Guided sedation has been widely developed in recent years. However, to the best of our knowledge, there are no large-cohort studies describing a positive effect on the duration of mechanical ventilation in a paediatric population. Thus, we developed a sedation protocol and evaluated its influence on practices.
The median age of the population in each period was 2 and 3 years, respectively, which is similar to that of other studies [15, 18]. The mortality rate in our study population (17%) was higher than in our overall PICU population (5%). However, this higher mortality rate could be expected as ventilated patients are likely to be more severely ill, as illustrated by the high mean PELOD scores in both groups: 18.8 and 17.5.
Population targeted by the protocol
We limited the nurses’ actions to the group “adequate sedation” with scores between 11 and 17. In case of deep sedation required for medical reasons, expected scores ranged from 7 to 11. In this setting, any modification of sedation or analgesia had to be validated by the physician. Recently, Gaillard-Le Roux et al.  showed in a paediatric study that a nurse-driven protocol was also applicable for deeply sedated patients, with lower target scores. Generalising the use of the protocol to all patient categories is certainly a good manner to enhance its acceptability and its efficacy.
Impact on duration of mechanical ventilation
All recent large-cohort studies chose duration of mechanical ventilation as the primary outcome, but none were able to show a significant reduction in mechanical ventilation [15,16,17,18]. Except for the study by Curley et al., which was a randomised, controlled trial in children with ARDS, the other studies were designed as prospective or retrospective before and after studies.
Either with nurse-driven sedation protocol or with daily interrupted sedation, obtaining significant reduction in mechanical ventilation appeared to be difficult in PICU. This may be linked to various factors which may contribute to delay mechanical ventilation weaning, especially in younger children: difficulties to evaluate pain and comfort, fear of extubation failure, hemodynamic complications with sedation drugs, and occurrence of withdrawal symptoms. A wide weight range of children is admitted in PICUs (i.e. infants weighting less than 10 kg as well as children and teenagers weighing sometimes more than 80 kg), and there are probably important differences, especially pharmacologic ones, according to the age and to the reason for admission. A recent French paediatric study showed in a subgroup analysis a shorter length of mechanical ventilation among children older than 12 months after implementation of the protocol . The only large RCT published on this subject  included patients with ARDS only, who present specific characteristics: long respiratory recovery, use of deep sedation and neuromuscular blockades; in consequence, their results could not be generalised to the overall PICU population. However, we did not performed subgroups analysis as the power of our study did not allow to. Future larger studies could bring interesting results on the use of guided sedation and analgesia in general population in PICUs and in specific groups of patients according to age or admission diagnosis.
Impact on drug prescriptions
After implementation of a nurse-driven sedation protocol, Keogh et al.  showed a reduction in the duration of morphine administration of 19 h, and Neunhoeffer et al.  described a significant reduction in benzodiazepine cumulative doses after application of the protocol. Deeter et al.  compared patients treated under a nurse-driven protocol to a historic cohort and highlighted that the protocol enabled the duration of sedative drugs to be reduced from 7 to 5 days. For our part, we showed a downward trend in the duration of sufentanil administration and daily doses of ketamine in the post-implementation period, results not confirmed by segmented regression analysis.
Impact on withdrawal symptoms
In children, withdrawal symptoms occurred frequently when the analgesics and sedatives are stopped. The incidence of withdrawal reported in the literature varies from 10% , 37% , and even 48% . Withdrawal symptoms are linked to discontinuation of either opioids or benzodiazepines and result from central nervous system irritability, gastrointestinal dysfunction, and autonomic nervous system dysfunction. The WAT-1 score is now used routinely in our PICU to detect occurrences of withdrawal [25, 26]. In their before and after study design including 165 and 172 patients, respectively, Neunhoeffer et al.  showed that the use of a sedation protocol enabled significant reductions in the incidence of withdrawal symptoms from 23.6% before and 12.8% after implementing the protocol (p = 0.005). In our population, withdrawal symptoms tended to decrease without reaching significance.
Impact on evaluation of sedation and analgesia
One of our main positive results was the improvement in sedation and analgesia evaluation. A fear could be that guided sedation led to insufficient sedation. In their randomised clinical trial, Curley et al.  described more reports of agitation and elevated pain scores in the group under guided sedation. In our experience, we recorded no increase in the number of COMFORT-B scores >17 or in unplanned endotracheal extubations.
Limitations of the study
Our study has several limitations. First of all, it was not a randomised controlled trial but a “before and after” observational study performed within a single PICU, as majority of precedent studies [15, 17]; thus, results could possibly be influenced by temporal trend rather than to the protocol efficiency itself. This phenomenon was well described when interventions diffuse into widespread practice in an uncontrolled way while studies evaluating them are under way; for example, implementation of guidelines could have a positive impact before their official publication [33, 34]. Nevertheless, there was no change in general treatment or ventilator care strategies during the two study periods. Even so, we cannot exclude that implementation of the protocol could have made physicians and nurses more careful and aware of the importance of reducing unnecessary sedation and ventilation. However to limit this bias, we performed a segmented regression analysis.
Oversedation persisted in the post-implementation period, and this is probably one of the major limitations of our work. The proportion of scores <11 per patient found in our study is more important than reported in literature . This result could be explained by the fact that any patient receiving more than 24 h of sedation was included in our study, even patients for whom the protocol was not applied, like patients under deep sedation for medical reasons and patients under neuromuscular blockade. Compared to other studies, patients under neuromuscular blockade were not excluded from our study; the use of this treatment was often limited in time, mainly in patients with respiratory disease (i.e. ARDS), and the protocol was successfully applied after the neuromuscular blockade discontinuation. Even if the proportion of patients receiving neuromuscular blockades is comparable between the two groups, the assessments under these medications may partly explain the high proportion of oversedation observations.
Another point to explain oversedation was probably the imperfect compliance to the protocol and the reluctance of nurses to decrease the drugs especially in a calm and comfortable patient. Thus, lack of compliance could have a major negative impact on our results. Unfortunately, in this study, the protocol compliance was not assessed; it should be the aim of a further study; for example, it could be interesting to analyse the discordances between the COMFORT-B scores and the nurses’ actions.
Implementing the protocol corresponded with major changes in practice and considerable modification of the medical staffs’ and nurses’ roles, giving more responsibility and independence to the PICU nurses. In previous studies, nurses seemed the best able to evaluate the patient’s sedation state and to adapt the therapeutics . We certainly could have enhanced the quality of our work with realising a quality evaluation of our protocol. Keogh et al.  carried out a staff survey after their protocol implementation; they demonstrated that negative points were 1—difficult comprehension of the protocol, 2—need for full concentration with attention to details, 3—practice to become familiar with, and 4—lack of medical leadership. In our study, we should wonder if the 4-month period of adaptation was long enough, if all the nurses had received clear information and if sufficient continuous training was done after the implementation period.
Recently, Deeter’s team re-evaluated the effect of a nurse-driven protocol 5 years after its implementation . Whereas in 2008, just after the protocol implementation, they showed a reduction in the median total sedation days from 7 to 5 days, in 2013, 5 years later, they obtained longer duration of total sedation days, mechanical ventilation, and PICU length of stay. The authors involved lack of routine feedback and of ongoing education programme, loss of interest, and redirection of priorities to other clinical concerns. With our guided sedation protocol, we replaced sedation and analgesia at the centre of our care; in the future, it will require good communication and in-service training in order to ensure long-term adherence to the protocol and to sustain efforts over time.
Lastly, we unfortunately did not assess the impact of our nurse-driven sedation protocol on outcomes such as delirium and post-traumatic stress disorder (PTSD) which are important issues for patients admitted to intensive care units. Indeed, the reported prevalence of paediatric delirium in PICU varies between 4 and 29%  and may be attenuated by better sedation and analgesia management. Regarding PTSD, a recent review  reported an incidence in adults of 17–34% one year after intensive care unit discharge. Risk factors seemed to be anterior psychological disorders, benzodiazepine administration, and delirious memories facilitated by pain or agitation. In children, Colville et al.  questioned 102 children (7–17 years of age) three months after their hospital discharge, according to the “Children’s Impact of Event Scale” and found an incidence of PTSD of 28. Administration of opioids or benzodiazepines for more than 48 h was significantly associated with the existence of delirious memories. The risk of occurrence of PTSD was 3 times greater when delirious memories were related. In the future, it could also be relevant to evaluate whether or not guided sedation protocols could influence delirium prevalence and long-term psychological complications in children.