Introduction

The provision of adequate sedation and analgesia to critically ill children is an important aspect of care in the pediatric intensive care unit. Sedatives and analgesics reduce anxiety, pain, and agitation, enhance synchronization with mechanical ventilation, and enable invasive procedures to be performed. Adequate sedation is defined as the level of sedation at which patients are asleep but easily arousable [1]. Oversedation delays recovery, as greater sedatives consumption is associated with longer duration of ventilation as well as extubation failure [2]. Oversedation also induces tolerance and withdrawal syndrome [3, 4]. Undersedation, on the other hand, may lead to increased distress and adverse events such as unintentional extubation or displacement of catheters. All this may also lead to a longer ICU stay.

Children are usually sedated through a combination of hypnotics (e.g., midazolam) and analgesics (e.g., morphine or fentanyl) [5–7]. Regrettably, there is little evidence from randomized trials on the efficacy of these drugs for sedation in critically ill children [8]. Nevertheless, efforts are being made to improve sedation management, for example with the use of sedation algorithms and standardized sedation management [9, 10].

To achieve the optimal level of sedation in individual patients, doses of sedatives are individually titrated to effect. This process is guided by scores on a variety of observational sedation scales [5]. The COMFORT score or COMFORT behavior scale and the Hartwig sedation scale are widely used and validated for this setting [11, 12]. Other scales used are the Ramsay scale [13], Richmond Agitation Sedation Scale (RASS) [14], State Behavior Scale (SBS) [15], and the University of Michigan Sedation Scale (UMSS) [16]. In addition, methods derived from the electro-encephalogram (EEG), such as the Bispectral Index (BIS) and middle latency auditory-evoked potential index (AEP), are applied, although their use is not validated in young children [17].

The aim of this systematic literature review is to evaluate the reported incidences of under-, optimal, and oversedation in pediatric intensive care patients and to determine to what extent the goal of adequate sedation is met [18].

Methods

Search strategy

A systematic literature search was performed in the PubMed and Embase databases from inception to July 2012, using the terms sedation, child, intensive care unit, and sedation quality/sedation level. We used a comprehensive search strategy to identify all published articles evaluating the level of sedation, measured with an observational scale, in pediatric intensive care patients. For Embase, appropriate search terms were applied. Full details of the search strategy are presented in Appendix 1. Furthermore, reference lists of retrieved articles were searched to identify other relevant papers that complied with the inclusion criteria.

Selection criteria

We used the following inclusion criteria:

  1. 1.

    Study population of PICU patients (0–18 years) on mechanical ventilation and receiving continuous sedation.

  2. 2.

    Reporting level of sedation and/or the incidence of under-, over-, and optimal sedation, as defined in the study.

Studies published in any language with an English-language abstract were eligible for review.

Exclusion criteria were:

  1. 1.

    Procedural sedation

  2. 2.

    Preterm patients

  3. 3.

    Patients treated with muscle relaxants, which preclude the use of sedation scores

  4. 4.

    Studies using only the BIS monitor in children aged <1 year, since this method is not validated for this patient group [17].

Two authors (NV, EI) independently reviewed titles and abstracts of all retrieved citations to identify eligible studies. Of all included studies, the full-text articles were again reviewed to ensure that they met inclusion criteria. Disagreements between reviewers were resolved by consensus.

Data extraction

Two authors (NV, EI) each independently recorded the following data: country of origin, study objective, study design, study population, age of patients, sample size, details of study intervention (if applicable), sedatives used (drug, dose), length of sedation, sedation scale used, and the incidence of optimal sedation, and under- and oversedation. We used the definitions for optimal sedation as used by the researchers in the individual studies (as percentage of number of observations, patients or time) to be able to pool the data, despite different sedation assessment methods (Table 1).

Table 1 Summary of included studies reporting the incidence of optimal, under-, and oversedation
Full size table

Quality assessment

Study quality was determined with the “Quality Assessment Tool for Quantitative Studies” by the McMaster University, School of Nursing [19] as strong, moderate, or weak.

Statistical analysis

We analyzed studies separately on study design, sedation scale used, and proportion of under-, over-, and optimal sedation. Proportion was expressed as percentage of number of observations, patients or time (h). If similar outcome measures were used, the results of individual studies were quantitatively pooled to calculate a weighted mean, using descriptive statistics. The large heterogeneity in study aims and study designs precluded further statistical analysis.

Results

Study selection

After filtering out duplicate studies, our search yielded 392 potentially relevant articles. Of these studies, 348 were excluded on the grounds of information in title and abstract (Fig. 1). Of the remaining 44 articles, the full-text was retrieved and assessed for eligibility. Nineteen studies were excluded for lack of quantitative data on sedation level or incidence of optimal-, under-, or oversedation, or for absence of a definition of optimal sedation. Details of the remaining 25 studies are presented in Table 1.

Fig. 1
figure 1

Flowchart search results

Full size image

Study characteristics

One study was a randomized controlled trial (comparing two sedative regimens); 22 studies were prospective observational studies; and two were retrospective studies on a sedative drug. Of all 25 studies, only two determined the level of sedation as primary study outcome [20, 21]. Fifteen studies investigated one or more sedation scales or sedation monitoring systems (such as the BIS) [11, 12, 15, 22–33]; six studies investigated a sedative drug [34–39]; one was a pharmacokinetic-pharmacodynamic study [40]; and one study described the effect of implementation of a sedation protocol on amount of sedatives administered [9]. Although assessment of level of sedation was not the primary objective in the latter 23 studies, they reported incidences of under-, optimal-, and oversedation.

Since sedation practices may differ between countries, we also looked at the country of origin. Of the 25 studies, eight were conducted in the United States, 16 in six European countries, and one in Brazil.

All studies together investigated a total of 1,163 critically ill children. The most frequently used drugs were benzodiazepines (midazolam, in 22 studies) and opioids (morphine, in 14 studies). Other drugs used were fentanyl, ketamine, clonidine, propofol, barbiturates, chloral hydrate, first-generation antihistamines, and dexmedetomidine in different combinations.

Quality assessment

Only two studies had level of sedation as their primary outcome, all other studies varied by aim and study design. Therefore, assessment of study quality with the “Quality Assessment Tool for Quantitative Studies” was not possible, and this makes direct comparison between the studies difficult.

Sedation scales

Across all studies, 12 different observational sedation scores were used, of which four were validated for the PICU setting, i.e., the COMFORT-score, the COMFORT-B scale, the Hartwig sedation scale, and the State Behavior Scale. Most frequently (11/25) used were the COMFORT-score and COMFORT-behavior scale (COMFORT-B), followed by the Ramsay score, the State Behavioral Scale, and the Hartwig sedation scale. Six studies (23 %) used the BIS monitor. In 13 studies two or more sedation scales or monitors were used.

All studies defined optimal sedation in terms of cut-off values (Table 1). The definition of optimal sedation differed between studies, even when the same sedation scale was used. For example, a COMFORT score between 17 and 26 is thought to indicate adequate sedation [20]. However, one study applied the 13–23 range to define adequate sedation [35]. This range was chosen a priori to target a level of sedation that would produce a patient who was under analgesics, calm, with minimal risk of self-extubation, but able to maintain an appropriate cough reflex and spontaneous respiratory effort to achieve ventilator synchrony. Furthermore, different cut-off values for the Ramsay score were used: i.e., 2–3 [25]; 2–4 [26]; and 1–5 [28, 30]. Assessment frequency also varied considerably between studies; from once daily to hourly.

Level of sedation

Reported incidences of optimal, under-, and oversedation are presented in Table 1.

Studies varied in the way incidence was reported (as a proportion of observations, patients or hours). Fifteen studies reported the incidence as a proportion of observations, as summarized in Fig. 2. Optimal sedation was ascertained in 15–93 % of observations, undersedation in 0–22 %, and oversedation in 0–82 % of observations. In these 15 studies, patients were optimally sedated in 57.6 % of the observations, undersedated in 10.6 % of the observations, and oversedated in 31.8 % of the observations.

Fig. 2
figure 2

Incidence of under-, optimal, and oversedation (% of observations)

Full size image

Two studies reported proportions of patients; in these two studies together, 68.6 % of patients were oversedated at any time during admission (Fig. 3).

Fig. 3
figure 3

Incidence of under-, optimal, and oversedation (% of patients)

Full size image

The two studies that used both an observational score and the BIS score reported considerably different results [28, 30]. The incidence of oversedation measured with the BIS was lower than that measured with a validated observational scale (56 vs. 92.9 % and 65 vs. 82 %).

Discussion

This review shows that the level of sedation in critically ill children is often suboptimal during their ICU stay, at least in ICUs that apply sedation assessment in daily practice. Patients are optimally sedated in only 60 % of assessments. Under- and oversedation occur in 10 and 30 % of the assessments, respectively. However, across all studies, there is a large variation in incidence of oversedation, i.e., from 0 to 82 % of assessments. Most studies, however, report incidence in the range of 40 to 65 %, which corresponds to that reported in adult ICU patients [41–43].

Our results indicate that in critically ill children oversedation is more common than undersedation. We suggest several reasons for the relatively high incidence of oversedation. First, there may be a tendency to avoid undersedation at all cost, as this may lead to discomfort and potential adverse effects as self-extubation and removal of lines and catheters. Since children, especially preverbal infants, cannot clearly communicate their well-being and are often bewildered by the ICU setting, nurses and doctors may also tend to avoid undersedation. Second, nurses believe that mechanical ventilation is uncomfortable and stressful, and this perception might lead to higher sedation level than necessary [42, 44]. Third, sedation protocols are not fully adhered to, so that sedatives are not tapered off when possible [45]. These tendencies are unwanted, as oversedation may be even more detrimental to patients.

Continuous sedation as such is an independent predictor of prolonged mechanical ventilation in adults, and consequently leads to longer ICU and hospital stay [46]. Oversedation, in addition, is also associated with tolerance, withdrawal, and delirium. Especially longer duration of use and high drug doses are risk factors for development of withdrawal symptoms in children [4]. Moreover, longer use of sedatives has been associated with symptoms of depression and post-traumatic stress symptoms in adults [47]. In a study in children, almost one-third of children reported delusional memories, and these were the children with the longest duration of administration of opiates/benzodiazepines and the highest risk of posttraumatic stress [48]. The administration of sedatives to children may also be associated with adverse neurodevelopmental outcomes at later age, probably by inducing neuroapoptosis [49–51].

The implementation of sedation algorithms aimed at less sedation has led to shorter duration of mechanical ventilation, ICU stay, and hospital stay in adults [52]. Also, daily sedation interruption significantly improved short- and long-term outcomes in adults [53]. A more recent “no-sedation” protocol is even more promising in this respect [54]. All evidence indicates that the use of sedative drugs should be reduced. In children, daily sedation interruption seems feasible and safe, but effectiveness needs to be demonstrated in large trials [55].

This review also shows a great variety of assessment instruments used in clinical practice. No more than four of the 12 observational sedation scores have been validated for PICU patients, i.e., the COMFORT-score, the COMFORT-B scale, the Hartwig sedation scale, and the State Behavior Scale. This is remarkable, as there is consensus that the level of sedation should be assessed and documented using a validated sedation assessment scale [5]. The reliability of the other scales is questionable. Furthermore, six studies used the BIS monitor. There is insufficient evidence, however, to support the use of the BIS monitor, or any other neurophysiological sedation scoring technique, such as auditory evoked potentials, in children below the age of 6 months [56]. The suitability of the adult-derived EEG algorithm to assess children’s BIS values is doubted. Furthermore, pre-awakening BIS values in children aged <1 year are lower than in older children [57]. This could explain why in some pediatric studies BIS monitoring resulted in a lower incidence of oversedation than did application of the COMFORT score [28, 30].

In all studies the authors defined optimal level of sedation. Remarkably, different studies applied different cut-off values of the COMFORT score and Ramsay score [25, 26, 28, 30]. This variation may be explained by the uncertainty in what constitutes optimal sedation, but may also be the result of patient-specific factors. For example, a deeper level of sedation is often aimed for in patients with pulmonary hypertension, traumatic brain injury or difficult airway. Playfor et al. [21] used a clinical sedation score based on the response to tracheal suction, categorizing the response on a five-point scale. A score of 1 (no response to tracheal suction) was considered as the desired level of sedation for children with severe head injury; a score of 2 for children receiving a high level of intensive care with frequent invasive procedures, and a score of 4 for children prior to extubation.

In addition, the relatively high incidence of suboptimal sedation shown in this review reflects the fact that titrating the correct amount of sedation for each child can be complex. There may be several reasons for this. First, PICU populations are quite heterogeneous with respect to disease type and severity, age, and neurodevelopmental stage, so optimal sedation management may differ widely. Second, pharmacokinetics and pharmacodynamics, largely insufficiently studied, may be unpredictable, particularly in patients with multiorgan failure [58]. Dosing regimens are often based on healthy adult volunteers and do not take into account factors such as altered protein binding, distribution, and clearance in critically ill children. Also, sedation requirements may change over the course of illness [59].

With the risks of oversedation and the difficulties of reaching adequate sedation in mind, a critical appraisal of sedation strategies in critically ill children is needed. Optimal sedation could perhaps be achieved with the use of validated sedation scales and standard sedation protocols and by studying promising interventions such as daily sedation interruption. These studies are needed in pediatric intensive care.

Conclusions

This review shows that optimal sedation for critically ill children remains challenging for health professionals. These children are often oversedated and consequently run the risk of adverse outcomes. It is high time to find conclusive evidence on optimal sedation strategies in the PICU setting.