ABSTRACT
Background
Preventive measures for postoperative vomiting (POV) in pediatric strabismus surgery are essential. Previous experimental studies have shown the independent antiemetic effects of propofol‐based total intravenous anesthesia (TIVA), dexamethasone (DEX), and ondansetron (OND). However, the real‐world outcomes of POV following the combined use of DEX and OND with propofol/opioid TIVA remain unknown.
Aims
To evaluate the longitudinal incidence of POV across three phases of different anesthesia regimens.
Methods
This retrospective observational study was conducted at a single tertiary‐care children's hospital in Japan, including children aged < 18 years who underwent strabismus surgery and had no major comorbidities. The primary outcome was either POV or the use of antiemetics within 24 h or by discharge. Changes in the levels and time‐trend slopes of POV were evaluated using interrupted time series analysis among three phases: (1) sevoflurane with pentazocine, (2) propofol/opioid TIVA with DEX, and (3) propofol/opioid TIVA with DEX and OND.
Results
Of the 2378 children, the POV incidence in Phases 1, 2, and 3 was 109/471 (23.1%), 87/1260 (6.9%), and 28/647 (4.3%), respectively (p < 0.001). A significant level change in POV occurrence was observed from Phase 1 to Phase 2, while no significant level change was found from Phase 2 to Phase 3. The time‐trend changes in POV occurrence showed no significant difference during Phases 2 and 3.
Conclusions
Real‐world departmental‐level data showed a decrease in POV occurrence after transitioning from sevoflurane‐based anesthesia with pentazocine to propofol/opioid TIVA with DEX. However, no significant decrease in POV occurrence was found by adding OND to propofol/opioid TIVA with DEX. Further studies are needed to improve the generalizability of evaluating the real‐world antiemetic effect of combining antiemetic medications on propofol/opioid TIVA.
Keywords: child, general anesthesia, interrupted time series analysis, postoperative nausea and vomiting, strabismus
Abbreviations
- 95% CI
95% confidence interval
- DEX
dexamethasone
- OND
ondansetron
- POV
postoperative vomiting
- TIVA
total intravenous anesthesia
1. Introduction
Postoperative vomiting (POV) occurs in approximately 30% of cases, of which up to 80% are in high‐risk children [1, 2]. POV potentially results in complications such as electrolyte imbalance, dehydration, wound dehiscence, aspiration [2, 3, 4, 5], psychological burden on children and families [2, 3, 4, 5], and increased medical expenses due to prolonged hospital stays [6, 7, 8]. Therefore, anesthesiologists must provide appropriate POV prophylaxis during general anesthesia in high‐risk children.
According to guidelines by Gan et al., pediatric strabismus surgery is considered high‐risk for POV, warranting the use of dexamethasone (DEX) and ondansetron (OND) with propofol‐based total intravenous anesthesia (TIVA) when additional POV risks are present [9]. Previous randomized controlled studies have demonstrated the benefits of using propofol‐based TIVA, DEX, and OND individually, as well as the combined use of DEX and OND [4, 10, 11, 12, 13].
Real‐world evidence regarding the impact of anesthesia regimens on POV occurrence is limited. A previous real‐world study on pediatric strabismus surgery found that implementing opioid‐free anesthesia regimens reduced the POV rate [14]. However, the real‐world prophylactic effects of combining these agents with propofol/opioid TIVA have not been well established. At our institution, the POV prophylaxis policy was launched in July 2017, recommending a shift from sevoflurane‐based anesthesia to propofol/opioid TIVA for strabismus surgery. In January 2022, following national health insurance approval for intraoperative use, OND was added to propofol/opioid TIVA with DEX for pediatric strabismus surgeries.
This study aimed to evaluate the real‐world departmental‐level impact of these prophylactic regimen changes in children undergoing strabismus surgery. We hypothesized that the incidence of POV would decline progressively across two key phases: (1) the introduction of propofol/opioid TIVA with DEX, and (2) the addition of OND to this regimen.
2. Materials and Methods
2.1. Ethical Considerations
The Institutional Ethics Committee of Aichi Children's Health and Medical Center approved this study (approval number: 2024088; January 7, 2025), which was conducted according to the principles of the Declaration of Helsinki. Written informed consent was waived, and an opt‐out procedure was implemented, allowing family members to decline participation per the decision of the local institutional review board.
2.2. Study Design, Setting, and Population
This retrospective observational study was conducted at a single tertiary‐care children's hospital in Japan from February 2016 to November 2024. Children aged < 18 years who underwent strabismus surgery during the study period were included. Exclusion criteria were as follows: a diagnosis of chromosomal abnormalities, use of desflurane for anesthesia maintenance, an American Society of Anesthesiologists Physical Status (ASA‐PS) classification of ≥III, or refusal to participate in the study.
2.3. Data Collection
For data collection, researchers retrospectively reviewed the electronic medical record system (HAPPY ACTIS, Canon Medical Systems, Fujitsu Corporation, Tokyo, Japan; HOPE EGMAIN‐EX Web Edition) to assess the incidence of POV and postoperative antiemetic use. The electronic anesthesia record system (Fortec ORSYS, Royal Philips Corporation, Amsterdam, the Netherlands) was reviewed to extract patient details (age, sex, medical record number, comorbidities, history of cerebral complications, body weight, and ASA‐PS), surgical data (date of surgery, primary surgeon, types of manipulated extraocular muscles, surgery duration, and laterality of manipulated eyes), and anesthesia‐related information (premedications, anesthetics for induction and maintenance, intraoperative analgesics, total intraoperative fluid volume, and anesthesia duration). Data analysis began in February 2016, as no electronic anesthesia records existed before this date.
All children were admitted to the hospital the day before surgery and discharged on postoperative day 1. During admission, POV occurrence following strabismus surgery is routinely evaluated and recorded in nursing notes. Researchers identified POV presence through physician and nursing diagnostic lists or narrative documentation in nursing records. Documentation of antiemetic usage is considered reliable, as nurses verify medication names and dosages used during general anesthesia on a daily basis. Investigators also reviewed anesthesia and medical records, as well as national health insurance claims, to confirm postoperative antiemetic administration.
2.4. Definitions
The primary endpoint was the occurrence of POV or the rescue use of antiemetic medications (such as OND, metoclopramide, and domperidone) in the post‐anesthesia care unit or general ward within 24 h or until hospital discharge. Details of research term definitions are described in Supporting Information Text S1.
2.5. Change of Antiemetic Anesthesia Regimens
In our institution, anesthesia for children is performed based on their tolerance for peripheral venous access placement—either via inhalational induction with sevoflurane with or without 40% nitrous oxide in oxygen or intravenous induction with propofol and opioids.
Regarding anesthesia maintenance, sevoflurane with pentazocine and without DEX was primarily used until June 2017. In July 2017, the anesthesia department changed its antiemetic anesthesia regimens policy for pediatric strabismus surgery, strongly recommending a switch from sevoflurane‐based anesthesia to propofol/opioid TIVA with DEX. In January 2022, OND received national health insurance approval for intraoperative use in pediatric surgery in Japan, prompting the department to recommend its addition to propofol/opioid TIVA with DEX. Our department stated that anesthesiologists were to provide propofol/opioid TIVA with DEX (and OND) for pediatric strabismus surgery.
The same chief surgeon either performed or supervised all strabismus surgeries during the study period, with no changes to surgical procedures or notable events affecting the study population. The details of surgical procedures are described in Supporting Information Text S1.
2.6. Statistical Analysis
Descriptive statistics for continuous variables, both normally and non‐normally distributed, were presented as means, standard deviations, medians, and interquartile ranges (IQRs), while categorical variables were expressed as numbers and percentages. The normality of continuous data was assessed using the Shapiro–Wilk test and visually with a Q–Q plot. Analysis of variance or the Kruskal–Wallis test was used for continuous variables, and the chi‐square test or Fisher's exact test for categorical variables to compare data among three anesthesia regimens: sevoflurane‐based anesthesia, propofol/opioid TIVA with DEX, and propofol/opioid TIVA with DEX and OND.
Multilevel logistic regression analysis assessed the antiemetic effects of propofol/opioid TIVA, DEX, and OND in this study population. The models were adjusted for potential confounders (i.e., patient, anesthesia, and surgery characteristics) as fixed effects. In addition, the model accounted for variations in surgical approaches by clustering surgeons and treating them as random intercepts [15]. Confounders were selected based on the clinical expertise of board‐certified pediatric anesthesiologists and variables significantly associated with POV in previous models, including age, sex, medical history, anesthesia type, use of nitrous oxide, pentazocine, remifentanil, dexmedetomidine, total amount of intraoperative fluid administration, inferior oblique muscle (IOM) manipulation, and surgery duration [16]. Oculocardiac reflex was excluded from the models owing to its lack of association with POV in a previous study [16]. Multicollinearity was assessed using the variance inflation factor, with values below five considered acceptable. An array‐based approach was applied to address an unmeasured confounder (Supporting Information Text S2) [17, 18].
This study applied interrupted time series analysis (ITSA) using segmented linear regression models to evaluate the real‐world departmental‐level impact of implementing different anesthesia regimens on POV occurrence. Further explanation of the ITSA applied in this study is described in Supporting Information Text S2 [19, 20].
Data were analyzed using Stata V.18.0 (StataCorp, College Station, TX), with a two‐tailed p‐value of < 0.05 as the criterion for rejecting the null hypothesis in each analysis.
3. Results
3.1. Study Population
Between February 2016 and December 2024, 2786 children underwent strabismus surgery at the study institution. Of these, 408 children were excluded, resulting in a final sample of 2378 children who were analyzed (Figure 1).
FIGURE 1.
Inclusion diagram.
A total of 471 (19.8%), 1260 (53.0%), and 647 (27.2%) children were observed in Phase 1, Phase 2, and Phase 3, respectively (Table 1). The median (IQR) ages were 6 [4, 9], 6 [4, 10], and 7 [5, 10] in Phases 1, 2, and 3, respectively. The distribution of female children was similar across the three phases. Children in Phases 1 and 2 were more likely to have cerebral complications than those in Phase 3 (Table 1).
TABLE 1.
Characteristics of patients grouped according to the different antiemetic protocol periods (n = 2738).
| Patient characteristics | Sevoflurane‐based anesthesia | Propofol/opioid TIVA + DEX | Propofol/opioid TIVA + DEX + OND | p |
|---|---|---|---|---|
| (n = 471) | (n = 1260) | (n = 647) | ||
| Age, years (n, %) | 0.004 | |||
| ≤ 2 | 58 (12.3) | 161 (12.8) | 57 (8.8) | |
| 3–6 | 215 (45.7) | 501 (39.8) | 265 (41.0) | |
| 7–10 | 123 (26.1) | 307 (24.4) | 174 (26.9) | |
| ≥ 11 | 75 (15.9) | 291 (23.1) | 151 (23.3) | |
| Sex (n, %) | 0.94 | |||
| Male | 216 (45.9) | 588 (46.7) | 298 (46.1) | |
| Female | 255 (54.1) | 672 (53.3) | 349 (53.9) | |
| Body weight, kg (median, IQR) | 20.0 (15.0, 29.0) | 21.8 (16.2, 33.6) | 22.5 (17.2, 35.3) | < 0.001 |
| Year of surgery (n, %) | < 0.001 | |||
| 2016 | 329 (69.9) | 0 (0) | 0 (0) | |
| 2017 | 142 (30.2) | 192 (15.2) | 0 (0) | |
| 2018 | 0 (0) | 307 (24.4) | 0 (0) | |
| 2019 | 0 (0) | 278 (22.1) | 0 (0) | |
| 2020 | 0 (0) | 224 (17.8) | 0 (0) | |
| 2021 | 0 (0) | 259 (20.6) | 0 (0) | |
| 2022 | 0 (0) | 0 (0) | 245 (37.9) | |
| 2023 | 0 (0) | 0 (0) | 184 (28.4) | |
| 2024 | 0 (0) | 0 (0) | 218 (33.7) | |
| History of cerebral complications a (n, %) | 18 (3.8) | 66 (5.2) | 16 (2.5) | 0.016 |
Note: Summaries are presented as the mean (standard deviation [SD]), median (interquartile range [IQR]), or no. (%).
Abbreviations: ASA‐PS, American Society of Anesthesiologists Physical Status; DEX, dexamethasone; IQR, interquartile range; OND, ondansetron; TIVA, total intravenous anesthesia. No data on age, sex, body weight, years since surgery, ASA‐PS score, or history of cerebral complications were missing.
History of cerebral complications included at least one of hydrocephalus, cerebral tumor, intraventricular hemorrhage, major brain trauma requiring hospital admission, epilepsy, and cerebral palsy.
Table 2 shows the anesthetic and surgical characteristics for the three phases. Anesthesia maintenance with sevoflurane was used in 83.4% (393/471) of children in Phase 1, while propofol/opioid TIVA was used in 72.2% (910/1260) and 98.0% (634/647) of children in Phases 2 and 3, respectively. Anesthesia maintenance with combined sevoflurane and propofol infusion was observed in 16.6% (78/471) of children in Phase 1, while either single sevoflurane or combined sevoflurane and propofol infusion for anesthesia maintenance was found in 27.8% (250/1260) and 2.0% (13/647) of children in Phases 2 and 3, respectively.
TABLE 2.
Characteristics of anesthesia and surgery grouped according to the different antiemetic protocol periods (n = 2738).
| Anesthetic and surgical characteristics | Sevoflurane‐based anesthesia | Propofol/opioid TIVA with DEX | Propofol/opioid TIVA with DEX/OND | p |
|---|---|---|---|---|
| (n = 471) | (n = 1260) | (n = 647) | ||
| Type of anesthetics for maintenance (n, %) | < 0.001 | |||
| Sevoflurane | 393 (83.4) | 157 (12.5) | 3 (0.46) | |
| Sevoflurane + Propofol a | 78 (16.6) | 193 (15.3) | 10 (1.55) | |
| Propofol b | 0 (0) | 910 (72.2) | 634 (98.0) | |
| Premedications (n, %) | < 0.001 | |||
| Midazolam | 379 (80.5) | 738 (58.6) | 342 (52.9) | |
| Diazepam | 84 (17.8) | 340 (27.0) | 106 (16.4) | |
| None | 8 (1.70) | 182 (14.4) | 199 (30.8) | |
| Pentazocine usage c (n, %) | 399 (84.7) | 93 (7.4) | 0 (0) | < 0.001 |
| Fentanyl usage d (n, %) d | 16 (3.40) | 1110 (88.1) | 639 (98.8) | < 0.001 |
| Remifentanil usage e (n, %) | 0 (0) | 786 (62.4) | 591 (91.3) | < 0.001 |
| Nitrous oxide usage for anesthesia maintenance (n, %) | 218 (46.3) | 284 (22.5) | 6 (0.93) | < 0.001 |
| Ocular peripheral nerve block usage (n, %) | < 0.001 | |||
| Peripheral ocular nerve block | 24 (5.1) | 80 (6.4) | 0 (0) | |
| Sub‐Tenon's block | 32 (6.8) | 1056 (83.8) | 644 (99.5) | |
| Dexamethasone usage f (n, %) | < 0.001 | |||
| None | 458 (97.2) | 128 (10.2) | 8 (1.24) | |
| < 0.15 mg·kg−1 | 12 (2.55) | 410 (32.5) | 260 (40.2) | |
| ≥ 0.15 mg·kg−1 | 1 (0.21) | 722 (57.3) | 379 (58.6) | |
| Ondansetron usage g (n, %) | 0 (0) | 0 (0) | 537 (83.0) | < 0.001 |
| IV midazolam usage h (n, %) | 2 (0.42) | 25 (1.98) | 4 (0.62) | 0.008 |
| Dexmedetomidine usage i (n, %) | 0 (0) | 181 (14.4) | 51 (7.88) | < 0.001 |
| Hydroxyzine usage j (n, %) | 9 (1.9) | 10 (0.79) | 5 (0.77) | 0.092 |
| Intraoperative adjunctive analgesics usage k (n, %) | < 0.001 | |||
| None | 54 (11.5) | 15 (1.2) | 1 (0.15) | |
| Acetaminophen | 331 (70.3) | 939 (74.6) | 358 (55.3) | |
| NSAIDs | 86 (18.3) | 100 (7.9) | 0 (0) | |
| Acetaminophen + NSAIDs | 0 (0) | 205 (16.3) | 288 (44.5) | |
| Total amount of intraoperative fluid administration, mL/kg/h (mean, SD) l | 3.5 (2.1) | 5.2 (3.3) | 6.0 (2.9) | < 0.001 |
| Type of manipulated extraocular muscles (n, %) | ||||
| Inferior oblique muscle | 163 (34.6) | 398 (31.6) | 230 (35.6) | 0.17 |
| Medial rectus muscle | 141 (29.9) | 391 (31.0) | 190 (29.4) | 0.74 |
| Lateral rectus muscle | 263 (55.8) | 776 (61.6) | 411 (63.5) | 0.027 |
| Inferior rectus muscle | 9 (1.9) | 19 (1.5) | 16 (2.5) | 0.33 |
| Superior rectus muscle | 118 (25.1) | 204 (16.2) | 108 (16.7) | < 0.001 |
| Duration of surgery, hour (mean, SD) | 0.49 (0.25) | 0.54 (0.26) | 0.64 (0.29) | < 0.001 |
| Duration of general anesthesia, hour (mean, SD) | 0.97 (0.29) | 1.10 (0.30) | 1.20 (0.32) | < 0.001 |
| Bilateral surgery (n, %) | 311 (66.0) | 752 (59.7) | 375 (58.1) | 0.018 |
| Presence of OCR m (n, %) | 55 (9.1) | 281 (22.3) | 270 (41.7) | < 0.001 |
Note: Summaries are presented as the mean (SD) or no. (%).
Abbreviations: DEX, dexamethasone; IQR, interquartile range; IV, intravenous; NSAIDs, non‐steroidal anti‐inflammatory drugs; OCR, oculocardiac reflex; OND, ondansetron; SD, standard deviation.
Sevoflurane was administered alongside propofol infusion. The mean (SD) doses of propofol infusion used with sevoflurane for anesthesia maintenance during the three phases (sevoflurane‐based anesthesia, propofol/opioid TIVA with dexamethasone, and propofol/opioid TIVA with dexamethasone and ondansetron) were 1.36 (0.56), 1.53 (0.66), and 2.00 (0.82) mg·kg−1·h−1, respectively.
The mean (SD) doses of propofol infusion for anesthesia maintenance in Phases 2 and 3 (propofol/opioid TIVA with dexamethasone and propofol/opioid TIVA with dexamethasone and ondansetron) were 2.91 (0.31) and 2.93 (0.30) mg·kg−1·h−1, respectively.
The mean (SD) doses of intraoperative IV pentazocine in Phases 1 and 2 (sevoflurane‐based anesthesia and propofol/opioid TIVA with dexamethasone) were 0.25 (0.069) and 0.22 (0.066) mg·kg−1, respectively.
The total amount of fentanyl administered included doses from the start of anesthesia induction until discharge from the post‐anesthesia care unit. The mean (SD) doses of fentanyl usage in the three phases (sevoflurane‐based anesthesia, propofol/opioid TIVA with dexamethasone, and propofol/opioid TIVA with dexamethasone and ondansetron) were 1.82 (0.43), 2.13 (0.88), and 2.79 (1.09) μg·kg−1, respectively.
The mean (SD) doses of remifentanil used in Phases 2 and 3 (propofol/opioid TIVA with dexamethasone, and propofol/opioid TIVA with dexamethasone and ondansetron) were 0.12 (0.041) and 0.14 (0.088) μg·kg−1·min−1, respectively.
The mean (SD) doses of dexamethasone used in the three phases (sevoflurane‐based anesthesia, propofol/opioid TIVA with dexamethasone, and propofol/opioid TIVA with dexamethasone and ondansetron) were 0.091 (0.038), 0.18 (0.067), and 0.17 (0.081) mg·kg−1, respectively.
The mean (SD) doses of ondansetron used in Phase 3 (propofol/opioid TIVA with dexamethasone and ondansetron) were 0.096 (0.041) mg·kg−1.
The median (IQR) doses of IV midazolam in the three phases (sevoflurane‐based anesthesia, propofol/opioid anesthesia with dexamethasone, and propofol/opioid anesthesia with dexamethasone and ondansetron) were 0.011 (0.0067, 0.016), 0.050 (0.044, 0.056), and 0.079 (0.053, 0.12) mg·kg−1, respectively.
The mean (SD) doses of dexmedetomidine in Phases 2 and 3 (propofol/opioid anesthesia with dexamethasone, and propofol/opioid anesthesia with dexamethasone and ondansetron) were 0.51 (0.20) and 0.48 (0.25) μg·kg−1, respectively.
The median (IQR) doses of hydroxyzine used in the three phases (sevoflurane‐based anesthesia, propofol/opioid anesthesia with dexamethasone, and propofol/opioid anesthesia with dexamethasone and ondansetron) were 0.42 (0.25, 0.50), 0.44 (0.41, 0.44), and 0.50 (0.48, 0.63) mg·kg−1, respectively.
Intraoperative adjunctive analgesic use includes one missing value.
The total amount of intraoperative fluid administered includes 54 missing values.
The presence of OCR was defined as either: (1) a sudden drop of heart rate by ≥ 20% from the baseline level immediately preceding the drop, or (2) atropine administration for a sudden heart rate drop, as indicated by the anesthesia record showing the drop before atropine administration or comments from the anesthesiologist confirming atropine use for OCR.
Pentazocine was used in most cases in Phase 1; however, its use decreased in Phases 2 and 3, where fentanyl was mainly used intraoperatively. DEX was used in 2.8% (13/471) of children in Phase 1; however, this increased to 89.8% (1132/1260) and 98.8% (639/647) in Phases 2 and 3, respectively. OND was used in 83.0% (537/647) of children in Phase 3. The proportions of peripheral ocular nerve block were 5.1% (24/471) and 6.4% (80/1260) in Phases 1 and 2, respectively, with no cases in Phase 3. Sub‐Tenon's block was performed in 6.8% (32/472), 83.8% (1056/1260), and 99.5% (644/647) of children in Phases 1, 2, and 3, respectively. The proportions of remifentanil, ocular peripheral nerve blocks, and intraoperative adjunctive analgesics usage increased from Phase 1 to Phase 3. The proportions of IOM manipulation were similar among the three phases.
A single ophthalmologist performed 1630/2420 (67.3%) of the surgeries, and ophthalmology trainees (fellows) performed 790/2420 (32.6%) during the study period.
3.2. Independent Antiemetic Effect of Propofol/Opioid TIVA and Administration of DEX and OND
POV occurred in 224 out of 2378 (9.4%) children during the study period. The use of DEX and OND showed a decreased association with POV occurrence independently, after adjusting for potential confounders. However, propofol/opioid TIVA was not associated with a decreased incidence of POV compared with sevoflurane usage for anesthesia maintenance (Table 3). As a sensitivity analysis, we applied the array‐based approach to adjust for an unmeasured confounder (such as family history of POV). The adjusted ORs for the use of DEX, OND, and propofol/opioid TIVA were 0.40, 0.70, and 1.30, respectively. Fentanyl was excluded from the final models owing to multicollinearity.
TABLE 3.
Multilevel logistic regression analysis to evaluate the antiemetic effects of anesthetics and antiemetic adjuncts (n = 2378).
| Variable | OR (95% CI) | p |
|---|---|---|
| Dexamethasone usage a | 0.37 (0.20–0.68) | 0.002 |
| Ondansetron usage b | 0.59 (0.35–0.98) | 0.042 |
| Type of anesthetics for maintenance | ||
| Sevoflurane | 1 (reference) | |
| Sevoflurane + Propofol | 0.72 (0.45–1.17) | 0.19 |
| Propofol | 1.17 (0.56–2.42) | 0.68 |
| Sex, male | 1.04 (0.77–1.40) | 0.79 |
| Age, years | ||
| ≤ 2 | 1 (reference) | |
| 3–6 | 3.68 (1.91–7.07) | < 0.001 |
| 7–10 | 3.11 (1.56–6.20) | 0.001 |
| ≥ 11 | 1.61 (0.67–3.19) | 0.34 |
| History of cerebral complications c | 1.49 (0.75–3.45) | 0.22 |
| Nitrous oxide usage for anesthesia maintenance | 1.18 (0.81–1.72) | 0.38 |
| Pentazocine usage d | 2.31 (1.28–4.17) | 0.005 |
| Remifentanil usage e | 1.19 (0.65–2.20) | 0.58 |
| Dexmedetomidine usage f | 1.21 (0.65–2.27) | 0.55 |
| Total amount of intraoperative fluid administration, mL/kg/h | 1.0006 (0.94–1.06) | 0.98 |
| IOM manipulation | 1.70 (1.26–2.29) | 0.001 |
| Duration of surgery, hour | 0.87 (0.49–1.54) | 0.63 |
Note: A multilevel logistic regression model was used to adjust for potential confounders, considering the surgeon as a random effect. Odds ratios and 95% CIs were calculated using multilevel logistic regression. Complete case analysis included 2324 patients in the model. The area under the curve of the regression model was 0.75 (95% CI, 0.72–0.79). The maximum value of the variance inflation factor was 3.49, which is < 5.
Abbreviations: CI, confidence interval; IOM, inferior oblique muscle; OR, odds ratio; SD, standard deviation.
The variable Ondansetron usage was dichotomous. The mean (SD) dose in the administered group was 0.096 (0.041) mg/kg.
The variable dexamethasone usage was dichotomous. The mean (SD) dose in the administered group was 0.18 (0.073) mg/kg.
A history of cerebral complications included at least one of the following: hydrocephalus, cerebral tumor, intraventricular hemorrhage, major brain trauma requiring hospital admission, epilepsy, or cerebral palsy.
The variable Pentazocine was dichotomous. The mean (SD) dose of pentazocine in the administered group was 0.24 (0.069) mg/kg.
The variable Remifentanil was dichotomous. The mean (SD) dose of remifentanil in the administered group was 9.3 (5.5) μg/kg.
The variable Dexmedetomidine was dichotomous. The mean (SD) dose of dexmedetomidine in the administered group was 0.50 (0.21) μg/kg.
3.3. Transitions of Anesthesia Regimens and Antiemetic Use
Figures [Link], [Link] show the proportions of anesthetics and antiemetics administered during anesthesia maintenance. The anesthesia department announced a recommendation to switch from sevoflurane‐based anesthesia with pentazocine to propofol/opioid TIVA with DEX in July 2017 (Month #17), and intraoperative administration of OND was approved in Japan in January 2022 (Month #71). Propofol/opioid TIVA use reached 80% at 13 months (Month #30), and 80% compliance for six consecutive months was achieved at 21 months (Month #38) after the departmental announcement. Intraoperative use of DEX and OND reached 80% at 5 months (Month #22) and 6 months (Month #77), and 80% compliance for six consecutive months was achieved at 10 months (Month #27) and 11 months (Month #82), respectively, after the departmental announcement (Figure S2). Pentazocine use decreased by 80% after 4 months (Month #21) following the department announcement, and an 80% reduction for six consecutive months was achieved after 9 months (Month #26). Fentanyl and remifentanil use reached 80% at 27 months (Month #44) and 21 months (Month #38), and 80% compliance for six consecutive months was achieved at 32 months (Month #49) and 38 months (Month #55), respectively, after the introduction of propofol/opioid TIVA (Figure S3).
3.4. Incidence of POV After a Departmental Policy Change in Anesthesia Regimens
The incidence of POV in Phases 1, 2, and 3 was 109/471 (23.1%), 87/1260 (6.9%), and 28/647 (4.3%), respectively (p < 0.001). The change in anesthesia regimens from Phase 1 to 2 showed a significant change in the level of POV occurrence (ꞵ‐coefficient, −17.6; 95% CI, −25.5 to −9.69; p < 0.001), while no significant change in the level of POV occurrence was found from Phase 2 to 3 (ꞵ, −1.69; 95% CI, −7.25 to −3.88; p = 0.55) (Table 4, Figure 2). The change in the time‐trend slope of POV occurrence showed no significant difference during Phases 2 (ꞵ, −0.0077; 95% CI, −0.11 to 0.12; p = 0.90) and 3 (ꞵ, −0.0053; 95% CI, −0.26 to 0.16; p = 0.62) (Table 4, Figure 2). The comparisons of changes in the time‐trend slope of POV occurrence showed no significant difference between Phases 1 and 2 (ꞵ, −0.18; 95% CI, −0.85 to 0.48; p = 0.58) and Phases 2 and 3 (ꞵ, −0.060; 95% CI, −0.30 to 0.18; p = 0.61) (Table 4, Figure 2).
TABLE 4.
Results of an interrupted time series analysis across different phases (n = 2378).
| Parameters | ꞵ‐coefficient (95% CI) | p |
|---|---|---|
| Immediate effect by changes in anesthesia regimens (intercept) | ||
| Sevoflurane + pentazocine (Phase 1) to propofol/opioid TIVA + DEX (Phase 2) | −17.6 (−25.5, −9.69) | < 0.001 |
| Propofol/opioid TIVA + DEX (Phase 2) to propofol/opioid TIVA + DEX + OND (Phase 3) | −1.69 (−7.25–3.88) | 0.55 |
| Trend change through each phase (slope) | ||
| Propofol/opioid TIVA + DEX (Phase 2) | −0.0077 (−0.11–0.12) | 0.90 |
| Propofol/opioid TIVA + DEX + OND (Phase 3) | −0.0053 (−0.26–0.16) | 0.62 |
| Difference in trend change compared different phases (slope) | ||
| Sevoflurane + pentazocine (Phase 1) to propofol/opioid TIVA+DEX (Phase 2) | −0.18 (−0.85–0.48) | 0.58 |
| Propofol/opioid TIVA + DEX (Phase 2) to propofol/opioid TIVA + DEX + OND (Phase 3) | −0.060 (−0.30–0.18) | 0.61 |
Note: Percentages of occurrence of POV and 95% CIs were calculated using interrupted time series analysis. Cumby–Huizinga test for autocorrelation showed no significant autocorrelation in 12 lags (minimal p‐value was 0.0619).
FIGURE 2.
Observed percentages of postoperative vomiting (POV) in the three periods: (1) sevoflurane‐based anesthesia with pentazocine, (2) propofol/opioid total intravenous anesthesia (TIVA) with dexamethasone, and (3) propofol/opioid TIVA with dexamethasone and ondansetron. Black plots show the percentage of POV occurrence each month; the black line shows the outcome estimation by an interrupted time series analysis.
4. Discussion
This study provides real‐world evidence regarding the departmental‐level impact of changing anesthesia policies to prevent POV in children undergoing strabismus surgery. The effect of a combination of antiemetic anesthesia regimens, rather than the sole antiemetic effect of DEX and OND, was also investigated. This study showed an immediate decrease in POV incidence following the change from sevoflurane with pentazocine to propofol/opioid TIVA with DEX. However, adding OND to propofol/opioid TIVA with DEX did not result in an immediate decrease in POV occurrence. No significant trend changes in POV occurrence were observed during each phase with different anesthesia regimens.
Previous randomized controlled studies have shown an apparent antiemetic effect with the combination of OND and DEX compared with the single use of DEX across different regimens and doses of administration [21, 22]. However, most of these previous studies maintained anesthesia with inhalational agents (including halothane, nitrous oxide, isoflurane, and sevoflurane). The individual‐level prophylactic effects of propofol‐based anesthesia, DEX, and OND have been well‐documented in experimental pediatric studies [10, 11, 12]. Furthermore, Weible S. et al. reported that the combination of DEX and OND had a beneficial effect for POV prophylaxis compared with the single use of either DEX or OND or compared with placebo [13]. There is limited evidence regarding the combination effect of antiemetic medications administered with propofol/opioid TIVA. Our study found that adding OND to propofol/opioid TIVA with DEX reduced POV occurrence by 1.7% without a statistically significant difference. These results suggest that adding OND to DEX during propofol/opioid TIVA may not be as effective as when used with inhalational anesthetics. Further prospective studies with broader generalizability are needed to evaluate the prophylactic effect of adding OND to propofol/opioid anesthesia with DEX.
The pharmacological effect of preventing POV has primarily been evaluated under experimental conditions. Chiem et al. reported real‐world evidence that implementing opioid‐free anesthesia reduced postoperative nausea and vomiting (PONV) incidence in pediatric strabismus surgery [14]. However, there is limited real‐world evidence regarding the adoption of propofol/opioid TIVA with DEX and OND in children. Our data showed that the departmental action of transitioning regimens resulted in an immediate reduction in POV incidence after shifting from sevoflurane‐based anesthesia with pentazocine to propofol/opioid TIVA with DEX, but not after adding OND to propofol/opioid TIVA with DEX. The immediate effect (level change) demonstrated the pharmacological effect of the new antiemetic regimens compared with the regimen before the transition (such as sevoflurane‐based anesthesia with pentazocine). Protocol adherence is essential to interpret the immediate effects of the adopted anesthesia regimen. In our study, 80% compliance for OND use was achieved 6 months after the departmental announcement. This long period to achieve high compliance could have caused the nonsignificant level change between Phases 2 and 3. Earlier transitions may have increased the immediate effect, suggesting that more vigorous departmental actions could have encouraged anesthesiologists to achieve high compliance with the new anesthesia regimens. The time‐trend effect can show a departmental‐level effect for the outcome over time, which can be majorly affected by compliance rate. In our study, the time‐trend effect during Phases 2 and 3 did not show beneficial effects over time, which may be attributed to the variability in adherence to the recommended anesthesia regimens in both Phases 2 and 3; the time‐trend change could have been more prominent with a slower transition to the new anesthesia regimens. Therefore, implementation of high‐compliance evidence‐based antiemetic strategies may maximize the clinical impact on POV prophylaxis over time.
A minor beneficial effect in preventing POV was observed when OND was added to propofol/opioid TIVA with DEX, raising the question of the cost‐effectiveness of OND when administered with other antiemetic adjuncts in propofol‐based anesthesia. Our study, based on retrospective data from three Japanese institutions, including our hospital, found that this benefit was relatively less cost‐effective [23]. The incremental cost‐effectiveness ratio for adding OND was €400.6 per averted POV, suggesting that propofol/opioid TIVA with DEX alone was more cost‐effective than propofol/opioid TIVA with DEX and OND under the current Japanese national schedule fee framework [23]. Sensitivity analyses identified OND cost and POV incidence as key determinants of cost‐effectiveness. While these findings reflect the economic context of Japan, given the international differences in healthcare systems [23], further discussions regarding health economics are warranted from the viewpoint of each healthcare system. In addition, the length of hospital stay was not assessed as an outcome in our study. The cost of prolonged hospital stay outweighs that of single OND administration. Therefore, further investigation addressing hospital stays is needed.
This study has some limitations. First, this study collected data from a single institution, limiting the generalizability of the results. Therefore, the findings should be interpreted cautiously owing to their limited external validity. However, conducting a multi‐center study to evaluate outcomes with changes in anesthesia regimens might be less feasible because different anesthesia departments have their own clinical policies and anesthesia regimens. Portnoy et al. reported low adherence to the Fourth Consensus Guidelines for POV management in pediatric populations [24], while our study showed more than 80% adherence to the propofol/opioid TIVA with DEX in Phase 2 and adding OND to the propofol/opioid TIVA in Phase 3. However, 80% compliance for propofol/opioid TIVA for six consecutive months was achieved after 8 months following the first achievement of 80% compliance. This variability in adherence might bias the results toward the null hypothesis; inconsistent compliance may prevent improvement in POV outcome. Second, reporting bias could be present due to the nature of the retrospective observational study design. Missing data regarding the outcome may have occurred, potentially leading the results toward a null finding. To address this, we reviewed nurses' records to check for the occurrence of POV. We also reviewed national health insurance claims to identify the use of postoperative antiemetic medications, including cases of POV, even when descriptions of POV were missing. This study did not include nausea as an outcome. This omission could lead to under‐evaluation of the PONV outcome, which is commonly used in adult studies, potentially leading to insufficient data to achieve statistical significance. However, assessing nausea was considered unreliable in younger children who may not be able to verbalize their complaints effectively. Therefore, we used only objective data, which could have led the results to lean toward a null finding owing to reduced power. Third, there were potentially unadjusted confounders in our logistic regression models. Unmeasured confounders include sevoflurane use at anesthesia induction and past and family history of PONV, motion sickness, and vertigo, which could have biased the results. Our sensitivity analysis, applying an array‐based approach, showed supporting results after adjusting for a potential confounder (family history of POV). Fourth, insufficient power might have led to non‐significant results in the ITSA. From Phases 2 to 3, adding OND to propofol/opioid TIVA with DEX reduced POV incidence by 1.7% as an immediate change effect but without statistical significance, which requires careful interpretation. Fifth, the results need to be interpreted cautiously to avoid ecological fallacy. This study found a real‐world departmental‐level impact of implementing policy changes in anesthesia regimens to decrease POV occurrence. Therefore, our findings must be interpreted differentially from the previous personal‐level studies. Finally, other factors could have influenced POV occurrence during the study period. ITSA requires strong assumptions regarding the absence of time‐progressing factors that could affect the outcomes, aside from the interventions. Our institution did not experience marked changes in surgical procedures or patient populations during the study period. However, surgical time and anesthesia duration increased over time, and an increase in peripheral nerve blocks and opioid use in anesthesia induction and maintenance might have affected POV outcome, causing a bias that negated the beneficial effects of the changes in anesthesia regimens during Phases 2 and 3. In addition, the secular trends of POV in our institution could have been affected by external factors (e.g., practical guidelines). Gan et al. published the Fourth Consensus Guidelines for the Management of Postoperative Nausea and Vomiting in 2022, which could have impacted the clinical practices of anesthesiologists. This guideline recommended the use of DEX and OND with propofol/opioid TIVA, which were included in the interventions of our study. Regarding other antiemetic strategies to reduce baseline POV risks in the guidelines [9], (1) the total amount of fentanyl between Phases 2 and 3 was similar, (2) intraoperative fluid administration in the three phases was comparable, and (3) no cases involved the administration of neostigmine. However, the proportions of nitrous oxide use for maintenance decreased as the phases progressed, which may have biased the results of the ITSA.
In conclusion, switching from sevoflurane‐based anesthesia with pentazocine to propofol/opioid TIVA with DEX resulted in a decreased POV incidence as an immediate effect. However, adding OND to propofol/opioid TIVA with DEX resulted in a decreased incidence of POV. Further longitudinal departmental‐level studies assessing generalizability are needed to evaluate the real‐world antiemetic effect of the combined use of DEX and OND with propofol/opioid TIVA.
Author Contributions
Taiki Kojima: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing the original draft, and reviewing the final version of the manuscript. Yusuke Yamauchi: conceptualization, data curation, formal analysis, investigation, methodology, visualization, writing the original draft, and reviewing the final version of the manuscript. Sayuri Yasuda: conceptualization, writing the manuscript, reviewing, and accepting the final version of the manuscript. Soichiro Obara: conceptualization, methodology, visualization, reviewing, and accepting the final version of the manuscript. Takashi Fujiwara: conceptualization, methodology, reviewing, and accepting the final version of the manuscript. Aya Sueda: conceptualization, methodology, visualization, reviewing, and accepting the final version of the manuscript.
Ethics Statement
The Institutional Ethics Committee for Aichi Children's Health and Medical Center approved this study (approval number: 2024088; January 7, 2025; Chair of the Ethics Committee: Komei Ito).
Consent
Written informed consent was waived. Opt‐out procedures were implemented to allow the patients' family members to decline study participation based on the decision of the local institutional review board.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information
Figure S1. Types of anesthetics used for anesthesia maintenance.
Figure S2. Types of antiemetic adjuncts used during general anesthesia.
Figure S3. Types of opioids used during general anesthesia.
Text S1. Supporting Information.
Text S2. Supporting Information.
Acknowledgments
We thank Editage (https://www.editage.jp/services/english‐editing) for their professional English editing services.
Kojima T., Yamauchi Y., Yasuda S., Obara S., Fujiwara T., and Sueda A., “Real‐World Impact on Postoperative Vomiting by Changing Anesthesia Regimens in Children Undergoing Strabismus Surgery: An Interrupted Time Series Analysis,” Pediatric Anesthesia 36, no. 1 (2026): 47–56, 10.1111/pan.70041.
Funding: This research received no specific grant from any funding agency in the public, commercial, or not‐for‐profit sectors.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Tan H. S., Dewinter G., and Habib A. S., “The Next Generation of Antiemetics for the Management of Postoperative Nausea and Vomiting,” Best Practice & Research. Clinical Anaesthesiology 34 (2020): 759–769. [DOI] [PubMed] [Google Scholar]
- 2. Ames W. A. and Machovec K., “An Update on the Management of PONV in a Pediatric Patient,” Best Practice & Research. Clinical Anaesthesiology 34 (2020): 749–758. [DOI] [PubMed] [Google Scholar]
- 3. Schaefer M. S., Kranke P., Weibel S., Kreysing R., Ochel J., and Kienbaum P., “Total Intravenous Anesthesia vs Single Pharmacological Prophylaxis to Prevent Postoperative Vomiting in Children: A Systematic Review and Meta‐Analysis,” Paediatric Anaesthesia 27 (2017): 1202–1209. [DOI] [PubMed] [Google Scholar]
- 4. Pankiv E., Albornoz A. E., Nghiem J., et al., “Prophylaxis for Pediatric Postoperative Nausea and Vomiting: A Scoping Review of Clinical Trials,” Canadian Journal of Anaesthesia 70 (2023): 1978–1988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Kovac A. L., “Postoperative Nausea and Vomiting in Pediatric Patients,” Paediatric Drugs 23 (2021): 11–37. [DOI] [PubMed] [Google Scholar]
- 6. Rojas‐García A., Turner S., Pizzo E., Hudson E., Thomas J., and Raine R., “Impact and Experiences of Delayed Discharge: A Mixed‐Studies Systematic Review,” Health Expectations 21 (2018): 41–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Micallef A., Buttigieg S. C., Tomaselli G., and Garg L., “Defining Delayed Discharges of Inpatients and Their Impact in Acute Hospital Care: A Scoping Review,” International Journal of Health Policy and Management 11 (2022): 103–111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Green Z., Woodman N., McLernon D. J., PATRN , and Engelhardt T., “Incidence of Paediatric Unplanned Day‐Case Admissions in the UK and Ireland: A Prospective Multicentre Observational Study,” British Journal of Anaesthesia 124 (2020): 463–472. [DOI] [PubMed] [Google Scholar]
- 9. Gan T. J., Belani K. G., Bergese S., et al., “Fourth Consensus Guidelines for the Management of Postoperative Nausea and Vomiting,” Anesthesia and Analgesia 131 (2020): 411–448. [DOI] [PubMed] [Google Scholar]
- 10. Schraag S., Pradelli L., Alsaleh A. J. O., et al., “Propofol vs. Inhalational Agents to Maintain General Anaesthesia in Ambulatory and In‐Patient Surgery: A Systematic Review and Meta‐Analysis,” BMC Anesthesiology 18 (2018): 162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. G. S. De Oliveira, Jr. , Castro‐Alves L. J., Ahmad S., Kendall M. C., and McCarthy R. J., “Dexamethasone to Prevent Postoperative Nausea and Vomiting: An Updated Meta‐Analysis of Randomized Controlled Trials,” Anesthesia and Analgesia 116 (2013): 58–74. [DOI] [PubMed] [Google Scholar]
- 12. Tramèr M. R., Reynolds D. J., Moore R. A., and McQuay H. J., “Efficacy, Dose‐Response, and Safety of Ondansetron in Prevention of Postoperative Nausea and Vomiting: A Quantitative Systematic Review of Randomized Placebo‐Controlled Trials,” Anesthesiology 87 (1997): 1277–1289. [DOI] [PubMed] [Google Scholar]
- 13. Weibel S., Schaefer M. S., Raj D., et al., “Drugs for Preventing Postoperative Nausea and Vomiting in Adults After General Anaesthesia: An Abridged Cochrane Network Meta‐Analysis,” Anaesthesia 76 (2021): 962–973. [DOI] [PubMed] [Google Scholar]
- 14. Chiem J. L., Donohue L. D., Martin L. D., and Low D. K., “An Opioid‐Free Anesthesia Protocol for Pediatric Strabismus Surgery: A Quality Improvement Project,” Pediatric Quality & Safety 6 (2021): e462. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Austin P. C. and Merlo J., “Intermediate and Advanced Topics in Multilevel Logistic Regression Analysis,” Statistics in Medicine 36 (2017): 3257–3277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Kojima T., Yamauchi Y., Fujiwara T., et al., “Clinical Impact of Specific Extraocular Muscle Manipulation and the Oculocardiac Reflex on Postoperative Vomiting in Pediatric Strabismus Surgery: A Multicenter, Observational Study,” Paediatric Anaesthesia 35 (2025): 163–174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Schneeweiss S., “Sensitivity Analysis and External Adjustment for Unmeasured Confounders in Epidemiologic Database Studies of Therapeutics,” Pharmacoepidemiology and Drug Safety 15 (2006): 291–303. [DOI] [PubMed] [Google Scholar]
- 18. Eberhart L. H. J., Geldner G., Kranke P., et al., “The Development and Validation of a Risk Score to Predict the Probability of Postoperative Vomiting in Pediatric Patients,” Anesthesia and Analgesia 99 (2004): 1630–1637. [DOI] [PubMed] [Google Scholar]
- 19. Mathes T., Röding D., Stegbauer C., Laxy M., and Pieper D., “Interrupted Time Series for Assessing the Causality of Intervention Effects. Part 35 of a Series on Evaluating Scientific Publications,” Deutsches Ärzteblatt International 121 (2024): 800–804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Linden A. and Adams J. L., “Applying a Propensity Score‐Based Weighting Model to Interrupted Time Series Data: Improving Causal Inference in Programme Evaluation,” Journal of Evaluation in Clinical Practice 17 (2011): 1231–1238. [DOI] [PubMed] [Google Scholar]
- 21. Splinter W. M., “Prevention of Vomiting After Strabismus Surgery in Children: Dexamethasone Alone Versus Dexamethasone Plus Low‐Dose Ondansetron,” Paediatric Anaesthesia 11 (2001): 591–595. [DOI] [PubMed] [Google Scholar]
- 22. Shen Y. D., Chen C. Y., Wu C. H., Cherng Y. G., and Tam K. W., “Dexamethasone, Ondansetron, and Their Combination and Postoperative Nausea and Vomiting in Children Undergoing Strabismus Surgery: A Meta‐Analysis of Randomized Controlled Trials,” Paediatric Anaesthesia 24 (2014): 490–498. [DOI] [PubMed] [Google Scholar]
- 23. Obara S., Kojima T., Yamauchi Y., Fujiwara T., Sueda A., and Takahashi R., “Cost‐Effectiveness Analysis of Anaesthesia Regimens for Paeditrc Strabismus Surgery Based on a Multicentre Retrospective Cohort Data From Japan,” BJA Open 14 (2025): 100404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Portnoy Y., Glebov M., Orkin D., Katsin M., and Berkenstadt H., “Postoperative Nausea and Vomiting in Pediatrics: Incidence and Guideline Adherence‐A Retrospective Cohort Study,” Anesthesia and Analgesia 141 (2025): 77–85. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1. Types of anesthetics used for anesthesia maintenance.
Figure S2. Types of antiemetic adjuncts used during general anesthesia.
Figure S3. Types of opioids used during general anesthesia.
Text S1. Supporting Information.
Text S2. Supporting Information.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.