The perioperative period is often an extremely traumatic time for the young child undergoing surgery. Three clinical phenomena have been described in children undergoing surgery: preoperative anxiety, emergence delirium, and new-onset postoperative maladaptive behavioral changes.
Up to 65% of all children undergoing anesthesia and surgery develop intense anxiety and fear in the preoperative holding area and during induction of anesthesia (1). This anxiety can be attributed to separation from parents and uncertainty about the anesthesia, surgery, and outcome of the procedure (2). Postoperative emergence delirium has been described to occur in 12%–18% of all children undergoing anesthesia and surgery (3,4). Factors such as age, anesthetic (e.g., desflurane), and postoperative pain have been suggested as possible etiologies (3,5). Of great importance also are new-onset maladaptive behaviors exhibited by the child after surgery. New maladaptive behaviors such as general anxiety, nighttime crying, enuresis, separation anxiety, and temper tantrums have been described to occur up to 50% of children undergoing surgery (1).
Previous studies have suggested that the three clinical phenomena described above may indeed be related. That is, Kain et al. (1,6) reported that increased levels of preoperative anxiety are associated with an increased incidence of postoperative behavioral changes and Aono et al. (7) found that elevated levels of preoperative anxiety are associated with increased incidence of emergence delirium. No studies, however, have examined the concept that preoperative anxiety, emergence delirium, and postoperative maladaptive behavioral changes are closely related clinical phenomena. Thus, we decided to examine this issue using data that were obtained by our laboratory over the past 6 yr.
Over the past 6 years, our laboratory has conducted a series of prospective studies related to preoperative anxiety, emergence status, and postoperative behavioral changes. The subjects involved were children with a physical class of ASA I–II and who were undergoing surgery with general anesthesia. The IRB approved all studies, and assents (if appropriate) and informed consents were obtained from participants. None of the patients in these studies had a history of chronic illness, prematurity, psychiatric illness or developmental delay.
Data from all these studies were combined in a central database. For the present investigation we included in our analysis only children who fulfilled the following criteria: outpatient surgery, no midazolam preoperatively, and a O2/N2O/sevoflurane anesthetic induction. Based on these criteria, data from eight present and previous studies conducted by our laboratory are included in this present report. It should be noted that although some patients who were included in the present analysis received parental presence during induction of anesthesia, this intervention has been shown to have no relationship to preoperative anxiety (8), emergence status (9), or the onset of postoperative behavioral changes (10). Nonetheless, we examined the data both ways: including all patients and including only patients who did not receive parental presence.
Data obtained using the following measures were included in the present investigation. Trained psychologists who are part of our laboratory administered all psychological instruments to children and parents. It is important to note that the same investigators performed all data collection using the same tools.
The State-Trait Anxiety Inventory (STAI) is a 40-item parental self-report measuring state anxiety and trait anxiety. Total scores for state and trait portions separately range from 20 to 80; higher scores denote higher levels of anxiety (11).
The EASI Scale of child temperament (EASI) assesses a child’s baseline temperament and includes 20 items in 4 subcategories: emotionality, activity, sociability, and impulsivity. Completed by the parents, this instrument has good reliability and validity (12).
The modified Yale Preoperative Anxiety Scale (mYPAS) is an observation measure of anxiety previously developed by our laboratory. The instrument consists of 5 domains and 27 behaviors and is widely used in the perioperative literature. This measure has good reliability and validity when compared with both the C-STAI and with cortisol levels (13,14).
Emergence Status Measure.
Emergence delirium status was determined using an observer measure that behaviorally assesses symptoms of emergence delirium, yielding a score of 1, 2, or 3. A score of 1 denotes no symptoms of emergence delirium, 2 represents mild symptoms (e.g., occasional movement or crying, no need for restraint), and a score of 3 represents marked symptoms of emergence delirium (e.g., thrashing and/or needs restraint and/or constant crying) (15).
The Post Hospital Behavior Questionnaire (PHBQ) is designed to evaluate maladaptive behavioral responses and negative behavioral changes in children after surgery (16). The PHBQ is considered to be the standard procedure for postoperative behavioral assessment. The PHBQ consists of 27 items and 6 categories of anxiety: General Anxiety, Separation Anxiety, Sleep Anxiety, Eating Disturbances, Aggression Against Authority, and Apathy/Withdrawal. This instrument has acceptable test-retest reliability, and good agreement with psychiatric interviews with parents (16).
Although procedures for each of the eight studies included in this investigation differed in some ways, they shared the following basic protocol.
Preoperative Waiting Area.
On the day of surgery, state anxiety of the child (mYPAS) and trait and state anxiety of the parent (STAI) were evaluated.
Child’s state anxiety (mYPAS) was again evaluated on entering the operating room (OR) and on introduction of the anesthesia mask. Parental anxiety (STAI) was assessed immediately after separation from their child. Children were brought into the OR and placed on the table. Next, a Spo2 probe was placed on the child’s hand and a scented anesthesia mask was introduced. Anesthesia was induced using a controlled O2/N2O/sevoflurane technique. Once anesthesia was induced, an IV cannula was inserted and 0.1 mg/kg IV vecuronium was administered to facilitate tracheal intubation. Anesthesia was maintained with O2/N2O and isoflurane, and IV fentanyl (1–5 μg/kg) was administered. Subjects undergoing pressure equalization tube placement underwent an exclusive mask anesthetic with O2/N2O/sevoflurane technique. Drugs such as ketamine, atropine, and droperidol were not allowed as part of the anesthetic regimen.
Postanesthesia Care Unit (PACU).
Emergence status was evaluated during the child’s stay in the PACU.
Postoperative Days 1, 2, 3 and Weeks 1 and 2.
The PHBQ was administered via telephone to parents on each of postoperative days (POD) 1, 2 and 3 and at postoperative weeks 1 and 2. This extensive follow-up method maximized retention and minimized recall bias.
Statistics and Data Analysis
Measurements of the child’s anxiety, emergence delirium, and postoperative behaviors were made as outlined in Figure 1. To facilitate the regression analysis, we combined each measure of child’s state anxiety into one number that reflects the value of each mYPAS score yet is also inclusive of changes in anxiety over time. The PHBQ was administered in 4 of the 8 studies included in this investigation (n = 333). Items were scored as positive if the parent reported that the child exhibited this behavior on the day of the report and “more than before the surgery.” This score was used in the repeated measures analyses and to compare the number of maladaptive behaviors exhibited by children with high or low preoperative anxiety. A categorical score denoting the presence or absence of any maladaptive behavior changes was used in the regression analysis.
Regression analysis was used to assess whether or not preoperative anxiety and emergence status would predict the presence or absence of postoperative maladaptive behavioral changes. Patients with emergence status of 1 or 2 did not significantly differ from each other in terms of mYPAS scores or PHBQ scores, and therefore were combined into a single referent group for the regression analysis.
We next used two mixed effects models with a random effect being the original study assignment. The first model used emergence status (marked versus nonmarked) as the dependent variable with anxiety as the predictors. In the second model, the PHBQ category is used as the dependent variable and emergence status and anxiety were the predictors. The model is as follows:
where Yijk is the kth categorized PHBQ score for the jth child in the ith study, Xijl and Xij2 are dummy variables and equal 1 if the emergence status for the jth child in the ith study belongs to the first or second nonreferent group, respectively; Xij3 and Xij4 are the starting point and the mYPAS change score respectively for the the jth child in the ith study; bli is a random effect for the ith study, and εijk represents measurement error.
Descriptive statistics provide an overview of the characteristics of both children and parent variables. Data are presented as mean ± sd and were appropriate for linear analysis. Differences between groups were examined using inferential statistics, including Student’s t-tests and analysis of variance.
A total of 1613 patients were included in our database. We eliminated all who did not have data collected regarding emergence delirium, leaving 1279 patients. Next, we eliminated all subjects who had received an intervention such as a sedative premedication (e.g., midazolam) or psychological preoperative preparation, resulting in 791 subjects who are included in the present report. It should be noted that partial data regarding some of these subjects were included in previous publications (17,18) by our laboratory (n = 80 and n = 61, respectively). Baseline psychological and demographic characteristics of the children and parents in this investigation are shown in Table 1.
Children in this report showed increased anxiety on induction of anesthesia as compared with preoperative holding [F(2,1576) = 185.5, P = 0.000]. Repeated-measures analysis of variance showed that emergence status was significantly associated with changes in preoperative anxiety over time [F(4,1572) = 10.75, P = 0.000] (Figure 2). That is, children with more intense preoperative anxiety were more likely to show symptoms of emergence delirium. Post hoc tests showed that children in each emergence classification category significantly differed from the children in each other emergence classification category at all three preoperative anxiety measurement points (i.e., in the holding area, on entrance to the OR, and on introduction of the anesthesia mask) (Table 2).
Children with intense preoperative anxiety in the holding area (upper 25%) were compared with children with low anxiety in the holding area (lower 25%) on each of the 5 selected PODs (POD1, POD2, POD3, POD7, POD14). Repeated-measures analysis showed that children with intense preoperative anxiety showed significantly more maladaptive behavior changes after surgery than did children with less preoperative anxiety [F (4,91) = 7.21, P = 0.0001]. That is, although the number of maladaptive behaviors exhibited by children with both high and low preoperative anxiety decreased from POD1 to POD14 [F (4,91) = 22.8, P = 0.00001], children with intense preoperative anxiety consistently exhibited significantly more numbers of maladaptive until POD14 (Fig. 3).
Based on these results, we posited a relationship linking preoperative anxiety, emergence delirium, and postoperative maladaptive behavior changes. To test this relationship, we used mixed effects regression models (see Methods). This analysis showed that children’s preoperative anxiety in the holding area (t = −3.93, P = 0.0001), children’s overall preoperative anxiety as captured in the m-YPAS change score (t = 4.69, P = 0.0001), and emergence status (t = −2.57, P = 0.01) were all significant predictors of the presence or absence of maladaptive behavior changes on the PHBQ.
Mixed-effect model analysis showed that children’s state anxiety was significantly associated with postoperative emergence status. Specifically, the odds of having marked symptoms of emergence delirium were increased by approximately 10% as a result of an increment of 10 points in the mYPAS anxiety score (95% CI = 1.0017– 1.0171, P = 0.0168). Regarding emergence status, we found that the odds ratio of having a new-onset maladaptive behavior change was 1.43 (95% CI = 1.09–1.88; P = 0.0104) for children with marked emergence status as compared with children with no symptoms of emergence delirium. Also, a 10-point increase in mYPAS anxiety scores led to a 12.5% increase in the odds that the child would have a new-onset maladaptive behavioral change after the surgery (95% CI = 1.0049–1.0168; P = 0.0001).
This report included children who did as well as children who did not receive parental presence; χ2 analysis showed that this variable was not associated with emergence status or preoperative anxiety (P = 0.3 and P = 0.67, respectively). We also analyzed the subset of children who did not receive parental presence during induction of anesthesia (PPIA) (n = 342) and found although the smaller number of subjects resulted in a decreased power, the relationship nonetheless remained similar, showing that in patients who did not receive PPIA, children’s preoperative anxiety and emergence status each remained significant predictors (P = 0.006, P = 0.012 respectively) for the presence or absence of maladaptive behavior changes on the PHBQ.
We next selected two groups of participants: a high-risk group (n = 57) that consisted of all children with marked emergence symptoms (score of 3) and either significantly intense overall preoperative anxiety (1 sd above the mean) or significantly large numbers of maladaptive behaviors postsurgery (1 sd above the mean), or both. The Low-Risk group (n = 104) consisted of all children with no emergence symptoms and significantly low overall preoperative anxiety (1 sd below the mean), and significantly few maladaptive behaviors (1 sd below the mean). As can be seen from Table 3, high-risk children were younger, more emotional, more impulsive, and less social. When parents of these 2 groups were compared, we found that parents of high-risk children were significantly more anxious in the holding area and on separation to the OR (Table 3).
We found that the odds of having marked symptoms of emergence delirium increased by approximately 10% as a result of an increment of 10 points in children’s state anxiety score (mYPAS). We also found that the odds ratio of having one or more new-onset postoperative maladaptive behavior changes is 1.43 for children with marked emergence status as compared with children with no symptoms of emergence delirium. In addition, a 10-point increase in state anxiety scores led to a 12.5% increase in the odds that the child would have one or more new-onset maladaptive behavioral changes after the surgery. We identified characteristics of children who are at high risk of developing all of these clinical phenomena: they are younger, more emotional, more impulsive, and less social. In addition, the parents of these children are significantly more anxious in the holding area and more anxious on separation to the OR.
Based on these results, however, we cannot suggest a cause-effect relationship. That is, although it is possible that extreme anxiety during the preoperative period results in more emergence delirium symptoms and more postoperative behavioral changes, the design of this present study does not support such a hypothesis. A cause-effect relationship could only be supported through a randomized controlled study that would assure yoking children with identical temperaments to 2 experimental groups wherein one group would be exposed to extreme upset during induction (“brutane induction”) and the other group would be exposed to a peaceful induction. Obviously, such a study is neither feasible nor ethical and would never be conducted. The alternative hypothesis is one of association, as investigated in the present study. That is, in the general population there is a group of children who are more at a risk to develop preoperative anxiety, emergence delirium, and behavioral changes.
It is likely that the relationship between preoperative anxiety, emergence and postoperative new-onset maladaptive behavioral changes is one that is reflective of underlying temperament constructs that relate to stress and adaptability. In other words, there is some substrate within each child that will result, more or less, in a fearful reaction to outside stimuli, depending on the child, the environment, and the interaction between the two. One construct that has gained some prominence in the field of developmental psychology is that of reactivity, which describes the “excitability, responsivity, or arousability” of the child (19). Perhaps this reactivity is the underlying causal substrate from which each of this study’s three indicators (preoperative anxiety, emergence, and postoperative new-onset maladaptive behavioral changes) arise.
It should be emphasized that we previously documented that reducing anxiety during induction of anesthesia reduced the incidence of postoperative behavioral changes but not emergence delirium (20). However, drawing conclusions from this previous study about the relationship between anxiety, emergence, and postoperative behavioral changes is confounded because we used midazolam in that study to reduce preoperative anxiety. Clearly, as the etiology of emergence delirium is likely to be a combination of variables including the child’s underlying temperament, preoperative anxiety, various anesthetic and other drugs, and other conditions such as pain. Thus, this conceptual framework is complex.
One of our important findings is that parents of the children at high risk of preoperative anxiety, emergence delirium, and maladaptive postoperative behavioral changes are more anxious in the holding area. This finding underscores the importance of identifying ways to mitigate this parental anxiety, such as developing preoperative preparation programs directed at parents.
Several methodological issues related to this study have to be addressed. First, data used for this study were obtained from subjects that were part of various continuing and past studies in surgical centers. As the same investigators performed all data collection using the same tools prospectively, we do not believe that the quality of the data is thereby hindered. Further, on data extraction we ensured that only patients who received a mask induction with sevoflurane and who received no midazolam, atropine, ketamine, or droperidol, were included in this study.
This was done to assure that these potential confounding variables were controlled for. Also, although it is true that other variables such as postoperative pain were not controlled for, we do not think that this had an impact on the study results because of the very large sample (and power) of this study. Also, because pain may play a role in the etiology of emergence delirium, we examined the pain scores of a subgroup of our subjects on presentation to the PACU. We found a very high correlation between pain scores (as assessed by the Children’s Hospital of Eastern Ontario Pain Score) and the emergence score (r = 0.78). This high correlation underscores the difficulties in differentiating between pain and emergence delirium. Because of the high correlation, our statistical experts strongly advised us against including pain in the regression models used.
In conclusion, this is the first study that indicates the odds of having marked symptoms of emergence delirium and development of negative postoperative behavioral changes based on measurement of children’s state anxiety before surgery, as well as the odds ratio of having a new-onset postoperative maladaptive behavior change for children with marked emergence status as compared with children with no symptoms of emergence delirium. Obviously, this finding is of importance to the clinician, who can now better predict the development of adverse postoperative phenomena in children based on the child’s preoperative anxiety.
1. Kain ZN, Mayes LC, O'Connor TZ, Cicchetti DV. Preoperative anxiety in children: predictors and outcomes. Arch Pediatr Adol Med 1996;150:1238–45.
2. Kain ZN, Caldwell-Andrews AA, Wang S-M. Psychological preparation of the parent and pediatric surgical patient. Anesthesiology Clin N Am 2002;20:69–88.
3. Voepel-Lewis T, Malviya S, Tait AR. A prospective cohort study of emergence agitation in the pediatric postanesthesia care unit. Anesth Analg 2003;96:1625–30.
4. Eckenhoff JE, Kneale DH, Dripps RD. The incidence and etiology of postanesthetic excitement. Anesthesiology 1961;22:667–73.
5. Davis P, Greenberg J, Gendelman M, Fertal K. Recovery characteristics of sevoflurane and halothane in preschool-aged children undergoing bilateral myringotomy and pressure equalization tube insertion. Anesh Analg 1999;88:34–8.
6. Kain Z, Mayes L, Caramico L, Hofstadter M. Distress during induction of anesthesia and postoperative behavioral outcomes. Anesth Analg 1999;88:1042–7.
7. Aono J, Mamiya K, Manabe M. Preoperative anxiety is associated with a high incidence of problematic behavior on emergence after halothane anesthesia in boys. Acta Anaesthesiol Scand 1999;43:542–4.
8. Kain Z, Mayes L, Wang S, et al. Parental presence and a sedative premedicant for children undergoing surgery: a hierarchical study. Anesthesiology 2000;92:939–46.
9. Kain Z, Mayes L, Wang S, et al. Parental presence during induction of anesthesia versus sedative premedication: which intervention is more effective? Anesthesiology 1998;89:1147–56.
10. Kain ZN, Mayes LC, Caramico LA, et al. Parental presence during induction of anesthesia: a randomized controlled trial. Anesthesiology 1996;84:1060–7.
11. Spielberger CD. Manual for the State-Trait Anxiety Inventory (STAI: Form Y) Palo Alto: Consulting Psychologists Press, 1983.
12. Buss AH, Plomin R. Theory and measurement of EAS temperament: early developing personality traits. Hillsdale: L. Erlbaum Associates, 1984:98–130.
13. Kain Z, Mayes L, Cicchetti D, et al. The Yale preoperative anxiety scale: how does it compare to a gold standard? Anesth Analg 1997;85:783–8.
14. Kain Z, Mayes L, Cicchetti D et al. Measurement tool for preoperative anxiety in children: the Yale Preoperative Anxiety Scale. Child Neuropsychol 1995;1:203–10.
15. Keegan N, Yudkowitz F, Bodian C. Determination of the reliability of three scoring systems to evaluate children after general anaesthesia. Anaesthesia 1995;50:200–2.
16. Vernon DT, Schulman JL, Foley JM. Changes in children's behavior after hospitalization. Am J Dis Child 1966;111:581–93.
17. Kain Z, Hofstadter M, Mayes L, et al. Midazolam: effects on amnesia and anxiety in children. Anesthesiology 2000;93:676–84.
18. Kain ZN, Caldwell-Andrews AA, Mayes LC, et al. Parental presence during induction of anesthesia: physiological effects on parents. Anesthesiology 2003;98:58–64.
19. Rothbart MK, Ahadi SA, Evans DE. Temperament and personality: Origins and outcomes. J Personality Soc Psychol 2000;78:122–35.
© 2004 International Anesthesia Research Society
20. Kain Z, Mayes L, Caramico L, et al. Postoperative behavioral outcomes in children: effects of sedative premedication. Anesthesiology 1999;90:758–65.