Introduction
The management of pain, agitation, and delirium is a key point in the care of critically ill patients [
1]. Once triggering conditions have been dealt with, pharmacological treatment becomes necessary. After adequate analgesia, sedative drugs, usually given by continuous intravenous (IV) infusion, ensure comfort and allow life-saving procedures, constituting an invaluable tool during the ICU stay. However, they have several side effects [
2,
3]. International guidelines [
4,
5] suggest using the lowest effective doses for early achievement [
6,
7] and constant maintenance of a light level of sedation even in the most severe conditions [
8,
9]. Several strategies have been proposed to avoid deeper-than-needed [
10] levels of sedation, aiming for the goal of keeping ICU patients “calm, conscious, and cooperative” [
11‐
13].
However, even if unjustified [
8], a large proportion of ICU staff still tend to consider this unfeasible [
14] because of the risk of self-removal of invasive devices [
15], the fear of greater stress/discomfort among patients, and the increased workload for operators. Despite the widespread use of validated scoring systems for sedation, like the Richmond Agitation and Sedation Scale (RASS) [
16], a very large proportion of ICU patients are kept at a sedation level deeper than desired [
10,
17], quite likely causing avoidable side effects.
With continuous intravenous (IV) infusion one can predict the duration of the drug effect using pharmacokinetic calculations. This approach, safe for patients with a short ICU stay, could be useless or dangerous in patients needing mechanical ventilation (MV) for more than 3 days. In these cases, it may cause over-administration [
18] even with adequate sedation targets. Moreover, the daily awakening trials [
15] could induce non-physiological neurological fluctuations, preventing the formation of factual memories, and becoming a precipitant cause of delirium [
19], leading to post-ICU cognitive dysfunction [
20].
Analgesic [
21] and sedative [
22] drugs are rarely administered enterally (EN) because of their slower onset of effect and unpredictable pharmacokinetics, even when intestinal absorption is adequate from ICU admission [
23]. Our “EN sedation” protocol [
24] uses hydroxyzine (a first-generation antihistaminic drug, with antiemetic and gastric antisecretory properties) and allows the addition of low doses of lorazepam (a medium-half-life benzodiazepine) if necessary. Melatonin is continuously used [
25] as a physiological sleep-inducer, with antioxidant, anti-inflammatory, analgesic, and immune-modulating properties [
26,
27].
The longer onset and offset time of EN administration make this approach difficult. However, at the same time, this route ensures a more stable level of consciousness with less neurological fluctuation and fewer cardiorespiratory side effects. EN drugs cost much less than IV ones [
22], are less likely to lead to deep sedation, and are similarly effective as judged by nurses, if an awake target is desired [
12].
The hypothesis of the present study is that an unusual EN sedative drugs administration protocol could reach and maintain light and effective sedation, compared to the more common IV continuous infusion. The main outcome was achievement of the target sedation level in ICU patients needing MV for more than 72 h.
Discussion
This study compared two very different approaches for the management of agitation in critically ill patients, using different sedative drugs administered by the unusual EN route compared to the more common IV route. No real differences were found in the most important clinical outcomes.
In agreement with international guidelines [
1,
4,
5], the target was a conscious patient for more than 80% of ICU days. This target was set more frequently in the EN group, but was achieved equally in both groups.
Among neurological indicators, studied throughout the ICU stay, the EN sedation protocol resulted in a similar incidence of delirium, while the RASS observed was slightly higher (Fig.
2b), which means a lower incidence of coma, but greater psychophysical agitation too. Indeed, the few self-removals of endotracheal tubes - about 5% among all participants - were more prevalent in the EN group. However, none of these caused death or serious complications, and the need to replace the tube was not different between groups, as reported elsewhere [
15].
The feasibility of EN sedation was lower, as this strategy was associated with a higher incidence of protocol violations. It is impossible to say whether these violations were due to higher sedation targets being set by physicians, inadequate drug dosage, or timing of administration. This last point was frequently reported as a problem in centers not used to managing EN sedatives. The raw number of violations has to be considered together with the very different amounts of unplanned drug administered, at times reaching up to one third of the planned amounts for propofol. The absolute difference in the proportions of work shifts with violation was 19.2% in the EN group and 10.7% in the IV group. The EN group had a smaller number of violations in a larger number of patients (46.6% vs 4.2%), meaning that the reasons for violating the protocol were not the same throughout the ICU stay. Perhaps for this reason, nurses judged the EN sedation as being as adequate as IV sedation (89.7 vs 92.4%). From these figures, the separation between EN and IV sedation may seem an academic question and in many cases a combination of both might provide a more rational approach.
In managing psychophysical agitation, one must consider the pros and cons of physical and pharmacological means of contention (restraints and drugs) [
32]. The cultural evolution [
33] in the management of conscious critically ill patients involves greater consideration of their surveillance. An updated approach should integrate the MV mode and weaning process, body posture and physiotherapy, nutrition, and communication strategies - also involving relatives at the bedside in ICUs that are open to family visitors.
Interestingly, in the EN group there was a significant tendency to a lower impact on organ function: MV was more assisted than controlled, urinary output was higher, infection signs were weaker, gastrointestinal motility worked better (Additional file
1: Table E5).
The present study brings to light the need for clinical/cultural change [
33] on two key points regarding the management of sedation therapy. First, despite great efforts to recommend aiming for the same RASS target in both arms, this decision was unexpectedly influenced by the group assignment (target RASS = 0 in 93.3% of the EN vs 82.9% of the IV group). Since this study is part of an educational research project, specific online medical education courses [
34] were offered. All staff members were invited to increase their knowledge and to use validated tools to evaluate pain, sedation, and delirium. Moreover, since they had to simultaneously manage two different protocols, a phone counseling service from the coordinating center was always available. Despite this, different sedation targets remained, probably because of different knowledge and expertise in the use of the two protocols, requiring the titration of drugs with different pharmacokinetics. The “fear” of a lighter sedation target probably increased when IV drugs with a short half-life were used: since their effect could run out in a few minutes, the patient might become suddenly agitated. On the other side, a “fear of oversedation” due to accumulation of oral drugs could have played a role in targeting lighter sedation in the EN group.
Second, the habitual use of IV sedation led physicians to plan and administer smaller amounts of enteral nutrition, probably because they know its side effects on gastrointestinal motility. In this context, the challenge to accept less powerful drugs (like hydroxyzine) and to keep patients more awake might serve as a means for introducing good clinical practices.
The adequacy of nutrition and drugs administered through nasogastric/nasojejunal tubes strongly depends on the ICU staff teamwork and problem-solving attitudes. In order to obtain the best results with such EN drugs with slow onset and offset, we recommended starting with the highest doses in the first 24 h on ICU, to withdraw the IV drugs early. Thereafter, the drugs could be accurately titrated by using validated tools to measure the results, together with a constant effort to decrease/suspend the drugs as early as possible.
Hospital charges for the drugs are altogether very low in relation to other ICU costs. The charges for planned sedatives were lower in the EN group and were higher for unplanned drugs. Considering the costs for neuroactive drugs altogether, there was a significant difference (IV 4.15 vs EN 2.39 €/MV day), meaning both that charges for antipsychotics were not increased, and charges for analgesics were slightly lower, probably because of melatonin’s pain-relieving effect [
35]. These charges are much lower than those reported in the literature; hospital costs could be significantly higher with respect to new drugs and approaches, like dexmedetomidine or sevoflurane [
25].
Study limitations and strengths
The unexpected difference in RASS targets was a significant limitation of this study, which might have favored the IV sedation protocol: as the patients were wanted to be more sedated, reaching such a target was easier. There are also several other limitations, like the single-blind design of the study, the data recording by clinical staff, the lack of anamnestic data on alcohol or substance abuse, the non-protocolized weaning from MV, the lack of a long-term cognitive outcome evaluation, and the sedation assessment over a whole nursing shift (prevalent RASS), which is very subjective. Moreover, some clinical practices were being introduced for the first time (EN protocol, conscious target, use of validated tools for neurological monitoring) in a substantial proportion of centers when the study was started. Data were gathered some years ago, and different skills and habits among intensivists in the use of sedatives could have played some role, particularly in the use of benzodiazepines, which is discouraged nowadays. Last, half the patients in the EN group had protocol violations, meaning the groups were not adequately separated.
The strengths of the present study are its design coherent with guidelines, always suggesting an early conscious sedation target [
1]. Rather than making it different in the two groups [
31,
36], two separate strategies were compared in the achievement of the same shared goal: a calm, conscious, and cooperative critically ill patient. Moreover, the lack of homogeneity among participant centers could render the results generalizable. Even with the large number of violations, the two strategies seem to be comparable: non-skilled centers can immediately use the EN strategy too.
Acknowledgements
The authors thank all the staff members of the participating ICUs for their essential cooperation. The corresponding author would like to dedicate the present study to the memory of Marco Rambaldi (1954-2014), friend and master, Steering Committee member and head of the Modena center, who was an insightful example of researcher, physician, and man. This publication endorses the “Humanization to Enhance Recovery On Intensive Care bundle” (
www.heroicbundle.org). We are grateful to J D Baggott for language editing.
Complete list of SedaEN investigators
Steering Committee members
Gaetano Iapichino (Study Chair), Alberto Morabito (Chair of Statistics, Dipartimento di Scienze cliniche e di comunità, Università degli Studi di Milano), Martin Langer (Dipartimento di Oncologia e Onco-Ematologia, Università degli Studi di Milano and Fondazione IRCCS Istituto Nazionale dei Tumori, Milano), Roberto Malacrida (Fondazione Sasso Corbaro, Bellinzona, Switzerland), Franco Valenza (Dipartimento di Fisiopatologia medico-chirurgica e dei trapianti, Università degli Studi di Milano), Marco Rambaldi (Nuovo Ospedale Civile Sant'Agostino Estense, Modena), Giovanni Mistraletti (Principal Investigator).
Investigators from the 12 participating centers
ASST Santi Paolo e Carlo, Ospedale San Paolo – Polo Universitario, Milano: Davide Chiumello, Giovanni Mistraletti, Michele Umbrello, Paolo Formenti, Paolo Spanu, Stefania Anania, Elisa Andrighi, Alessandra Di Carlo, Federica Martinetti, Serena Barello, Andrea Noto, Gianfranco Capello, Bruno Sabatelli, Giovanni Brenna, Morena Astori, and Pietro Placido.
IRCCS Ospedale Maggiore Policlinico, Milano: Luciano Gattinoni, Alessandro Protti, Paolo Cadringher, Riccarda Russo, Francesca Pagan, Virna Berto, and Paola Roselli.
ASST Monza, Ospedale di Desio (MI): Giulio Ronzoni, Eduardo Beck, Silvia Francesconi, and Maurizio Gaiotto.
ASST Ovest Milanese, Ospedale Nuovo di Legnano (MI): Danilo Radrizzani, Luca Ferla, Federico Valdambrini, Riccardo Giudici, and Laura Merlini.
ASST Monza, Ospedale San Gerardo, Monza (MB): Antonio Pesenti, Giacomo Bellani, Alessia La Bruna, Emanuele Rezoagli, and Alberto Lucchini.
IRCCS San Matteo, Pavia: AR1 Antonio Braschi, Alessandra Palo, Thekla Niebel, Marina Selvini, Sergio Cortesi, and Attilio Quaini; AR2 Giorgio Iotti, Francesca Riccardi, Enrico Contri, and Antonella Sacchi.
AO San Giovanni Bosco, Torino: Sergio Livigni, Giuseppe Naretto, Enrica Ferretti, Alessandro Deprado, and Virna Venturi degli Esposti.
AOU San Luigi Gonzaga, Orbassano (TO): Pietro Caironi, Giulio Radeschi, Maurilio Festa, Lorenzo Odetto, Daniele Ferrero, Stefano Cognolato, Roberto Penso, and Roberta Vacchelli.
AO Ospedale Cardinal Massaia, Asti: Silvano Cardellino, Edda Bosco, Anna Maria Gado, Anna Bresciani, Ivana Pozzo, Annachiara Alessio, Vanessa Clarindo Rodrigues, and Edna Biase.
AON SS.Antonio e Biagio e Cesare Arrigo, Alessandria: Nicoletta Vivaldi, Martina Taverna, and Antonella Nava.
Nuovo Ospedale Civile Sant’Agostino Estense, Modena: Marco Rambaldi, Cristina Pinna, Francesco Ponzetta, Lucilla Bavutti, Paola Martina, Beatriz Palacios, and Giancarla Bergonzini.
Take-home message
In 348 randomized ICU patients, an unusual enteral sedation protocol (hydroxyzine, lorazepam, and melatonin) gave no differences in clinical outcomes but lower costs than the more common intravenous sedation (propofol, midazolam). Mechanically ventilated patients may require different strategies to manage pharmacological sedation best: “gentle” enteral administration of drugs with longer half-life, accurately titrated to the lowest effective doses, could be an option to aim for, and to maintain them at a light level of sedation.
140-character Tweet
Enteral sedation (hydroxyzine, lorazepam, melatonin) was not superior to intravenous (propofol, midazolam), and cost less in ICU patients.