Emergence agitation (EA) involves restlessness, disorientation, excitation, non-purposeful movement, inconsolability, thrashing, and incoherence during early recovery from general anesthesia [1]. The incidence of EA varies, from approximately 0.25% to 90.5%, with age, assessment tool used, definitions, anesthetic techniques, type of surgery, and time of EA assessment during recovery [2–6]. The clinical consequences of EA are similarly varied. It is typically short lived and resolves spontaneously, and its clinical consequences are often considered minimal [7,8]. However, it may have clinically significant consequences, such as injury to the affected patient or their medical staff, falling out of bed, bleeding at the surgical site, accidental removal of drains or intravenous catheters, unintended extubation, respiratory depression, and increasing medical care costs [9–11].

Emergence delirium (ED) is an acute confusion state during recovery from anesthesia; patients with ED may present with disorientation, hallucination, restlessness, and purposeless hyperactive physical behavior [8,12]. ED is not fully equivalent to EA; ED can involve hypoactive signs or mixed forms and hyperactive signs similar to agitation [13–15]. Nevertheless, the terms EA and ED have been used interchangeably in several studies [16,17]. Moreover, the same assessment tools (e.g., Riker Sedation-Agitation Scale or Richmond Agitation-Sedation Scale) have been used for both conditions [18–21]. EA and ED should be differentiated from postoperative delirium. Postoperative delirium involves ED; ED represents the early onset of postoperative delirium in the operating room or on arrival at the postanesthesia care unit (PACU) immediately after the anesthesia period [18,21,22]. EA and ED in the PACU are strong predictors of postoperative delirium, which is associated with prolonged hospital stay and increased morbidity (e.g., pulmonary complications), mortality, and the need for institutionalization of adult patients [2,23]. The terms EA and ED are used interchangeably in this review, as in previous studies [16,17,24,25].

This review discusses the important themes of EA research, issues that remain unresolved, and future research directions.

The precise pathophysiological mechanism of EA after general anesthesia is unknown [19,20]. In children, proposed causes of EA include high levels of anxiety regarding surgery, new environments, separation from parents, and encounters with unfamiliar medical staff [9,26]. These may lead to increased sympathetic tone and prolongation of the excited state during anesthesia recovery [27].

The advent of volatile agents with low blood solubility, such as sevoflurane and desflurane, has increased the incidence of EA in children [11,12,28]. A proposed explanation for this is that sevoflurane and desflurane cause differential recovery rates in brain function, due to differences in clearance of inhalational anesthetics from the central nervous system [12,29]; whereas audition and locomotion recover first, cognitive function recovers later, resulting in EA. In addition, elevated lactate and glucose concentrations in the parietal cortex due to sevoflurane anesthesia, and the occurrence of clinically silent sevoflurane-induced epileptogenic activity have been proposed to induce EA [16,30,31].

Functional magnetic resonance imaging has been used to study the mechanisms underlying the alteration of consciousness during anesthesia [32,33]. Studies have reported that alterations of brain network connectivity vary with the level of sedation. During emergence from general anesthesia, thalamocortical connectivity in sensory networks, and activated midbrain reticular formation are preserved. However, delayed recovery of impaired functionality of subcortical thalamoregulatory systems could contribute to defects in cortical integration of information, which could lead to confusion or an agitated state [33].

The etiology of EA is multifactorial [3]. It is important to identify the causes and risk factors of EA, and modify them, when applicable, to reduce incidence and prevent adverse consequences. Results from previous studies have been inconsistent due to the application of different assessment tools, definitions, and study designs (e.g., prospective randomized controlled studies, prospective observational studies, or retrospective studies). In addition, proposed risk factors of EA have been different for children and adults. Potential risk factors for EA in children are as follows: preschool age (2–5 years), no previous surgery, hospitalization or high number of previous interventions, poor adaptability, attention-deficit hyperactivity disorder, patient pre-existing behavior, psychological immaturity, preoperative anxiety, parental anxiety, patient and parent interaction with healthcare providers, lack of premedication (with midazolam), paradoxical reaction to midazolam stated in child’s medical history, type of surgery, use of inhalational anesthetics with low blood–gas partition coefficients (e.g., sevoflurane and desflurane), excessively rapid awakening (in a hostile environment), and pain [6,12,17,29,34–36].

The proposed risk factors for EA in adults are age, sex, obesity (body mass index ≥ 30 kg/m²), African ethnicity, number of intubation attempts, type of surgery, emergency operation, method of anesthesia (inhalation anesthesia), duration of surgery or anesthesia, pre-existing mental health problems (e.g., psychiatric problems or cognitive impairment), chronic lung disease, recent smoking, history of social drinking, substance misuse, anticholinergics, doxapram, premedication with benzodiazepines, voiding urgency, postoperative pain, postoperative nausea and vomiting, and the presence of invasive devices (e.g., urine catheter, chest tube, or tracheal tube) [2,3,8,18,21,37–44] (Table 1). We review common risk factors and other related issues presented in literature.

Prevention is preferred over treatment, for EA; EA can have serious consequences for patients, and increase the patient care burden [16,71]. Recently, Hino et al. [54] developed and validated the EA risk scale (consisting of four domains—age, Pediatric Anesthesia Behavior score, operative procedure, and anesthesia time) for children receiving sevoflurane anesthesia in a single-center study. The EA risk scale showed excellent predictive performance. Therefore, the EA risk scale may be used to predict and prevent EA after sevoflurane anesthesia in pediatric patients. However, the EA risk scale is not validated for use in patients anesthetized with drugs other than sevoflurane. Further studies are needed to demonstrate external validity in other hospitals. In addition, for effective prevention of EA, it would be helpful to identify a biomarker that could predict the occurrence of EA, based on preoperative blood sample examination. In elderly patients undergoing gastrointestinal surgery, the plasma level of brain-derived neurotrophic factor (BDNF) collected at skin closure via blood sampling was significantly increased in patients with EA [72]. However, the study showed that the level of plasma BDNF collected before induction of anesthesia did not differ between patients who did and did not show EA. The study included only a limited number of well-selected patients. Thus, larger-scale clinical trials are needed to ensure the validity of BDNF as a predictive biomarker for EA. In addition, if the occurrence of postoperative EA can be predicted through objective monitoring during surgery, it may contribute to improved postoperative outcomes by preventing the occurrence of EA. In a prospective observational study published in 2019, the occurrence of specific electroencephalogram patterns (burst suppression and emergence trajectory) during anesthesia was associated with PACU delirium [73]. The authors could not provide information regarding agitation during emergence because all patients underwent assessment for PACU delirium after the return of consciousness. However, they suggested that EA could be predicted through intraoperative patient monitoring.

Although several scales and their variants have been proposed as tools for assessing EA in children, the most commonly used in pediatric EA studies is the Pediatric Anesthesia Emergence Delirium (PAED) scale developed in 2004 (Table 2). It provides a score from 0 to 20 and reportedly shows validity for assessment of EA in children [74]. However, the PAED scale has disadvantages of inherent subjectivity in assessing each behavior item and suboptimal interrater reliability [75]. In addition, the cutoff point for defining the presence of EA is controversial. Bong and Ng [76] suggested that PAED score ≥ 10 was the ideal cutoff for EA. In contrast, Bajwa et al. [69] reported that PAED score > 12 had greater sensitivity and specificity than PAED score ≥ 10 in the assessment of EA PAED. In another study, PAED score ≥ 16 was adopted as an indicator of EA without an obvious rationale [68].

In adults, the Riker Sedation-Agitation Scale (RSAS, 7-point scale with three levels of agitation) [71], Richmond Agitation-Sedation Scale (RASS, 10-point scale with four levels of agitation) [77], Aono’s 4-point scale [78], Nurses Delirium scale [79], and the 3-point scale (graded as mild, moderate, or severe) [3] have been introduced for assessment of EA (Table 2). Although the RSAS and RASS have been commonly used, and show high interrater reliability in adult intensive care unit patients [80,81], none of the scales have been validated in the operating room and/or the PACU. There have been few studies of EA in intensive care unit patients [37]; the majority of EA studies have been performed in PACUs or operating rooms [19,20,82]. Consequently, the reported incidence of EA differed with the evaluation site (e.g., operating room vs. PACU), assessment tool (e.g., RSAS vs. RASS), and definition of EA (e.g., RASS ≥ +1 vs. ≥ +2 vs. ≥ +3). The reported incidence of EA was higher in the operating room when emerging from general anesthesia than in the PACU (e.g., 3.7% vs. 1.3% and 54.3% vs. 28.6%, respectively) [8,38,83]. The RSAS tended to show an incidence of EA that was similar to or higher than the incidence indicated by the RASS for the same patient group (13.8% by the RSAS vs. 11.2% by the RASS, respectively) [83] or same type of surgery (50% by the RSAS vs. 22% by the RASS, respectively) [39,82]. Fields et al. [2] used RASS ≥ +3 as an indicator of EA, while Jee et al. [20] and Ham et al. [84] adopted RASS ≥ +2 as an indicator of EA; most other groups defined RASS ≥ +1 as an indicator of EA [38,39,83].

Standardized clinical research practice guidelines are needed to reduce the variations of EA incidence with assessment tools and definitions among researchers. Furthermore, there is a need to develop a tool that can objectively evaluate the degree of agitation, in a manner similar to the bispectral index, which is an objective tool for assessing the depth of sedation during general anesthesia.

In this section, we review strategies to prevent EA, classified into pharmacological and non-pharmacological methods (Table 3). Caution is needed when interpreting the results of studies comparing the preventive effects of drugs or agents on EA; the same drugs may not have identical effects depending on the dose, method of administration (e.g., continuous infusion or single bolus), timing of administration, or patients (e.g., children, adults, or elderly patients) [20,85].

EA is a self-limiting phenomenon, which lasts for only a short period (1–15 minutes) [5,11]. The elimination of causative factors (e.g., pain, anxiety, presence of invasive devices) is the mainstay of EA management [3,39]. Differential diagnosis and prompt treatment should also be performed for conditions that can lead to disorientation, such as increased intracranial pressure, bladder distention, upper airway obstruction, hypo- and hyperglycemia, hypotension, hypoxia, and hypercarbia [11]. Two web-based surveys conducted by pediatric anesthesiologists in Canada and Germany [36,91] revealed that sedatives (e.g., propofol and midazolam) and opioids (e.g., fentanyl and morphine) were preferred therapeutic pharmacological treatments for EA. Rarely, some anesthesiologists chose “wait for spontaneous resolution” and/or “parental presence” as the first choice of therapy for EA [36]. To the best of our knowledge, there has only been one randomized controlled trial to evaluate the effects of therapeutic strategies on EA, which compared physostigmine and placebo as its treatments [135]. Further studies are needed to determine the efficacy of pharmacological or non-pharmaceutical interventions (e.g., parental presence) for treatment of established EA.

There have been few EA-related studies in adults compared to studies performed in children. Since the results of the studies in children cannot be extrapolated to adults, it is necessary to verify the consistency of the results for children by performing randomized controlled trials in adults. The establishment of standardized clinical research practice guidelines and the development of objective assessment tools for EA are needed to reduce the discrepancies of EA incidence among various studies, and facilitate better interpretation of the results from published studies. Effective EA prevention involves the identification of risk factors, elimination of correctable risk factors, and the application of pharmacological and non-pharmacological strategies on patients or during surgeries with high risks of EA. Strategies to reduce EA may vary with patient age (i.e., children vs. adults). Considerations for the dosage, timing, and method (e.g., bolus or continuous infusion) of administration of the agent should be made before applying pharmacological methods. In the future, prediction through objective indicators (before or during surgery) is expected to aid in preventing EA.