Keywords

Ensuring safe care for critically ill children necessitates, in part, effective analgesia and sedation. Yet, the use of analgesic and sedative medications also carries risks. Opioids and benzodiazepines are associated with respiratory depression, the development of tolerance, iatrogenic withdrawal syndrome, and delirium [1,2,3,4,5,6,7]. Ketamine can cause hypertension, increased oral secretions, emergence delirium, agitation, and severe hallucinations [6, 8]. Dexmedetomidine has cardiovascular effects, including bradycardia and hypo- or hypertension that may limit its use in select patient subpopulations. It has also been associated with withdrawal after prolonged infusions [8,9,10,11,12]. Several studies have reported that exposure to analgesic and sedative medications early in life has been linked to adverse neurodevelopmental outcomes [13,14,15]. Although these outcomes have not been reported consistently [16,17,18], they remain a legitimate cause of longer-term concern. Oversedation can lengthen the duration of mechanical ventilation, increase the risk of extubation failure, and increase healthcare costs, in addition to increased exposure to analgesic and sedative medications, further contributing to the development of tolerance, iatrogenic withdrawal syndrome and delirium [19,20,21]. Meanwhile, undersedation is associated with agitation; increased risk for adverse events, such as self-extubation or dislodgement of invasive devices; and psychological distress for both patients and their parents [22,23,24,25].

Achieving adequate analgesia and sedation in the pediatric intensive care unit (PICU) is an ongoing challenge. Evidence-based guidelines have increasingly advocated for the assessment of analgesia and sedation using standardized observational assessment tools and titrating depth of sedation to a prespecified goal [26,27,28]. However, surveys and observational studies of critical care practice indicate that the implementation of such practices is far from universal; in one survey, only 36% of respondents reported the use of sedation protocols [29], and in another, only 42% of units routinely set patient-specific sedation goals [30]. There is burgeoning evidence for the safety and role of daily sedation interruption and standardized, goal-directed sedation in reducing cumulative exposure to analgesic and sedative agents [31,32,33,34,35,36,37,38,39,40], thereby reducing the incidence of complications, such as withdrawal and delirium [37, 41,42,43,44]. Nevertheless, considerable variation exists in the dosing and administration of analgesic and sedative medications across institutions [2, 20, 29, 45,46,47], in part due to clinician biases and variable implementation of evidence-based practices [48,49,50,51,52,53,54,55]. With the goal of providing safe, developmentally appropriate analgesia and sedation to critically ill children, this chapter will discuss available assessment tools for monitoring sedation and analgesia in the PICU and their roles in the context of analgesia and sedation management, including evidence-based clinical practice guidelines.

Assessment Tools

Pain Assessment

It is increasingly understood that adequate analgesia must be achieved prior to attempting to titrate sedation in a critically ill child. Consequently, increasing sedation in the setting of ongoing pain represents a failure to recognize one key source of patient agitation and contributes to apparent sedation “failures.” Painful procedures are common in the PICU environment, with one recent study finding a median of 11 painful and stressful procedures occurring per patient per day [48]. Additionally, there is an increasing recognition that routine care practices in critically ill patients, such as endotracheal tube suctioning, repositioning, and mobilization, are significant sources of pain rather than mere agitation [56,57,58]. Pain expression also changes across age and developmental spectrums [59]; therefore, different tools have been developed and validated for use in specific age groups and for particular types of pain among critically ill infants and children. Providers must use their clinical acumen to select the appropriate assessment tool for a given patient, based on the child’s verbal and cognitive abilities, clinical status, and consideration of the applicable population and purpose of available tools. Any assessment tools that are chosen for use in a given institution should be implemented with extensive education of clinical staff on appropriate application and scoring of the tool, and procedures should be put in place for establishing and maintaining interrater reliability to ensure consistent assessment of pain across raters [60,61,62].

Self-Report Tools and Surrogate Reporting

Self-report of pain is widely recommended but challenging to achieve in the PICU. In addition to individual patients’ developmental capacity, the impact of underlying disease, use of sedation therapies, and endotracheal intubation and mechanical ventilation may all adversely impact the child’s ability to communicate regarding pain, verbally or otherwise. Despite these limitations, research suggests that children age eight and older can provide valid self-reports of pain using a numeric rating scale of 0–10 [63,64,65] or faces pain scale, such as the widely known Wong–Baker Faces Pain Rating Scale [66]. These scales have been shown to be valid in populations of children as young as 3 years and older who are able to engage in self-report [67]. Unfortunately, development and validation of numeric rating scales or faces pain scales specifically for use in intubated and mechanically ventilated children have received limited research. While recent studies have shown that children can safely be maintained in a more awake state during mechanical ventilation [31], which may facilitate the child’s self-report of pain using appropriate augmentative and alternative communication devices [68], further evidence is needed to explore the efficacy and validity of such approaches. Therefore, it is currently recommended that the assessment of pain through self-report be limited to children with verbal abilities who can utilize either a faces pain scale (age 3 and up) or a numeric rating scale (age 8 and up).

Surrogate reporting of pain by a parent or other caregiver familiar with the child’s typical pain behaviors can provide valuable information, particularly in nonverbal children or patients who may exhibit more subtle behavioral cues. However, significant discrepancies have been reported between parent, nurse, and patient (when able to self-report) ratings of pain. In a meta-analysis comparing these assessment sources, only moderate correlations between either nurse or parent ratings of pain and the child’s self-report were found [69]. A recent cohort study identified only 50–68% agreement between parent and child reports of pain [70], and another reported that parents of children with cerebral palsy tended to report significantly lower pain scores compared with both their children and other observers [71]. Thus, in children unable to self-report pain, it is recommended that a multimodal approach be utilized in which surrogate reporting is combined with other observational methods of pain assessment [72].

Observational Tools

Among neonates and infants, the Crying, Requires increased oxygen, Increased vital signs, Expression, Sleeplessness (CRIES) scale [73] and the Neonatal Pain, Agitation and Sedation Scale (N-PASS) [74, 75] have been developed to evaluate acute procedural and postoperative pain. The N-PASS has additionally been found to have good reliability and validity for assessing prolonged pain, such as that associated with a critical illness [74], and has moderate correlations with nursing assessments of pain and agitation [76]. Both tools can also be used to assess premature neonates. The CRIES has been reported effective in neonates down to 32 weeks gestational age, whereas the N-PASS has been evaluated in neonates as young as 23 weeks gestational age. The CRIES tool has five domains (see above), each scored between 0 and 2, resulting in a summary score ranging from 0 (no pain) to10 (most intense pain severity). Since the N-PASS measures both pain and sedation, its scoring is slightly more complex. It also has five domains (Crying/irritability, Behavior state, Facial expression, Extremity tone, and Vital signs), which are graded 0–2 for pain/agitation and 0, −1, or −2 for sedation, resulting in scores ranging from −10 (deeply sedated) to 10 (intense pain/agitation). Scores are corrected for gestational age [74]. Older consensus statements on pain assessment in nonverbal neonates and infants have particularly recommended the CRIES [72, 77], but those documents were released prior to the development of the N-PASS. A 2016 policy statement from the American Academy of Pediatrics simply recommends routine assessment of pain in critically ill neonates and infants, without endorsing the use of one particular tool [28]. A recent study comparing five available assessment tools for pain in neonates similarly concluded that the available tools have comparable and acceptable psychometric performance, and the authors recommend that clinical utility and practicality of tool implementation be used as the basis for selection of a specific tool [78].

The Face, Legs, Activity, Cry and Consolability (FLACC) scale was designed to evaluate acute procedural pain in children that are 2 months to 7 years of age in the postanesthesia care unit [79]. Its use was subsequently extended to the assessment of preverbal patients in the PICU [80] and for procedural pain in children that are 5–16 years old [81]. A modified FLACC has been found to be valid for the measurement of acute pain in critically ill and mechanically ventilated children up to 13 years of age [82, 83]. The FLACC tool has five domains scored between 0 and 2, resulting in a summary score ranging 0–10 and higher scores reflecting increasing pain severity. It relies solely upon behavioral indicators of pain, an important consideration since physiologic parameters, such as blood pressure, respiratory rate, and heart rate , have been found to be unreliable measures of pain and poor predictors of analgesic requirements [84, 85]. Consensus recommendations for pain assessment in nonverbal infants and children have encouraged the use of the FLACC [27, 72, 86], and it has been implemented in a variety of clinical settings [29, 45, 87, 88]. For children with cognitive impairment, a revised FLACC with individualized pain behaviors added by parents has demonstrated reliability and validity for postoperative pain assessment [83, 89].

Two additional scales have been published but are more limited in that they have only been validated for use during specific procedures or in specific populations. The Hartwig scale specifically assesses pain during endotracheal tube suctioning and is, thus, limited in generalizability. The scale utilizes five criteria (motor response, facial grimace, eye movement, respiratory pattern, and reaction to endotracheal tube aspiration), each graded from 1 to 5 for a possible score range of 5–25 [90]. In critically ill patients aged newborn to 15 years, the scale has demonstrated good internal consistency and validity [90, 91]. The Cardiac Analgesia Assessment Score (CAAS) was specifically developed to evaluate postoperative pain in mechanically ventilated children following cardiac surgical procedures [92]. It includes one behavioral (motor/respiratory response) and three physiologic (HR, BP, and pupillary size) components, each scored from 0 to 2 for a score range of 0–8. While it has been validated in cardiac surgical patients aged 0–19 years, the specificity of this population makes generalizability to broader PICU populations, surgical or otherwise, unclear.

Agitation and Sedation Assessment

Once adequate analgesia is ensured, the child’s level of sedation can be assessed, which typically requires providers to make subjective judgments or conclusions regarding behavioral responses to stimuli. Optimal sedation has been defined in clinical guidelines as a state in which the child is somnolent but responsive to the environment, while tolerating therapeutic procedures without excessive movements [20, 27]. Two instruments have been developed and are used in many ICUs in an attempt to standardize assessment and improve titration of sedative medications [29, 30, 45]: the State Behavioral Scale (SBS) score and the COMFORT score.

Similar to the Richmond Agitation-Sedation Scale (RASS) used in adult critical care, the SBS score describes a patient’s level of sedation across a continuum of behavioral responses, from −3 (unresponsive/comatose) to +2 (severely agitated), during normal care [93]. If a child is unresponsive during the initial, pre-stimulus observation period, a progressive stimulus is used to evaluate the patient’s level of sedation. The SBS has demonstrated good inter-rater reliability and constructs validity in assessing sedation and agitation among critically ill children across the age spectrum, though it was initially validated in children that are 0–6 years old. It has been incorporated into several published studies and standardized sedation protocols [31, 32, 94, 95].

The COMFORT score was originally developed as a tool with six behavioral dimensions and two physiologic dimensions, to be used for assessment during continuous analgesic and/or sedative infusions in the PICU [96, 97]. However, subsequent psychometric testing and the recognition that physiologic variables may be influenced by both common medications used in the PICU and by critical illness itself, led to the exclusion of physiologic items and development of the COMFORT-Behavioral (COMFORT-B) scale [85]. The COMFORT-B scale demonstrated improved internal consistency, is reliable and valid for the assessment of children that are 0–17 years of age, and discriminates between under- and oversedated children in the PICU, with good sensitivity to change in level of comfort following analgesic and sedative administration [85, 98, 99]. It has six domains scored between 1 and 5, with summary scores ranging 7–30 and higher scores reflecting increasing agitation. However, the tool’s length and simultaneous assessment of pain and sedation/agitation has hindered its use in some clinical settings, and the quality of evidence regarding its use is mixed [100]. A recent survey suggested that the SBS score is more commonly used for sedation assessment in the USA, whereas the COMFORT and COMFORT-B scales have been widely implemented internationally [30].

As indicated above, the RASS is commonly utilized in adult critical care while only a single validation study has been performed in critically ill children aged 2 months to 21 years [101]. Scores range from −5 (unarousable) to +4 (combative) with a score of 0 representing an awake, alert, and calm state. Despite this limited evaluation in critically ill children, use of the RASS is likely to continue to grow, as it is the score which both the pediatric Confusion Assessment Method for the ICU (pCAM-ICU) and Cornell Assessment of Pediatric Delirium (CAPD) tools use for determining eligibility to perform delirium screening in children [102,103,104].

Objective Monitoring Tools

The subjective nature of observational assessment tools has led many clinicians to desire an objective monitoring method that can be used to titrate sedation to more precisely fit an individual patient’s needs. From a safety standpoint, cardiorespiratory monitors and pulse oximetry should be considered standards of care in monitoring for sedation-induced effects on hemodynamic status and respiratory effort. For procedural sedation in non-intubated children, the use of end-tidal capnography is additionally recommended to detect hypoventilation that would not be detected by pulse oximetry [105]. Careful examination of the capnographic waveforms can additionally allow early identification of different sources of airway compromise [106]. Heart rate variability (HRV) is an emerging area of interest in analgesia and sedation monitoring. The analgesia nociception index (ANI) is a quantitative value derived from measures of HRV that may allow detection of lesser degrees of nociception during sedation or anesthesia which the above-discussed observational assessment tools are inadequately sensitive to identify [107,108,109]. Research in critically ill adults suggests that the ANI may be most useful for detection of pain during routine care [110]. However, to date, this tool has not been evaluated specifically in the PICU setting. Its utility may be further limited by the effects of sedative medications common in the critical care setting, such as dexmedetomidine, that affect sympathetic tone and, consequently, HRV [109].

Electroencephalography (EEG) sensitively measures electrical activity within the brain in the clinical setting. However, the complexity of the output typically requires in-depth training and experience to interpret accurately, and the volume of information gleaned generally precludes comprehensive, real-time analysis. Advanced mathematical modeling allows EEG signals to be broken down into frequency, phase, and amplitude components, which can then be interpreted using proprietary algorithms. Such processed EEG measures have been investigated for bedside neurological monitoring and sedation assessment, including the Bispectral Index score (BIS) and amplified EEG (aEEG) , which will be reviewed in the sections that follow.

Processed EEG

The BIS was developed and released in the mid 1990s as a brain function monitor to assist with monitoring depth of anesthesia in the operating room, but its use has subsequently been expanded to a number of other clinical areas [111]. The BIS is a processed EEG measure that uses a proprietary and unpublished algorithm to distill cerebral electrical activity into a numeric scale with values ranging from 0 (isoelectric EEG) to 100 (fully awake) [111,112,113]. Those values can be further stratified into four clinically relevant categories reflecting increasing depth of sedation: from fully awake or lightly sedated with potential for recall, to very deep sedation similar to general anesthesia. Additional parameters provided along with the BIS value are the suppression ratio, representing the cumulative percentage of burst suppression corresponding to cortical silence over the previous 65 seconds; electromyographic (EMG) activity, reflecting high frequency from muscle movement; and signal quality index (SQI), a global parameter reflecting electrode impedance and artifacts, including EMG activity. Considering the BIS value alongside the suppression ratio, EMG activity and SQI is meant to help the clinician evaluate the quality of the BIS signal and determine whether it is a reliable indicator of the patient’s level of sedation.

Though originally developed and tested in adults, several studies have examined the use of BIS monitoring in the pediatric critical care setting in children from neonates to 18 years old [114,115,116,117,118,119,120,121,122,123,124]. BIS values have demonstrated reasonable sensitivity and specificity in differentiating between inadequate and adequate levels of sedation in older children [115, 118, 125, 126], whereas the burst suppression ratio and other EEG parameters have been shown to change with increasing age as a function of cerebral maturation and should be interpreted with caution [118, 126]. In general, it appears that the BIS is more accurate in differentiating inadequate from adequate sedation than it is for differentiating adequate from excessive sedation [114, 119, 120, 125, 126], with low to moderate correlations with sedation scores, such as the COMFORT score [116, 117, 124, 125]. Given the trend towards avoiding deep sedation except when clinically necessary, this drawback is a significant hindrance to extending the use of BIS monitoring. EMG activity can lead to interference that artificially elevates BIS scores [127], which may or may not reflect inadequate analgesia and sedation depending on the source [113, 128]. Elimination of EMG interference by neuromuscular blockade may not improve the accuracy of BIS monitoring [123, 124, 128,129,130], and in fact reliance on BIS values seems to lead to oversedation [130, 131]. Thus, BIS monitoring may have challenges in both cases of under- and oversedation. BIS monitoring during the use of ketamine and dexmedetomidine may not accurately quantify sedation as those agents appear to affect the EEG differently than other sedatives [111,112,113, 126, 132], which may be a significant limitation in the context of increasing use of both medications in current sedation practice [2, 6, 133]. Finally, abnormal brain activity as a consequence of delirium, encephalopathy, trauma, or other focal neurologic processes may all be reflected in EEG changes that could be misinterpreted as sedation, depending on electrode placement [113, 116, 128].

Amplitude-integrated EEG (aEEG) filters continuous EEG output to remove both low-frequency (<2 Hz) and high-frequency (>15 Hz) signals, smoothing and reformatting the signal for display in a single time-compressed waveform. In some neonatal ICUs, aEEG has become standard for prognostication after hypoxic–ischemic injury [134], but its use for sedation monitoring is less-well studied. Sedative and antiepileptic medications lower the background EEG amplitude and alter sleep–wake cycling (SWC). However, the results across studies of aEEG use during sedation are mixed: midazolam was shown to delay onset of normal SWC and fentanyl caused continuous low voltage in newborns undergoing cardiac surgery [135], whereas in another study of neonates undergoing noncardiac surgery, neither morphine nor midazolam altered background aEEG pattern [136]. Giordano et al. (2018) reported that aEEG was able to differentiate between no sedation and either light or deep sedation in ventilated neonates, but, similar to the BIS, it could not sensitively differentiate between light and deep sedation, and it was not superior to assessment using the N-PASS tool [119]. All aEEG parameters were depressed by increasing doses of analgesics and sedatives. Therefore, there is insufficient evidence to recommend the routine use of aEEG alone for sedation monitoring in the PICU, particularly outside the neonatal population given the lack of data in this older pediatric population.

Conventional EEG

Conventional continuous EEG monitoring is currently indicated for use in the PICU for children with refractory status epilepticus, acute encephalopathy, intracranial pressure management, acute brain injury, including traumatic brain injury and hypoxic–ischemic brain injuries, and altered mental status of unknown etiology [137]. There is also evidence that children who have undergone surgery for congenital heart disease are at risk for postoperative seizures [138,139,140], and continuous EEG monitoring in that population is increasing. For children undergoing sedation with or without neuromuscular blockade, continuous EEG monitoring offers the advantage of identifying nonconvulsive electrographic seizures; the majority of which may not have visible clinical signs but are associated with poor outcomes [137]. Reported practice among neonatologists is to perform continuous EEG monitoring for 1–2 days if no seizures are detected [141], though it has been suggested that duration of monitoring be tailored to the patient’s age, clinical status, and cause of acute encephalopathy if seizures may reasonably be expected to arise with changing clinical status (e.g., during rewarming after therapeutic hypothermia) [137]. The most common use of continuous EEG data in altering the plan of care is initiating, titrating, or discontinuing anticonvulsant medications due to seizure identification (or lack thereof).

Evaluation of continuous EEG background features, such as burst suppression, discontinuity, attenuation, reactivity, and periodic or multifocal epileptiform discharges, has been used to aid in prognostication in patients who have sustained cardiac arrest, status epilepticus, or hypoxic–ischemic injuries [142, 143]. However, the need for real-time interpretation of EEG data and limited availability of neurologists during acute changes in patient status has led to increased interest in quantitative EEG algorithms that can be interpreted by bedside clinicians [134]. While such algorithms are commercially available, their use is subject to substantial variability in correct identification of seizure activity dependent upon user experience, and to date their use in practice is infrequent [137]. Given the difficulties with implementing an easily interpreted quantitative EEG in the PICU for seizure detection, it is unlikely that continuous EEG could feasibly be used for titration of sedation in the PICU, particularly given the complexities of sedative medication impact on EEG spectra in various brain regions [132, 144,145,146] and significant interindividual and age-related variability in calculated EEG parameters [147].

Conclusions

There is insufficient evidence for any of the currently available objective monitoring technologies to be used as stand-alone modalities for monitoring sedation in the PICU and, when used, information from these devices should be weighed in conjunction with observational assessment tools to form a global impression of the patient’s level of sedation [112, 128]. In particular, it is important to consider the effect of the chosen analgesic and sedative regimen on EEG activity [112, 113], as well as altered pharmacokinetics and pharmacodynamics of analgesic and sedative medications during critical illness that may not be reflected in EEG or other objective monitoring parameters [6]. The exception to this may be the child in whom continuous neuromuscular blockade is being utilized and in whom BIS monitoring in particular may be helpful in trending depth of sedation. More research is needed to establish the role and utility of EEG-based monitoring devices, and this work is in progress; a working group of clinical and engineering experts is currently designing a “neuroPICU” display that would support rapid review of neurologic and physiologic data in specialized visualizations which would be modifiable based on the specialty of the reviewer [148]. However, such a device will require extensive testing prior to implementation to determine its efficacy specifically for sedation monitoring, as well as its effect on relevant patient outcomes, such as sedative exposure and iatrogenic complications. Furthermore, use of objective monitoring devices should not supersede the implementation of evidence-based practices, including setting sedation targets and utilizing standardized, goal-directed sedation protocols that incorporate validated pain and sedation assessment tools.