales to rate separately the severity of clouding of consciousness and of altered alertness. Thus, patients with a given severity of clouding of consciousness might be relatively somnolent or relatively alert (or even hyper-alert). The former type we would categorize overall as
TABLE 1. Characteristics of Delirious Patients Number Age (mean rt SD) Education (years) MMSE Digit span Clouding of consciousness score Somnolence score
58 52.8 10.4 14.3 4.4 5.6 4.2
rt rt rt rt k
+
18.2 3.7 7.5 2.4 2.4 2.6
Delirium: Phemmnologic and Etiologic Subtypes
137
“somnolent” and the latter type as “activated.” corresponding to Lipowski’s terms of hypoalert-hypoactive and hyperalert-hyperactive, respectively. We investigated whether positive symptoms such as hallucinations, illusions, and delusions were more common in one or the other subtype. Finally, we attempted to determine whether these phenomenologic subtypes, based on an assessment of alertness, were related to etiology.
METHODS Fifty-eight adult general medical and surgical inpatients (see Table 1) were ascertained in one of three ways. Investigators checked several medical floors each day, asking the nursing staff if there were any patients who appeared confused or disoriented. Other patients were referred by the psychiatry consultation team. Finally, some patients with chronic liver disease were referred from the gastroenterology service. Most patients (52/58) were interviewed by two psychiatrists (CAR and CEP, or CAR and IS). In the initial portion of the interview, efforts were made to rouse the patient as much as possible, so that patients who were simply sleepy would be fully alert. The patient was asked about his subjective level of alertness and clarity of thinking, and his level of alertness was observed. The patient was also asked questions about his present illness, with particular attention to the time course of symptoms in order to gauge his understanding of his current circumstances, causal relationships, and time sequences. He was then asked about major past life events and his educational level in order to assess his remote memory and to determine thc extent of his education. While the general outline of questioning was similar for each patient, the specific questions were determined by the responses to previous questions. A structured interview was not used. The patient’s ability to focus on questions and answer appropriately without distraction from other stimuli was carefully assessed. His cognitive status was evaluated by administration of the Mini-Mental Status Examination (MMSE) (Folstein, Folstein, & McHugh, 1975) and forward digit span. At this point, the two investigators independently completed analog scales (Folstein & Luria, 1973) of(a) severity ofclouding of consciousness and (b) dcgrcc of diminished alertness. In the first scale clear consciousness was defined as 0, and maximally clouded consciousness, i.e., maximally impaired ability to focus and shift attention, was defined as 10, with intermediate degrees of impairment indicated by marking a point on a lOmm line (“visual analog scale,” Folstein & Luria, 1973). Raters considered the degree of difficulty in engaging the paticnl’s attention and the degree of impairment of the patient’s ability to focus his attention on the examiner’s questions without distraction by extraneous stimuli. A similar scalcof“Globa1 Accessibi1ity”has been shown to be useful in the rating of delirium (Anthony, LeResche, Von Korff, Niaz, & Folstein, 1985). We therefore also rated patients on the Global Accessibility scale. Somnolence was rated purely on thc
138
C. A . Ross et al.
basis of how alert or somnolent the patient appeared to be. Full alertness was defined as 0, and stupor as 10. Patients were rated on several other analog 10-point scales: hyperactivity, agitation, roving eye movements, mood lability, rapid speech, and incoherence. Patients were also scored independently by each investigator as either somnolent or activated overall. The investigators then asked about hallucinations, delusions, and illusions. Patients were included whom both investigators agreed had an alteration of attention, and who were cooperative in the interview. All patients gave informed consent, and the protocol was approved by The Johns Hopkins Joint Committee on Clinical Investigation. A screening neurological exam was performed (CAR). Patients with focal deficits (hemiplegia, hemisensory deficits, or cranial nerve deficits) were excluded. Finally, pupillary diameter (in mm) was estimated, and blood pressure, pulse, and temperature were recorded. After the patient interview, medical charts were reviewed. Etiology of delirium was determined from this chart review and consultation with house staff. Patients with a previous history of dementia, as determined from chart review and interviews with family members, were excluded. Patients with history of focal CNS damage were also excluded, as were patients with focal deficits (hemiplegia, hemisensory deficits, or cranial nerve deficits) on a screening neurological exam. In addition, presence or absence of abnormal behaviors within the prcvious two days was recorded after reviewing the patient’s chart and consulting with the nursing staff. Agitated behavior was recorded when patients behaved in an excitcd, disturbed fashion, shouting out, or actively resisting staffdircctions for no apparcnt reason. “Hallucinatory behavior” was recorded if paticnts had bccn noted to appcar to respond to stimuli not apparent to the observer, such as seeming to carry on a conversation when no one else was present. “Delusional behavior” was rccordcd if patients were noted to behave in accordance with consistent false beliefs, such as a belief that nursing staff were trying to poison them. Eighteen control patients were also interviewed in a similar fashion in order to obtain scores on the subscales of the MMSE in this population. Control paticnts were matched on age, sex, race, and by floor. They were excluded if they appeared confused or disoriented. Other exclusion criteria were the same as for delirious patients.
RESULTS Characteristics of the patients are presented in Table 1 . The mean MMSE scorc of 14.3 F 7.5 (mean k SD) and digit span of 4.4 k 2.4 indicate that paticnts wcrc moderately cognitively impaired. Control patients (n = 18) matched on age, racc, and sex who were not delirious had MMSE scores of 29.6 f l . 9 (mean t- SD) and digit spans of 6.8 k 1.3. MMSE subscale scores for controls, mildly delirious (MMSE 2&30, mean 21.8), moderately delirious (MMSE 10-20, mean 15.8), and severely delirious
Delirium: Phemmemlogic and Etiologic Subtypes
139
(MMSE 0-10, mean 4.2) patients are shown in Figure 1 . Subscales relatively most affected in the mildly impaired patients were calculation, orientation, and recall. Subscales relatively preserved in the severely impaired patients were naming and registration. There were no differences in the subscales between patients with
MMSE SUBSCALE
1
MMSE SUBSCALE
MMSE SUBSCALE
Figure 1. MMSE J S A f ~~ ~ ~ ~ ~ ~ r n ~ ,tomtrols. a n Note that Orientation to t h e calculation and dare mo5t affected in mild delirium. Registration, naming,and following a command are m t preservedfor severedelirium.
d
~
v
C . A. Ross et 01.
140
activated or somnolent delirium, or between patients with or without hallucinations or delirium. Agreements between the two raters for clouding of consciousness and somnolence ratings are indicated in Table 2. The kappa for the two raters characterizing patients overall as activated or somnolent was 0.87. There was a high correlation between severity of clouding of consciousness and degree of abnormality on the MMSE ( r = -0.77, n = 58, p < 0.001). There was also a high correlation between scores of cloudingof consciousness and global accessibility (Anthony et al., 1985) ( r = 0.88, n = 58, p < 0.001). Almost one-third of all delirious patients had psychosis (defined as hallucinations or delusions or both) at the time of examination (TabIe 3). Patients with hallucinatory or delusional behavior or both weremore likely to have been recorded on the chart as being agitated during these two days than patients without (Table 4). In addition,these agitated patients were more likely to have a psychosis at the time of exam (Table 4). Patients with activated or somnolent delirium had similar ages, mean MMSE scores, digit spans, and observer ratings of clouding of consciousncss {Table 5). MMSE subscale scores also did not differ between the two groups (data not shown). There were significant differences, however, in the presence of psychotic symptoms. The activated patients had more hallucinations, delusions, and illusions on examination than the somnolent patients, and had more hallucinatory and delusional behavior in the two days prior to examination (Table 5). Activated patients were also more likely to have been agitated in the two days prior to examination.
TABLE 2. Reliability of DeIirium Rating Scales Clouding of consciousness (Rater I vs. rater 2) Somnolence (Rater 1 vs. rater 2)
P
r 0.78
n 52
