Greater Incidence of Emergence Agitation in Children after Sevoflurane Anesthesia as Compared with Halothane: A Meta-analysis of Randomized Controlled Trials

We conducted a systematic review according to the Quality of Reporting of Meta-analyses (QUOROM) recommendations for improving the quality of meta-analyses.6 

A comprehensive literature search was performed using MEDLINE, EMBASE, American College of Physicians Journal Club database, Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, and Database of Abstracts of Reviews of Effects. Every effort was made to find studies reporting on emergence agitation or equivalent state after sevoflurane or halothane anesthesia in children. Although there seemed to be no widely accepted definition of emergence agitation after general anesthesia, we generally agreed on the definition proposed by Sikich and Lerman7: a disturbance in a child’s awareness of and attention to his or her environment, with disorientation and perceptual alterations including hypersensitivity to stimuli and hyperactive motor behavior in the immediate postanesthesia period. The following text searches, search headings, and their combinations were used: sevoflurane , halothane , child , agitation , and delirium . A manual search of references listed in reports and reviews was also performed. Only articles written in English were included. The date of the most recent search was April 15, 2007.

The two authors independently accessed each article to meet the following inclusion criteria. Disagreements were resolved by consensus. To enter our analysis, the studies had to be a prospective randomized trial, compare halothane and sevoflurane, report the results of ambulatory procedures on children aged younger than 12 yr, and report the incidence of emergence agitation or equivalent state after general anesthesia. We were careful to avoid including data from duplicate publications. We did not include articles from any studies with insufficient data; however, we tried to contact the corresponding author to collect unpublished data for our analysis. We also excluded studies regarding neurologically impaired patients.

Unmasked quality assessment on the selected published studies was performed by two investigators on composite aspects of study quality (six aspects in total, with scores 0 or 1: randomization, standardized anesthesia protocol, blindness of outcome measurement, comparability, withdrawals, definition of emergence agitation). Differences in opinion were settled by consensus. Data abstraction was also performed independently by two authors using standardized data collection forms. Data extracted from eligible studies included the following items: patient age, type of surgical procedure, premedication, and use of regional anesthesia in supplement to general anesthesia. Dichotomous data on the incidence of emergence agitation after sevoflurane or halothane anesthesia was also extracted from eligible studies.

All statistical analysis was performed using RevMan 4.2.10 (The Cochrane Collaboration, Oxford, United Kingdom).

An analysis of the incidence of emergence agitation was performed using the odds ratio (OR). The OR represents the odds of emergence agitation occurring in the sevoflurane group compared with the halothane group. An OR of more than 1 means that sevoflurane is expected to cause more frequent emergence agitation than halothane does. It is considered statistically significant at the P < 0.05 level if the 95% confidence interval (CI) does not include value 1.0. Because eligible studies showed clinical and methodologic diversity, the heterogeneity of collected data was assessed using a homogeneity test based on the χ²test and I². The I²statistic was used to assess the impact of heterogeneity on the results. This statistic indicates the percentage of the variability in effect estimates that is due to heterogeneity rather than sampling error.8Because of the low power of this test, especially when trials have a small sample size or are few in number, we determined a minimum cutoff P  value of 0.10 and I²value of 50% as a threshold of homogeneity to avoid false-negative results; P < 0.10 and I²> 50% indicated heterogeneity and prevented us from relying on the combination of the study results. If significant heterogeneity was denied statistically, the Mantel-Haenszel fixed-effect method was used to calculate the pooled OR. The studies that contained a zero in one cell for the number of events resulted in problems with the computation of the ratio measurement, so a value of 0.5 was added to both groups of that particular study.

Sensitivity analysis was performed by recalculating the pooled OR using data with a study quality rating greater than 4. Because we were aware that age, preoperative anxiety, surgery type, and postoperative pain could be significant confounding factors on the incidence of emergence agitation, we classified the following subgroups and analyzed them separately:

1. Study patients aged younger than 7 yr

2. Study protocols including routine premedication with benzodiazepines

3. Study patients anesthetized for myringotomy tube insertion

4. Study patients anesthetized for minor inguinal or urologic surgery, all with regional blocks appropriate for surgical analgesia

Publication bias with unpublished studies that show no significant difference can limit the validity of meta-analyses. To assess the potential for publication bias, a funnel plot was constructed of OR against associated SEs.9An asymmetrical funnel plot can reflect the publication bias, in which the OR estimates suggesting strong associations in an expected direction are preferentially published. The Begg test10was used to assess asymmetry of the funnel plot.

Using electronic databases, we initially identified 271 articles for review. Of those, 228 studies were excluded because they were unrelated studies, review articles, editorials, letters, or non-English literature. The other 43 articles were thoroughly checked to meet our inclusion criteria. Six studies were excluded because they did not match the inclusion criteria. Incidence of emergence agitation was not available in 14 studies, and failed to obtain data from respective investigators. Therefore, 23 studies1–5,11–28were identified by the defined search strategy and fulfilled the inclusion criteria, containing the necessary data for the planned comparison. The process of identifying eligible studies is illustrated in figure 1, and the details of selected trials are summarized in table 1. As shown in table 1, the included trials compared sevoflurane and halothane anesthesia in minor surgical or diagnostic procedure in children; hence, we considered it appropriate that the results of the respective studies were combined for analysis.

In 23 trials, a total of 1,252 patients anesthetized with sevoflurane and 1,111 patients anesthetized with halothane were evaluated for the incidence of emergence agitation. The pooled OR of the incidence of emergence agitation with sevoflurane was 2.21 (95% CI, 1.77–2.77; P < 0.0001). The major results of the meta-analysis comparing sevoflurane and halothane are summarized in figure 2. The test for heterogeneity gave χ²= 29.69, I²= 25.9%, and 22 degrees of freedom (P = 0.13). These results warranted the use of the Mantel-Haenszel method to calculate the pooled OR.

Sensitivity analysis was used by recalculating the pooled OR using high-quality studies. The pooled OR of 1.82 (95% CI, 1.37–2.41; P < 0.0001) was obtained when subgroup analysis was performed for the 14 studies that had a study quality rating greater than 5. Furthermore, we conducted subgroup analyses to explore the effects of known confounding factors on the incidence of emergence agitation (table 2). When the pooled analysis was restricted to the studies of patients aged younger than 7 yr, the pooled OR was 1.88 (95% CI, 1.39–2.54; P = 0.0001). In 12 studies, the study protocol included sedative premedication for all patients. In the subgroup of routine premedication, the pooled OR was 1.77 (95% CI, 1.26–2.47; P = 0.0009). We found 6 studies of myringotomy procedure. In myringotomy studies, the pooled OR was 1.79 (95% CI, 1.26–2.53; P = 0.001). Five trials studied emergence agitation in children who had inguinal or minor urologic surgery and appropriate regional block with local anesthetics. When we calculated the pooled OR of pain-treated inguinal surgery, sevoflurane anesthesia also showed a higher incidence of emergence agitation (OR = 3.20; 95% CI, 1.65–6.22; P = 0.0006).

A funnel plot was used to detect possible publication bias in the meta-analysis (fig. 3). In this study, the existence of major publication bias was not supported statistically by the Begg test (Kendall correlation coefficient = 0.0988; Z value = 0.6603; P = 0.5091).

Our meta-analysis revealed that sevoflurane anesthesia more often resulted in emergence agitation than did halothane in pediatric patients. The results of our analysis are consistent with the current consensus among pediatric anesthesiologists. Several previous studies reported that sevoflurane anesthesia caused children to have a higher incidence of emergence agitation as compared with halothane, but some prospective studies performed to evaluate this point found no significant difference between the two anesthetics. Indeed, 20 of 23 trials that met our inclusion criteria yielded an OR of greater than 1 for emergence agitation due to sevoflurane, whereas only 8 studies showed a statistically significant difference (fig. 2). The reasons for conflicting results include differences in study design and quality. Some trials could have insufficient power to adequately explore the differences between two anesthetics. The use of meta-analytic techniques allowed the inclusion of 2,363 patients, of which 1,252 had sevoflurane anesthesia and 1,111 had halothane anesthesia. This sample size would otherwise be impossible to accumulate in a reasonable time in a randomized control trial, and the large sample size made it possible to detect the significant differences of incidence between the two anesthetics.

The sensitivity analyses showed the strength of the evidence of higher incidence of emergence agitation after sevoflurane anesthesia (table 2). Because we were aware that the different types of surgical procedure could be a significant confounding factor on the incidence of emergence agitation, we analyzed myringotomy procedures separately as a subgroup. Myringotomy is a minimally invasive surgery and has the least significant difference in terms of anesthesia and surgical methods among institutions. As shown in table 2, the pooled results of 6 myringotomy studies showed a significantly higher incidence of emergence agitation in sevoflurane anesthesia. Aono et al.  2reported that preschool children have a significant risk of emergence agitation. We found 11 studies that included children aged younger than 6 yr, and we analyzed them separately as a subgroup. The pooled OR indicates that sevoflurane anesthesia results in a higher incidence of emergence agitation than does halothane in the preschool age group.

The etiology of emergence agitation derives from multiple factors, including pain, preoperative anxiety, type of surgical procedures, personal character of the patient, too rapid awakening, and type of anesthetics. No sole factor can explain the etiology of emergence agitation. Although pain is definitely a major reason for emergence agitation, screaming as a result of pain should be distinguished from emergence agitation. However, especially in younger children, it is sometimes not possible to distinguish between them. It is widely believed that reducing or eliminating pain decreases the incidence of emergence agitation after sevoflurane anesthesia. Several studies demonstrated that regional block,14,29opioids,17,30and nonsteroidal antiinflammatory drugs4decrease the incidence of emergence agitation. However, emergence agitation often occurs even after adequate pain treatment or after procedures that are not associated with pain. Indeed, our subgroup analyses clearly indicated that sevoflurane anesthesia still causes a higher probability of emergence agitation than halothane, even if pain treatment strategy is provided. In addition, if we consider the fact that anesthesia for essentially pain-free procedural sedation can cause emergence agitation,16complete pain treatment does not guarantee a calm awakening after sevoflurane.

The reasons for a higher incidence of emergence agitation after sevoflurane are not fully understood. Sevoflurane in particular situations may exert an irritating side effect on the central nervous system, because epileptiform seizure activity in previously nonepileptic patients has been observed with electroencephalography during sevoflurane anesthesia.31,32The mechanism of cortical epileptogenicity by sevoflurane is largely unknown.33Because volatile anesthetics with low blood solubility generally tend to cause a higher incidence of emergence agitation, rapid awakening has been posited as a cause for this phenomenon. However, rapid awakening after propofol has not been associated with emergence agitation.34Therefore, rapid awakening per se  does not seem to be a factor in the causation of emergence agitation. Giving sedative premedication before anesthesia has been tried to ameliorate preoperative anxiety with hopes that it would decrease the incidence of emergency agitation. Although some studies18,35showed the usefulness of sedative premedication to reduce agitation, our subgroup analysis of the routine benzodiazepine administration group indicated that premedication does not reduce a higher incidence of emergence agitation to the same level as the incidence after halothane.

Our data demonstrate that sevoflurane in children has an increased risk of emergence agitation, and anesthesiologists treating children should be aware of the risk. Restless recovery after anesthesia is not absolutely harmless; rather, it may cause self-injury and the accidental removal of dressing or catheters. Parents feel extreme anxiety to see their child cry inconsolably. Although emergence agitation is typically self-limited and resolves spontaneously within 15 min, additional medications to address retractable excitation may delay recovery room discharge and add additional cost. Pediatric anesthesiologists should consider methods to reduce the risk of emergence agitation after sevoflurane anesthesia. Complete analgesia is an absolute requirement for comfortable emergence. A variety of medications, including opioids,36benzodiazepines,18and α²agonists,37,38were tried to reduce the incidence of emergence delirium with various success. Intravenous anesthesia using propofol reduces the risk of emergence agitation.34,39The efficacy of switching to isoflurane maintenance is debatable.40,41,42The strategies regarding prevention and treatment of emergence agitation were discussed in detail in a previous review.43 

It is important to address some limitations of meta-analysis. First, each study has a different study protocol. This can elicit significant heterogeneity, although significant heterogeneity in our study was not suspected by statistical analysis. It is debatable whether it is justified to combine the results of different protocols in the calculation of the pooled OR and in drawing conclusions. Second, emergence agitation was the focus of this meta-analysis; however, measurement of emergence agitation in each respective study was not always performed using a validated and reliable tool. This may preclude comparisons among the clinical trials. Most studies used a self-made, nonvalidated rating scale to evaluate emergence agitation, which focused more on behavioral than psychometric factors. None used a validated scoring system such as the Pediatric Anesthesia Emergence Delirium scale.7Furthermore, the emergence agitation was evaluated by a blinded specialist in some studies; in other studies, however, blindness was not warranted. These factors can cause serious errors in measurement of emergence agitation and significantly bias the results of this study. Third, because the meta-analysis is based on published articles, there is a possibility of publication bias. In this study, the number of unpublished, nonindexed, or non-English articles could affect our conclusions, although the funnel plot analysis showed no evidence of publication bias. Although we limited our analysis to the literature in English, the effect of excluding non-English trials on the results of a meta-analysis is equivocal. Some data suggest that exclusion of trials not published in English may actually result in a more conservative estimate of the treatment effect.44This may be related in part to the presence of publication bias where only positive findings are published, primarily in English-language journals.45 

In summary, meta-analysis of the currently available randomized controlled trials that compared the incidence of emergence agitation in children after sevoflurane and halothane anesthesia indicates that sevoflurane results in a higher probability of emergence agitation than halothane, with a pooled OR of 2.21 (95% CI, 1.77–2.77; P < 0.0001). Various sensitivity analyses strengthened the reliability of the results. Anesthesiologists who care for children should be aware of the higher risk of excitatory emergence after sevoflurane anesthesia and should consider methods to prevent emergence agitation in order to provide high-quality anesthetic care for children.

The authors thank Toshiya Shiga, M.D., Ph.D. (Assistant Professor, Department of Anesthesiology, Toho University Ohashi Medical Center, Tokyo, Japan), Ryuichiro Araki, Ph.D. (Assistant Professor, Community Health Science Center, Saitama Medical University, Saitama, Japan), Yoshitaka Tsubono, M.D., Ph.D. (Professor, Division of Health Policy, Tohoku University School of Public Policy, Sendai, Japan), and Shinichi Kuriyama, M.D., Ph.D. (Associate Professor, Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan), for their statistical assistance.