Frequency and predictors of palliative sedation among patients with cancer who died in a specialist inpatient palliative care unit: a retrospective study

Abstract

Background

Palliative sedation (PS) is a therapeutic measure used in end-of-life care to alleviate the suffering caused by refractory symptoms. The objectives of our study were to determine the frequency of PS in a specialized inpatient acute palliative care unit (APCU), describe the characteristics of patients who received PS, compare patients who required PS with those who did not, and compare patients who received a single medication for PS with those who received multiple medications.

Methods

This was a retrospective study of 444 patients with cancer who died in a specialized inpatient APCU of a tertiary public hospital.

Results

A total of 167 patients received PS (38%). The most frequent indication was delirium (64%). The mean duration of PS was 49 h. Patients who received PS were significantly younger (p = 0.001), had higher anxiety (p = 0.024), had longer hospital admissions (p = 0.001), were more likely to have a spouse as their primary caregiver (p = 0.003), were more aware of their prognosis (p = 0.024), and had more advance directives (p = 0,001) than those who did not receive PS. Forty-six patients (28%) required two or three drugs for PS. They had a longer PS (p = 0.002) and received more parental hydration (p = 0.015) than did those who received one drug. Patients receiving midazolam as first-line treatment were less likely to receive a second drug (p = 0.003).

Conclusions

A total of 38% of patients with cancer receiving specialized palliative care required PS before death. Patients who received PS were younger, had a longer hospital stay, were more likely to have their spouse be their primary caregiver, were better informed and had more advance directives. 28% of sedated patients required more than one drug, which was associated with a longer PS, the maintenance of parental hydration during PS and the initiation of PS with levomepromazine instead of midazolam.

Background

Palliative sedation (PS) is defined as the monitored use of drugs to decrease the level of consciousness, with the aim of relieving the otherwise unmanageable burden of symptoms (refractory suffering) in a way that is ethically acceptable to the patient, their family, and the clinical team [1]. In this context, refractory suffering refers to the presence of an intolerable condition or symptom experienced by the patient as extremely distressing, which cannot be alleviated by a multidisciplinary team with experience in palliative care, in a given environment, and within a sufficient time frame [2]. The concept of proportionality in PS has recently gained emphasis. It involves adjusting depth and timing, titrating sedation to the minimum effective dose for symptom relief while preserving interaction, and favouring intermittent sedation early in the illness for temporary relief or respite before potential reawakening [3].

The frequency of PS varies greatly across different studies, with a wide range of prevalence rates that can fluctuate between 1% and 88%. Globally, approximately 20–30% of patients at the end of life receive PS. This wide range present in the literature can be explained by various factors, such as the very definition of PS, differences in the medications used, clinical or cultural patterns of the environment, the resources available for symptom control in earlier phases, or the place of death [4]. PS in acute palliative care units (APCUs) within hospital settings appears to be more common, with a reported prevalence exceeding 50%. However, data on the frequency and characteristics of PS in these settings are limited [5]. There are few studies on PS at home, which show a slightly lower prevalence than that reported in the hospital setting [6, 7].

The most common refractory symptoms that lead to PS are delirium, dyspnoea, and pain. In most studies, delirium accounts for 60% of the causes of PS [8]. Other less prevalent symptoms include vomiting, seizures, and psychological symptoms, although the latter indication is not without controversy [9]. PS must have the explicit consent of the patient, or if the patient is unable to provide it owing to their clinical condition, consent from the family [10]. The average survival of sedated patients varies between 1 and 4 days across different studies, and it has been shown that PS does not reduce survival [10].

Sedation should be titrated to the minimum level needed for symptom relief, following the principle of proportionality [3]. A reduction in consciousness can be achieved with various classes of drugs and is intentional, defining it as PS [1]. Benzodiazepines, particularly midazolam, are the first-line choice, especially in the absence of delirium [4, 11, 12]. If midazolam fails, antipsychotics such as levomepromazine are used, especially when sedation is required for delirium. Some guidelines consider it the drug of choice when PS is indicated for delirium; once it has been initiated as a treatment for delirium itself, its dosage is progressively increased, and consequently, the level of sedation is increased [13]. In rare cases, third-line agents such as propofol are needed, although clinical experience remains limited, as these agents account for only 4% of reported cases [14–16]. Sedation is considered complex when multiple drugs are needed [17].

There is no, nor should there be, a universally accepted benchmark regarding how frequently PS is used, as its indication must be based on individual patient needs, clinical judgment, and contextual factors. Additionally, there is no consensus among different guidelines on the preferred first-line medications in cases of delirium. Furthermore, existing studies have not clearly established how often PS requires the use of more than one drug [4]. Our objectives were to determine the frequency of PS in a specialized APCU, describe the characteristics of the patients who received PS, compare patients who needed PS with those who did not, and compare those who received one drug for PS with those who received more than one drug.

Methods

Design and setting

A retrospective observational study was conducted in an APCU of a tertiary public hospital: La Paz University Hospital (Madrid, Spain). All patients cared for between January 1, 2014, and December 31, 2016, were included. The medical records of all patients admitted during this period were reviewed.

Patients

Patients with cancer who died in the APCU during the study period and required PS.

Palliative sedation

In the event of refractory symptomatic distress refractory to other therapeutic measures, in the context of end-of-life care, the possibility of PS was jointly discussed by the clinical team (palliative medicine specialists, nurses, and psychologists), the patient (if the clinical situation allowed), and the family (primary caregiver). The drugs used were midazolam, levomepromazine, and propofol, which were administered parenterally (subcutaneously or intravenously), with the goal of reducing the level of consciousness to the point that it was clinically and subjectively considered sufficient for symptom relief. This was continuously monitored by the clinical team in charge of the APCU. Secondary medications were administered continuously to improve the maintenance of PS.

We only classified an intervention as PS when this intent was explicitly documented in the medical record, in order to distinguish it from the use of the same medications for general symptom control. At our hospital, it is standard and mandatory clinical practice to clearly document the intention to initiate PS in the medical record whenever it is carried out.

Outcomes

Upon admission, the following variables were recorded: date of admission, reason for admission, clinical and demographic characteristics (sex, age, tumour type, metastatic disease), functionality according to the Palliative Performance Scale (PPS) and Barthel Index, levels of anxiety and depression according to the Hospital Anxiety and Depression Scale (HADS A and D), presence of advance directives, awareness of prognosis, CAGE test, relationship of the primary caregiver, and whether the patient lived alone, with company, or in a care facility.

Upon the initiation of PS, the following variables were recorded: the primary indication for PS (the main refractory symptom that led to the decision to initiate sedation), secondary indications (other coexisting symptoms that contributed to patient distress but were not the primary trigger for PS), and the drugs used.

During PS and after death, the following variables were collected: date of death, length of stay (days), duration of PS (hours), final dose of the drug(s) used for PS (total dose received, including both the continuous maintenance dose and any additional doses administered on an as-needed basis), and concomitant use of benzodiazepines, antidepressants, adjuvants for pain treatment, antipsychotics, antiepileptics, corticosteroids, opioids, and parenteral hydration.

Statistical analyses

A basic descriptive analysis of the variables of interest from the collected data was performed. For this purpose, the Statistical Package for the Social Sciences (SPSS) version 24.0 for Windows was used. A p value of < 0.05 was considered to indicate statistical significance. Qualitative variables are represented by their frequency distributions and percentages, whereas quantitative variables are represented by their medians. Comparisons between the different variables were carried out via the chi-square test.

Results

During the study period, 1176 patients were admitted to the APCU, 1040 (88%) of whom were patients with cancer. Among the patients with cancer, 444 (43%) died. Figure 1 represents the general outline of the study.

Fig. 1.

Patients treated in the Palliative Care Unit between 01/01/14 and 31/12/16

A total of 38% (n = 167) of the patients with cancer received PS. The pharmacy data for two patients were not available, so 165 patients were analysed. Table 1 summarizes the characteristics of the group of patients who received PS compared with those who died without receiving PS. Between the two groups, there were no differences in tumour type, sites of metastasis, functional status according to the PPS and Barthel Index, depression levels as measured by the HADS-D, CAGE test results, or living situation (alone, with company, or in a care facility).

Table 1.

Characteristics of patients who did not receive PS vs. those who received PS

	No PS (n = 277) N (%)	PS (n = 167) N (%)	P Value
Mean Age (years) ± SD	72 ± 13	65 ± 15	0.001
Sex
Female	110 (40)	59 (35)	0.357
Tumour type
Central nervous system	2 (1)	7 (4)	0.102
Lung	52 (19)	41 (25)
Gastrointestinal	94 (34)	53 (32)
Gynecological	32 (12)	21 (13)
Urothelial	36 (13)	19 (11)
Head and neck	8 (3)	4 (2)
Other	53 (19)	22 (13)
Reason for admission
Pain	59 (21)	50 (30)	0.04
Dyspnea	55 (20)	27 (16)
Fatigue	35 (13)	17 (10)
Bowel obstruction	11 (4)	14 (8)
Delirium	25 (9)	19 (11)
Other	92 (33)	40 (24)
Mean length of stay (days) ± SD	8 ± 7.5	11 ± 10	0.001
Metastases
Liver	112 (45.5)	67 (46)	0.945
Central nervous system	33 (13)	24 (16)	0.402
Mean Barthel Index ± SD	25 ± 21	29 ± 26	0.144
Mean PPS ± SD	36 ± 11	36 ± 1	0.963
Mean HADS ± SD
A	7 ± 4	9 ± 2	0.024
D	8 ± 3	9 ± 2	0.271
Advance directives
Yes	49 (18)	53 (32)	0.001
Awareness of prognosis
Yes	109 (39)	84 (50)	0.024
Relationship of the primary caregiver
Partner	123 (51)	93 (66)	0.017
Child	83 (34)	28 (20)
Other	31 (13)	16 (11)
No relative	5 (2)	3 (2)
Lives
Alone	32 (12)	15 (9)	0.139
Company	223 (83)	144 (89)
Care facility	12 (5)	3 (2)
CAGE positive	29 (11.5)	17 (11)	0.903

Percentages may not total 100 because of rounding. SD denotes Standard Deviation. PPS denotes Palliative Performance Status. HADS denotes The Hospital Anxiety and Depression Scale

Patients receiving PS were significantly younger (p = 0.001), had a longer hospital stay (p = 0.001), were better informed about their prognosis (p = 0.024), and had advanced care planning more frequently (p = 0.001). It was also more common for the primary caregiver to be a spouse (p = 0.003) and less common for it to be a child (p = 0.003). Patients who received PS had higher HADS-A scores (p = 0.024).

The most frequent indication for PS was delirium (64%), followed by dyspnoea (20%). The mean duration of PS was 49 h. Among the patients who received PS, 72% (n = 119) required only one drug (midazolam or levomepromazine), and 28% (n = 46) required two or more drugs (midazolam, levomepromazine, or propofol). 2% (n = 3) required a third dose of propofol.

Table 2 summarizes the characteristics of the group of patients who received one drug compared with those who received two or more drugs. Patients who received two or more drugs had significantly longer PSs (p = 0.02), and parenteral hydration was more frequently maintained during PS (p = 0.015). Delirium was the most frequent primary indication among those who received more than one drug for PS (n = 35, 78%). However, no statistically significant differences were found in the group of patients who received only one drug (59%) (p = 0.078). The differences in the use of other medications, either prior or concomitant, between the two groups (benzodiazepines, antidepressants, adjuvants for pain treatment, antipsychotics, anticonvulsants, corticosteroids, and opioids) were investigated, revealing only statistically significant differences in the use of antidepressants, which were more common in the group of patients who needed more than one drug for PS (p = 0.078). No other differences were found between the two groups.

Table 2.

Characteristics of patients who received one drug vs. two or more drugs

	Patients who received one drug (n = 119)	Patients who received two or more drugs (n = 46)	P Value
	N (%)	N (%)
Mean Age (years) ± SD	66 ± 15	63.9 ± 13	0.418
Sex
Female	40 (34)	18 (39)	0.506
Tumour type
Central nervous system	5 (4)	2 (4)	0.325
Lung	32 (27)	9 (20)
Gastrointestinal	39 (33)	14 (30)
Gynecological	10 (8)	10 (22)
Urothelial	15 (13)	3 (6.5)
Head and neck	3 (2.5)	1 (2)
Other	15 (13)	7 (15)
Reason for admission
Pain	30 (26)	20 (43.5)	0.346
Dyspnea	21 (18)	6 (13)
Fatigue	13 (11)	3 (6.5)
Bowel obstruction	10 (8.5)	3 (6.5)
Delirium	14 (12)	6 (13)
Other	29 (25)	8 (17)
Mean length of stay (days) ± SD	9 ± 6	10.5 ± 7	0.125
Metastases
Liver	47 (46)	20 (48)	0.866
Central nervous system	15 (15)	9 (21)	0.325
Mean Barthel Index ± SD	29 ± 26	29 ± 21	0.849
Mean PPS ± SD	35 ± 11	37 ± 10	0.296
Mean HADS ± SD
A	9 ± 4	8 ± 2	0.522
D	9 ± 3	10 ± 2	0.194
Advance directives
Yes	25 (23)	8 (20.5)	0.715
Awareness of prognosis
Yes	50 (43)	23 (51)	0.337
Relationship of the primary caregiver
Partner	63 (64)	29 (71)	0.423
Child	22 (22)	6 (15)
Other	10 (10)	6 (15)
No relative	3 (3)
Lives
Alone	12 (10)	3 (7)	0.759
Company	102 (88)	41 (91)
Care facility	2 (2)	1 (2)
CAGE positive
Positive	12 (11)	5 (12)	0.876
Maintained Parenteral hydration	21 (18)	29 (62)	0.015
Mean duration of PS (hours) ± SD	42 ± 41 (11–208)	68 ± 50.5 (2-216)	0.002
Indication for PS
Delirium	65 (59)	35 (78)	0.077
Dyspnea	27 (24)	4 (9)
Pain	4 (4)	3 (7)
Psychological distress	5 (4.5)	2 (4)
Other	10 (9)	1 (2)
Secondary indication for PS
Delirium	7 (23)	4 (20)	0.592
Dyspnea	8 (26)	3 (15)
Pain	6 (19)	3 (15)
Psychological distress	6 (19)	8 (40)
Other	4 (13)	2 (10)

Percentages May not total 100 because of rounding. SD denotes standard deviation. PPS denotes palliative performance status. HADS denotes the hospital anxiety and depression scale. PS denotes palliative sedation

Among the 46 patients who needed two or three drugs to achieve PS, 67% (n = 31) received midazolam as the first drug, and 33% (n = 15) received levomepromazine. Patients who received midazolam first needed a second drug for PS less frequently than those who received levomepromazine first did (p = 0.003). Among the patients who received two or three drugs for PS indicated by delirium, 66% (n = 23) received midazolam as the first drug, and 34% (n = 12) received levomepromazine.

The mean dose of each drug at the end of PS was as follows: 56 mg/24 h for midazolam; 80 mg/24 h for levomepromazine; and 278 mg/24 h for propofol. The dose of midazolam required by patients who received PS with two or three drugs was significantly greater than that required by those who received only one drug (64 mg vs. 52 mg; p = 0.03). No further differences were found between these two groups regarding drug doses.

The mean age of the three patients who received propofol was 15 years younger than that of the other patients who received PS (50 vs. 65). The indication for PS in all three patients was delirium, and all had metastatic disease. No other significant differences were observed.

Discussion

Under the care of an interdisciplinary team led by palliative medicine specialists, 38% of patients with cancer admitted to the APCU of a tertiary public hospital required PS. Among them, one in four needed more than one drug to control symptoms. Patients who received PS were significantly younger, had longer admissions, were more likely to have their spouse as the main caregiver, were better informed about their prognosis, and more frequently had advanced directives than those who did not receive PS. The main factors associated with the need for more than one drug for PS were a longer PS duration, maintaining parenteral hydration during PS, and starting PS with levomepromazine instead of midazolam. These findings highlight that PS has a major role in alleviating severe distress among patients with cancer close to the end of life.

The average duration of sedation (49 h) reported in our study is similar to that reported in the literature (19–82 h) [4]. The frequency of PS use in our study is similar to findings from other hospital-based studies [18–21]. There is no universally accepted standard—and arguably, there should not be one—for the frequency of PS, which must always be considered on the basis of patient characteristics, the setting, and access to healthcare resources, among other factors [22]. A recent systematic review indicated that, within the healthcare setting, hospitalized patients are more likely to receive palliative sedation than those receiving care at home or in a nursing facility [23]. Since APCUs admit patients with a high level of distress, it is likely that the frequency of PS will be lower in other settings, such as regular inpatient units or community settings. Another explanation for the higher frequency of PS in the hospital setting could be the multidisciplinary environment and the continuous presence of qualified healthcare professionals. This setting may facilitate the management of the ethical and communication challenges inherent to PS for both families and healthcare providers [24, 25].

With respect to the differences between patients who received PS and those who died without PS, those who received PS were younger, a finding common to many previous studies [4, 7]. Likewise, patients who receive PS are better informed about their prognosis, although several studies have identified a lack of prognostic information as a contributing factor to the need for PS [7, 26]. We also observed that patients who received PS more frequently had advanced directives, which may be related to a better understanding of their disease prognosis; however, we did not find studies reporting this association. One possible explanation is that younger patients may be more likely to seek or receive information about their illness; this could partly account for the higher proportion of well-informed individuals observed among those who received PS [27, 28]. However, this remains a hypothesis and should be interpreted with caution, as we did not find statistical evidence to confirm this correlation in our study.

In our sample, spouses were more often primary caregivers in the PS group, while adult children assumed this role more frequently in the non-PS group. This may reflect age-related family dynamics, with younger patients (PS group) more likely to have a partner as a caregiver, and older patients (non-PS group) relying on their children. However, this interpretation remains speculative and should be explored further in future studies.

Patients who received PS had significantly higher HADS-A (anxiety) scores than those who were not sedated (p = 0.024). Emotional distress may contribute to refractory suffering, leading to PS. Additionally, as previously described, this could also be explained by the younger average age of the PS group, since younger age may be associated with greater emotional distress [14].

We did not find a relationship between previous alcohol consumption and a greater frequency of PS. Similarly, in our experience, alcohol consumption was not associated with a greater need to administer more than one drug to achieve PS. This finding supports the idea that drug doses should not initially be adjusted for PS in patients with a positive CAGE questionnaire [4].

Midazolam was the most commonly used drug at our centre, followed by levomepromazine and propofol. These data are similar to those in the literature, where midazolam accounts for approximately 70% of PS cases and propofol, 4% [4]. These medications also align with the recent European Association for Palliative Care (EAPC) recommended framework on palliative sedation, which advocates for a stepwise pharmacological approach: step 1 with midazolam (or lorazepam as an alternative), step 2 with levomepromazine (or chlorpromazine), and step 3 with propofol [3].

The frequency of the use of midazolam for PS, as the first-line drug, both in monotherapy and in combination with neuroleptics, has increased in studies conducted specifically in APCUs, where only a small subgroup of patients do not receive midazolam at the beginning of PS [12]. The choice of drug for PS should be personalized not only on the basis of patient characteristics but also on the basis of the clinical context and location of the PS.

In our series, those who received midazolam as a first-line drug were less likely to need a second drug (p = 0.003) than those who received levomepromazine first. The scientific literature is quite varied regarding recommendations for PS drugs. Most authors do not consider levomepromazine a first-line drug, recommending starting with midazolam for all PS indications, reserving levomepromazine for combination with midazolam or as a second-line option when midazolam fails, especially in cases of neuropsychiatric symptoms [12]. However, some clinical practice guidelines consider levomepromazine as a first-line option when delirium is the cause of PS, on the basis of the reasoning that benzodiazepines, as initial treatments for delirium, may worsen symptoms [29]. Nevertheless, there is no robust evidence supporting this indication for levomepromazine [13]. Our study may reflect the potential superiority of midazolam over levomepromazine in symptom control for all PS indications, including delirium, which suggests that midazolam is the most effective drug for PS, regardless of the indication. This idea is supported by recent research favouring the use of benzodiazepines, in combination with neuroleptics, in patients with refractory agitation [30–32].

The dose of midazolam was greater in the group that received two or three drugs, which is expected since those who needed more than one drug had more difficult PS, and the midazolam dose was maximized before adding another drug. However, the dose of levomepromazine was similar in both groups. There is no established consensus on the optimal levomepromazine dosage for PS, with wide dose ranges reported across different studies [13].

Notably, our results revealed a relationship between maintaining parenteral hydration and the need for two or three drugs (p = 0.015). Several previous studies discuss the role of hydration in the last days of life, where it has been shown not to improve symptoms or prolong survival; however, sometimes, discontinuing hydration can cause emotional distress for the patient and family [33]. The latest clinical guidelines recommend independent decision-making for hydration in the context of PS [3]. Our study may support the withdrawal of parenteral hydration in the last days of life on the basis of this association with the need for more than one drug for PS. However, more prospective studies are needed to better explore the role of parenteral hydration at the end of life and its relationship with PS.

Patients with complicated PS more frequently had delirium as the main indication than those who required only one drug, although this difference was not statistically significant. Delirium was the main cause of PS in most previous studies, regardless of the type and number of drugs used [4].

With respect to survival differences after the initiation of PS, patients who required two or three drugs had a longer PS duration (p = 0.002), suggesting that PS is not associated with hastened death, which is consistent with previous literature [8]. Our work aligns with a Cochrane review, which mentioned that 12 of the 13 reviewed studies reported longer survival in the PS group [4].

The main limitation of this study is its retrospective nature, which results in the loss of main data and the exclusion of variables not reflected in patients’ medical records. Another limitation is that it was conducted in a specialized APCU where patients are often referred due to the high severity of symptomatic distress, which may be associated with a higher frequency of PS. However, this study included a large number of patients, allowing conclusions with reasonable clinical validity to be drawn.

In conclusion, in our study, 38% of patients with cancer who died in an acute specialized APCU required PS. These patients were younger, had longer hospital stays, with their spouse as the primary caregiver, were better informed, and more frequently had advanced directives. 28% of sedated patients required more than one drug, which was associated with longer PS, maintaining parenteral hydration during PS, and starting PS with levomepromazine instead of midazolam. These results reinforce the value of the PS as a frequent and important technique for patients with cancer facing high symptomatic distress.

More studies are needed to determine the optimal frequency of PS, to identify factors that predispose patients to require more than one drug for PS, and to demonstrate that midazolam should be the drug of choice in all cases. Importantly, research on PS is very challenging because of the impracticality of randomized controlled trials with a control group and because observational studies are more susceptible to bias.

Abbreviations

PS

Palliative sedation

APCU

Acute palliative care unit

PPS

Palliative Performance Scale

HADS

Hospital Anxiety and Depression Scale

SPSS

Statistical Package for the Social Sciences

JTT

Juan Luis Torres Tenor

PVC

Paula Villalba Cuesta

Author contributions

All authors made substantial contributions to the study. JTT and PVC were responsible for data acquisition, analysis, and interpretation. The initial draft of the manuscript was written by JTT, and all authors reviewed and provided substantive revisions. All authors have approved the submitted version (and any substantially modified version involving their contribution) and agree to be personally accountable for their own contributions. Furthermore, they commit to ensuring that any concerns regarding the accuracy or integrity of any part of the work, even those in which they were not directly involved, are appropriately investigated, resolved, and documented in the literature.

Funding

This study did not receive any funding.

Data availability

The datasets used and analysed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The project was approved by the Ethics and Clinical Research Committee of La Paz University Hospital (code: PI-3556). Given the characteristics of the study, a waiver of informed consent was requested for the included patients, which was granted. The study was conducted in accordance with the principles of the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.