Postoperative recovery in preschool‐aged children: A secondary analysis of a randomized controlled trial comparing premedication with midazolam, clonidine, and dexmedetomidine

Preoperative anxiety in pediatric patients can worsen postoperative outcomes and delay discharge. Drugs aimed at reducing preoperative anxiety and facilitating postoperative recovery are available; however, their effects on postoperative recovery from propofol‐remifentanil anesthesia have not been studied in preschool‐aged children. Thus, we aimed to investigate the effects of three sedative premedications on postoperative recovery from total intravenous anesthesia in children aged 2–6 years.


| INTRODUC TI ON
In the early postoperative period after anesthesia, preschool-aged children are at risk of developing and experiencing adverse outcomes such as emergence delirium (ED), anxiety, pain, and/or nausea/vomiting, which might result in prolonged emergence and delayed recovery and discharge. 1Preoperative anxiety activates a stress response that increases the risk of these events.Thus, in preschool-aged children proper premedication to reduce preoperative anxiety is crucial. 2 Midazolam, a short-acting benzodiazepine, has traditionally been used in pediatric anesthesia to reduce perioperative anxiety. 3Side effects of midazolam are increased risk of ED and amnesia. 4However, the effect of midazolam on the risk for postoperative delirium remains unclear.Furthermore, some reports suggest that midazolam may even prevent such events. 5Clonidine and dexmedetomidineα 2 -agonists-are widely used as alternatives to midazolam for premedicating children. 4 The α 2 -agonists are reported to reduce episodes of ED and postoperative pain when compared to midazolam. 4Further, clonidine seems superior at preventing postoperative nausea and vomiting (PONV) compared to midazolam. 6The use of total intravenous anesthesia with propofol-remifentanil is associated with a shorter recovery, less ED, and less PONV when compared to volatile anesthesia. 7However, knowledge and consensus regarding the optimal premedication regimen to be used along with propofol-remifentanil to reduce adverse postoperative outcomes in preschool-aged children is limited and unclear.
The aim of this study was to compare postoperative recovery in preschool-aged children randomized into three different premedication groups: midazolam, clonidine, or dexmedetomidine.We aimed to assess the recovery time from total intravenous anesthesia after ear, nose, and throat (ENT) surgery, using the criteria for discharge as outcomes.Consequent evaluations of sedation levels, occurrence of postoperative episodes of ED, anxiety, pain, and PONV until reaching the discharge criteria were performed using validated observation tools.We hypothesized that clonidine and dexmedetomidine are associated with a shorter recovery time and lower risk of ED, anxiety, postoperative pain, and PONV in the post anesthesia care unit (PACU) when compared with midazolam.

| Study design
This study was a preplanned secondary analysis of a doubleblinded clinical trial evaluating premedications in preschool-aged children undergoing anesthesia for ENT surgery at Sunderby Hospital, Luleå, Sweden, from February 2017 to May 2019.
Aspects of preoperative anxiety and perioperative cardiac response in the trial cohort were previously published. 8,9The study protocol followed Good Clinical Practice and was approved by the

| Participants
Inclusion criteria were children of both sexes aged 2-6 years without severe comorbidities (American Society of Anesthesiologists Physical Status (ASA-PS) I-II), weight of ≤30 kg, scheduled for elective tonsillotomies, tonsillectomies, adenotomies or adenectomies, propofol-remifentanil anesthesia, and written informed consent from their parents.The exclusion criteria were ASA-PS > II, severe comorbidities, use of psychotropic medications, and history of recent surgery (within a year).had a better recovery profile than short-acting benzodiazepines; although the overall recovery time in the postoperative care unit was not affected.

K E Y W O R D S
child, clonidine, dexmedetomidine, midazolam, postoperative period, premedication

What is already known?
Preoperative anxiety and stress in children are associated with postoperative anxiety, emergence delirium, pain, and nausea/vomiting.Premedication regimens may influence postoperative outcomes, but there is no consensus regarding the optimal premedication regimen.

What this article adds?
This study showed that premedication with midazolam was associated with increased emergence delirium, anxiety, and pain than premedication with clonidine or dexmedetomidine alone.There were no differences in recovery time between the regimens.
The doses, administration route, and premedication timing were based on published pharmacokinetic profiles and were chosen to be safe with clinically relevant effects.The rationale for different administration times was discussed in the first publication of the trial. 9

| Randomization and blinding
Eligible children were randomly assigned in a 1:1:1: ratio to one of the three study groups.The study was randomized using sequentially numbered opaque envelopes containing group assignments.
An independent statistician generated a randomization list with a block size of 15 using the randomization function in Microsoft Excel.

Randomization of children and administration of the study drugs
were performed by a nurse independent of the research team and data collection.The patients, parents, department staff, and researchers were blinded to the interventions.
Sixty minutes before the planned induction of anesthesia, all included children were orally administered a fluid containing clonidine (CLO-group) or sterile water (MID-and DEX-groups).Twenty minutes later, a fluid was administered intranasally using a mucosal atomization device (MAD Nasal™, Teleflex, USA) containing dexmedetomidine (DEX-group) or 0.9% NaCl (CLO-and MID-groups), followed by an orally administered fluid containing midazolam (MIDgroup) or sterile water (CLO-and DEX-groups) (Table 1).

| Endpoints and assessment
The main endpoint for this secondary analysis was recovery time in the PACU.We used the Post Anaesthestic Discharge Scoring System (PADSS) and considered a PADSS-score of ≥9 as recovered. 10planatory endpoints were conceivable postoperative outcomes affecting recovery time, including vital signs, sedation levels, ED events, anxiety, pain, and PONV.
An independent observer (ÅB), not responsible for providing anesthesia or postoperative care to the children, documented the baseline assessments and evaluated the children in the PACU every 15 min for 2 h after arrival in the PACU.The observer was trained to read, learn, and test the instruments during normal clinical encounters, followed by pilot runs of the complete protocol.None of the questions were directly addressed to the children.

| Recovery/readiness for discharge/PADSS
The PADSS has high validity and reliability and is a tool for providing a safe discharge from the PACU. 10 The PADSS is based on the following five main domains: vital signs, activity and mental status, pain/ PONV, surgical bleeding, and intake/output.The maximum score on the PADSS is 10, with a maximum of two points in each subdomain (Appendix S1).
The five domains of the PADSS-score were assessed using standard monitoring in the PACU and validated observational instruments.The domain vital signs included assessments of respiratory rate, peripheral oxygen saturation, heart rate, capillary refill time, and body temperature.Activity and mental status were assessed using the Ramsay Sedation Scale (RSS), and a PADSS-score of two was achieved when the RSS-score was 2 or 3 (awake or drowsy). 10Pain was assessed using the Face Legs Activity Cry Consolability scale (FLACC) and PONV as observations of nausea, retching, or vomiting.
When FLACC score was ≤3 in the absence of PONV, a PADSS-score of two (2) was achieved for the domain.Input/output was assessed as oral intake (one point) and the ability to void (one point).
Time to readiness for discharge/recovery was defined as the time from arrival at the PACU to when the PADSS score was 9 or 10 on two consecutive assessments, and the only sub-score allowed to remain unfulfilled was the ability to void.

| Sedation/RSS
Sedation levels were assessed using the Ramsay Sedation Scale (RSS).High scores denote high levels of sedation, including the following states of consciousness: 1, anxious, restless, or agitated; 2, cooperative; 3, responds to commands only; and 4-6, different levels of unconsciousness, where 6 indicates no response to a light glabellar tap or loud auditory stimulus. 11The children were considered recovered from anesthesia at an RSS score of 2 or 3 (awake or drowsy); hence, RSS levels 1 and 4-6 were synonymous with not having recovered from anesthesia.The Paediatric Anaesthesia Emergence Delirium scale (PAED) was used to define ED events and is considered a reliable and valid measure of ED in children. 12,13The PAED scale has been applied in efforts to differentiate between ED and pain in children aged 2-6 years in the early postoperative period after tonsillectomy, adenoidectomy, or both. 14The PAED scale consists of five characteristics (the child makes eye contact with a caregiver, makes purposeful actions, is aware of their surroundings, is restless, and is inconsolable), each of which is scored using a 5-point Likert scale.

TA B L E 1
The individual scores for each item were added to determine the composite score at a particular moment.For the presence of pediatric ED, we defined the threshold score as a PAED score ≥10, which has 64% sensitivity and 86% specificity for ED, whereas a score >12 yielded 100% sensitivity and 94.5% specificity for ED. 12

| Anxiety/mYPAS
Anxiety levels were assessed using a validated Swedish version of the modified Yale Preoperative Anxiety Scale (mYPAS).The instrument contains 27 items in the following five categories (representing five domains of anxiety): activity, emotion, expressivity, state of arousal, vocalization, and parental use. 15,16The scale ranges from 0 to 100, and the score is normalized for different numbers of steps in each item.High scores denote high levels of anxiety, and based on prior characterization of the scale, an mYPAS ≤30 is classified as not anxious. 15The instrument has good-to-excellent observer reliability, with high concurrent and construct validity for assessing children's anxiety during the perioperative period. 15,16

| Pain/FLACC
The FLACC is recommended for evaluating postoperative pain in children aged 1-6 years as young children might have difficulties in expressing their level of pain. 17With this observational tool (instead of self-valid measurements) we could identify signs of pain in sedated and drowsy children.
Each of the five behavioral categories in the FLACC-facial expression, leg movement, bodily activity, crying or verbalization, and consolability-was rated on a scale of 0 to 2 to provide an overall pain score ranging from 0 to 10. Merkel et al. 18 suggested the following overall interpretations based on the score: 0, relaxed and comfortable; 1-3, mild discomfort; 4-6, moderate pain; and 7-10, severe discomfort or pain.

| Nausea/vomiting
Since nausea is a subjective phenomenon, small children may not be able to address and describe this symptom.Thus, objective symptoms of retching and vomiting were used as the endpoints.
Situations in which children verbally expressed nausea were also considered as events of PONV.
Perioperatively, local anesthetic was injected in the surgical area, and intravenous morphine (1 mg/mL, 0.1 mg/kg) was administered at the end of surgery.Betamethasone (4 mg/mL, 0.2 mg/kg) was administered intravenously at the induction of anesthesia to prevent PONV.

| Statistical analysis
No specific sample size calculation was performed for this planned secondary analysis as the size of the cohort in this clinical trial was based on the primary analysis. 9 Statistical analyses were performed using SPSS Statistics for PC, version 26.0 (IBM Corporation, Armonk, NY, USA, released 2018).
Categorical variables are presented as numbers with percentages, normally distributed continuous variables as mean ± standard deviation (SD), and non-normally distributed variables as medians with interquartile ranges (IQR) and minimal and maximal values.
To assess the overall differences among the three treatment groups, Pearson's chi-square test with Yates correction was used for categorical variables, and Kruskal-Wallis test was used for continuous variables (non-normal distribution).Significance level was set to 5%.To describe the effect size of continuous variables, Hodge-Lehman method was used to estimate differences in medians, including the 95% confidence interval (pairwise comparison of the groups). 19o compare the treatment groups in recovery and behavioral measures in the PACU, we used logistic regression with a mixed model approach to account for repeated measurements.Measures were observed/sampled at every 15th minute of an observation period of 120 min, and each measure was categorized as recovered or not recovered in the model.If a child was discharged from the PACU within 120 min, the child was considered to have recovered at the missing time points in the model.Main outputs from the model were as follows: (1) a regression model, where the intercept is common for all groups and slope is relative to the MID-group and (2) effect size described as estimated marginal means (EMM), where the effect size is relative to a standardized value of 0.5 standing for "no effect." Values are presented as 95% confidence intervals, and the significance level was set at 5% in the model.The details of the model are provided in the Appendix S2.

| RE SULTS
A total of 239 children scheduled for ENT surgery were screened for eligibility, and 90 were enrolled in the study.Eight children were excluded from this secondary analysis: six refused the study drugs (CLO-group, n = 3; MID-group, n = 3), one child in the CLO-group received sevoflurane for the maintenance of anesthesia, and one child in the MID-group had surgical complications.The final study cohort comprised 82 children (Figure 1).
The characteristics of the children and their perioperative management are presented in Table 2.

| Recovery/readiness for discharge assessed by PADSS
There were no differences between the study groups in terms of time to readiness for discharge (PADDS-score ≥9) (Table 3).

| Sedation in the PACU
After 60 min in the PACU, 38% of children (n = 31) recovered from sedation (RSS-score of 2 or 3), and there were no differences between the study groups (Table 4).4).

| Pain in the PACU
Eighteen percent of the children (n = 15) had signs of clinically significant pain in the PACU (FLACC score >3).There were significant differences between the MID-and DEX-group (p = .03).Nine children (35%) had a FLACC-score >3 in the MID-group, five (19%) in CLOgroup, and one (3%) in DEX-group.No differences were found in the number of children who were administered rescue opioids (p = .19).
After the first 60 min in the PACU, only one child had a FLACC score >3 (CLO-group, n = 1) (Tables 4 and 5).

| PONV in the PACU
Four children showed signs of vomiting, retching, or nausea in the PACU.There were no significant differences between the groups (p = .52)(Table 5).

| DISCUSS ION
This randomized study investigated the characteristics of postoperative recovery in children aged 2-6 years premedicated with midazolam, clonidine, or dexmedetomidine.We did not find any major differences in the recovery time in the PACU.However, more ED and pain was experienced with midazolam than with dexmedetomidine, and more anxiety with midazolam than with both dexmedetomidine and clonidine.
We found no differences in the overall recovery time from the PACU between the groups.A recent randomized controlled study comparing premedication with midazolam or dexmedetomidine in preschool-aged children also found no differences in PACU length of stay. 20The authors concluded that the delayed emergence from anesthesia might be associated with a smoother overall recovery with fewer episodes of ED.Consistent with that study, we found a better recovery profile with dexmedetomidine and clonidine.
ED is reported to affect every fifth child after general anesthesia. 21Preoperative anxiety, pain, and volatile anesthesia (sevoflurane) are associated with ED in children, 1 but the evidence for the risk estimate is weak.Compared to no premedication at all, previous studies have shown that midazolam reduces the risk of ED. 22 However, more recent studies report that α 2 -agonists are superior to midazolam in reducing the risk of ED in young children. 6,20In congruous with these reports, we found that both dexmedetomidine and clonidine were associated with lower risk of ED (3% and 12%, respectively) than midazolam (31%).High levels of preoperative anxiety have been reported to be a risk for postoperative anxiety. 1 In our previous analysis of the cohort, we found that midazolam resulted in better preoperative anxiolysis compared to both clonidine and dexmedetomidine. 9Despite better preoperative anxiolysis after midazolam, we found an increased risk for early delirium and anxiety in the midazolam group compared with dexmedetomidine group.A possible explanation is that the short duration of midazolam causes rapid emergence from anesthesia, whereas dexmedetomidine and clonidine provide an extended and smoother emergence. 20stoperative pain is known to be associated with adverse postoperative outcomes and delayed recovery, 23 and children with a high level of preoperative anxiety experience more postoperative pain. 1 In contrast, we found that children who received α 2 -agonists experienced higher levels of preoperative anxiety 9 but lower levels of postoperative pain than those who received short-acting benzodiazepine.However, several studies have reported that premedication with α 2 -agonists decreases the requirement for postoperative opioids, indicating their analgesic properties. 24dazolam, clonidine, and dexmedetomidine have been described as having antiemetic properties. 20In our study, only four children (5%) experienced PONV in the PACU, with no differences between the groups.Generally, the risk of PONV in children is estimated to be 30%. 25The low risk in our study might reflect the fact that all children received propofol as part of the total intravenous anesthesia and that all children also received PONV prophylaxis with betamethasone.No further nausea was observed after the first postoperative oral dose.

| Strengths and limitations
This study was a preplanned secondary analysis of a clinical trial.A major limitation is the sample size, as the size of the study cohort was justified for the previously published primary report. 9However, for this secondary analysis, we must be cautious in the conclusions, as the sample size might not be appropriate.We found no evidence

| Conclusions
Our study showed no differences in the postoperative recovery time in the PACU between the premedication regimens.Given the small sample size, our results indicate that premedication with midazolam results in a more rapid emergence from anesthesia compared with premedication with clonidine or dexmedetomidine, as children administered midazolam had more episodes of postoperative delirium, anxiety, and pain than those administered clonidine or dexmedetomidine.For smoother emergence and recovery after anesthesia, the use of clonidine or dexmedetomidine as premedication might be a better option than midazolam.
Ethic Review Board in Umeå (Dnr 2016-46-31M, March 30, 2016, chair: A Iacobaeus) and the Swedish Medical Products Agency (Dnr 5.1-2016-17854, May 13, 2016).Written informed consent was obtained from parents on the day of the procedure.All children were accompanied by a parent throughout the preoperative and postoperative periods, and one parent was allowed to attend during anesthesia induction.The trial protocol was registered in the European Union Drug Regulation Authorities' Clinical Trials Database (EudraCT 2015-003676-70) and adhered to the CONSORT guidelines for reporting the study.

Fifteen
percent of children (n = 12) had ED in the PACU (PAED-score ≥10).The MID-group had significantly more ED (31%, n = 8) than the DEX-group (n = 1).None of the children experienced ED after the first hour in the PACU (Table for a difference in the primary objective for this analysis (time to recovery), but future studies must challenge our conclusion and might use our data to estimate a proper study size.This study had few strengths.First, the blinding process, which allowed us to administer the premedications through different routes and at different time points.Second, the use of validated observational instruments that were adapted and developed for use in children during the postoperative period.Third, the study cohort of children aged 2-6 years who underwent ENT surgery is a clinically common, challenging, and relevant group of patients.However, the single-center study design with a limited number of children reduces the generalizability of the results.Additionally, dose-response relationships of the interventions were not examined in this study, but the doses used were standard and based on previous studies and clinical routines.
Administration and blinding of the study drugs according to the allocation.

Study group MID n = 26 CLO n = 26 DEX n = 30
Patient and perioperative characteristics.
TA B L E 2

Numbers of children not recovered in the domain MID n = 26 CLO n = 26 DEX n = 30
Values are the number of children (%) not recovered in a specific domain every 15th minute during the first 2 h in the PACU.Differences between the study groups were analyzed using logistic regression with a mixed model approach to account for the repeated measurements.In the model, intercept is common for the three groups, but the slope is relative to the reference (MID-group).Children discharged from PACU were considered fully recovered and remained in the model as recovered.Postoperative rescue analgesia and PONV at PACU.Values are numbers (percent).Differences between groups tested with Pearson's Chi-2 test including Yate's correction.Abbreviations: CLO, clonidine; DEX, dexmedetomidine; i.v., intravenous; MID, midazolam; PACU, post anesthesia care unit; PONV, postoperative nausea and vomiting.