The problems associated with sedation in end-of-life care and the different attitudes among clinicians and palliative care experts are reflected in inconsistent terminology [5], variation in techniques used to induce, maintain and monitor sedation [6], duration of and frequency of application, different concepts on the time of administering sedation and on intentions [7]. While the ambiguous term "terminal sedation (TS)" [8,9] is used most often, other terms more clearly reflect the different viewpoints: "sedation for intractable distress in the imminently dying [10,11];", end-of-life sedation [12] "slow-euthanasia" [13], "palliative sedation" [14], "total sedation" [15], "sedation in the final phase", "palliative sedation therapy" [16].

Sedation in the terminal or final stages of life can be defined as the use of sedative drugs (usually benzodiazepines with or without complementary opioids given by the intravenous or by the subcutaneous route) to reduce the level of consciousness sufficiently deep to provide comfort for the patient until death occurs. According to the EAPC Ethics Task Force "terminal' or 'palliative' sedation in those imminently dying must be distinguished from euthanasia. In terminal sedation the intention is to relieve intolerable suffering, the procedure is to use a sedating drug for symptom control, and the successful outcome is the alleviation of distress. In euthanasia the intention is to kill the patient, the procedure is to administer a lethal drug and the successful outcome is immediate death. In palliative care mild sedation may be used therapeutically but in this situation it does not adversely affect the patient's conscious level or ability to communicate [17]. The use of heavy sedation (which leads to unconsciousness) may sometimes be necessary to achieve identified therapeutic goals. The intention is the relief of otherwise intractable and refractory distress [18]. Target symptoms include persisting pain, delirium, dyspnoea, nausea and vomiting, massive haemorrhage, agitated anxiety and other forms of psychological distress. Sedation can be classified into mild to deep, intermittent to continuous, primary to secondary, sudden to slow [19]. Controversies exist about the time limit and proximity to death [20], the clinical implications (e.g. the withdrawal of nutrition and hydration [21,22]), patient monitoring, the stage and circumstances of the illness in which sedation can be offered or employed, and the informed consent process with patients and surrogates [23]. Requests for sedation seem to become of increasing relevance in advance directives, but the ethical implications may rise to conflicts between patients' wishes to hasten death and physicians' intentions to provide the best care and not to shorten life.

In particular, there is controversy on the issues of dehydration in sedation for existential suffering [24]. The increasing acceptance and use of "terminal sedation" [25] in end-of-life care makes it necessary to scrutinize guidelines, which may help to continuously consider and reconsider the needs and wishes of patients and surrogates, as well as intentions and concerns of caregivers.

The medical charts of all patients (n= 548), who died during the observation period where investigated in order to find out those, who had sedation in the final stage of their life. A ranking of symptoms on admission, during treatment and in the last 48 hrs was made according to our symptom assessment scale with the items: pain, gastrointestinal (nausea/emesis/intestinal obstruction), dyspnoea, anxiety/depression, fatigue/cachexia, cognitive disorder/delirium (drowsiness/agitation), bleeding, skin problems (ulcerations, oedema), neurological and others. Data collection, rating and analysis was made by a single member of the medical staff. Reduction of rating and ranking errors in doubtful cases was achieved by complementary interviews with those involved more personally in the medical care of the patients. In this study the single rater method (by the clinically most experienced physician) was preferred to reduce disagreement on symptom definitions in trust on the experience and accuracy of an individual point of view though recollection of data sometimes was difficult to obtain. Charts of patients with sedation were reviewed and evaluated systematically and discussed with those involved in the decision-making on sedation in the terminal or final phase. A ranking was made to identify a single predominant symptom, which mainly lead to sedation and concomitant others. 6 predominant symptoms for sedation were differentiated: dyspnoea, pain, delirium/agitation, gastrointestinal, bleeding, anxiety/psychological distress. The recorded indications for sedation were classified into "mainly physical" when refractory physical symptoms predominated and "mainly psychological", when intolerable suffering by panic anxiety, refractory insomnia or affective decompensation persisted. Agitated delirium was separated due to the diagnostic uncertainty to relate this symptom to a more physical or more psychological origin retrospectively. Attention was given to the stage of disease at the time of admission, prevalence of symptoms on admission and in the final phase, changes of symptoms during treatment, type and duration of sedation until death, survival time after administration of the sedatives, provision of concomitant therapy, nutrition supply and fluids, state of consciousness and communication skills. Special interest was given to patients' attitudes and frequency of request for sedation in the final phase as documented in the medical records or in advance directives. Patients' characteristics were related to all patients admitted.

- Sedation in the last 48 hrs was performed in 14,6 % (n = 80) of all patients who died in the PCU of GK Havelhoehe (n = 548) with significantly increasing frequency in the years 2000–2002 (χ² = 8.57; p = 0.003) (Figure ​(Figure11).

- No difference was found in the duration of stay in our unit until death between those patients who died with, and those who died without sedation. Mean age of patients, who finally had sedation was less with 54 years vs. 64 years in those without sedation (p = 0.001, Student's t-test).

- All patients had effective and sufficient pain control, but in those patients who were sedated in the last 48 hrs of life, burdensome dyspnoea, panic-stricken anxiety and agitated delirium had increased during the stay (Figure ​(Figure22).

- The indications for performing sedation due to predominant psychological distress (anxiety, refractory insomnia, decompensation) and due to refractory symptoms (dyspnoea, gastrointestinal, pain) did not differ significantly during the two observational periods. Interestingly, there was a tendency that the main indications for sedation during the last three years shifted more and more to psychological distress.

- Mean survival time after administration of sedation was 63 ± 58 hrs during the whole observation period. In the years 2000–2002 patients with more psychological distress had a longer survival time than those with otherwise resistant control of more physical symptoms (Table ​(Table33).

- In the years 2000–2002 we noticed an increase in the request for sedation in the final days of life from 19% to 34% by the patients themselves, documented in personal statements or advance directives. In those patients in which sedation was given also "on request", the mean survival time was found to be slightly shorter (52 ± 42 hrs) in the years 2000–2002, but not in the years before (Table ​(Table33).

- In most cases, sedation was performed with slowly increasing doses of midazolam 0,5 mg – 8 mg/h iv, aimed at achieving effective symptom control. When this was obtained, doses were reduced and sedation was continued intermittently with the documentation of the level of consciousness, comfort, eating, drinking and communication skills. 48 patients had continuous sedation (mean duration 53 hrs), 32 intermittent (mean duration 77 hrs). In 53 patients (66%) oral supply of fluids and in 10 patients also oral nutrition intake during sedation was reported in the records. 27 patients (33.8%), had no oral fluid or nutrition supply after sedation was introduced either because of the deepness of the sedation or because of refusal of oral fluid and nutrition intake. Infusion of fluids was continued in all patients with regard on comfort according to clinical signs of thirst with restriction of volumes in cases of oedema, ascites and pulmonary congestion. Special attention was given to establish a good sleeping period at night and a more patient controlled sedative state with communicative skills and reports of comfort during the day.

- Abilities to communicate like asking for help or answering questions on pain, comfort, thirst etc. and preserved forms of cooperation under sedation were reported in 40 cases (50%).

- The following case of a 40 yrs old hospice worker, in whom oropharyngeal cancer was diagnosed during pregnancy, may illustrate the procedure of sedation in our PCU but also the difficulties associated with drawing the line in consideration of patients needs, personal intentions and adequacy of sedative procedures for refractory symptom control:

The intention of this survey is to contribute to the international discussion on sedation in end-of-life situations. The analysis was primarily done to reconsider our internal guidelines by reflecting our clinical practice Certainly a survey like this suffers from the drawbacks of a retrospective study. Nevertheless our results can be understood as a contribution to more transparency in this difficult issue. The increasing incidence and also the finding that sedation was more often used in situations with psychological distress and also requested by the patients' themselves may indicate an increasing awareness and respect for the needs of the patients. It also seems to signify that in the care of patients with physical symptoms and psychological distress in situations near death needs and deficits exist, for which sensitive medical strategies must be combined and integrated with personal support, commitment and understanding. In our guidelines we tried to relate the possible benefits of sedation to the basic concerns of palliative care by using it as a reversible therapeutic procedure. Concentrating on communicated comfort as the main indicator for the value of this option means also, that in the care of the sedated patients personal support played an important role.

Our decisions for the administration of sedation were restricted to terminal and final situations when the course of the disease had so progressed that death due to the underlying disease could be expected within the next 48 hrs with or without this procedure. Nevertheless mean survival time after administration of sedation was 63 hrs. It is, in general, difficult to predict death in a particular patient because of individual variations and since "we do not know the exact borderline between life and death" [27]. To a certain extent the point in time to accept dying – at least in medical institutions – depends on empirically based decisions to abstain from potentially life-sustaining interventions e.g. pharmacological, technical or other artificial support of organ function "to prevent death" at least for limited moments of time [28,29]. In our internal guidelines we grouped patients according to the staging system of Jonen-Thielemann into rehabilitative, preterminal, terminal and final phases [30]. With relation to intentions of care, life expectancy according to the progress of disease and empirical statistics, consideration on perhaps burdensome diagnostic and therapeutic procedures and communication strategies with patients and proxies a systematic staging of patients in palliative care on expected outcome seems to be useful. The predominant classification into the preterminal stage of most of our patients on admission signifies also the diagnostic uncertainty and difficulty with such a staging system with relationship to responsible decision-making and consequences of care.

We considered sedation as a an palliative therapeutic option of last resort in the last stages of life. Many experts prefer the term "palliative sedation" instead of "terminal sedation". Though the term "palliative sedation" seems to be more appropriate, it must be distinguished from those situations in other phases of palliative care, when palliative sedation is used for burdensome diagnostic or painful therapeutic procedures (e.g. wound debridement and dressing of exulcerating tumours, insertion of catheters) and usually does not exceed the time in which the procedure is undertaken. Among the most controversial discussed issues of "terminal sedation" are the questions of when and for which forms of intractable suffering sedation may be indicated. Another central point refers to the question of how to distinguish clearly the crucial intention of relief providing sedation from (slow) euthanasia by the application of lethal doses of sedatives [31].

In our study we found that dyspnoea, delirium and anxiety increased during the treatment period in those patients who finally had sedation while pain was significantly reduced by adequate treatment during the progressing course of the disease (Figure ​(Figure2).2). This finding is in confirmation with other studies, in which besides acute bleeding and gastrointestinal symptoms, distressing dyspnoea, delirium and agitated anxiety also where found to be the main reasons for sedation in end-of-life care [32]. The adequate treatment and care of patients of patients with otherwise refractory symptoms certainly is a great challenge in palliative care and improvement in this field is needed urgently. Our target in sedation is calming and comfort without lowering the level of consciousness deep enough to loose communication. This also means, that sedation cannot compensate personal palliative support and care – it must be combined. Our study indicates also, that psychosocial distress contributed to the decision making, especially when patients had made an advance directive with requests for sedation in the last phase of life. In our study the shorter survival time after administration of sedation in patients who had made an request for sedation might be due to the fact that in those patients the medical indications and preconditions were considered more intensively while the underlying disease progressed.

The ethical dilemmas and the possibilities to abuse sedation by turning it in a form of medicalised killing [33,34] are important aspects that seem to restrict the administration of sedatives in terminally ill patients. In studies on survival time of patients with and without sedation no significant differences were found. [35-37]. Certainly also the cultural background determines attitudes and view points to the goals of care and in which situations of "intolerable suffering" sedation is ethically acceptable [38]. In several retrospective and prospective surveys the reported frequency of sedation due to intractable symptoms in different palliative care settings ranged from 7–52% with increasing incidence especially for "existential suffering" [39-41]. Diagnostic criteria and clinical preconditions for the consideration of sedation in patients with far advanced disease have been described by several authors [42-44], but aims, targets, types of sedation and decision making remain conflicting issues.

In modern Western culture, the elements of a "good death" include, besides freedom of pain, death at home with the family members being around, amongst others also the element of "awareness". However, the technical possibilities of palliative care with concentration on quality of life also brought about a "diminished emphasis on the good death" [45]. Consequently, the increasing requests of patients "to die in sleep" by terminal sedation could be an comprehensible alternative to euthanasia or physician-assisted suicide and must be reflected in the discussion on the value of sedation in end-of-life care. Though we could not make a clear differentiation between primary (when patients want to be sedated approaching the moment of death) and secondary sedation (when sedation results as a "side-effect" in otherwise refractory symptom control [46], it must be clear to all, that the intention of sedation in the terminal and final phase is not a concealed form of euthanasia [47]. Under ethical aspects the requests for sedation to reduce consciousness to a state of unawareness need an individual and balanced approach, especially when the wish for sedation is not connected with the intolerability of physical or psychological distress This makes it necessary to discuss the problem of "terminal sedation" like other forms of therapeutic and existential support with patients and family members at an early stage in palliative care. Sedation should be restricted as a reversible therapeutic option to otherwise refractory suffering by burdensome symptoms in the terminal and final phase, which may, but should not intentionally, hasten death. With regard to symptom relief the level of sedation must not necessarily need "deep sleep" or bring about unconsciousness. The recent consensus guidelines for dying intensive care patients released in Canada provide a clear, but certainly also problematic, definition of terminal sedation which distinguishes between "real" terminal sedation for therapeutic intentions and terminal sedation as a possible form of euthanasia by lethal doses of sedatives. The Canadian guidelines define "Terminal Sedation" as "sedation with continuous IV narcotics and/or sedatives until the patient becomes unconscious and death ensues from the underlying illness". Emphasis is added that the balance between possible euthanasia and inadequate palliative care is achieved by titrating the dosage of sedatives so that one avoids over- or under medication to allow death to result from the underlying disease in a state of continuing unconsciousness [48].

IA participated in the design and manuscript editing

TJ participated in the design and manuscript editing.