Abstract
Delirium is a common complication in terminally ill cancer patients. Identification of underlying pathologies and prediction of clinical features may improve effective symptom alleviation. This study aims to clarify precipitating factors and their associations with clinical features of terminal delirium. Consecutive hospice inpatients who developed delirium were prospectively evaluated following a structured protocol. Among 237 patients followed until death, 245 episodes of delirium were identified in 213 patients. Precipitating factors for delirium were disclosed in 93% of the 153 cases in which investigations were completed. Mean number of etiologies was 1.8 ± 1.1 per patient, and two or more factors were recognized in 52%. The main pathologies identified were hepatic failure, medications, prerenal azotemia, hyperosmolality, hypoxia, disseminated intravascular coagulation, organic damage to the central nervous system, infection, and hypercalcemia. Occurrence of hyperactive delirium and the requirement for symptomatic sedation significantly correlated with hepatic failure, opioids, and steroids, while dehydration-related pathologies were significantly associated with hypoactive delirium. Complete recovery was frequently achieved in cases with medication- and hypercalcemia-induced delirium, whereas a low remission rate was related to hepatic failure, dehydration, hypoxia, and disseminated intravascular coagulation. In conclusion, standard examinations can confirm factors potentially contributing to delirium and thereby predict the severity of agitation and clinical outcomes.
Keywords
Introduction
Delirium is one of the most common psychiatric disorders in terminally ill cancer patients. Recent studies revealed that 28–44% of cancer patients demonstrate delirious symptoms on admission to palliative care units and 68–88% are delirious just before death.
1
, 2
, 3
, 4
, 5
Agitated delirium causes severe distress for both patients and their family members, and intensive medical interventions, including symptomatic sedation, is required in 7–29% of dying patients.6
, 7
, 8
, 9
, 10
, 11
Although many authorities stress the importance of identifying underlying pathologies for effective symptom alleviation of delirium,
12
, 13
, 14
, 15
empirical studies have not always succeeded in establishing these etiologies in terminal patients.1
, 3
, 16
, 17
, 18
In a pioneering study by Bruera et al., biological causes of cognitive failure could be established in only 44%.1
Several retrospective or preliminary studies suggest that the underlying pathologies of delirium in advanced cancer patients are often multiple and occasionally unspecified.16
, 17
, 18
In contrast, a recent report from the Edmonton palliative care team identified some factors precipitating delirium in almost of all patients receiving intensive palliative care.2
These conflicting findings are due mainly to methodological inconsistencies among studies. All studies, except that of Lawlor et al.,
2
failed to clarify the criteria by which clinicians determined whether each pathology did or did not contribute to delirium, possibly resulting in a significant detection bias.1
, 3
, 16
, 17
In some studies, delirium was diagnosed based on a general cognitive function test or a retrospective chart review, not using validated diagnostic criteria, and patients with various cognitive disorders might have been included in the study populations.1
, 17
, 18
In the recent prospective study by Lawlor et al.,2
although rigorous standardized methods were adopted to overcome these shortcomings, their conclusions might not be consistently applicable to delirium occurring just before death because they analyzed only the first episodes of delirium with respect to their biological etiologies.2
To our knowledge, although identification of determinants of delirium subtypes is clinically important in planning palliative care suitable for each patient, no empirical studies have examined the potential effects of underlying pathologies on the clinical features of delirium, such as severity of agitation. This study was designed to identify factors precipitating terminal delirium and explore their associations with clinical features using a structured protocol in an inpatient hospice care setting, where the majority of patients die under the close and direct observation of clinicians.
Methods
All consecutive terminally ill cancer patients admitted to our palliative care unit from February 1998 to April 2000 were regarded as potential participants in this study. Patients were referred to hospice mainly for end-of-life care based on the physician judgment that they were unlikely to live for six months.
During the study period, primary physicians clinically monitored the symptoms suggestive of delirium, such as attention disturbance, disorientation, hallucinations, and delusions, for each patient from admission to discharge or the patient's death. Although we did not use any screening tools to detect delirium, the primary physicians investigated their responsible patients at least twice a day, and the principal investigator (T.M.) saw all patients every week. When the diagnosis of delirium was established on the basis of the Diagnostic and Statistical Manual of Mental Disorders (4th edition) by the primary physicians,
19
if necessary with assistance of the principle investigator, patients underwent the following structured evaluation.First, the physician recorded the patient background,that is, sex, age, tumor locations, and general condition. General condition was measured using the Palliative Performance Scale (PPS),
20
a modified Karnofsky Performance Scale (KPS), which graded patient conditions as 0 (death) to 100 (normal).Second, the physician assessed the severity of delirium using the Memorial Delirium Assessment Scale (MDAS) and the Delirium Rating Scale (DRS) within 24 hours after the diagnosis.
21
, 22
The MDAS is a ten-item scale to specifically measure the severity of delirium in cancer patients.21
To avoid an extraordinary burden on severely ill patients, we used the modified MDAS in this study, which excluded 3 items requiring active patient cooperation from the original scale (items 2, 3, and 4). The total score of the modified MDAS was calculated by prorating the scores of the items completed, such that it ranged from 0 to 30. This prorated method was approved in the original validation study, although the validity was not fully established.21
The DRS is a 10-item observer-rating scale with a possible range of 0–32.22
Although the DRS is more suitable for diagnostic aims, quantification of symptom severity using this scale is widely accepted in clinical settings.Third, the physician was requested to identify factors precipitating delirium on the basis of physical, biochemical and radiological examinations, and careful review of medical histories. Laboratory tests, including serum calcium, ammonia, osmolality, and oxygen saturation (or arterial blood gas), were encouraged to be completed in all cases as a part of our routine practice. Electrocardiographic and radiographic studies of the chest and abdomen were performed as clinically required. Brain computerized tomography (CT) or magnetic resonance imaging (MRI) was considered when no potential pathologies were detected by the above basic investigations or when intracranial pathologies were suspected from physical findings. If the etiologies of delirium were obvious by physical findings (e.g., overt icterus), or patients and/or their family members did not want medical interventions, the examinations were not performed.
To standardize physicians' judgments, whether or not a factor contributed to delirium was defined by operational criteria. That is, a factor was determined to be a potential etiology for delirium if 1) a specific pathology known to cause delirium was identified, and 2) there was a clear temporal association between occurrence of the pathology and the course of delirium. This corresponded to a “possible” or “probable” etiology by Francis's definitions.
23
Changes in delirium severity in association with changes in etiologic factors were not included in the definition, because repeated blood sampling was often impossible due to extremely poor patient conditions in this setting. The degree of abnormality which was regarded as a potential cause of delirium was defined as in Table 1 These cutoff points were determined on clinical grounds, because the degree of each abnormality which impairs patient cognitive function has not been established.24
Structural damage to the central nervous system was supported by positive findings on brain CT or MRI, and disseminated intravascular coagulation was diagnosed based on the criteria proposed by the Research Committee of the Ministry of Health and Welfare.25
Etiologic values of medications, infection, hypovolemic shock due to bleeding, and poorly-palliated physical discomfort were determined on the basis of both the presence of positive physical findings (medical history) and clear temporal associations with onset of delirium.Table 1Definitions of Biological Etiologies of Delirium
| Hepatic failure | Total bilirubin ≥43 μmol/L or ammonia ≥47 μmol/l |
|---|---|
| Intrinsic renal failure | BUN ≥8 mmol/l and BUN/creatinine <0.037 (no evidence of intestinal bleeding) |
| Prerenal azotemia | BUN ≥8 mmol/l and BUN/creatinine ≥0.037 (no evidence of intestinal bleeding) |
| Hyperosmolality | Osmolality ≥300 mOsm/kg |
| Hypernatremia | Sodium ≥160 mmol/l |
| Hyponatremia | Sodium ≤125 mmol/l |
| Hypercalcemia | Corrected calcium ≥2.9 mmol/l |
| Hypoxia | Oxygen saturation ≤90%, oxygen partial pressure ≤60 mmHg, or requirement for an oxygen flow ≥2 l/min |
| Hypercapria | Carbon dioxide partial pressure ≥75 mmHg |
| Anemia | Hemoglobin ≤6.0 g/dl |
| Hyperglycemia | Blood sugar ≥25 mmol/l |
| Hypoglycemia | Blood sugar ≤2.8 mmol/l |
| Normal ranges: total bilirubin, <21 μmol/L; ammonia, <41 μmol/L; BUN, 1.3–3.7 mmol/l; creatinine, 53–115 μmol/l; osmolality, 275–295 mOsm/kg; sodium, 136–148 mmol/l; calcium, <2.6 mmol/l; hemoglobin, 11.3–15.2 g/dl; blood sugar, 3.1–6.1 mmol/l. | |
Patients were treated employing flexible guidelines proposed by standard textbooks.
12
, 13
, 14
, 15
We principally encouraged supportive nursing care and treated reversible causes, for example, withdrawal of responsible medications, bisphosphonate for hypercalcemia, and antibiotics for infection. Parenteral haloperidol, rectal mianserin, or chlorpromazine was used as the first line psychopharmacological treatment.12
, 13
, 14
, 15
, 26
, 27
, 28
When the first intervention failed, benzodiazepines and barbiturates were considered as second and third line therapies, respectively, to achieve symptomatic relief by intermittent or continuous sedation.12
, 13
, 14
, 15
If an opioid was regarded as a major cause of delirium, dose reduction or rotation to alternative opioids, mainly morphine to fentanyl, was indicated with a combination of rehydration therapy for patients without fluid retention signs.29
, 30
Final treatment levels and whether or not complete remission of delirious symptoms was eventually achieved were recorded by the primary physician. Complete remission was defined as disappearance of all delirious symptoms and attainment of normal consciousness and cognitive ability continuing for at least three days. Survival was defined as the interval from the initial assessment to death.
Physicians participating in this study were all attending physicians in Seirei Hospice, who were directly involved in patient care. The principal investigator had been in a training program for liaison-consultation psychiatry, and the other authors had been trained to diagnose delirium and use the MDAS and the DRS for another study.
31
All clinical and research decisions were validated by full agreement of the authors in team conferences, regularly held twice per a week.This study was not referred to the ethics committee of our hospital because the interventions applied were within our routine care and extraordinary burdens were not placed on these patients.
Analyses
As the primary purpose of this study was elucidation of the delirium that occurred near death, analyses were performed on patients who died in our institution within 6 months. To explore the possibility that differences in underlying pathologies influenced the clinical features of delirium, the frequency of hyperactive delirium, severity of hyperactive symptoms, and frequency of symptomatic sedation were compared in patients with various etiologies of delirium. Hyper- and hypoactive delirium was characterized by ratings for a single item of the MDAS, namely, decreased or increased psychomotor activity (item 9). For quantitative analyses, the scores were converted to −3 (severe hypoactivity), −2 (moderate hypoactivity), −1 (mild hypoactivity), 0 (normal), +1 (mild hyperactivity), +2 (moderate hyperactivity), and +3 (severe hyperactivity). Hyperactive or hypoactive delirium was defined as a score ≥1, or ≤−1, respectively, and if a patient was diagnosed as having mixed-type delirium, they were included in the hyperactive category. Symptomatic sedation was defined as temporal or continuous use of benzodiazepines or barbiturates.
To identify independent determinants of the clinical features of delirium, two types of multiple regression analyses were conducted. One was a linear multiple regression analysis in which the item 9 score of the MDAS was used as a continuous dependent variable, and the other was a logistic regression analysis in which patients receiving symptomatic sedation were compared with the others. In both analyses, underlying pathologies observed in 7.5% or more of the cases, use of steroids, and tumor locations (lung, liver, peritoneum, bone) were initially screened as potential contributing factors by univariate analyses. All significant contributing factors were then entered into the equation in a backward elimination fashion as independent variables after being categorized as absent (0) or present (1).
Univariate analyses were performed using the chi square test (Fisher's exact method), the unpaired two-sided t-test, or the Mann–Whitney U-test, where appropriate. All analyses were performed using the Statistical Package for the Social Sciences (ver. 9.0).
Results
Among 237 patients who died during this study period (of 265 admitted patients), 245 episodes of delirium were identified in 213 patients (90% of 237 patients). Laboratory examinations were not performed in 92 episodes (38%) due to an obvious lack of clinical necessity for the examinations and failure to obtain consent. Table 2 summarizes the characteristics of the delirious patients who did and did not have complete examinations. The patients who received full evaluations were significantly more likely to be male and have better performance status than those who did not. There were no significant differences in patient age, primary diseases, severity of delirious symptoms, or types of delirium between the two groups. Further analyses were performed on 153 episodes for which complete evaluations were undertaken. The median interval from admission to death was 25 days.
Table 2Patient Backgrounds
| With examinations | Without examinations | P | |
|---|---|---|---|
| Number of episodes (patients) | 153 (130) | 92 (83) | |
| Sex (male, %) | 98 (64%) | 41 (45%) | <0.01 |
| Age (mean ± SD) | 65 ± 13 | 66 ± 15 | 0.60 |
| Primary tumor sites | 0.24 | ||
| Lung | 28 (18%) | 24 (26%) | |
| Liver/biliary system/pancreas | 20 (13%) | 18 (20%) | |
| Colon/rectum | 29 (19%) | 9 (9.8%) | |
| Stomach | 23 (15%) | 13 (14%) | |
| Genito-urinary system | 14 (9.2%) | 13 (14%) | |
| Neck | 11 (7.2%) | 2 (2.2%) | |
| Esophagus | 8 (5.2%) | 5 (5.4%) | |
| Breast | 7 (4.6%) | 4 (4.3%) | |
| Others | 13 (8.5%) | 4 (4.3%) | |
| Palliative Performance Scale (mean ± SD) | 26 ± 11 | 18 ± 8.4 | <0.01 |
| Clinical Features | |||
| Modified MDAS (mean ± SD) | 15 ± 4.9 | 16 ± 4.7 | 0.072 |
| DRS (mean ± SD) | 14 ± 4.4 | 15 ± 3.7 | 0.56 |
| Hyperactive # | 27 (18%) | 17 (18%) | 0.87 |
| Final treatment levels | 0.61 | ||
| No treatment | 63 (41%) | 39 (42%) | |
| First line | 49 (32%) | 26 (28%) | |
| Second line | 39 (25%) | 27 (29%) | |
| Third line | 2 (1.3%) | 0 | |
| MDAS = Memorial Delirium Assessment Scale; DRS = Delirium Rating Scale; # = defined by ratings on item 9 of MDAS. | |||
a chi square test (Fisher' exact test).
b unpaired two-sided t-test.
c soft tissue (N = 7), unknown (N = 4), blood, skin (N= 1).
d blood (N = 2), soft tissue, unknown (N= 1).
The numbers of delirious episodes per patient were one (n = 130), two (n = 19), and three (n = 4). Delirium occurred an average of 19 ± 29 days (median = 10) before death, and the 30- and 50-day mortality rates were 83% and 91%, respectively.
The symptoms were alleviated in 55 episodes (36% of 153 episodes) by the first-line pharmacological therapy and 63 episodes (41%) required no psychopharmacological treatment. Symptomatic sedation was required in the other 35 episodes (23%), using benzodiazepines (n = 33) or barbiturates (n = 2). Sedation was eventually provided as an intermittent form in 26 cases (17% of 153 episodes) and as a continuous treatment in 9 cases (5.9%). The major psychotropics prescribed one day after diagnoses were; parenteral haloperidol, 6.2 ± 5.7 mg/day (n = 43); rectal diazepam, 7.1 ± 3.9 mg/day (n = 11); rectal mianserin, 42 ± 16 mg/day (n = 5); parenteral midazolam, 51 ± 37 mg/day (n = 4).
Table 3 shows the potential etiologies of delirium identified. Precipitating factors were revealed in 93% of the cases and 7.2% were diagnosed as having “delirium not otherwise specified,” in which no responsible pathologies were identified. The mean number of identified factors was 1.8 ± 1.1 (median = 2, maximum = 6) per patient, and two or more factors were recognized in 52% (n = 80). The main pathologies were hepatic failure, medications, prerenal azotemia, hyperosmolality, hypoxia, disseminated intravascular coagulation, organic damage to the central nervous system, infection, and hypercalcemia. Metabolic abnormalities were detected in 44% (n = 68). Opioids were regarded as a participating factor in 21%. Opioids were prescribed in 65% of the cases (n = 100), with a mean dose of 189 ± 431 mg/day oral morphine equivalent milligrams (median = 20, maximum = 2800) just before occurrence of delirium.
Table 3Potential Etiologies of Delirium
| Unknown | 11 (7.2%) |
|---|---|
| Identified | 142 (93%) |
| Organic damage to the central nervous system | 19 (12%) |
| Neoplasms in brain | 14 |
| Meningeal carcinomatosis | 3 |
| Vascular disease | 1 |
| Inflammation | 1 |
| Medication | 38 (25%) |
| Opioids | 32 |
| Psychotropics | 6 |
| Others | 6 |
| Hepato-renal dysfunction | |
| Hepatic failure | 45 (29%) |
| Intrinsic renal failure | 5 (3.3%) |
| Fluid imbalance | |
| Prerenal azotemia | 32 (21%) |
| Hyperosmolality | 32 (21%) |
| Electrolyte imbalance | |
| Hypernatremia | 1 (0.7%) |
| Hyponatremia | 5 (3.3%) |
| Hypercalcemia | 13 (8.5%) |
| Blood gas imbalance | |
| Hypoxia | 24 (16%) |
| Hypercapnia | 1 (0.7%) |
| Hematological disturbance | |
| Disseminated intravascular coagulation | 19 (12%) |
| Anemia | 9 (5.9%) |
| Disturbance in blood sugar level | |
| Hyperglycemia | 2 (1.3%) |
| Hypoglycemia | 4 (2.6%) |
| Infection | 17 (11%) |
| Hypovolemic shock due to bleeding | 7 (4.6%) |
| Unpalliated physical distress | 10 (6.5%) |
a cerebral infarction.
b brainstem encephalitis.
c 5 cases had two or more potentially causative medications.
d morphine (n = 27), buprenorphine (n = 3), pentazocine, fentanyl.
e haloperidol, prochlorperazine, promethazine, methylphenidate, hydroxygine, phenobarbiturate.
f ketamine, carbamazepine, lidocaine, digitalis, betamethasone, metoclopramide.
g lung (N= 9), bile duct (N = 3), urinary tract, colon, peritoneum, pulmonary tuberculosis, unknown.
h neuropathic pain (N = 4), incidental pain due to bone matastases (N = 3), increased bronchial secretions (N = 2), rectal pain, dyspnea, nausea (N = 1).
Of 11 cases in which no precipitating factors were discovered, delirium was clinically attributed to nonspecific medical factors related to advanced age in 4 cases, that is, cerebral atrophy, multiple cerebral infarction, and sensory deprivation. In the remaining 7 cases, clear responsible etiologies were “unknown.” Brain CT scans were performed in 4 of these cases.
The complete remission rate was 20% overall (n = 31). It was higher in delirium related to medications (37%, n = 14) and hypercalcemia (38%, n = 5), relatively low in delirium due to infection (12%, n = 2), and extremely low with the other pathologies (less than 10%).
Parenteral hydration (≥500 ml/24 hours) was performed in 18 episodes (47%) of delirium related to medications, but was not indicated in 12 episodes (32%) because of symptomatic fluid retention. In addition, 13 and 12 delirious patients with prerenal azotemia and hyperosmolality received artificial fluid therapy (41%, 38%, respectively), while 15 and 16 patients did not receive rehydration due to symptomatic fluid retentions (47%, 50%, respectively).
The effects of differences in precipitating factors on the clinical features of delirium are summarized in Table 4. There were no significant effects of patient sex, tumor location in the liver and peritoneum, or pathologies of hypoxia and intracranial damage. Hyperactive symptoms were significantly associated with drug-induced delirium, while dehydration-related pathologies (prerenal azotemia and hyperosmolality) were significantly associated with hypoactivity. Also, elderly patients (≥70) had lower scores on item 9 of the MDAS, with marginal statistical significance (−1.4 ± 1.6, n = 68 vs. −1.1 ± 1.4, n = 85; P = 0.055). Using a multivariate regression model, medication-related pathology and hyperosmolality were identified as independent determinants of item 9 scores on the MDAS (beta ± S.E. [95% C.I.] = 1.2 ± 0.26 [0.68∼1.7] (P < 0.01), −0.91 ± 0.29 [−1.5∼−0.34] (P < 0.01), respectively; F = 16.4; R2 = 0.19).
Table 4Effects of Underlying Pathologies on Clinical Features of Delirium
| Hyperactive delirium | P | Mean Score of Item 9 of MDAS | P | Symptomatic Sedation | P | ||
|---|---|---|---|---|---|---|---|
| Medications | Presence (N = 38) | 15 (39%) | <0.01 | −0.21 ± 1.7 | <0.01 | 7 (18%) | 0.51 |
| Absence (N = 115) | 12 (10%) | −1.5 ± 1.3 | 28 (24%) | ||||
| Hepatic failure | Presence (N = 45) | 5 (11%) | 0.17 | −1.3 ± 1.3 | 0.80 | 16 (36%) | 0.016 |
| Absence (N = 108) | 22 (20%) | −1.1 ± 1.6 | 19 (18%) | ||||
| Prerenal azotemia | Presence (N = 32) | 2 (6.3%) | 0.069 | −1.8 ± 1.3 | 0.050 | 8 (25%) | 0.75 |
| Absence (N = 121) | 25 (21%) | −1.0 ± 1.5 | 27 (22%) | ||||
| Hyperosmolality | Presence (N = 32) | 2 (6.3%) | 0.069 | −2.0 ± 1.1 | <0.01 | 6 (19%) | 0.53 |
| Absence (N = 121) | 25 (21%) | −0.98 ± 1.5 | 29 (24%) | ||||
| Hypercalcemia | Presence (N = 13) | 0 | 0.13 | −1.5 ± 0.52 | 0.81 | 0 | 0.040 |
| Absence (N = 140) | 27 (19%) | −1.2 ± 1.6 | 35 (25%) | ||||
| DIC | Presence (N = 19) | 2 (11%) | 0.53 | −1.4 ± 1.2 | 0.84 | 8 (42%) | 0.033 |
| Absence (N = 139) | 25 (18%) | −1.2 ± 1.5 | 27 (19%) | ||||
| Infection | Presence (N = 17) | 6 (35%) | 0.043 | −0.82 ± 2.1 | 0.94 | 2 (12%) | 0.36 |
| Absence (N = 136) | 21 (15%) | −1.2 ± 1.4 | 33 (24%) | ||||
| DIC = disseminated intravascular coagulation. | |||||||
a chi-square test (Fisher's exact test).
b Mann-Whitney U-test.
Symptomatic sedation was significantly associated with hepatic failure and disseminated intravascular coagulation. No patients with hypercalcemia required palliative sedation therapy (Table 4). Patients with lung and bone malignancies were significantly less likely to be sedated than patients without these tumor locations (10/63 [16%] vs. 24/77 [31%], P = 0.036; 2/35 [5.7%] vs. 32/111 [29%]; P < 0.01; respectively). Also, sedation was required more frequently for patients receiving steroids than for patients not on steroids (21/70 [30%] vs. 14/83 [17%]; P = 0.054). By multivariate logistic regression analysis, lung tumors, hepatic failure, and steroid use were identified as independent determinants of the necessity for symptomatic sedation therapy (Odds ratio [95% C.I.] = 0.11 [0.022∼0.53] (P < 0.01), 4.3 [1.5∼11.7] (P < 0.01), 4.2 [1.6∼11.3] (P < 0.01), respectively).
Discussion
Previous studies have revealed that delirium occurs in 68–88% of cancer patients with impending death.
1
, 2
, 3
, 5
Bruera et al.'s study showed that delirium was observed an average of 16 ± 6 days before death,1
and several studies identified delirium as a significant factor indicating a poor prognosis in advanced cancer.32
, 33
, 34
Consistent with these findings, the frequency of delirium just before death was very high in our populations, and median survival from the onset of delirium was only 10 days. Thus, for terminally ill cancer patients, delirium is either a very common psychiatric complication, or rather can be interpreted as “a natural part of the dying process.”13
Intensive pharmacological management is often necessary for symptom relief of delirium in cancer patients. Olofsson et al. demonstrated that 66% of delirious patients were managed by haloperidol alone and 28% required additional benzodiazepines.
16
In the palliative care literature, 26% of terminal delirium was not manageable without benzodiazepine-induced sedation,17
and 7.0–29% of all cancer patients ultimately received palliative sedation therapy for symptoms of delirium.6
, 7
, 8
, 9
, 10
, 11
In this study, 41% of the patients needed no pharmacological treatment, and 36% received successful first-line psychiatric therapy, mainly haloperidol. While symptomatic sedation was performed in 23% of terminal delirium, only 5.9% of all required a continuous type of sedation. These findings are generally consistent: supportive measures with or without use of haloperidol can palliate delirious symptoms in two-thirds to three-fourths of patients, while the remainder requires palliative sedation therapy for acceptable symptom relief with mainly intermittent, or, in small number of patients, continuous administrations of benzodiazepines.Overall improvement of delirious symptoms was achieved in 29–33% of cancer patients when delirium was diagnosed on admission to palliative care units
1
, 5
and in 49% of those admitted to a tertiary palliative care unit.2
Although the complete recovery rate in our sample, 20%, seems smaller than those in previous studies, we believe the difference is due mainly to variations in patient selection, rather than inappropriate medical treatment. That is, in the above studies, the patients investigated were those who developed delirium on admission, not always just before death,1
, 5
or patients who eventually died within 48 hours after the occurrence of delirium were excluded from analyses.1
In another study,2
the proportion of terminal delirium to all delirious episodes is assumed to be smaller than in this study, because 1) their patients were referred for intensive symptom control rather than terminal care, 2) the frequency of opioid medications as an etiology of delirium, which was identified as a factor in treatable delirium in our study, was notably high (76%), and 3) the general condition of the patients was better (half of patients were discharged with a higher 50-day survival rate than ours [31%]). Therefore, we emphasize that, even in the latest stage of cancer, a significant proportion of patients can achieve total remission of delirious symptoms if underlying pathologies are appropriately treated.With respect to the associations between underlying pathologies and reversibility, metabolic etiologies caused by progressive underlying disease were, not surprisingly, associated with poor outcomes, whereas delirium related to hypercalcemia and medications was more likely to be amenable to complete recovery. These findings are generally consistent with previous observations.
1
, 2
It is stressed that efforts to screen treatable etiologies of delirium are of great importance for minimizing the risk of undertreatment, because delirium itself is intrinsically a reversible complication.12
, 13
, 14
Establishment of an underlying etiology can contribute to effective psychiatric management. However, while one well-designed study revealed factors precipitating delirium in almost all cases (70 of 71 cases),
2
other studies from palliative care settings failed to confirm an underlying pathology in 56–74%.1
, 17
This study disclosed underlying pathologies in 93% of the patients with simple examinations, while multiple precipitating factors were recognized in 52%. Moreover, an additional chart review demonstrated that the proportion of cases where medical causes of delirium were attributable to an “unknown” origin was very small, 7 of 153 cases. This discrepancy can be explained by differences in the criteria by which clinicians determined whether each etiology did or did not contribute to development of delirium. That is, pioneer studies did not clarify the threshold levels of each etiology that could cause delirium,1
, 3
, 16
, 17
while Lawlor et al.'s2 study and ours, in which responsible etiologies were established in more than 90% of cases, adopted explicit criteria. These findings suggest that, although it is difficult to establish a single cause of delirium in terminally ill cancer patients, basic screening examinations can at least discover several precipitating factors to this syndrome in a major portion of patients.No empirical studies have systematically investigated possible effects of underlying pathologies on clinical features of terminal delirium. In this study, medication-induced delirium often showed a hyperactive feature, but complete remission was frequently achieved. Medications were determined as a potential cause of delirium in 25% of our patients, and morphine was commonly identified as a precipitating factor. These results, generally consistent with Lawlor et al.'s findings,
2
support recent empirical findings that opioids—especially morphine—can cause terminal agitation through substantial accumulation of their metabolites.35
, 36
, 37
Therefore, we strongly recommend that delirium related to opioid toxicities should be aggressively managed with intensive palliative methods such as opioid rotation to minimize patient distress.29
, 30
, 35
, 36
, 37
This study also revealed concomitant use of steroids as an independent risk factor for requiring symptomatic sedation therapy. Many articles reported steroid-induced psychiatric complications, including agitated delirium,
38
and steroids are prescribed for a number of reasons such as appetite loss, general malaise, pain, and nausea in terminally ill cancer patients.39
This study suggests that dose reduction of steroids should be considered in the final stage of cancer, when steroid therapy no longer provides any benefit for symptom palliation. The potential role of steroids in the development of agitated delirium should be confirmed in a well-designed observational or intervention study.The interpretation of the finding that agitated delirium was less frequently observed in the patients with dehydration-related pathologies is complicated. These findings correspond to traditional clinical observations by hospice practitioners that dehydration itself rarely becomes a cause of severe distress for patients and rather facilitates patient comfort through decreased consciousness.
40
, 41
However, Lawlor et al. identified dehydration as a significant determinant of the treatability of delirium, although its independent effects were not confirmed by multivariate analyses.2
This inconsistency may, we speculate, be due to differences in diagnostic criteria for dehydration, variability in study subjects described above, and possibly different hydration policy among institutions. That is, Lawlor et al.'s study required positive treatment effects by parenteral rehydration as evidence of the “probable” etiologic role of dehydration, although patients could receive various co-effective treatments such as opioid rotation.2
For instance, a dehydrated patient for whom parenteral rehydration did not eventually succeed in symptom alleviation was not regarded as having a “probable” dehydration-related etiology. Therefore, we feel that their criteria are more likely to select intrinsically treatable acute dehydration as an etiology of delirium rather than chronic dehydration related to the dying process.In our settings, hydration is principally applied to patients without fluid retention symptoms, because artificial rehydration might exacerbate patient distress such as peripheral edema and bronchial secretions, which could not be balanced against the potential benefit of decreased delirium when death was impending.
40
, 42
Thus, although rehydration was encouraged to be considered as a choice in palliative treatment, approximately half of our delirious patients with dehydration-related pathologies did not receive rehydration therapy because existing fluid retention was problematic. Our experiences can be verified, we believe, by Lawlor et al.'s conclusions that dehydration is not always an essential cause of delirium, but rather a facilitating factor of opioid toxicities.2
From these findings, we recommend that clinicians not overlook treatable dehydration, especially in pre-terminal patients with minimum fluid retention symptoms and large doses of morphine, because dehydration and accumulation of opioid metabolites can accelerate agitated delirium. Simultaneously, however, we consider it justified not to perform artificial fluid therapy for extremely ill patients with dehydration-related hypoactive delirium, if they have severe fluid retention and require low-doses of opioids. To confirm the potential benefits of rehydration therapy for delirious dehydrated patients, an intervention trial with comprehensive quality-of-life assessments, including both delirious and fluid retention symptoms, is strongly encouraged.This study has several limitations. First, as the study population was limited to Japanese cancer patients, the results may not be generalized to cancer patients in other countries with different types of malignancies. Second, as this study was performed within routine clinical practice, 38% of the cases did not complete laboratory examinations and were excluded from the analyses. As the patients excluded were in significantly worse condition than those analyzed, the results might not be generalized to delirium developing much closer to death. Third, some uncommon causes of delirium such as vitamin deficiencies and endocrine disorders were not investigated, and brain imaging was not routinely performed. Fourth, clinical features of delirium were assessed only cross-sectionally in this study. The finding that hepatic failure frequently required eventual symptomatic sedation despite its nonsignificant association with hyperactive feature of delirium on the initial MDAS ratings can be interpreted to mean that clinical features of delirium might change during the intervals. Because delirium subtypes may not be constant even for a short period,
43
a longitudinal observation study is needed to confirm the findings. Finally, because this study was conducted in a single inpatient hospice, the primary aim of which is end-of-life care, the findings might not be applicable to other settings such as acute palliative care units.In conclusion, delirium is very frequently observed in terminally ill cancer patients. Although it is difficult to establish a single cause of terminal delirium, simple investigations can confirm multiple factors contributing to the syndrome, which are associated with clinical severity and outcomes. Establishment of underlying pathologies can help clinicians to identify reversible causes of delirium and plan treatment strategies suited to an individual patient.
Acknowledgements
The authors would like to thank to Tatsuo Akechi, MD, PhD, Psycho-Oncology Division, National Cancer Center Research Institute East, for his useful comments.
References
- Cognitive failure in patients with terminal cancer.J Pain Symptom Manage. 1992; 7: 192-195
- Occurrence, causes, and outcome of delirium in patients with advanced cancer. A prospective study.Arch Intern Med. 2000; 160: 786-794
- Delirium in terminally ill cancer patients.Am J Psychiatry. 1983; 140: 1048-1050
- Psychiatric morbidity in terminally ill cancer patients. A prospective study.Cancer. 1996; 78: 131-137
- The frequency and clinical course of cognitive impairment in patients with terminal cancer.Cancer. 1997; 79: 835-842
- Symptom control during the last week of life on a palliative care unit.J Palliat Care. 1991; 7: 5-11
- Sedation for uncontrolled symptoms in a South African hospice.J Pain Symptom Manage. 1998; 16: 145-152
- Sedation for delirium and other symptoms in terminally ill patients in Edmonton.J Palliat Care. 2000; 16: 5-10
- Sedation for symptom control in Japan.J Pain Symptom Manage. 1996; 12: 32-38
- A comparison of the use of sedatives in a hospital support team and in a hospice.Palliat Med. 1997; 11: 140-144
- Symptom prevalence and control during cancer patients' last days of life.J Palliat Care. 1990; 6: 7-11
- Practice guideline for the treatment of patients with delirium.Am J Psychiatry. 1999; 156: 1-20
- Delirium in the terminally ill.in: Chochinov H.M. Breitbart W. Handbook of psychiatry in palliative medicine. Oxford University Press, New York2000: 75-90
- Delirium. Optimizing management.Br Med J. 2001; 322: 144-149
- Psychopharmacology in supportive care of cancer.Support Care Cancer. 2000; 8: 89-97
- A retrospective study of the psychiatric management and outcome of delirium in the cancer patient.Support Care Cancer. 1996; 4: 351-357
- Acute confusional states in patients with advanced cancer.J Pain Symptom Manage. 1992; 7: 94-98
- Acute encephalopathy in patients with systemic cancer.Ann Neurol. 1992; 32: 288
- Diagnostic and statistical manual of mental disorders. 4th ed. American Psychiatric Association, Washington1994
- Palliative Performance Scale (PPS).J Palliat Care. 1996; 12: 5-11
- The Memorial Delirium Assessment Scale.J Pain Symptom Manage. 1997; 13: 128-167
- A symptom rating scale for delirium.Psychiatry Res. 1998; 23: 89-97
- A prospective study of delirium in hospitalized elderly.JAMA. 1990; 263: 1097-1101
- Harrison's principles of internal medicine. 14th ed. McGraw-Hill, New York1998
Aoki N, Hasegawa H. Auxiliary clinical results and findings for diagnosis in the DIC diagnostic criteria. Ministry of Health and Welfare Research Committee on Specific Disease, Blood Coagulation Abnormalities, Research Report for Fiscal Year 1987. 1988:37–41.
- Usage of haloperidol for delirium in cancer patients.Support Care Cancer. 1996; 4: 390-392
- A double-blind trial of haloperidol, chlorpromazine, and lorazepam in the treatment of delirium in hospitalized AIDS patients.Am J Psychiatry. 1996; 153: 231-237
- Mianserin suppositories in the treatment of post-operative delirium.Human Psychopharm. 1997; 12: 595-599
- Subcutaneous fentanyl and sufentanil infusion substitution for morphine intolerance in cancer pain management.Pain. 1995; 63: 263-269
- Fentanyl by continuous subcutaneous infusion for the management of cancer pain.J Pain Symptom Manage. 1998; 16: 323-326
- Communication Capacity Scale and Agitation Distress Scale to measure the severity of delirium in terminally ill cancer patients.Palliat Med. 2001; 15: 197-206
- Estimate of survival of patients admitted to a palliative care unit.J Pain Symptom Manage. 1992; 7: 82-86
- Impact of delirium on the short-term prognosis of advanced cancer patients.Cancer. 2000; 89: 1145-1149
- The Palliative Prognostic Index.Support Care Cancer. 1999; 7: 128-133
- Changing pattern of agitated impaired mental status in patients with advanced cancer.J Pain Symptom Manage. 1995; 10: 287-291
- Opioid rotation for toxicity reduction in terminal cancer patients.J Pain Symptom Manage. 1995; 10: 378-384
- Attenuation of morphine-induced delirium in palliative care by substitution with infusion of oxycodone.J Pain Symptom Manage. 1996; 12: 182-189
- Corticosteroids in cancer.Cancer Invest. 1989; 7: 479-491
- Steroids in advanced cancer.Br Med J. 1992; 305: 999
- Dehydration in terminally ill patients. Is it appropriate palliative care?.Post Grad Med. 1993; 93: 201-208
- Comfort care for terminally ill patients. The appropriate use of nutrition and hydration.JAMA. 1994; 272: 1263-1266
- Three dimensions of the rehydration-dehydration problem in a palliative care setting.J Pallit Care. 1999; 15: 60-61
- Phenomenology of delirium and its subtypes in advanced cancer patients.J Palliat Care. 1998; 14: 106
Article info
Publication history
Accepted:
February 26,
2001
Identification
Copyright
© 2001 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc.
