Palliative Sedation in a Specialized Unit for Acute Palliative Care in a Cancer Hospital: Comparing Patients Dying With and Without Palliative Sedation

Introduction 

At the end of life, many patients suffer from severe symptoms.1, 2 Sometimes, sedation until death is needed to control symptoms that cannot be relieved with conventional measures. This practice is often referred to as “palliative sedation.” The degree of sedation necessary to relieve suffering may vary from superficial to deep.3 The use of continuous deep sedation until death is considered to be the most far-reaching. This practice has been extensively discussed among professional medical organizations, in politics, and in the media. The debate mostly focuses on the conditions under which it is acceptable, and whether such sedation can shorten life.4, 5, 6, 7 The latter is often not considered to be true when it is used for patients with a life expectancy of less than one to two weeks.

Several studies have described the practice of palliative sedation.8, 9, 10, 11, 12, 13 It is known that the use of palliative sedation is a rather frequently occurring practice. A European study reported that continuous deep sedation until death was practiced in 2.5%–8.5% of all deaths in the studied countries.9 Within palliative care settings, incidence estimates of the use of sedatives prior to death range from 15% up to more than 60% of the patients.10, 12, 13, 14, 15 Dutch nationwide studies showed that palliative sedation is increasingly used in the Netherlands, up to 8.2% of all deaths in 2005.16, 17

It is known that palliative sedation is mostly practiced in hospitals and used for cancer patients.8, 9 Further, it is known that the most common indications for palliative sedation are refractory terminal restlessness and dyspnea.14 However, little is known about whether and to what extent patients who receive sedation differ from patients who die without sedation. This is also true for specialized care settings, such as palliative care units (PCUs), where patients with complex symptoms are admitted for a multidisciplinary approach.

The present study describes the practice of palliative sedation at a specialized unit for palliative care in an academic cancer hospital. Further, the study compares patients who were sedated prior to death with patients who were not sedated by studying characteristics of the patients and the course of disease.

Methods 

We performed a systematic retrospective analysis of the medical and nursing records of all cancer patients who died at a specialized acute PCU in an academic cancer hospital between October 2001 and October 2005. The main purpose of this acute PCU is to provide symptom control for cancer patients in an advanced stage of their disease and to discharge them after adequate levels of symptom control have been reached. At this PCU, daily multidisciplinary meetings are held with nurses, medical oncologists, an anesthesiologist, a psychiatrist and a neurologist. During the study period, 157 of 753 admitted patients died (21%).

The nursing records consist of different care plans focusing on a particular symptom. At the moment the dying phase is recognized, the nursing and medical records are combined into one single registration system: the Liverpool Care Pathway for the Dying Patient.The Liverpool Care Pathway for the Dying Patient is a standardized registration method to monitor care and its results during the dying phase.18 The guideline regarding the use of sedating medications in the acute PCU stated that sedating medications should be titrated to the cessation of symptoms. For almost all of the patients for whom sedation was used prior to death, continuous deep sedation until death was necessary. In this study, a patient was considered to have received palliative sedation when there was an annotation in the medical records of the use of “continuous deep sedation.”

The records of the patients who had received palliative sedation were analyzed with the use of a checklist that scored characteristics of the decision-making process (discussion with patient, relatives, and other caregivers), the indications to start palliative sedation, the drugs used for sedation, and the duration of the sedation. The indications for palliative sedation were literally taken from the records. Later, these were recoded into pain, dyspnea, anxiety, terminal restlessness (confusion, restlessness or agitation), and other reasons. More than one indication for sedation was possible.

Patients who had received palliative sedation were compared with patients who had not, with respect to sex, age, primary tumor site, survival after admission, survival after identification of the primary tumor, and survival after identification of metastases. We also checked the patients' symptom profile at admission and during the last 48hours of life by looking at whether pain, constipation, dyspnea, nausea/vomiting, delirium, and anxiety had been present. Contrary to the notion of symptoms, which were based on patients' complaints and/or clinical observations only, delirium was only scored to be present when a physician (mostly a psychiatrist) had clinically diagnosed it.

The analysis of records was performed by LVDW. Half of these records were randomly verified, partly by a member of the nursing staff (HVV) and a member of the medical staff (LVZ). The data were analyzed using SPSS version 11.

Results 

Of all 157 deceased patients, 68 (43%) had received palliative sedation. Terminal restlessness (62%) and dyspnea (47%) were the most frequently mentioned indications for palliative sedation (Table 1). Pain (26%) and anxiety (6%) were less common indications. For most patients, palliative sedation was started on the last day before death (68%), with a mean duration of 19hours (range, 1–125hours). Midazolam, sometimes combined with propofol, was the most commonly used drug to induce sedation. The records indicated that in 91% the decision to use palliative sedation was discussed either with the patient or the patient's relatives. In the remaining 9%, these data were missing. Reasons not to discuss this with the patient were mostly that the patient was not competent to make decisions anymore (data not reported). The physician always discussed the decision to use sedation with other caregivers, mostly in a multidisciplinary team meeting (65%).

Table 1. Characteristics of the Palliative Sedations

		Sedated Patients; n=68
		n	%
	Indications for palliative sedationa
	Pain	19	28
	Terminal restlessness	42	62
	Dyspnea	32	47
	Anxietyb	4	6
	Other	10	15
	Number of indications (median, min–max)c	1 (1–4)
	Start palliative sedation
	0–24hours before death	46	68
	24–48hours before death	13	19
	>48hours before death	9	13
	Duration of sedation, hours (median, min–max)	19 (1–125)
	Drugs used
	Midazolam	51	75
	Midazolam and other benzodiazepine	1	1
	Midazolam and propofol	6	9
	Propofol only	10	15
	Sedation discussedd
	With patient and relatives	39	58
	With patient only	1	2
	With relatives only	21	31
	Unknown whether discussed with patient or relatives	7	10
	Decision made ind
	Multidisciplinary meeting	44	65
	Meeting between medical oncologist and nurse	23	34

a More than one indication possible. b Anxiety was for one patient the only indication for palliative sedation. c One indication: n=36; two indications: n=28; three indications: n=2; four indications: n=2. d For one patient, information was missing.

There were no differences between sedated and nonsedated patients in sex, but patients who received sedation were on average younger (55 years) than patients who did not receive sedation (59 years) (P=0.04) (Table 2). Patients who received sedation more often had cancer of the digestive tract compared to patients who did not receive sedation. Further, there were no differences between sedated and nonsedated patients with respect to survival after admission to the acute PCU (eight days and seven days, respectively; P=0.12). Survival after the diagnosis of the primary tumor (18 months vs. 26 months) and diagnosis of the metastases (12 months vs. 18 months) was shorter for patients who received sedation compared to patients who did not, although this only reached statistical significance for the latter.

Table 2. Characteristics of Sedated and Nonsedated Patients

		Sedated Patients; n=68		Nonsedated Patients; n=89		P-value
		n	%	n	%
	Sex					0.94
	Male	31	46	40	45
	Female	37	54	49	55
	Primary tumor
	Lung	15	22	12	13	0.19
	Gastrointestinal tract	14	21	5	6	<0.01
	Breast	11	16	22	25	0.16
	Genitourinary tract	7	10	17	19	0.13
	Head and neck	5	7	5	6	0.66
	Melanoma	8	12	6	7	0.27
	Sarcoma	5	7	7	8	0.91
	Other	3	4	15	17	0.02
	Age (mean [SD])	55 [12]		59 [12]		0.04
	Survival after primary cancer diagnosis (months) (median [range])	18 [1–292]		26 [1–230]		0.21
	Survival after metastases diagnosis (months) (median [range])	12 [0–244]		18 [0–201]		0.05
	Survival after admission (days) (median [range])	8 [0–38]		7 [0–38]		0.12

Boldfaced numbers = significant results.

Table 3 shows that there were no significant differences between sedated and nonsedated patients' symptoms at admission. Pain was present in 79% of the patients who received palliative sedation and in 87% of the patients who did not. Constipation, dyspnea and nausea/vomiting were present in 23%–48% of the patients, whereas less than 15% of the patients were diagnosed with delirium or clinically relevant anxiety at admission. On average, patients in both groups had two out of the six listed symptoms.

Table 3. Symptom Prevalence of Sedated and Nonsedated Patients at Admission

		Sedated Patients; n=68		Nonsedated Patients; n=89		P-value
		n	%	n	%
	Pain	69	79	59	87	0.23
	Constipation	35	40	32	48	0.46
	Dyspnea	28	32	20	29	0.77
	Nausea/vomiting	20	23	17	25	0.77
	Delirium	9	11	8	12	0.78
	Anxiety	6	7	6	9	0.66
	Number of symptoms (median [range])	2 [0–4]	2 [0–4]	0.32

Table 4 shows that during the interval of 48–25hours before death, sedation was started for 13 patients, and during the interval of 24–0hours before death, sedation was started for 45 patients. The table further presents the patients' symptoms on the day on which sedation was started and the symptoms of patients for whom on the same day (relative to the moment of their passing away) sedation was not (yet) started. Pain was a common symptom in both patient groups. It was present in the interval of 48–25hours before death in 69% (sedation started) and 61% (sedation not [yet] started) of the patients, and in the interval of 24–0hours before death in 42% and 49%, respectively. In both intervals, patients for whom sedation was started significantly more frequently suffered from dyspnea (69% in 48–25hours before death and 50% in 24–0hours before death) than patients for whom sedation was not (yet) started (38% in 48–25hours before death and 31% in 24–0hours before death). In the interval of 24–0hours before death, delirium was more frequently present in patients for whom sedation was started than in the nonsedated group (29% vs. 13%, P=0.04). This difference was also observed (though not significantly) in the interval of 48–25hours before death (31% vs. 23%, P=0.54).

Table 4. Symptom Prevalence in Patients for Whom Sedation Was Started Compared to Patients Who Were Not Sedated

		48–25 hours before deatha					24–0 hours before deathb
		Sedation Started in This Period; n=13		Not sedated; n=121		P-Value	Sedation Started in This Period; n=45		Not Sedated; n=89		P-value
		n	%	n	%		n	%	n	%
	Pain	9	69	74	61	0.63	19	42	44	49	0.33
	Constipation	5	38	41	34	0.82	1	2	4	4	0.68
	Dyspnea	9	69	46	38	0.04	22	50	28	31	0.05
	Nausea/vomiting	2	15	23	19	0.72	4	9	16	18	0.17
	Delirium	4	31	28	23	0.54	13	29	12	13	0.04
	Anxiety	0	0	15	21	—	7	16	10	11	0.49
	Number of symptoms		2		2			2		2

Boldfaced numbers = significant results.

a Number of patients in this period: 10 patients were already sedated (in seven patients, PS was started >72hours prior to death) for 13 patients, symptom prevalence is missing because these patients were not yet admitted. b Number of patients in this period: 23 patients were already sedated.

Discussion 

In the studied acute PCU, palliative sedation was used for almost half of all deceased patients prior to their death. Based on an extensive review of medical and nursing records of all patients who died, our study provides an overview of the characteristics of the patients who received palliative sedation and compares them to patients who did not. Also, characteristics of the palliative sedations are described.

Three findings of our study deserve particular attention. First, the finding that for 43% of the patients dying at the acute PCU, palliative sedation was used prior to death. This seems a rather high percentage; most other studies report lower frequencies regarding the use of palliative sedation in a PCU or hospice. In a Taiwanese PCU, palliative sedation was used for 28% of all deceased patients.13 A multicenter study conducted in Israel, South Africa and Spain reported frequencies of 15%–36%.12 Further, Muller-Busch et al. described that 15% of the deceased patients in a German PCU were given palliative sedation in their last 48hours of life.10 In a hospice in the UK, sedation was given to 48% of the patients; however, patients who were mildly sedated were also included in this study and the sedation was not necessarily deep or until death.11 Our study found a higher frequency of palliative sedation, which can be explained by the fact that this study was conducted at an acute PCU in a specialized cancer center. The main purpose of the studied acute PCU is to provide symptom control for cancer patients in an advanced stage of their disease, and to discharge them after adequate levels of symptom control have been reached. Since most patients (79% during the study period) are discharged to receive end-of-life care in an outpatient or hospice setting, only patients with the worst conditions died in the acute PCU. It can be expected that refractory symptoms are more common among such patients compared to patients dying in hospices in general, resulting in a higher frequency of palliative sedation.

Second, it was striking that in our study the duration of palliative sedation was relatively short, with a median of 19hours; 68% of the patients died within 24hours. Most other studies in PCUs show larger continuous periods of palliative sedation, mostly varying between two and four days.10, 11, 12, 13, 14 There are several potential explanations for this short duration of palliative sedation. It could be that the availability of specialized palliative treatments at the acute PCU may put off the moment when palliative sedation is a true “last resort” treatment. Another explanation may be that the most important indication for palliative sedation, terminal restlessness, typically has an onset in the period very close to death. For those patients, an indication for palliative sedation would only occur in the last days to hours before death. Lastly, the short duration of sedation may indicate that the sedating medications or the forgoing of life-sustaining treatments hastened the patient's death. However, survival after admission to the acute PCU did not differ between sedated and nonsedated patients, which suggests that palliative sedation does not hasten death when used for patients with a limited life expectancy. This finding is in line with other studies as well.11, 13, 19

A last finding that deserves attention is the differences and similarities between sedated and nonsedated patients. Patients who received sedation were, on average, younger, suffered more often from cancer of the digestive tract, and had a shorter survival after discovery of the tumor and metastases. This may indicate that sedation is, on average, preceded by a relatively aggressive disease trajectory. Further, at admission, both groups did not differ in the prevalence of symptoms: symptoms at admission apparently do not predict whether a patient ultimately needs sedation. However, patients who ultimately received sedation more often suffered from dyspnea and delirium prior to the onset of sedation compared to nonsedated patients at a similar moment before death. This higher prevalence of dyspnea and delirium before death seems to indicate a higher complication rate during admission. Nevertheless, a considerable proportion of patients who were not sedated still suffered from dyspnea or delirium. This suggests that other factors that were outside the scope of our retrospective review of records may have played a role too, such as the severity of symptoms or other nonregistered patient characteristics.

Finally, it seems noteworthy to explain that terminal restlessness was the most frequently mentioned indication for the use of sedation (in about two-thirds of the patients) but delirium was mentioned, in only one-third of sedated patients. This can be explained by the fact that a delirium was only noted in the records when a physician had clinically diagnosed it. This is not true for the notion of symptoms, which are based on patients' complaints and/or clinical observations only. Obviously, formally diagnosing a delirium is not always feasible and helpful shortly before death.

There are potential limitations to these findings. First, it can be questioned to what extent medical and nursing records reflect true practice and are filled in adequately. Furthermore, they provide rather summary information and only give limited insight into, for example, the decision-making process regarding palliative sedation. However, the care plans related to the different symptoms in the nursing records were designed to give detailed insight into the prevalence of the symptoms and the results of the given care. Second, we have no insight into the severity of symptoms; we merely know whether or not a symptom was present. Lastly, it is unsure whether these findings are generalizable to other palliative care settings because our study was concerned with an acute PCU in a tertiary cancer setting where palliative sedation was relatively frequently used.

We conclude that in the studied acute PCU, patients who received palliative sedation did not present different symptoms than nonsedated patients at admission, but on the day that the sedation was started, they more often had a delirium and dyspnea. This may have practical implications. Timely anticipation of palliative sedation is important for a careful decision-making process, which may ultimately provide patients and their relatives with a good death. For anticipating the decision of palliative sedation, our study shows that it seems important to be especially attentive to signs of delirium and dyspnea in the last few days of life. Further research is needed that adds to our understanding of other factors that are associated with the decision to use palliative sedation, including the severity of the patients' symptoms.