Abstract

Background: Palliative sedation therapy (PST) is indicated for and used to control refractory symptoms in cancer patients undergoing palliative care. We aimed to evaluate whether PST has a detrimental effect on survival in terminally ill patients.

Methods: This multicenter, observational, prospective, nonrandomized population-based study evaluated overall survival in two cohorts of hospice patients, one submitted to palliative sedation (A) and the other managed as per routine hospice practice (B). Cohorts were matched for age class, gender, reason for hospice admission, and Karnofsky performance status.

Results: Of the 518 patients enrolled, 267 formed cohort A and 251 cohort B. In total, 25.1% of patients admitted to the participating hospices received PST. Mean and median duration of sedation was 4 (standard deviation 6.0) and 2 days (range 0–43), respectively. Median survival of arm A was 12 days [90% confidence interval (CI) 10–14], while that of arm B was 9 days (90% CI 8–10) (log rank = 0.95, P = 0.330) (unadjusted hazard ratio = 0.92, 90% CI 0.80–1.06).

Conclusion: PST does not shorten life when used to relieve refractory symptoms and does not need the doctrine of double effect to justify its use from an ethical point of view.

introduction

Some far advanced and terminally ill cancer patients experience symptoms that cannot be controlled even by the best supportive and palliative therapies. For these ‘refractory symptoms’, it is necessary to turn to palliative sedation therapy (PST), defined as ‘the use of sedative medications to relieve intolerable suffering from refractory symptoms by a reduction in patient consciousness’ [1, 2].

Many clinicians argue that palliative sedation does not necessarily mandate sedation to total unconsciousness and suggest that PST can vary in terms of level (mild, intermediate, and deep), duration (intermittent or continuous), and pharmacological characteristics (primary, by drugs not proven to be effective to relieve the underlying symptom, or secondary, by medications pharmacologically effective for the relief of underlying distress, with secondary somnolence) [3]. Other authors classify sedation as sudden or proportional on the basis of whether it is established rapidly [4]. ‘Emergency sedation’ is made in immediately preterminal patients with overwhelming symptoms [5] for catastrophic events such as massive bleeding, severe dyspnea, agitated delirium, or pain. A further, highly specific but potentially useful subtype of PST is ‘respite sedation’, a procedure involving temporary and time-limited sedation [6]. Finally, the possibility of using ‘routine’, ‘infrequent’ or ‘extraordinary’ sedation has also been put forward [7]. It is therefore clear that deep, continuous sedation represents one of the several forms of PST, namely that used to manage the very last stages of the disease.

The definition of refractory symptom proposed by Cherny and Portenoy [8] in 1994 is now widely accepted: ‘symptom for which all possible treatment has failed, or it is estimated that no methods are available for palliation within the time frame and the risk–benefit ratio that the patient can tolerate’. Psychological and existential distress as an indication for palliative sedation is a controversial issue [9, 10], and some authors have recently identified useful strategies for dealing with feelings of hopelessness or futility in patients approaching end of life [11–13].

An inappropriate and excessive use of palliative sedation is directly linked to limited experience in palliative care or to operator burnout and fatigue [14, 15]. The use of sedation for relief of symptoms is, in our opinion, open to abuse and it cannot be denied that some physicians ostensibly administer medication to relieve symptoms, but with a covert intention of hastening the patient's death [16–18].

Some authors have hypothesized a negative impact of PST on survival, going so far as to label this approach ‘slow euthanasia’ or terminal sedation [19]. Even if such an impact were present, the use of PST could nevertheless be ethically justified by referring to the ‘doctrine of double effect’ [20]. This doctrine says that if doing something morally good has a morally bad side-effect, it is ethically acceptable to do it provided the bad side-effect was not intended. This is true even if it is foreseen that the bad effect would probably happen.

The results of some original works and of some literature reviews [21–29] show that, in populations of patients undergoing palliative care, sedation does not have a negative impact on survival and that the doctrine of double effect is not needed to justify the use of palliative sedation [30, 31]. Obviously, the study method with the highest level of clinical evidence, i.e. the randomization of patients into sedation or nonsedation arms, cannot be used in this clinical situation.

As far as we know, ours is the first study to prospectively match sedated patients (cohort A) with nonsedated patients (cohort B) in such a way that the two arms differ only in terms of one characteristic, i.e. sedation. The primary hypothesis of the study was that PST does not have a detrimental effect on survival, and our main aim was to verify this hypothesis.

methods

study population

This multicenter, observational, prospective, nonrandomized population-based study was conducted from March 2005 to December 2006. Two cohorts of hospice patients were enrolled during this period, one submitted to PST (cohort A) and the other managed without this procedure (cohort B). Cohort A was comprised of all far advanced cancer patients of either gender and any age group consecutively admitted to the study hospices for psychosocial reasons or because of uncontrolled symptoms or problems relating to the terminal phase of the disease. The control group (cohort B) was recruited from the same hospices and matched to cohort A for gender, age class (≤65 and >65 years), reason for admission (psychosocial, uncontrolled symptoms, and terminal phase), and Karnofsky performance status (KPS) subdivided into three classes (10–20, 30–40, and ≥50). Recruitment of cohort B was feasible because cohort A patients were registered at the time of beginning sedation, together with their matching variables. Constantly updated lists of cohort A patient characteristics were regularly sent to the centers so that patients fulfilling matching variable criteria could be included in cohort B. The therapeutic approach for each patient was clinically driven.

The criteria used to indicate the need for PST, based on the presence of refractory symptoms according to Cherny's definition [8], were discussed, compared, and standardized by center coordinators in the protocol designed before the study was activated. The decision to begin PST was made in each center following a meeting between the members of the hospice team, composed of attending physicians, nursing staff, a psychologist, and a spiritual counselor. All teams had received training in palliative care and had had at least 5 years experience in a hospice proposing this type of therapy to patients or, when necessary, to their primary caregiver.

Demographic characteristics, KPS, activities of daily living (ADLs), symptoms at study entry, information on whether or not the patient lived alone or with his/her family, primary tumor and metastatic sites, date and reason for admission, and date of death were routinely registered for each patient enrolled on to the study. From October 2005 onward, the Palliative Prognostic Score (PaP Score), a prognostic index proposed and validated by our group in various independent case series, was calculated [32–35]. The following information was collected for patients undergoing palliative sedation: start/finish dates of treatment, refractory symptoms, and type and dose of drugs. Decision-making criteria for all therapies, including those with opioids and neuroleptics, and life support interventions, such as rehydration, were deemed homogeneous in the two patient cohorts because a requisite of the study protocol was that the best palliative care should be provided for both groups of patients. As this was a clinical observational study, physicians were free to choose the drugs used for sedation as well as sedation procedures, having only agreed upon the criteria indicating the need for PST [8].

The duration of PST was calculated in days (start/finish date, with only full days taken into consideration). The study was approved by the ethics committee of each participating institution.

statistical analysis

The primary objective was to evaluate the overall survival (OS) of two patient cohorts matched for clinical characteristics, one treated with PST (cohort A) and the other managed as per routine hospice practice (cohort B). The secondary objective was to assess the OS of the two cohorts classified into subgroups as a function of the PaP Score.

The primary efficacy hypothesis of the study was that the OS of cohort A patients would not be inferior to that of cohort B. The hazard ratio (HR) was the ratio between the hazard rate for death in cohort A and B patients. Palliative sedation was not considered to be inferior to standard clinical practice with respect to duration of survival if the upper boundary of the 90% confidence interval (CI) for HR was <1.30. Hypothesizing that the median survival in cohort B would be 9 days, an HR = 1.30 would correspond to a median survival of 7 days for cohort A patients. The difference between these two values is the largest that could be judged as being clinically acceptable. Setting α = 0.05 (one-tailed test) and in order to have a 90% power of rejecting the null hypothesis that the HR is 1, it was estimated that 518 patients were needed.

Survival time was calculated as the number of full days from the last hospice admission until the date of death from any cause. OS was estimated using the Kaplan–Meier life-table method [36] and survival curves were compared by the log-rank test [37]. An exploratory Cox multiple regression analysis [38] was also carried out including all the matched variables. Unadjusted and adjusted HRs for OS were estimated according to the Cox model, together with their relative 90% CIs.

P’ values were not adjusted for the multiplicity of the tests carried out. Statistical analyses were carried out using SAS statistical software (release 9.1; SAS Institute, Cary, NC).

results

Five hundred and eighteen patients were enrolled on to this study, of whom 267 were part of the sedation group (cohort A) and 251 were part of the nonsedation group (cohort B). The two cohorts were recruited in four hospices of the Emilia–Romagna region using the same study entry and admission criteria. Patient characteristics are reported in Table 1. The overall prevalence of PST on the patients admitted to the participating hospices was 25.1%, with low variability among the centers (±5%).

Table 1.

Patient characteristics

Variable Total (n = 518)
 
Cohort A (PST) (n = 267)
 
Cohort B (no PST) (n = 251)
 
 n n n 
Age (years)       
    Median value (range) 70 (22–100)  68 (22–88)  72 (29–100)  
Gender       
    Male 287 55.4 149 55.8 138 55.0 
    Female 231 44.6 118 44.2 113 45.0 
Karnofsky performance status       
    10–20 102 19.7 50 18.7 52 20.7 
    30–40 286 55.2 154 57.7 132 52.6 
    ≥50 130 25.1 63 23.6 67 26.7 
PaP Scorea       
    Group A (0–5.5) 48 13.4 23 11.7 25 15.4 
    Group B (5.6–11.0) 171 47.8 101 51.3 70 43.2 
    Group C (11.1–17.5) 139 38.8 72 36.7 67 41.4 
    Not applicableb 27  16  11  
ADL       
    Mean value (SD) 1.8 (1.9)  2.0 (1.9)  1.6 (1.9)  
Reason for admission       
    Uncontrolled symptom 274 52.9 151 56.6 123 49.0 
    Terminal phase 212 40.9 102 38.2 110 43.8 
    Psychosocial 32 6.2 14 5.2 18 7.2 
Living with family       
    Yes 462 89.2 236 88.4 226 90.0 
    No 56 10.8 31 11.6 25 10.0 
Primary site of tumor       
    Lung 103 19.9 68 25.5 35 13.9 
    Colorectum 73 14.1 42 15.7 31 12.3 
    Stomach 69 13.3 25 9.4 44 17.5 
    Breast 37 7.1 21 7.9 16 6.4 
    Pancreas 36 7.0 21 7.9 15 6.0 
    Urogenital tract 29 5.6 11 4.1 18 7.2 
    Hematological cancers 29 5.6 15 5.6 14 5.6 
    Female genital tract 27 5.2 14 5.2 13 5.2 
    Central nervous system 24 4.6 10 3.7 14 5.6 
    Liver 22 4.2 11 4.1 11 4.4 
    Male genital tract 18 3.5 1.5 14 5.6 
    Head and neck 17 3.3 3.4 3.2 
    Others 34 6.6 16 6.0 18 7.1 
Metastatic sites       
    Liver 177 34.2 90 33.7 87 34.7 
    Serosal involvement 168 32.4 91 34.1 77 30.7 
    Soft tissue lymph nodes 159 30.7 83 31.1 76 30.3 
    Bone 124 23.9 69 25.8 55 21.9 
    Lung 98 18.9 58 21.7 40 15.9 
    Locally advanced disease 68 13.1 39 14.6 29 11.6 
    Central nervous system 45 8.7 18 6.7 27 10.8 
Variable Total (n = 518)
 
Cohort A (PST) (n = 267)
 
Cohort B (no PST) (n = 251)
 
 n n n 
Age (years)       
    Median value (range) 70 (22–100)  68 (22–88)  72 (29–100)  
Gender       
    Male 287 55.4 149 55.8 138 55.0 
    Female 231 44.6 118 44.2 113 45.0 
Karnofsky performance status       
    10–20 102 19.7 50 18.7 52 20.7 
    30–40 286 55.2 154 57.7 132 52.6 
    ≥50 130 25.1 63 23.6 67 26.7 
PaP Scorea       
    Group A (0–5.5) 48 13.4 23 11.7 25 15.4 
    Group B (5.6–11.0) 171 47.8 101 51.3 70 43.2 
    Group C (11.1–17.5) 139 38.8 72 36.7 67 41.4 
    Not applicableb 27  16  11  
ADL       
    Mean value (SD) 1.8 (1.9)  2.0 (1.9)  1.6 (1.9)  
Reason for admission       
    Uncontrolled symptom 274 52.9 151 56.6 123 49.0 
    Terminal phase 212 40.9 102 38.2 110 43.8 
    Psychosocial 32 6.2 14 5.2 18 7.2 
Living with family       
    Yes 462 89.2 236 88.4 226 90.0 
    No 56 10.8 31 11.6 25 10.0 
Primary site of tumor       
    Lung 103 19.9 68 25.5 35 13.9 
    Colorectum 73 14.1 42 15.7 31 12.3 
    Stomach 69 13.3 25 9.4 44 17.5 
    Breast 37 7.1 21 7.9 16 6.4 
    Pancreas 36 7.0 21 7.9 15 6.0 
    Urogenital tract 29 5.6 11 4.1 18 7.2 
    Hematological cancers 29 5.6 15 5.6 14 5.6 
    Female genital tract 27 5.2 14 5.2 13 5.2 
    Central nervous system 24 4.6 10 3.7 14 5.6 
    Liver 22 4.2 11 4.1 11 4.4 
    Male genital tract 18 3.5 1.5 14 5.6 
    Head and neck 17 3.3 3.4 3.2 
    Others 34 6.6 16 6.0 18 7.1 
Metastatic sites       
    Liver 177 34.2 90 33.7 87 34.7 
    Serosal involvement 168 32.4 91 34.1 77 30.7 
    Soft tissue lymph nodes 159 30.7 83 31.1 76 30.3 
    Bone 124 23.9 69 25.8 55 21.9 
    Lung 98 18.9 58 21.7 40 15.9 
    Locally advanced disease 68 13.1 39 14.6 29 11.6 
    Central nervous system 45 8.7 18 6.7 27 10.8 
a

Data available from October 2005 onward.

b

Renal tumors, myelomas, hematological cancers, and patients undergoing chemotherapy.

PST, palliative sedation therapy; PaP Score, Palliative Prognostic Score; ADL, activities of daily living; SD, standard deviation.

The distribution of the main clinical characteristics was superimposable as the two patient cohorts were matched for gender, age class, KPS, and reason for hospice admission. The predicted survival of both groups was comparable for PaP Score values. Furthermore, there were no statistically significant differences between the two cohorts in relation to the number of patients living alone or with family, the site of the primary tumor, or metastases.

PST characteristics for our study are reported in Table 2. A predominance of proportional sedation (87.6%) compared with sudden sedation (12.4%) and of primary (85.8%) compared with secondary sedation (14.2%) can be observed. The ratio between mild and deep sedation was approximately 2 : 1, while an intermittent duration of sedation was only slightly more prevalent than the continuous form (56.2% versus 43.8%, respectively). Sixty-three patients underwent continuous deep sedation (23.6%).

Table 2.

PST characteristics

Variable Cohort A (PST)
 
 n 
Type of sedation   
    Proportional 234 87.6 
    Sudden 33 12.4 
    Mild 166 62.2 
    Deep 101 37.8 
    Intermittent 150 56.2 
    Continuous 117 43.8 
    Primary 229 85.8 
    Secondary 38 14.2 
Drugs administered for sedation [mean dose in mg/day (SD, range)] 
    Lorazepam 101, 4.9 (3.8, 1–20) 37.8 
    Chlorpromazine 101, 55.8 (49.7, 2.5–350) 37.8 
    Midazolam 20, 41.7 (24.8, 2.5–110) 7.5 
    Promethazine 63, 49.3 (51.6, 2.5–350) 23.6 
    Haloperidol 61, 3.6 (2.2, 0.5–10) 22.8 
    Morphine 68, 41.2 (40.3, 5–180) 25.5 
    Diazepam 24, 25.5 (11.1, 3–40) 9.0 
    Others 11 4.1 
Number of drugs administered for sedation 
    1 132 49.4 
    2 87 32.6 
    3 45 16.9 
    >3 1.1 
Duration of sedation (days)   
    Mean value (SD) 4.0 (6.0)  
    Median value (range) 2.0 (0–43)  
Variable Cohort A (PST)
 
 n 
Type of sedation   
    Proportional 234 87.6 
    Sudden 33 12.4 
    Mild 166 62.2 
    Deep 101 37.8 
    Intermittent 150 56.2 
    Continuous 117 43.8 
    Primary 229 85.8 
    Secondary 38 14.2 
Drugs administered for sedation [mean dose in mg/day (SD, range)] 
    Lorazepam 101, 4.9 (3.8, 1–20) 37.8 
    Chlorpromazine 101, 55.8 (49.7, 2.5–350) 37.8 
    Midazolam 20, 41.7 (24.8, 2.5–110) 7.5 
    Promethazine 63, 49.3 (51.6, 2.5–350) 23.6 
    Haloperidol 61, 3.6 (2.2, 0.5–10) 22.8 
    Morphine 68, 41.2 (40.3, 5–180) 25.5 
    Diazepam 24, 25.5 (11.1, 3–40) 9.0 
    Others 11 4.1 
Number of drugs administered for sedation 
    1 132 49.4 
    2 87 32.6 
    3 45 16.9 
    >3 1.1 
Duration of sedation (days)   
    Mean value (SD) 4.0 (6.0)  
    Median value (range) 2.0 (0–43)  

PST, palliative sedation therapy; SD, standard deviation.

Of the drugs used for PST, neuroleptics (84.2%) were predominant, followed by benzodiazepines (54.3%) and opioids (25.5%). These varying typologies can be attributed to the different practices of the four participating centers. The mean doses, standard deviations (SD), and dose ranges for each drug administered are reported in Table 2. The mean and median duration of sedation was 4 (SD = 6.0) and 2 days, respectively (range 0–43 days). The higher mean value was due to the number of patients whose PST lasted a relatively long time (29 patients, 10.8% ≥10 days and nine patients, 3.4% ≥20 days). Patients with a sedation >10 days showed similar parameters to those of the entire sedated group, with the exception of the following: fewer hospice admissions due to ‘uncontrolled symptoms’ (13.8% versus 56.6%) and more admissions because of ‘terminality’ (65.5% versus 38.2%); more ‘mild’ (89.7% versus 62.2%) and ‘secondary’ (27.6% versus 14.2%) sedations; less delirium (48.3% versus 78.7%) and dyspnea (3.4% versus 14.5%), and greater psychological distress (10.3% versus 6.0%).

Refractory symptoms for which PST was indicated are described in Table 3. Delirium and/or agitation were the main reasons for PST in 78.7% of cases, followed by dyspnea in 19.5%, pain in 11.2%, and vomiting in 4.5%. A fairly high percentage of patients (18.7%) required PST for refractory psychological distress. However, the majority of this group (34 of 50 patients) also presented physical refractory symptoms. Psychological distress alone was treated with PST in 16 (6%) patients. Parameters in these patients did not differ substantially from those of the entire sedated group, with the exception of the following: patients were younger (62 versus 68 years), more males than females (62.5 versus 55.8%), better performance status (KPS ≥ 50 in 56.3% versus 23.6%) and ADL (3.2 versus 2.0), and less often hospitalized for reasons of terminality (12.5% versus 38.2%). Mild sedation was more common (81.2% versus 62.2%); average duration of treatment was slightly longer (6.9 versus 4.0 days), as was median treatment duration (5.5 versus 2.0 days).

Table 3.

Refractory symptoms requiring PST

Variable Cohort A (PST)
 
 n 
Refractory symptom   
    Delirium and/or agitation 210 78.7 
    Dyspnea 52 19.5 
    Pain 30 11.2 
    Vomiting 12 4.5 
    Psychological and physical distress 50 18.7 
    Only psychological distress 16 6.0 
    Others 10 3.7 
Number of refractory symptoms   
    1 168 62.9 
    2 89 33.3 
    3 3.4 
    4 0.4 
Variable Cohort A (PST)
 
 n 
Refractory symptom   
    Delirium and/or agitation 210 78.7 
    Dyspnea 52 19.5 
    Pain 30 11.2 
    Vomiting 12 4.5 
    Psychological and physical distress 50 18.7 
    Only psychological distress 16 6.0 
    Others 10 3.7 
Number of refractory symptoms   
    1 168 62.9 
    2 89 33.3 
    3 3.4 
    4 0.4 

PST, palliative sedation therapy.

The sum of the refractory symptoms for which PST was administered was >100 because several of the treated patients exhibited more than one symptom. Indeed, 168 patients (62.9%) were sedated for a single symptom, 89 (33.3%) for two, nine (3.4%) for three, while one patient (0.4%) actually showed four refractory symptoms.

The OS of the two cohorts is reported in Figure 1. The two curves are superimposable, without statistically significant differences. The median survival of the control cohort from time of admission to the hospice was 9 days (90% CI 8–10), while that of the sedated cohort was 12 days (90% CI 10–14) (log rank = 0.95, P = 0.330) (unadjusted HR = 0.92; 90% CI 0.80–1.06).

Figure 1.

Kaplan–Meier survival curves for cohort A [palliative sedation therapy (PST)] and cohort B (no PST).

Figure 1.

Kaplan–Meier survival curves for cohort A [palliative sedation therapy (PST)] and cohort B (no PST).

Multivariate analysis, taking into account all the matching factors, showed an adjusted HR of 0.86 (90% CI 0.74–1.00), which was fairly similar to the unadjusted HR. Evaluation of PaP Score values showed that this index maintained its prognostic capacity in both patient cohorts [median OS (90% CI) in sedated patient cohort A: 25 (16–26), 12 (10–15), and 5 (4–8) days; nonsedated patient cohort B: 27 (13–40), 11 (8–14), and 4 (4–6) days for PaP Score groups A, B, and C, respectively]. Furthermore, in each PaP Score group, the ‘sedation’ variable did not influence outcome.

discussion

The frequency of ‘palliative sedation’ for patients at end of life varies considerably from 1% [39] to 88% [40] in populations undergoing palliative care. Such heterogeneity is dependent on several factors, in particular the lack of a universally accepted definition of PST, which indicates the need to integrate the many existing definitions of the treatment with a more detailed description. Furthermore, the retrospective nature of many studies, lack of consensus on the definition of a refractory symptom, ethical and moral concerns, and cultural and ethnic diversities also play a role in complicating the issue [41]. The frequency of PST reported in the reviews by Porta Sales [4] and Sykes and Thorns [30] was ∼45% for the ‘proportional’ form of sedation and ∼16% for the ‘sudden’ form of PST. Two epidemiological reviews also provide important information on the prevalence rate of palliative sedation. In the survey of Peruselli et al. [42] on the final week of life in an Italian setting, it was reported that 25% of patients in home palliative care were totally pharmacologically sedated, albeit with enormous variability (0%–60%) due to the habits of patients’ doctors rather than to the clinical conditions of the patients themselves. More recently, Miccinesi et al. [43], in a work conducted by the EURELD group, reported that out of a total of 20 480 death certificates retrieved, the percentage of continuous deep sedation carried out in six European nations ranged from 2.5% (Denmark) to 8.5% (Italy). It is worthy to note that both reviews referred exclusively to continuous deep sedation. In the present study, 25.1% of all the patients undergoing palliative care during the same period in the four hospices taking part received PST, while 23.6% of the sedated subgroup (and 5.9% of hospice admissions) were submitted to continuous deep sedation. It was confirmed, therefore, that sedation to complete unconsciousness is a less common procedure than more moderate sedation.

With regard to duration of PST, multicenter study of Fainsinger et al. [44], which included one Israeli, one Spanish, and two South African centers, showed considerable homogeneity in its results. The work reported the mean, median, and range of duration of PST of the four centers. The mean duration of therapy varied from 1.9 to 3.2 days, median duration ranged from 1 to 3 days, and the range of duration for all four centers was between 1 and 6 days. Similarly, in the review of Sykes and Thorns [30], the mean duration of PST in the 10 studies evaluated was 2.8 days. Our data are similar to those reported in the literature, even though the mean duration of treatment in our population (4 days, SD = 6.0) was twice that of the median duration (2 days, range 0–43), probably due to the presence of a small subgroup of patients (10.8%) who underwent sedation for at least 10 days.

The symptoms most frequently responsible for a refractory state requiring sedation are delirium (55%–65%), dyspnea (26%–27%), pain (14%–18%), and vomiting (4%) [1, 4, 30]. As reported in the ‘Introduction’ section, PST administered for psychological or existential distress is still a widely discussed issue because of its unique ethical implications. Few authors focus on PST used for this reason. However, in the study of Porta Sales et al. [45], psychological or existential distress was present in 36% of the sedations carried out, representing the most frequent reason for PST. Similarly, in the article published by the previously mentioned Spanish group, the percentage of patients submitted to PST for psychological reasons was higher (9%) than that of other participating centers (1%) [44]. The authors hypothesized that, as the patients were Spanish and thus often poorly informed about the gravity of the disease, the disclosure of details about the terminal phase near the end of life could lead to some patients undergoing an existential crisis, which makes adequate counseling extremely difficult. However, a nonnegligible percentage of PST administered for psychological or existential reasons can also be observed in other studies carried out in non-Latin-origin centers [29, 46]. In our population, although the percentage of patients undergoing PST for these reasons seems high at 18%, psychoexistential distress as the sole reason for PST was only present in 6% of this subgroup.

The most widely used drugs for PST are benzodiazepines, especially midazolam [1, 30]. Psychotropic drugs are also frequently used, sometimes in conjunction with benzodiazepines. Chater et al. [46] reported that PST can be used for more than one reason in individual patients and that more than one drug can be used.

Our study focuses on the potential effect of PST on patient survival. Some authors have described this therapy as a slow and disguised form of euthanasia because death often comes fairly rapidly after sedation [19]. Even physicians who do not embrace this idea sometimes fear that PST may hasten death and tend to fall back on the doctrine of double effect to keep the two approaches separate and to justify the use of PST [20]. In actual fact, there are numerous reports in the literature which attest, directly or indirectly, to the absence of an impact of palliative sedation on survival duration [21–29]. The study of Morita et al. [47] did not reveal any important differences in survival between patients who received no sedatives and those who received different doses of sedatives. Furthermore, an analysis of the use of sedatives in the final 48 h of life in a multiple regression model for survival prediction showed no significant increase in predictability. The study of Sykes and Thorns [31] reported similar survival for sedated and nonsedated patients in the final 24 h of life (14.2 versus 14.3 days), while patients submitted to a 7-day sedation showed a significantly higher survival (36.6 days). This may, however, have been due to the fact that the 7-day sedation patients had been admitted to the hospice earlier.

We aimed to focus on methodological problems present in studies conducted in this setting, fully aware that the typology of patients and the therapeutic approach makes randomized studies impossible for ethical reasons. However, the study design we adopted, within the context of observational studies, results in the lowest probability of selection bias. As we were able to conduct a matched cohort study for the different characteristics that may influence patient outcome, the a posteriori techniques developed to add validity to group comparisons from nonrandomized studies, e.g. propensity score utilizing the same variables, do not add any correction to potential selection bias [48–51]. However, a potential limitation of the work lies in the fact that, although administration of the best palliative care to all the patients was a requisite of the protocol and was monitored throughout the study, we did not record the use of neuroleptics with potential sedative side-effects in either group. Some authors argue that PST must, per se, be accompanied by an interruption in nutrition and hydration, leading inevitably to a reduction in survival. In actual fact, from an ethical point of view, it has been reported that decisions regarding the administration of PST and the withdrawal or withholding of nutrition and hydration must be considered separately [2, 13]. The interruption of hydration was not evaluated in the two groups. However, even if hydration had been interrupted in a higher number of the sedated patients with respect to the nonsedated group, this would, at most, have had a detrimental effect on former cohort; this effect was not observed.

Several scientific sources have also reinforced the clear distinction between euthanasia and PST reported by the European Association for Palliative Care Ethics Task Force [52] with respect to intention (relief of intolerable suffering in PST, killing the patient in euthanasia), procedure (use of a sedative drug for symptom control in PST, administration of a lethal drug in euthanasia), and outcome (alleviation of suffering in PST, immediate death in euthanasia).

In conclusion, the results from the first studies carried out on PST and from subsequent literature reviews are further confirmed by the findings in the present prospective cohort study, which attest to a lack of impact of PST on survival in sedated and nonsedated hospice patients. However, there is clearly a significant lack of research into this specific area. Most studies are retrospective, descriptive, and carried out on selected groups of patients. As randomized studies are not feasible in this setting, stronger evidence to support the benefit of PST will be difficult to obtain. However, we hope that our study actively contributes to enhancing the level of evidence in this area.

funding

Istituto Oncologico Romagnolo, Forlì and Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori.

The authors thank Gráinne Tierney for editing the manuscript. The following participating investigators and institutions are acknowledged: Manlio Monti, Istituto Scientifico Romagnolo per lo Studio e la Cura dei Tumori, Meldola; Federica Monduzzi, Marco Montanari, and Francesco Carrozza, Palliative Care Unit, S. Domenico Hospice and Department of Medical Oncology, Umberto I Hospital, Lugo; and Daniele Dini, Centro Oncologico, Hospice Ospedaliero, Modena, Italy.

References

1.
De Graeff
A
Dean
M
Palliative sedation therapy in the last weeks of life: a literature review and recommendations for standards
J Palliat Med
 , 
2007
, vol. 
10
 (pg. 
67
-
85
)
2.
Broeckaert
B
Olarte
JMN
Ten Have
H
Clark
D
Sedation in palliative care: facts and concepts
The Ethics of Palliative Care: European Perspectives
 , 
2002
Buckingham, UK
Open University Press
(pg. 
166
-
180
)
3.
Morita
T
Tsuneto
S
Shima
Y
Definition of sedation for symptom relief: a systematic literature review and a proposal of operational criteria
J Pain Symptom Manage
 , 
2002
, vol. 
24
 (pg. 
447
-
453
)
4.
Porta Sales
J
Sedation and terminal care
Eur J Palliat Care
 , 
2001
, vol. 
8
 (pg. 
97
-
100
)
5.
Cherny
NI
Sedation for the care of patients with advanced cancer
Nat Clin Pract Oncol
 , 
2006
, vol. 
3
 (pg. 
492
-
500
)
6.
Rousseau
P
Existential suffering and palliative sedation: a brief commentary with a proposal for clinical guidelines
Am J Hosp Palliat Care
 , 
2001
, vol. 
18
 (pg. 
151
-
153
)
7.
Levy
MH
Cohen
SD
Sedation for the relief of refractory symptoms in the imminently dying: a fine international line
Semin Oncol
 , 
2005
, vol. 
32
 (pg. 
237
-
246
)
8.
Cherny
NI
Portenoy
RK
Sedation in the management of refractory symptoms: guidelines for evaluation and treatment
J Palliat Care
 , 
1994
, vol. 
10
 (pg. 
31
-
38
)
9.
Cherny
NI
Sedation in response to refractory existential distress: walking the fine line
J Pain Symptom Manage
 , 
1998
, vol. 
16
 (pg. 
404
-
406
)
10.
Chochinov
H
Interventions to enhance the spiritual aspects of dying
Office of the Director (Ed), NIH State-of-the-Science Conference on Improving End-of-Life Care
 , 
2004
Bethesda, MD
National Institutes of Health
(pg. 
77
-
80
)
11.
Chochinov
H
Dignity and the essence of medicine: the A, B, C, and D of dignity conserving care
Br Med J
 , 
2007
, vol. 
335
 (pg. 
184
-
187
)
12.
Breitbart
W
Spirituality and meaning in supportive care: spirituality and meaning-centered group psychotherapy interventions in advanced cancer
Support Care Cancer
 , 
2002
, vol. 
10
 (pg. 
272
-
280
)
13.
National Ethics Committee, Veterans Health Administration
The ethics of palliative sedation as a therapy of last resort
Am J Hospice Palliat Med
 , 
2007
, vol. 
23
 (pg. 
483
-
491
)
14.
Morita
T
Akechi
T
Sugawara
Y
, et al.  . 
Practices and attitudes of Japanese oncologists and palliative care physicians concerning terminal sedation: a nationwide survey
J Clin Oncol
 , 
2002
, vol. 
20
 (pg. 
758
-
764
)
15.
Coyle
N
Adelhardt
J
Foley
K
Portenoy
RK
Character of terminal illness in the advanced cancer patient: pain and other symptoms during the last four weeks of life
J Pain Symptom Manage
 , 
1990
, vol. 
5
 (pg. 
83
-
93
)
16.
Rietjens
JA
van Delden
JJ
van der Heide
A
, et al.  . 
Terminal sedation and euthanasia: a comparison of clinical practices
Arch Intern Med
 , 
2006
, vol. 
166
 (pg. 
749
-
753
)
17.
van der Heide
A
Onwuteaka-Philipsen
BD
Rurup
ML
, et al.  . 
End-of-life practices in the Netherlands under the Euthanasia Act
N Engl J Med
 , 
2007
, vol. 
356
 (pg. 
1957
-
1965
)
18.
Lo
B
Rubenfeld
G
Palliative sedation in dying patients
JAMA
 , 
2005
, vol. 
294
 (pg. 
1810
-
1816
)
19.
Billings
JA
Block
SD
Slow euthanasia
J Palliat Care
 , 
1996
, vol. 
12
 (pg. 
21
-
30
)
20.
Latimer
EJ
Ethical decision-making in the care of the dying and its applications to clinical practice
J Pain Symptom Manage
 , 
1991
, vol. 
6
 (pg. 
329
-
336
)
21.
Ventafridda
V
Ripamonti
C
De Conno
F
, et al.  . 
Symptom prevalence and control during cancer patients’ last days of life
J Palliat Care
 , 
1990
, vol. 
6
 (pg. 
7
-
11
)
22.
Fainsinger
R
Miller
MJ
Bruera
E
, et al.  . 
Symptom control during the last week of life on a palliative care unit
J Palliat Care
 , 
1991
, vol. 
7
 (pg. 
5
-
11
)
23.
Kohara
H
Ueoka
H
Takeyama
H
, et al.  . 
Sedation for terminally ill patients with cancer with uncontrollable physical distress
J Palliat Med
 , 
2005
, vol. 
8
 (pg. 
20
-
25
)
24.
Chiu
TY
Hu
WY
Lue
BH
, et al.  . 
Sedation for refractory symptoms of terminal cancer patients in Taiwan
J Pain Symptom Manage
 , 
2001
, vol. 
21
 (pg. 
467
-
472
)
25.
Muller-Busch
HC
Andres
I
Jehser
T
Sedation in palliative care—a critical analysis of 7 years experience
BMC Palliat Care
 , 
2003
, vol. 
2
 pg. 
2
 
26.
Vitetta
L
Kenner
D
Sali
A
Sedation and analgesia-prescribing patterns in terminally ill patients at the end of life
Am J Hosp Palliat Care
 , 
2005
, vol. 
22
 (pg. 
465
-
473
)
27.
Waller
A
Bercovitch
M
Fainsinger
R
Adunsky
A
Symptom control and the need of sedation during the treatment in Tel Hashomer in-patient hospice
Book of Abstracts, 6th Congress of the European Association for Palliative Care, Geneva, Switzerland, 1999; 53
 
28.
Morita
T
Differences in physician-reported practice in palliative sedation therapy
Support Care Cancer
 , 
2004
, vol. 
12
 (pg. 
584
-
592
)
29.
Stone
P
Phillips
C
Spruyt
O
Waight
C
A comparison of the use of sedatives in a hospital support team and in hospice
Palliat Med
 , 
1997
, vol. 
11
 (pg. 
140
-
144
)
30.
Sykes
N
Thorns
A
The use of opioids and sedatives at the end of life
Lancet Oncol
 , 
2003
, vol. 
4
 (pg. 
312
-
318
)
31.
Sykes
N
Thorns
A
Sedative use in the last week of life and the implications for end-of-life decision making
Arch Intern Med
 , 
2003
, vol. 
163
 (pg. 
341
-
344
)
32.
Pirovano
M
Maltoni
M
Nanni
O
, et al.  . 
A new palliative prognostic score: a first step for the staging of terminally ill cancer patients
J Pain Symptom Manage
 , 
1999
, vol. 
17
 (pg. 
231
-
239
)
33.
Maltoni
M
Nanni
O
Pirovano
M
, et al.  . 
Successful validation of the palliative prognostic score in terminally ill cancer patients. Italian Multicenter Study Group on Palliative Care
J Pain Symptom Manage
 , 
1999
, vol. 
17
 (pg. 
240
-
247
)
34.
Maltoni
M
Caraceni
A
Brunelli
C
, et al.  . 
Prognostic factors in advanced cancer patients: evidence-based clinical recommendations—a study by the Steering Committee of the European Association for Palliative Care
J Clin Oncol
 , 
2005
, vol. 
23
 (pg. 
6240
-
6248
)
35.
Glare
P
Virik
K
Independent prospective validation of the PaP score in terminally ill patients referred to a hospital-based palliative medicine consultation service
J Pain Symptom Manage
 , 
2001
, vol. 
22
 (pg. 
891
-
898
)
36.
Kaplan
EL
Meier
P
Nonparametric estimation for incomplete observation
J Am Stat Assoc
 , 
1958
, vol. 
53
 (pg. 
457
-
481
)
37.
Lawless
JS
Statistical Models and Methods for Life-Time Data
 , 
1982
New York
John Wiley and Sons
38.
Cox
DR
Regression models and life tables
J R Stat Soc
 , 
1972
, vol. 
34
 (pg. 
187
-
220
)
39.
Fainsinger
RL
Use of sedation by a hospital palliative care support team
J Palliat Care
 , 
1998
, vol. 
14
 (pg. 
51
-
54
)
40.
Turner
K
Chye
R
Aggarwal
G
, et al.  . 
Dignity in dying: a preliminary study of patients in the last three days of life
J Palliat Care
 , 
1996
, vol. 
12
 (pg. 
7
-
13
)
41.
Rousseau
P
Palliative sedation in the management of refractory symptoms
J Support Oncol
 , 
2004
, vol. 
2
 (pg. 
181
-
186
)
42.
Peruselli
C
Di Giulio
P
Toscani
F
, et al.  . 
Home palliative care for terminal cancer patients: a survey on the final week of life
Palliat Med
 , 
1999
, vol. 
13
 (pg. 
233
-
241
)
43.
Miccinesi
G
Rietjens
JAC
Deliens
L
, et al.  . 
Continuous deep sedation: physicians’ experiences in six primary countries
J Pain Symptom Manage
 , 
2006
, vol. 
31
 (pg. 
122
-
129
)
44.
Fainsinger
RL
Waller
A
Bercovici
M
, et al.  . 
A multicentre international study of sedation for uncontrolled symptoms in terminally ill patients
Palliat Med
 , 
2000
, vol. 
14
 (pg. 
257
-
265
)
45.
Porta Sales
J
Ylla-Català Bore
E
Estibalez Gil
A
Estudio multicentrico catalano-balear sobre la sedacion terminal en cuidados paliativos
Med Pal
 , 
1999
, vol. 
6
 (pg. 
153
-
158
)
46.
Chater
S
Viola
R
Paterson
J
Jarvis
V
Sedation for intractable distress in the dying—a survey of experts
Palliat Med
 , 
1998
, vol. 
12
 (pg. 
255
-
269
)
47.
Morita
T
Tsunoda
J
Inoue
S
Chihara
S
Effects of high dose opioids and sedatives on survival in terminally ill cancer patients
J Pain Symptom Manage
 , 
2001
, vol. 
21
 (pg. 
282
-
289
)
48.
Newhouse
JP
McClellan
M
Econometrics in outcomes research: the use of instrumental variables
Annu Rev Public Health
 , 
1998
, vol. 
19
 (pg. 
17
-
34
)
49.
D'Agostino
RB
Jr
Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group
Stat Med
 , 
1998
, vol. 
17
 (pg. 
2265
-
2281
)
50.
Rosenbaum
PR
Discussing hidden bias in observational studies
Ann Intern Med
 , 
1991
, vol. 
115
 (pg. 
901
-
905
)
51.
Braitman
LE
Rosenbaum
PR
Rare outcomes, common treatments: analytic strategies using propensity score
Ann Intern Med
 , 
2002
, vol. 
137
 (pg. 
693
-
695
)
52.
Materstvedt
LJ
Clark
D
Ellershaw
J
, et al.  . 
Euthanasia and physician-assisted suicide: a view from an EAPC Ethics Task Force
Palliat Med
 , 
2003
, vol. 
17
 (pg. 
97
-
101
)