The findings of this study make it possible to better understand what are relatively rare sedation practices for patients hospitalized in cancer centers. Most indications are for refractory symptoms and acute complications that are immediately life-threatening, each accounting for almost 50% of cases. Sedation was maintained until the patient’s death in almost all cases (98%). In most cases sedation was implemented in accordance with HAS guidelines, with the exception of the low frequency of titration. Information provided to the loved ones, the collegiality of the decision making, and the traceability of the sedation were insufficient.
What this study adds
The prevalence of sedation in our study is lower than that found in the literature (14.6 to 66.7% in the review of Maltoni et al. [
11]), but these frequencies vary in large proportions because of differences between the definitions, indications and practices selected according to the studies [
11,
12]. These findings are nevertheless in line with the opinion of French experts, who consider that situations possibly warranting sedation are relatively rare [
13].
The most frequent indications in the litterature for sedation are delirium and dyspnea. While pain is scarcer in this regard [
11,
12,
14‐
26], it is a frequent reason for hospitalization in the COL. The indication for dyspnea was, however, predominant in the COL. There was also a significant proportion of indications related to hemorrhages (11.1% vis-a-vis 1.6 to 3.3% in the work of Benitez-Rosario [
16]). The differences can be explained by the fact that our study took into account all indications of sedation: while the published works are generally limited to deep and continuous sedations for refractory symptoms, our findings show that sedation is frequently used for acute complications that are immediately life-threatening. These indication is distinguished only by the SFAP (
Société française d’accompagnement et de soins palliatifs [French society for end-of-life and palliative care]) [
1] and does not exist in the other recommendations [
3‐
5]. Such complications often involve terminal hemorrhage and asphyxia-type respiratory distress. The purpose of sedation is to relieve the patient of the panic and terror that these situations cause, and which are also generally very stressful for the family and caregivers. In the terminal phase, sedation is an emergency procedure that can affect the time of death. Moreover, the frequency of information provided to patient and family was greater for immediate life-threatening complications than for refractory symptoms, these differences becoming significant only when known sedations were considered. The severity of these situations as well as the likelihood of a very rapid death may have prompted professionals to devote more time to their care.
The proportion of titrations was low (44.4%). In the majority of cases, the midazolam maintenance dose was directly introduced or increased arbitrarily and then adjusted in stages. With the exception of those of the SFAP [
1], no recommendation specifically describes the titration procedures, and many studies do not detail or explicitly use them [
11,
12,
14‐
24,
26‐
34]. However, the absence of titration exposes to risks of over or under dosages, because of a large inter-individual variability of the sensitivity to midazolam [
35,
36], as well as to a delay in relief [
35,
36]. Titration also has an important ethical value, demonstrating the principle of proportionality [
37]. This principle, by the use of adapted doses in order to obtain sedation, allows to differentiate between sedation and euthanasia, among other criteria.
In addition, our findings reveal two other arguments that titration is essential, whetherin routine clinical practice or in clinical studies.
The frequent absence of titration may explain the use of higher doses: 36 mg/d at the beginning of the sedation and 86.4 mg/d at the end in our study, whereas the usual reported doses are between 10 mg/d et 50 mg/d [
15,
16,
18,
21,
38], more rarely beyond this (up to 75 mg/d in the Caraceni study [
34]). Prior administration of midazolam was associated with higher doses of midazolam to initiate and maintain sedation as well as at the time of death. This observation is explained by the phenomenon of tachyphylaxis [
18,
36], but its physiopathology is not fully explained, and several mechanisms have been reported [
35,
36,
39].
Titration makes it possible to search for the minimum effective dosage, in the context of sedation that is proportionate to the severity of the symptoms, the objective being to reduce or eliminate the patient’s exposure to an unbearable situation, without necessarily becoming completely unconscious.
The information of the patient and the research of his consent were found to be inconsistent (respectively 40.7 and 31.5%). These frequencies are similar in several studies [
10‐
22,
24,
26‐
28]. Patients at the end of life often have impaired alertness, rendering the obtaining of consent impossible or inappropriate. The proportion of patients that was able to express an opinion about sedation in a palliative setting was about 50% in 2 published studies [
20,
26]. But even in the absence of an impairment of judgment, consenting to sedation remains a difficult decision for patients.
Notification was given to the family in a greater proportion (72.2%). However, it is often close to 90% in many publications [
12,
16,
19,
22,
26,
29]. This may be explained by a lack of information, a lack of traceability, or even insufficient foresight, as suggested by the low rate of anticipated prescriptions of 27.8%. The time to talk with loved ones was then probably shorter, hence more succinct and less established information.
The collegiality of the decision-making process was quite rare (37.0%). The frequencies of collegial meetings reported in the literature are higher, between 54 and 70% [
19,
23,
30]. The frequent participation of COL doctors in oncological multidisciplinary consultation meetings could lead to confusion between the decision for palliative care and that of sedation. It is also important to remember that for deep and continuous sedation maintained until death, the French law requires a meeting to ensure a collective procedure, but such meetings are simply recommended for other sedation practices. All the arguments put forward during this meeting and proof that the patient’s consent has been sought must be entered in the patient’s file [
10].
The sedations performed in the PCU were more often associated with information provided to the patient and the patient’s consent, and even more frequent information provided to the family when probable sedations were excluded. Anticipation of sedation was more common in the PCU. The use of an initial titration was also predominant in the PCU. The palliative care team had a priori better training and greater experience in sedation. Daily collegial meetings probably facilitated the exchange of information and identification of risk situations. Our results therefore confirm those of a study in the Netherlands in 2007 [
40]
, describing greater compliance with sedation recommendations when the prescribing doctor specializes in palliative care.
All cases of sedation for psychological or existential suffering included the trace of a discussion with the patient, his/her consent, as well as providing information to the relatives. Collegiality was systematically found, and the carrying out of sedation was reported in the end-of-hospitalization communication. It can be assumed that the complexity of these situations, the reflection and the time required for the decision have favored a richer discussion between the team, the patient and his family, as well as a better traceability. Deep and continuous sedations for psychological or existential refractory suffering is subject of discussion and needs to be clarified because no consensus is yet available within medical societies [
41]. When there is refractory psychological or existential suffering, the main goal of the psycho-oncology team is to look for differential diagnoses, such as depressive syndrome, demoralization, a desire to hasten death or a request for euthanasia [
42].
The main limitation of this work was its retrospective nature, with a risk of poor traceability and therefore an underestimation of the practices studied; and low standardization of the declarative data that required interpretation or recoding. These mainly impacted decision-making and information provided, and prevented the depth of sedation from being collected. The impact of this bias on the study of the prescriptions was more limited, as their recording was automated by the software.
This was a single center study conducted in the specific context of palliative cancer care in the north of France. It may therefore be difficult to extrapolate our findings to other healthcare fields and systems.
Perspectives
Our work involved identifying a certain number of probable sedation scenarios, corresponding to criteria very similar to routine sedation practices, but without being identified as such. The absence of any significant difference in their indications, methods and dosages suggested that they were indeed cases of sedation that fell within the scope of this study: it seemed unlikely that these prescriptions had been decided on without the prescriber being aware of the resulting decrease in alertness. However, the lack of an explicit appointment of sedation, its inefficiency and the more scarce information provided to the patient and his family made it seem less likely that the team would carry out these probable sedations, a confusion between implementation and result (carrying out sedation and managing to “sedate” the patient), a confusion in intention between anxiolysis and sedation, or a concealment of failure. However, given the low number of probable sedations, the sensitivity was poor. Assessment of comfort was not available for a significant proportion of patients (8/54), which could also bias the analysis. In a Canadian study of medication prescriptions used for sedation, 64.5% of cases had no explicit mention [
21].
The identification of these probable sedation cases illustrates the need for further training for care teams on this complex practice.
The French society for assistance and palliative care (SFAP) recently created a tool called SEDAPALL [
43] to describe and analyze end-of-life palliative sedation practices. It is both an educational tool and a research tool that helps to ascertain the intentions behind the decision. It makes it possible to verify whether or not the intention is actually put into effect. The intentionality of sedation is described by SEDAPALL based on 3 criteria:
Duration: transient, indeterminate or maintained until death.
Depth: proportionate or deep from the outset.
Level of consent: not obtained, obtained in advance, obtained at the onset of sedation, requested by the patient.
A sedation working group has been set up for the C3 - the three cancer centers in North West France (Caen, Rouen and Lille). Based on the findings of our study, which will illustrate sedation practices in cancer centers, and given that it is vital to characterize probable sedation scenarios, the objective of this working group will be to develop an educational activity regarding sedation practices to enable medical and healthcare teams to implement the SEDAPALL tool and to reflect on their daily clinical practices.