Delirium is a frequent complication of pediatric critical illness and is associated with substantial morbidity.
Refractory agitation in the pediatric intensive care unit is often a manifestation of delirium.
Hypoactive delirium is frequently missed.
Unit-wide screening for delirium is feasible with valid and reliable bedside tools.
There are modifiable risk factors for delirium.
Detecting, treating, and preventing pediatric delirium may improve short- and long-term outcomes in critically ill children.
Delirium is the behavioral manifestation of acute global brain dysfunction. It is defined by the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V), as a disturbance of awareness and cognition, characterized by an acute onset and a fluctuating course. Importantly, delirium is not a psychiatric diagnosis but is rather a complication of an underlying general medical condition.
There are three principal subtypes of delirium. Hyperactive delirium—previously referred to in the intensive care unit (ICU) community as ICU psychosis —is characterized by agitation, restlessness, and emotional lability. Hypoactive delirium, sometimes called encephalopathy of critical illness , is notable for apathy and decreased responsiveness. Mixed delirium represents fluctuation between symptoms of both hyperactive and hypoactive delirium. Hyperactive delirium is the subtype most readily recognized. Conversely, unless there is a high index of suspicion, hypoactive delirium is often missed. ,
In January 2013, the Society of Critical Care Medicine (SCCM) published “Clinical Practice Guidelines for the Management of Pain, Agitation, and Delirium in Adult Patients in the Intensive Care Unit.” The SCCM guidelines recommend routine monitoring for delirium in adult ICU patients, as “monitoring critically ill patients for delirium with valid and reliable delirium assessment tools enables clinicians to potentially detect and treat delirium sooner, and possibly improve outcomes.” This was the result of an explosion of delirium research, with thousands of articles published in peer-reviewed journals.
Delirium is endemic in adult ICUs. With an overall prevalence of greater than 30%, it is more common in the elderly as compared with younger patients, affecting up to 80% of adult patients on mechanical ventilation. Delirium in adults has been linked to increased mortality, increased ICU length of stay, increased hospital length of stay, long-term cognitive impairment, and post–intensive care syndrome. Median time to tracheal extubation is longer in delirious patients. In addition, delirium has been associated with various ICU morbidities, such as self-extubation and removal of catheters. The association with mortality is strong, with duration of delirium correlated with increased risk. Two well-designed cohort studies in adult ICUs reported that for each day spent delirious, there was an increase in mortality. , In addition, delirium has been associated with significantly increased healthcare costs.
Delirium in critically ill children has recently received considerable attention in the pediatric critical care literature. As in adults, it has been associated with significant morbidity. Importantly, ICU medical and nursing practices can affect delirium rates. Therefore, accurate and early recognition of delirium in critically ill patients is imperative.
The etiology of delirium is complex and the pathophysiology incompletely understood. An interruption of brain network connections occurs, with a failure of the integration and processing of sensory information and motor responses. Much research has demonstrated that regardless of the primary etiology, the final common pathway involves alteration in neurotransmission, with subsequent cognitive and behavioral changes in the affected individual. Neurotransmitter dysregulation includes deficient acetylcholine and melatonin; excess dopamine, glutamate, and norepinephrine; and variable changes in γ-aminobutyric acid (GABA), serotonin, and histamine. ,
The cholinergic system modulates attention and consciousness, sleep, and memory. , Animal models have shown decreased acetylcholine synthesis in animals subjected to hypoxia/hypoxemia and immobilization. Anticholinergic agents have long been implicated in the genesis of delirium, with the elderly at increased risk likely due to an age-related decrease in acetylcholine synthesis in the hippocampus and prefrontal cortex. , Notably, children under 5 years of age have decreased acetylcholine synthesis in these areas as well.
A literature review by Maldonado described a range of etiologic theories as to the processes underlying delirium. These mechanisms are interrelated and likely synergistic in critically ill patients. The neuroinflammatory hypothesis proposes that systemic inflammation from an underlying illness, with associated cytokine release, compromises the integrity of the blood-brain barrier. This results in CNS inflammation that leads to dysfunction of neurons and synapses. Numerous studies in adults have shown that delirious patients have elevation of proinflammatory cytokines as compared with nondelirious patients even after adjusting for multiple confounders.
The neuronal aging hypothesis proposes that the elderly brain is more vulnerable to experiencing delirium owing to a lack of physiologic reserve when the patient is seriously ill. In fact, dementia is the single strongest risk factor for delirium in the elderly. , In a damaged brain, microglia are primed to respond more vigorously to systemic inflammation. This may be relevant to the population of developmentally delayed children in the pediatric ICU (PICU), a group that has been shown to be at higher risk for developing delirium than developmentally typical children.
The oxidative stress hypothesis maintains that delirium represents cerebral insufficiency. This is a form of brain failure, due to either hypoxia or hypoperfusion, which generates reactive oxygen species that induce oxidative damage. Several studies have demonstrated a correlation between hypoxia and delirium in both adults and children. , ,
The neuroendocrine hypothesis suggests that physiologic stress leads to an elaboration of glucocorticoids that cause direct neuronal injury. The diurnal (or melatonin) dysregulation theory posits that disruption of sleep architecture can precipitate, and then perpetuate, delirium. ,
Clinically, delirium can be thought of as the result of three synergistic factors: (1) the underlying illness, (2) iatrogenic effects of the treatment for that illness, and (3) highly abnormal ICU environment ( Fig. 134.1 ). As an example, consider the child who has recently undergone surgical repair of a congenital cardiac anomaly using cardiopulmonary bypass. There may have been episodes of hypoxemia and hypotension in the perioperative period, which led to brain hypoperfusion. The postoperative state is characterized by inflammation and mechanical ventilation, both of which are associated with delirium. There has been exposure to anesthetic agents, and the patient is currently receiving opiates and sedatives, which have been strongly linked to delirium. The patient is bedridden in an ICU and may also be immobilized, sleep deprived, and exposed to lights and noises 24 hours a day. These circumstances act together to form the framework for the emergence of delirium in this child.
The prevalence of delirium in the PICU has been reported as ranging from approximately 10% to 66%. A multi-institutional point prevalence study, including 994 children in 25 different pediatric ICUs around the world, demonstrated an overall delirium rate of 25%. Delirium rates increased to 38% in children who had been in the ICU for more than 5 days. In children on invasive mechanical ventilation (IMV), rates were even higher, at 53%. This study included every child admitted to each of the participating PICUs on the study day. As such, it is likely a widely generalizable representation of the prevalence of delirium in pediatric critical care units.
Other studies have focused on specific patient subgroups. A study completed in 2010 (n = 68), during the validation of a pediatric delirium screening tool, demonstrated a rate of 12.3%, but this study included only children over 5 years of age, and only 6% were receiving mechanical ventilation. Thus, those children in the ICU most at risk for delirium may have been either excluded or underrepresented. A single-center study focused on hospitalized children with cancer (n = 319) and found a 19% delirium incidence in these noncritically ill children. Two studies focused specifically on children following surgery performed on cardiopulmonary bypass (n = 194, n = 99), and found delirium rates of 49% and 57%, respectively. , A longitudinal study in a PICU in Germany, focusing on postoperative delirium, showed a 66% delirium rate in 93 children. A single-center study (n = 300), focused specifically on infants and children between the ages of 6 months and 5 years, demonstrated a delirium rate of 47%.
It is important to note that, over time, delirium rates have decreased substantially in units that have implemented routine screening for delirium and have adopted delirium prevention practices. For example, in an academic PICU in a major urban center, delirium rates decreased from 21% to 17% over a 3-year period. , Similarly, in a single-center longitudinal study involving 1875 patients, a 39% reduction in the rate of delirium was demonstrated over a 22-month study period. This indicates that, with careful attention to this frequent complication of pediatric critical illness, we can meaningfully reduce the impact of delirium in critically ill children.
Similar to the approach taken by Inouye et al., it is useful to separate risk factors for pediatric delirium into predisposing (i.e., patient-related) and precipitating (i.e., treatment-related) categories. Precipitating factors can be further divided into modifiable and nonmodifiable features ( Table 134.1 ).
|Precipitating Risk Factors|
|Young age (<2 y) |
Cyanotic congenital heart disease
|Invasive mechanical ventilation||Depth of sedation |
Red blood cell transfusion
Pediatric intensive care unit environment
Eight separate pediatric studies have shown that younger children (age <5 years) are at greater risk for delirium, with the highest rates demonstrated in children younger than 2 years. , , , , , In the largest study (n = 1547), when delirium rates in older children were compared with those in children younger than 2 years, adjusted odds ratios (AORs) decreased incrementally with age. In children 2 to 5 years old, the OR was 0.6; in children older than 5 years, the OR was 0.4. As another example, a single-center study that included 99 children after surgery on cardiopulmonary bypass showed that delirium risk decreased with each additional month of age (OR 0.35; 95% confidence interval [CI], 0.2–0.6).
Children with baseline cognitive impairment are at higher risk for developing delirium in the setting of critical illness when compared with children with normal development. Three separate studies demonstrated more than tripling of delirium risk after controlling for multiple confounders (AOR 3.3, 3.4, and 3.5). , , As stated previously, this may be analogous to the elderly population with underlying dementia, in whom critical illness can be conceptualized as a “stress test” that unmasks those patients with the least cognitive reserve.
Other at-risk subgroups include children with cyanotic congenital heart disease (AOR 2.4; P = .019) and children with poor baseline nutritional status (more than doubling of delirium risk in three separate studies). ,
A major precipitating factor for delirium is IMV, which has been shown to be an independent risk factor for delirium irrespective of sedation practices or severity of illness. A prospective study including 1547 children demonstrated an OR of 1.6 ( P = .006) in children who required IMV after controlling for age, cognitive impairment, probability of mortality, and depth of sedation. In an international point prevalence study, the effect of IMV was even stronger, with an adjusted OR of 1.7 (95% CI, 1.1–2.7) after controlling for multiple confounders.
In contrast to IMV, which is most frequently not a modifiable risk factor, there are other treatment-related risk factors that are (at least potentially) modifiable. An important example is use of restraints. A multisite study of 994 subjects revealed a quadrupling of delirium risk (AOR 4.0; 95% CI, 2.0–7.7) after controlling for age, sedation, and IMV. Although causality cannot be attributed based on the study design, it is likely that restraint use proceeded delirium development.
Similar to adults, depth of sedation is an extremely important modifiable risk factor for pediatric delirium. A burgeoning pediatric literature demonstrates that an analgosedation approach—in which pain control is optimized and sedation minimized—is not only feasible in critically ill children but also improves outcomes. In addition to lowering the incidence of delirium, studies have also demonstrated decreased duration of mechanical ventilation and shorter PICU and hospital length of stay. Importantly, there has been no statistical increase in serious adverse events (including unplanned extubations) that is attributed to this minimalist approach to sedation. ,
When sedation is absolutely required, the pediatric literature favors avoidance of benzodiazepines in favor of alternative sedative classes (including the α-agonists). There are now seven separate pediatric studies that show an independent increase in delirium risk associated with benzodiazepine treatment, with AORs ranging from 2.2 (n = 994) to 5.2 (n = 1547). , , , , Two studies also demonstrate a dose-response effect, with an incremental increase in delirium risk based on benzodiazepine exposure. , One study noted that for every 1-log increase in benzodiazepines received, there was a 43% increase in subsequent delirium.
Rather than merely describing associations, a nuanced study by Mody et al. used sophisticated statistical modeling to estimate the causal effect of benzodiazepines. Marginal structural modeling creates a randomized sample from observational data to minimize both static and time-varying confounders. With this approach, benzodiazepines were shown to more than triple the risk for delirium development (OR 3.3; 95% CI, 1.4–7.8).
Finally, there has been one high-quality randomized double-blinded placebo-controlled trial comparing midazolam to dexmedetomidine in 42 critically ill adolescents after scoliosis surgery. Not only was midazolam use associated with increased delirium rates (31% vs. 12.5%; P < .05), total fentanyl use was significantly higher ( P = .002) and duration of mechanical ventilation was greater ( P < .05) in this group as well.
To summarize, the preponderance of evidence supports a causal inference between benzodiazepine use and delirium. With alternative sedative choices available, this offers an important opportunity for modifying our traditional prescribing practices in order to decrease delirium risk in the PICU.
Another strong and modifiable risk factor for delirium is red blood cell (RBC) transfusion practices. A single-center study demonstrated an independent association between receipt of RBCs (AOR 2.2; P = .001) and delirium development. This was not a function of anemia, as nadir hemoglobin prior to transfusion was not associated with delirium. A biological gradient was noted, with odds of developing delirium rising with RBC dosage (ORs 1.4, 1.9, and 2.6 for 5 mL/kg, 10 mL/kg, and 15 mL/kg, respectively). Adherence to strict transfusion criteria and transfusing as small a volume of RBCs as necessary may decrease delirium risk.
At this time, there is insufficient evidence to determine whether opiates, anticholinergics, and steroids independently increase delirium risk. Further research is needed to clarify these potential associations.
Pediatric delirium has been linked to delayed extubation, increased hospital and PICU length of stay, and increased hospital costs. Most significantly, delirium has been related to excess mortality in PICU patients.
Delirium is associated with increased duration of mechanical ventilation. In a cohort of 93 postoperative patients, early delirium (within the first 5 postoperative days) predicted delayed time to extubation. Findings were similar in a separate cohort of 99 children after cardiac bypass surgery. A longitudinal cohort of 1547 children admitted to a general PICU showed that the duration of mechanical ventilation was longer (median 4 vs. 1 day; P < .001) in children with delirium.
At least five studies (n = 93, n = 99, n = 194, n = 300, n = 1547) have demonstrated an independent association between delirium and PICU length of stay. , , In the largest study, the adjusted relative length of stay for children with delirium was 2.3 ( P < .001) after controlling for IMV and severity of illness on admission.
A prospective observational study investigating the effect of delirium on hospital costs included 464 consecutive PICU admissions. Children were screened for delirium daily. Hospital costs were analyzed (using cost-to-charge ratios in 2014 dollars). Results showed that median PICU costs increased 392% in children with delirium ( P < .001), with an incremental increase in costs for each day with delirium. After controlling for the usual drivers of ICU costs, delirium was independently associated with an 85% increased expense.
Finally, delirium has been linked to excess mortality in children. In fact, in a 1547 patient cohort, delirium was a stronger predictor of mortality (AOR 4.4) than the widely accepted Pediatric Index of Mortality (PIM3; AOR 3.2). Although this is a strong and independent association, it is important to recognize that this does not establish a causal relationship. Rather, it is likely that delirium is a sensitive marker for patients at high risk for poor outcome. Early recognition of delirium will provide practitioners with the opportunity to focus their efforts on these fragile children.
Further research is needed to explore the longer-term effects of delirium on PICU survivors. In adults, delirium has been linked to poor long-term outcomes in regard to cognitive and psychologic health. Studies in children exploring the lingering effects of delirium after PICU discharge are underway.
Pediatric delirium is complex and multifactorial, further complicated by developmental variability. Normal behavior in a 2-year-old would be considered highly abnormal behavior in a 12-year-old. Thus, the developmental context is integral to the diagnosis of delirium in children. In fact, in early pediatric delirium research, questions were raised regarding the reliability of the DSM criteria in the diagnosis of delirium in young children, particularly preverbal children. To address this issue, Silver and colleagues investigated the reliability of the criterion standard (delirium diagnosis by psychiatric examination using the DSM-IV criteria) by performing 38 paired psychiatric evaluations of children at risk for delirium by two different psychiatrists. Of the children in this cohort, 50% were younger than 2 years. Interrater reliability of the paired psychiatric assessments was excellent, with a κ of 0.95, demonstrating the reliability of the standard criteria for diagnosis of delirium in children.
The diagnosis of delirium may be challenging in the developmentally delayed child. As such, this at-risk subgroup is often omitted from research. This is unfortunate and unnecessary, as delirium can be distinguished from static encephalopathy. With attention to the child’s prehospital baseline and focus on fluctuation (a hallmark of delirium) in current symptoms, delirium can be reliably diagnosed even in this difficult-to-assess population. , ,
Children with delirium can present with a range of alterations in psychomotor activity, including fidgetiness and self-stimulating behavior or sluggish responses to interactions. Children and adolescents may have disordered emotional states, with extreme tearfulness, inconsolability, or inappropriate calmness relative to their circumstances. In addition, pediatric delirium is often marked by profound sleep disturbance that, when present, further worsens the delirium.
Delirium in children is most often of the hypoactive or mixed subtypes. A large prospective study of 1547 children showed that only 8% of children had hyperactive delirium; 46% had hypoactive delirium and 45% had mixed delirium. A study of 99 children in a cardiothoracic ICU had a similar phenotype distribution: 5% hyperactive, 52% hypoactive, and 43% mixed. A study focusing on younger patients (n = 300, ages 6 months–5 years) also showed that only 7% of delirious children presented with the hyperactive subtype.
Although hyperactive delirium represents only a small portion of pediatric delirium, it is the most readily recognized, as the child is actively interfering with the medical team’s interventions. In the past, these children frequently received large doses of sedatives, which further contributed to the evolving delirium. However, it is the docile child, often termed the “good patient,” who lies in bed without protest and allows the care team to perform multiple uncomfortable interventions, whose condition often goes unnoticed. This is markedly abnormal behavior for a preschool- and school-age child and likely represents hypoactive delirium. As discussed previously, hypoactive delirium is often missed without routine screening and has been linked to the worst prognosis in several studies in adult patients. ,
Pediatric delirium (similar to adult delirium) is often short-lasting, with a median duration of 2 days (interquartile range [IQR] 1–5). This short-lasting delirium cannot be dismissed, as even 1 day with delirium has measurable negative effects on outcomes. , ,
Epidemiologic studies have shown that pediatric delirium often occurs early in the PICU stay, highlighting the importance of screening every patient every day, even on the day of admission. , , Recurrent delirium is also common, with 27% of children with delirium having repeat discrete episodes within a PICU stay. With timely recognition and intervention, the duration of delirium can be decreased.
Until recently, the only way to diagnose delirium in children was via a psychiatric interview and examination using the DSM criteria. This is time-consuming, expensive, and relies on a limited resource (pediatric psychiatrists). This explains why, historically, only children with hyperactive delirium were recognized and received appropriate attention. The advent of bedside delirium screening tools presents an opportunity to assess all critically ill children for delirium. , , Two validated tools are currently available for use in the PICU.
The Pediatric Confusion Assessment Method for the ICU (pCAM-ICU) is a tool derived from the well-validated and widely used CAM-ICU. It uses yes/no questions, hand signals, and pictures and has been designed for use in children older than 5 years who do not have significant developmental delay. There is a preschool version available, the Preschool Confusion Assessment Method for the ICU (psCAM-ICU), for children 6 months to 5 years of age. The pCAM-ICU and psCAM-ICU are point-in-time interactive and cognitively oriented assessments. Delirium is scored as present or absent.
The Cornell Assessment of Pediatric Delirium (CAPD) is a strictly observational tool, with eight questions that are scored by the bedside nurse on a Likert scale. The elements of the tool are consistent with the diagnostic criteria for delirium. It is designed to be applicable to children of all ages and cognitive levels. A Developmental Anchor Points chart is available for the nurse to use as a point-of-care reference when administering the tool in the youngest preverbal patients. The screen is scored on a numeric scale, with a score of 9 or higher consistent with a diagnosis of delirium in a developmentally typical child. The individual’s score can be trended over time to determine response to interventions. The European Society of Pediatric and Neonatal Intensive Care (ESPNIC) has recommended routine delirium screening of all critically ill children with the CAPD twice daily throughout their PICU stay ( Fig. 134.2 ).