Abstract
Background
Intranasal dexmedetomidine is widely used in pediatric procedural sedation; however, its dose-related effect on delayed awakening remains insufficiently characterized.
Methods
This single-center retrospective cohort study included pediatric patients who received oral midazolam combined with intranasal dexmedetomidine for procedural sedation at the Sedation Center of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital between March 2022 and March 2024. The exposure was the intranasal dexmedetomidine dose (μg/kg). The primary outcome was delayed awakening, defined as a time of ≥ 90 min from the last sedative administration (including rescue/secondary sedatives when applicable) to full awakening documented by the anesthesiologist on duty using routinely recorded standardized criteria (Modified Aldrete score ≥ 9). Firth penalized multivariable logistic regression models were applied to evaluate the association between dexmedetomidine dose and delayed awakening, adjusting for age group, sex, weight, fasting duration, procedure type, and midazolam dose. Dose–response relationships and subgroup analyses restricted to baseline/procedural characteristics were also conducted.
Results
A total of 2,116 children were included, with an overall delayed awakening incidence of 2.6%. When stratified by tertiles of dexmedetomidine dose, the incidence of delayed awakening increased from 1.1% in the low-dose group to 2.9% in the medium-dose group and 4.1% in the high-dose group (P = 0.004). In multivariable analyses, each 1 μg/kg increase in intranasal dexmedetomidine dose was associated with a significantly higher risk of delayed awakening (adjusted odds ratio [OR] 3.665, 95% confidence interval [CI] 2.091–6.636; P < 0.001). Dose–response analysis demonstrated a positive linear association (no evidence of nonlinearity, P = 0.451) between dexmedetomidine dose and delayed awakening. This corresponds to an aOR of 1.91 per 0.5 μg/kg increase (and 1.99 per IQR increase); adjusted predicted probabilities were 0.96% at 2 μg/kg, 3.37% at 3 μg/kg, and 10.70% at 4 μg/kg.
Conclusions
In pediatric procedural sedation, higher intranasal dexmedetomidine dose is associated with delayed awakening and demonstrates a positive dose–response relationship. Younger children appear to be more susceptible to this adverse outcome. Clinically, dexmedetomidine dosing should be carefully individualized—particularly in younger patients—to balance effective sedation with recovery efficiency.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12871-026-03876-y.
Keywords: Dexmedetomidine, Intranasal administration, Pediatric sedation, Delayed awakening, Retrospective cohort study
Background
As a highly selective α₂-adrenergic receptor agonist, dexmedetomidine has been increasingly used in pediatric sedation in recent years, particularly in regimens combining noninvasive administration routes (e.g., intranasal) with benzodiazepines, demonstrating favorable sedative efficacy and safety [1, 2]. It provides sedation, anxiolysis, and analgesia without significant respiratory depression, and is therefore widely used for sedation management during outpatient diagnostic and therapeutic procedures such as magnetic resonance imaging (MRI) and auditory brainstem response (ABR) testing [2, 3]. However, the pharmacologic properties of dexmedetomidine may also lead to adverse effects such as delayed awakening. Multiple studies have shown that dexmedetomidine can prolong postoperative recovery time, including time to awakening and length of stay in the post-anesthesia care unit (PACU) [4–6]. For example, Joshi et al. found in pediatric cardiac catheterization that children receiving a dexmedetomidine–ketamine regimen had a significantly longer awakening time than those in the propofol–ketamine group (40.88 vs. 22.28 min) [5]; Dai et al. likewise reported that the intravenous dexmedetomidine group had a markedly longer awakening time than the intranasal group and the control group [4]. This evidence suggests that dexmedetomidine dose, route of administration, and its combination with other sedatives may have important effects on the awakening process.
In clinical practice, "oral midazolam + intranasal dexmedetomidine" has become a common pediatric sedation combination regimen, intended to balance rapid onset with smooth recovery [3, 7]. Midazolam has a rapid onset but may increase the risk of emergence agitation; dexmedetomidine has a slower onset yet can effectively reduce the incidence of emergence delirium [8, 9]. However, whether this combination is associated with clinically meaningful delayed awakening has not been systematically evaluated using large-sample real-world data. Although dexmedetomidine has clear advantages in preventing emergence delirium [10, 11], its potential to delay recovery may affect outpatient workflow efficiency and healthcare resource utilization. Recent real-world evidence has also reported associations between dexmedetomidine use and postoperative recovery and cost outcomes in children [6]. In addition, delayed awakening not only extends patients'time in the hospital but may also increase nursing workload and caregiver anxiety [12]. Therefore, while pursuing high-quality sedation, its potential impact on recovery efficiency should be weighed. Existing studies have mostly focused on single agents or specific procedure types (e.g., tonsillectomy, dental treatment). Epidemiologic evidence remains insufficient regarding the relationship between dexmedetomidine dose (in μg/kg) and delayed awakening in pediatric patients undergoing a variety of outpatient diagnostic and therapeutic procedures with the"oral midazolam + intranasal dexmedetomidine"regimen.
Accordingly, based on a single-center retrospective cohort dataset from 2,116 children who received this combined sedation regimen at the Sedation Center of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital between March 2022 and March 2024, this study aims to quantify the adjusted association between intranasal dexmedetomidine dose (μg/kg) and delayed awakening, while accounting for key baseline/procedural covariates (including age, sex, weight, fasting time, procedure type, and midazolam dose). This study is expected to provide evidence to optimize pediatric outpatient sedation regimens, support individualized adjustment of dexmedetomidine dosing, and reduce the risk of delayed awakening, thereby improving healthcare service efficiency while ensuring sedation safety and quality.
Methods
Study population
This study was a single-center retrospective cohort study based on electronic medical records (EMR) of sedation encounters occurring between March 2022 and March 2024 at the Sedation Center of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital. The study population included all pediatric patients who received the sedation regimen of"oral midazolam combined with intranasal dexmedetomidine"at this center during the study period. The study was approved by the Ethics Committee of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital [Approval No.: (2024–3)166], and the requirement for study-specific informed consent was waived due to the retrospective design and the use of de-identified data. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and complied with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Inclusion criteria were: age ≤ 12 years, receipt of the above sedation regimen, and complete records of exposure and outcome variables. Exclusion criteria included: allergy to contrast agents during the examination, missing information on dexmedetomidine dose or the outcome of delayed awakening, and severely incomplete clinical data preventing confirmation of key covariates. All data were obtained from the hospital electronic medical record system and independently extracted by two trained researchers. A standardized form was used to collect variables including demographic characteristics, medication information, procedure type, and time points, to ensure data consistency and completeness.
Between March 2022 and March 2024, 2,548 pediatric patients who received outpatient sedation at our center were initially screened. According to the predefined inclusion and exclusion criteria, 432 patients were excluded for the following reasons: (1) 12 patients did not meet the inclusion criteria because they were older than 12 years; (2) 4 patients developed allergic reactions to contrast agents during the examination, which interrupted the sedation and observation process; and (3) 416 patients were excluded due to incomplete clinical data, primarily missing key fields required to define exposure and/or the recovery interval (dexmedetomidine dose and/or timestamps for sedative administration and full awakening). Ultimately, 2,116 eligible children were included in the data analysis (see Fig. 1 for the screening flowchart).
Fig. 1.
Study flow diagram
Variables
The exposure variable was the intranasal dexmedetomidine dose (μg/kg), calculated as a continuous variable based on the actual administered dose recorded in the clinical documentation and the child's body weight.
The outcome variable was delayed awakening (0 = no, 1 = yes), defined as a time of ≥ 90 min from the last sedative administration to full awakening. The 90-min cutoff reflects our center's standardized post-sedation observation workflow, where failure to meet discharge readiness criteria by 90 min triggers extended observation and additional resource utilization. "Last sedative administration" was defined as the last sedative dose recorded in the EMR, including any rescue sedatives if administered after the initial regimen. Full awakening was operationalized as a Modified Aldrete score ≥ 9, which is the routine discharge readiness criterion in our center. This status was assessed and documented by the duty anesthesiologist in the medical progress notes. Children were monitored according to a standardized institutional protocol, including continuous SpO2 and ECG monitoring, with non-invasive blood pressure and consciousness assessments recorded every 15 min. Awakening was assessed by calling the child's name or applying gentle tactile stimulation.
Covariates included age (categorized as ≤ 1 year, > 1–3 years, and > 3–12 years), sex, weight (kg), fasting duration (hours), type of medical procedure (MRI, CT, auditory brainstem response test, superficial mass excision, and others), and midazolam dose (mg/kg). "Rescue sedation" was defined as the administration of rescue sedatives (e.g., intravenous propofol or inhalational sevoflurane) after the initial regimen when sedation was inadequate, as documented in the EMR. Variables such as time to fall asleep, awakening time, and total length of stay were treated as post-exposure process measures and were therefore not used as primary confounders in the main adjusted model. Awakening time (min) was defined as the continuous interval from the last sedative administration (including rescue/secondary sedatives when applicable) to the first documentation of full awakening (Modified Aldrete score ≥ 9). This variable is the continuous counterpart of the primary binary outcome, where delayed awakening was defined as awakening time ≥ 90 min.
Patients with incomplete clinical records were excluded during screening; therefore, the final analytical cohort had 0% missingness for all analyzed variables.
Statistical analysis
According to the distributional characteristics of the data, continuous variables are presented as mean ± standard deviation (Mean ± SD) or median (interquartile range) [Median (IQR)], and categorical variables are described as frequency (percentage) [n (%)]. For between-group comparisons, one-way analysis of variance (ANOVA) was used for continuous variables that met assumptions of normality and homogeneity of variance, and F values were reported;otherwise, nonparametric methods were used, with medians and interquartile ranges reported for non-normally distributed data. For categorical variables, comparisons across three or more exposure levels were performed using the chi-square test (χ2 test), and χ2 values were reported; comparisons between two groups used the independent-samples t test or chi-square test, depending on variable type. All tests were two-sided, with the significance level set at α = 0.05.
Given the low outcome incidence (56/2,116), the primary analyses used Firth penalized logistic regression to evaluate the association between dexmedetomidine dose (μg/kg) and delayed awakening (binary outcome: 0 = no, 1 = yes). Univariable logistic regression was used to calculate crude odds ratios (ORs) and 95% confidence intervals (CIs) for each covariate. The primary multivariable model estimated the adjusted OR and 95% CI for dexmedetomidine dose after adjusting for baseline/procedural covariates: age, sex, weight, fasting duration, type of medical procedure, and midazolam dose. Post-exposure process measures (time to fall asleep, awakening time, and total length of stay) were not included in the primary adjustment set to avoid potential overadjustment/collider bias.
Sensitivity analyses were conducted by (1) additionally adjusting for rescue sedation, (2) restricting the analysis to encounters without rescue sedation, and (3) restricting the analysis to the MRI-only subgroup to minimize confounding by procedure type.
To explore potential effect modification, subgroup analyses were restricted to baseline/procedural variables (e.g., age group, sex, weight group, fasting group, procedure category, and midazolam dose category). Within each stratum, regression models were refit to estimate the association between dexmedetomidine and delayed awakening, and interaction P values were calculated using the likelihood ratio test to determine whether effects differed across subgroups. Strata with sparse/zero outcomes were reported as not estimable and were not interpreted inferentially.
To examine whether a nonlinear dose–response relationship existed between dexmedetomidine and delayed awakening, a generalized additive model (GAM) was constructed using cubic B-spline basis functions (4 basis functions, degree = 3), fitted via iteratively reweighted least squares (IRLS), with adjustment for age, sex, weight, fasting duration, procedure type, and midazolam dose. The significance of the smooth term was evaluated using the Wald test and the likelihood ratio test.
All hypothesis tests were two-sided, and P < 0.05 was considered statistically significant. Statistical analyses were performed using R software (version 4.3.1). When the number of outcome events was 0 in certain categorical levels, model estimates could be unstable; related results are presented descriptively only and are not interpreted inferentially.
Results
Study population and data completeness
A total of 2,116 pediatric patients who received the"oral midazolam combined with intranasal dexmedetomidine"sedation regimen at the Sedation Center of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital between March 2022 and March 2024 were included. Patients with incomplete documentation of key fields required to define the exposure and/or recovery interval (dose and/or timestamps) were excluded during screening; therefore, the final analytical cohort had 0% missingness for all analyzed variables (Supplementary Table S1).
Incidence of delayed awakening and baseline characteristics by dose tertiles (Table 1)
Table 1.
Baseline characteristics stratified by intranasal dexmedetomidine dose tertiles (μg/kg)
| Variable | Overall | Low dose (0.30–2.73) | Medium dose (> 2.73–3.00) | High dose (> 3.00–4.00) | Test | P value |
|---|---|---|---|---|---|---|
| Sample size | 2116 | 708 | 819 | 589 | – | – |
| Delayed awakening (1 = yes, 0 = no) | χ2 = 11.239 | 0.004 | ||||
| 0 | 2060 (97.4%) | 700 (98.9%) | 795 (97.1%) | 565 (95.9%) | – | – |
| 1 | 56 (2.6%) | 8 (1.1%) | 24 (2.9%) | 24 (4.1%) | – | – |
| Age (≤ 1y/> 1–3y/> 3–12y) | χ2 = 188.801 | < 0.001 | ||||
| ≤ 1 year | 854 (40.4%) | 432 (61.0%) | 245 (29.9%) | 177 (30.1%) | – | – |
| > 1–3 years | 660 (31.2%) | 145 (20.5%) | 297 (36.3%) | 218 (37.0%) | – | – |
| > 3–12 years | 602 (28.4%) | 131 (18.5%) | 277 (33.8%) | 194 (32.9%) | – | – |
| Sex (male/female) | χ2 = 11.329 | 0.003 | ||||
| Male | 1425 (67.3%) | 443 (62.6%) | 576 (70.3%) | 406 (68.9%) | – | – |
| Female | 691 (32.7%) | 265 (37.4%) | 243 (29.7%) | 183 (31.1%) | – | – |
| Weight (kg) | 12.06 ± 5.60 | 8.00 (6.00–13.50) | 12.90 ± 5.11 | 12.80 ± 4.70 | F = 44.207 | < 0.001 |
| Fasting duration (h) | 1.00 (0.50–2.00) | 1.00 (0.50–2.00) | 1.00 (0.50–2.00) | 1.00 (0.50–2.00) | F = 3.656 | 0.026 |
| Medical procedure | χ2 = 42.563 | < 0.001 | ||||
| MRI | 1743 (82.4%) | 533 (75.3%) | 696 (85.0%) | 514 (87.3%) | – | – |
| CT | 302 (14.3%) | 145 (20.5%) | 99 (12.1%) | 58 (9.8%) | – | – |
| ABR testing | 30 (1.4%) | 14 (2.0%) | 10 (1.2%) | 6 (1.0%) | – | – |
| Excision of superficial masses | 33 (1.6%) | 11 (1.6%) | 13 (1.6%) | 9 (1.5%) | – | – |
| Other | 8 (0.4%) | 5 (0.7%) | 1 (0.1%) | 2 (0.3%) | – | – |
| Midazolam (mg/kg) | 0.46 ± 0.12 | 0.46 ± 0.13 | 0.45 ± 0.11 | 0.46 ± 0.13 | F = 0.212 | 0.809 |
| Rescue sedation (yes/no) | χ2 = 8.486 | 0.014 | ||||
| No | 1996 (94.3%) | 666 (94.1%) | 786 (96.0%) | 544 (92.4%) | – | – |
| Yes | 120 (5.7%) | 42 (5.9%) | 33 (4.0%) | 45 (7.6%) | – | – |
| Time to fall asleep (min)* | 20.35 ± 7.94 | 19.96 ± 8.72 | 20.27 ± 7.14 | 20.93 ± 8.01 | F = 2.448 | 0.087 |
| Awakening time (min)* | 80.83 ± 29.58 | 78.05 ± 32.31 | 81.90 ± 28.01 | 82.67 ± 28.04 | F = 4.814 | 0.008 |
| Time spent in hospital (min)* | 111.60 ± 33.38 | 109.52 ± 36.53 | 112.37 ± 31.91 | 113.03 ± 31.26 | F = 2.139 | 0.118 |
Continuous variables are presented as mean ± SD when mean > 2 × SD, otherwise median (IQR); categorical variables as n (%). Between-group comparisons used ANOVA (F) for continuous variables and χ2 tests for categorical variables; two-sided P values are shown
*Post-exposure process measures reported descriptively only and excluded from the primary adjustment set
Based on tertiles of the intranasal dexmedetomidine dose (μg/kg), the study population was divided into a low-dose group (0.30–2.73 μg/kg, n = 708), a medium-dose group (2.73–3.00 μg/kg, n = 819), and a high-dose group (> 3.00–4.00 μg/kg, n = 589). The overall incidence of delayed awakening—defined as time from the last sedative administration (including rescue sedatives when applicable) to full awakening ≥ 90 min, with full awakening operationalized as Modified Aldrete score ≥ 9 as documented by the anesthesiologist on duty—was 2.6% (56/2,116). By dose group, the incidence was 1.1% (8/708), 2.9% (24/819), and 4.1% (24/589), respectively, with a statistically significant difference among groups (χ2 = 11.239, P = 0.004) (Table 1).
Significant between-group differences were observed in age distribution (χ2 = 188.801, P < 0.001), sex composition (χ2 = 11.329, P = 0.003), weight (F = 44.207, P < 0.001), fasting duration (F = 3.656, P = 0.026), procedure type (χ2 = 42.563, P < 0.001), and the proportion receiving rescue sedation (χ2 = 8.486, P = 0.014). Notably, the proportion of infants ≤ 1 year was lower in the high-dose group (30.1%) than in the low-dose group (61.0%), whereas the proportion undergoing MRI was higher in the high-dose group (87.3%) than in the low-dose group (75.3%) (Table 1). Time-to-sleep, awakening time, and time spent in hospital were treated as post-exposure process measures and are reported descriptively only; they were not included in the primary adjustment set.
Univariable associations (Table 2)
Table 2.
Univariable logistic regression analysis (Outcome: delayed awakening)
| Variable | OR | 95% CI | P value |
|---|---|---|---|
| Dexmedetomidine dose (per 1 μg/kg) | 3.365 | (1.897, 5.970) | < 0.001 |
| Age group (ref: ≤ 1 year) | |||
| < 1–3 years | 0.737 | (0.415, 1.309) | 0.298 |
| < 3–12 years | 0.166 | (0.059, 0.472) | < 0.001 |
| Sex (ref: male) | |||
| Female | 0.749 | (0.412, 1.363) | 0.344 |
| Weight (per 1 kg) | 0.961 | (0.912, 1.012) | 0.130 |
| Fasting duration (per 1 h) | 0.932 | (0.686, 1.266) | 0.651 |
| Medical procedure (ref: MRI) | |||
| CT | 0.217 | (0.053, 0.895) | 0.035 |
| ABR testing | 2.323 | (0.539, 10.011) | 0.258 |
| Excision of superficial masses | NE | NE | NE |
| Other | NE | NE | NE |
| Midazolam dose (per 0.1 mg/kg increase) | 1.245 | (1.009, 1.526) | 0.041 |
Univariable logistic regression with delayed awakening (1 = yes, 0 = no) as outcome. OR Odds ratio, CI Confidence interval. NE indicates not estimable due to sparse/zero outcome events in that category (separation)
In univariable logistic regression, each 1 μg/kg increase in dexmedetomidine dose was associated with higher odds of delayed awakening (OR = 3.365, 95% CI 1.897–5.970; P < 0.001). Age 3–12 years (vs ≤ 1 year) was associated with lower odds of delayed awakening (OR = 0.166, 95% CI 0.059–0.472; P < 0.001), and CT (vs MRI) was also associated with lower odds (OR = 0.217, 95% CI 0.053–0.895; P = 0.035). Midazolam dose (per 0.1 mg/kg increase) was associated with increased odds (OR = 1.245, 95% CI 1.009, 1.526; P = 0.041) (Table 2).
Primary multivariable analysis (Firth penalized logistic regression) and clinical interpretability (Tables 3, 4 and 5)
Table 3.
Primary multivariable analysis (Firth penalized logistic regression)
| Covariate | aOR | 95% CI | P value |
|---|---|---|---|
| Dexmedetomidine dose (per 1 μg/kg) | 3.665 | (2.091, 6.636) | < 0.001 |
| Age group (ref: ≤ 1 year) | |||
| < 1–3 years | 0.257 | (0.124, 0.536) | < 0.001 |
| < 3–12 years | 0.034 | (0.007, 0.140) | < 0.001 |
| Sex (ref: male) | |||
| Female | 0.800 | (0.424, 1.436) | 0.464 |
| Weight (per 1 kg) | 1.117 | (1.018, 1.212) | 0.021 |
| Fasting duration (per 1 h) | 0.964 | (0.679, 1.279) | 0.821 |
| Medical procedure (ref: MRI) | |||
| CT | 0.211 | (0.037, 0.687) | 0.006 |
| ABR testing | 3.537 | (0.671, 12.304) | 0.121 |
| Excision of superficial masses | 0.751 | (0.006, 5.754) | 0.836 |
| Other | 2.580 | (0.019, 24.971) | 0.577 |
| Midazolam dose (per 0.1 mg/kg) | 1.070 | (0.981, 1.146) | 0.099 |
Primary model fitted using Firth penalized logistic regression due to low event count. Adjusted for baseline/procedural covariates: age group, sex, weight, fasting duration, procedure type, and midazolam dose. Reference categories: age ≤ 1 year; male; MRI
Table 4.
Clinically interpretable dose units and sensitivity analyses (dose effect)
| Model | OR per 1 μg/kg | OR per 0.5 μg/kg | OR per IQR |
|---|---|---|---|
| Primary_Firth | 3.665 | 1.914 | 1.991 |
| Sensitivity 1: add rescue sedation covariate | 3.429 | 1.852 | 1.922 |
| Sensitivity 2: restrict to no rescue sedation | 2.954 | 1.719 | 1.776 |
ORs are for intranasal dexmedetomidine dose, rescaled to common clinical units. Primary model adjusts for baseline/procedural covariates; sensitivity models additionally account for rescue sedation use (as specified) while retaining the same baseline/procedural covariates
Table 5.
Adjusted predicted probabilities of delayed awakening at representative doses
| Model | Dose (μg/kg) | Predicted probability |
|---|---|---|
| Primary_Firth | 2 | 0.0096 (0.96%) |
| Primary_Firth | 3 | 0.0337 (3.37%) |
| Primary_Firth | 4 | 0.1070 (10.70%) |
| Sensitivity 1: add rescue sedation covariate | 2 | 0.0104 (1.04%) |
| Sensitivity 1: add rescue sedation covariate | 3 | 0.0338 (3.38%) |
| Sensitivity 1: add rescue sedation covariate | 4 | 0.1011 (10.11%) |
| Sensitivity 2: restrict to no rescue sedation | 2 | 0.0113 (1.13%) |
| Sensitivity 2: restrict to no rescue sedation | 3 | 0.0320 (3.20%) |
| Sensitivity 2: restrict to no rescue sedation | 4 | 0.0853 (8.53%) |
Predicted probabilities are model-based adjusted estimates at representative doses
Because only 56 delayed-awakening events occurred, the primary analysis used Firth penalized logistic regression to reduce small-sample bias and improve estimate stability. The primary adjustment set included only prespecified baseline/procedural covariates: age group, sex, weight, fasting duration, procedure type, and midazolam dose.
In the primary model, each 1 μg/kg increase in dexmedetomidine dose was associated with a 3.66-fold higher odds of delayed awakening (aOR = 3.665, 95% CI 2.091–6.636; P < 0.001) (Table 3). This corresponds to an OR of 1.91 per 0.5 μg/kg increase and 1.99 per interquartile-range (IQR) increase (Table 4). Adjusted predicted probabilities of delayed awakening were 0.96% at 2 μg/kg, 3.37% at 3 μg/kg, and 10.70% at 4 μg/kg (Table 5). Given the marked baseline/procedural imbalances across dose tertiles suggestive of protocol-driven dosing, these results are interpreted as adjusted associations, and residual confounding by indication may remain.
Sensitivity analyses for rescue sedation and procedure type(Tables 4 and 5; Supplementary Table S2)
Two sensitivity analyses were conducted to assess robustness with respect to rescue sedation: (1) adding rescue sedation as a covariate; and (2) restricting the analysis to patients without rescue sedation. Both sensitivity estimates were directionally consistent with the primary analysis (Table 4). Predicted probabilities at representative doses (2, 3, 4 μg/kg) were similar across models (Table 5).
Additionally, to address potential confounding by procedure type (given the unequal distribution of MRI/CT across dose groups), a sensitivity analysis restricted to patients undergoing MRI (n = 1,743) was conducted using the same multivariable model. The association remained robust and directionally consistent (adjusted OR = 3.958, 95% CI 2.185–7.443; P < 0.001) (Supplementary Table S2).
Subgroup analyses restricted to baseline/procedural characteristics (Table 6)
Table 6.
Subgroup analyses (baseline/procedural strata only; dose effect within each stratum)
| Stratification variable | Stratum | n | Events | OR (per 1 μg/kg) | 95% CI | P value | P for interaction |
|---|---|---|---|---|---|---|---|
| Age group | ≤ 1 year | 854 | 33 | 4.396 | (2.033, 10.356) | < 0.001 | 0.255 |
| < 1–3 years | 660 | 19 | 1.894 | (0.694, 5.147) | 0.213 | – | |
| < 3–12 years | 602 | 4 | 2.862 | (0.357, 29.313) | 0.371 | – | |
| Sex | Male | 1425 | 41 | 3.228 | (1.626, 6.705) | 0.001 | 0.373 |
| Female | 691 | 15 | 5.047 | (1.819, 15.586) | 0.001 | – | |
| Weight group | ≤ 12 kg | 1121 | 41 | 3.230 | (1.642, 6.662) | < 0.001 | 0.557 |
| > 12 kg | 995 | 15 | 4.042 | (1.048, 20.041) | 0.041 | – | |
| Fasting group | ≤ 1 h | 1397 | 40 | 2.541 | (1.352, 5.098) | 0.003 | 0.311 |
| > 1 h | 719 | 16 | 6.769 | (2.266, 22.718) | < 0.001 | – | |
| Procedure | MRI | 1743 | 52 | 3.958 | (2.185, 7.443) | < 0.001 | 0.613 |
| CT | 302 | 2 | 0.465 | (0.0003, 15.612) | 0.681 | – | |
| ABR testing | 30 | 2 | 1.018 | (0.144, 12.413) | 0.985 | – | |
| Excision of superficial masses | 33 | 0 | NE | NE | NE | – | |
| Other | 8 | 0 | NE | NE | NE | – | |
| Midazolam (mg/kg) | Low ≤ 0.47 (0.19–0.47) | 1058 | 21 | 1.515 | (0.729, 3.447) | 0.277 | 0.002 |
| High > 0.47 (0.48–0.81) | 1058 | 35 | 9.356 | (3.972, 23.063) | < 0.001 | – |
Within each stratum, the association between dexmedetomidine dose (per 1 μg/kg) and delayed awakening was estimated using penalized logistic regression. NE indicates not estimable due to small n and/or 0 outcome events in that stratum
Subgroup analyses were restricted to baseline/procedural characteristics only (age group, sex, weight group, fasting group, procedure category, and midazolam group). In strata with sparse/zero events, estimates were not estimable (NE) (Table 6). The direction of association between dexmedetomidine dose and delayed awakening was consistent across subgroups. The association appeared more pronounced among infants ≤ 1 year (OR = 4.396, 95% CI 2.033–10.356; P < 0.001) and among children receiving a higher midazolam dose (OR = 9.356, 95% CI 3.972–23.063; P < 0.001) (Table 6). Subgroup analyses stratified by post-exposure process variables (time-to-sleep, awakening time, hospital stay duration) were removed.
Dose–response relationship (Fig. 2)
Fig. 2.
Dose–response relationship between intranasal dexmedetomidine dose and delayed awakening. The curve represents the adjusted predicted probability of delayed awakening estimated using a generalized additive model (GAM); shaded area indicates the 95% confidence interval. The model was adjusted for age group, sex, weight, fasting duration, procedure type, and midazolam dose. Nonlinearity was not observed (smooth-term Wald test P = 0.451; likelihood ratio test P = 0.451)
The nonlinear dose–response relationship was assessed using a generalized additive model (GAM). Nonlinearity was not observed (smooth-term Wald test P = 0.451; likelihood ratio test P = 0.451). Figure 2 displays the adjusted predicted probability of delayed awakening across the dexmedetomidine dose range, adjusted for age group, sex, weight, fasting duration, procedure type, and midazolam dose.
Discussion
This single-center retrospective cohort study examined the association between dexmedetomidine dose (μg/kg) and delayed awakening among children who received an"oral midazolam plus intranasal dexmedetomidine"regimen at the Sedation Center of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital. A total of 2,116 pediatric patients were included. In the primary analysis using Firth penalized multivariable logistic regression (given the low event count of 56/2,116), and adjusting for baseline/procedural covariates (age group, sex, weight, fasting duration, procedure type, and midazolam dose), each 1 μg/kg increase in intranasal dexmedetomidine dose was associated with higher odds of delayed awakening (adjusted OR = 3.66, 95% CI: 2.09–6.64; P < 0.001). For clinical interpretability, this corresponded to an adjusted OR of 1.91 per 0.5 μg/kg increase and 1.99 per IQR increase; adjusted predicted probabilities were 0.96% at 2 μg/kg, 3.37% at 3 μg/kg, and 10.70% at 4 μg/kg. We also observed that younger children had higher odds of delayed awakening than older children, consistent with age-related differences in dexmedetomidine pharmacokinetics.
Overall, our findings are broadly consistent with prior literature on the use of dexmedetomidine in pediatric sedation, but there are also several important differences. Multiple randomized controlled trials have confirmed that dexmedetomidine can effectively reduce emergence delirium and postoperative agitation [9, 10, 13], a mechanism attributed to its α2-adrenergic receptor agonism producing sedative, anxiolytic, and analgesic effects [12]. However, existing studies have mostly focused on recovery after general anesthesia, and often treat"delayed awakening"as a secondary outcome or adverse event. For example, in children undergoing tonsillectomy, intravenous dexmedetomidine has been shown to reduce opioid requirements and dose-dependently decrease emergence delirium (for each additional 0.5 μg/kg, the delirium risk ratio decreased by 87%) [14], but its effect on recovery time was not systematically evaluated. In contrast, a systematic review focusing on MRI sedation reported that dexmedetomidine alone is associated with longer recovery times compared to agents like propofol or midazolam, although combination regimens may mitigate this prolongation [15]. The uniqueness of our study lies in its focus on non-surgical procedural sedation (e.g., MRI, CT, ABR testing) using a combined regimen of oral midazolam plus intranasal dexmedetomidine, and in explicitly quantifying a positive association between dexmedetomidine dose and delayed awakening. Notably, in similar populations, prior research has reported that younger children (especially those aged 0–3 years) show higher sedation success rates after intranasal dexmedetomidine, but also experience longer recovery times [16]. A pharmacokinetic study by Li et al. confirmed that dexmedetomidine clearance in infants aged 3–36 months is significantly influenced by body weight and gestational-age maturity, suggesting that age-related pharmacokinetic differences (e.g., lower hepatic and renal clearance) may be a potential mechanism underlying the increased risk of delayed awakening [16]. In addition, a clinical trial for auditory brainstem response (ABR) testing showed that intranasal dexmedetomidine (3 μg/kg) combined with buccal midazolam (0.1 mg/kg) significantly improved sedation success compared with chloral hydrate alone (89.7% vs 69.5%) [17], but did not report recovery-time differences across age subgroups in detail. These findings align closely with our observation that younger children have a higher risk of delayed awakening.
This study has important clinical implications. First, it clarifies that dexmedetomidine dose is associated with delayed awakening in routine procedural sedation, providing quantitative evidence to support individualized dosing. Although delayed awakening rarely leads to severe complications, it can prolong observation time, increase consumption of medical resources, and cause parental anxiety [6]; therefore, optimizing dexmedetomidine dosing is crucial for improving sedation efficiency. Current evidence indicates that dexmedetomidine, whether used alone or in combination, can effectively increase sedation success and reduce emergence delirium [18], but its use is often accompanied by a notable prolongation of recovery time. For example, in MRI sedation, dexmedetomidine prolongs recovery time by an average of 5.22 min compared with propofol, ketamine, or midazolam [19]. In sedation for bone marrow aspiration in pediatric oncology patients, the dexmedetomidine pre-treatment group had a significantly longer awakening time (92 min) than the midazolam group (65 min) [20]. Collectively, these data suggest that the dose–recovery time relationship of dexmedetomidine should be incorporated into clinical decision-making frameworks. Second, our study helps fill a gap in the existing evidence: most current guidelines and systematic reviews emphasize dexmedetomidine's advantage in reducing emergence delirium [8, 18, 21], but pay less attention to its impact on awakening time in non-anesthetic sedation. Our data suggest that while pursuing high-quality sedation, clinicians must balance dexmedetomidine dose against recovery speed. From a public health perspective, as demand for noninvasive procedural sedation increases rapidly [2], establishing precise dosing strategies based on age and weight may improve health system efficiency. Pharmacokinetic studies have confirmed that dexmedetomidine clearance is significantly influenced by body weight and corrected gestational age [16], and that effective doses differ across pediatric age groups [22]. Future research should conduct prospective dose–response trials to identify optimal effective and safe dose windows for dexmedetomidine in different age groups, and evaluate interactive effects of combination therapy (e.g., with low-dose midazolam) on awakening time. Notably, although the dexmedetomidine–midazolam combination can improve sedation success [18], existing evidence indicates it does not significantly improve recovery time [18], further highlighting the need to optimize single-agent dosing strategies.
Methodologically, this study has several strengths. First, it is based on real-world electronic medical record data from a large sample (n = 2,116); after screening exclusions, the final analytical cohort had 0% missingness for analyzed variables. Second, both the exposure (dexmedetomidine dose) and outcome (delayed awakening) were defined in a standardized manner and recorded in routine care, improving measurement reliability. Specifically, full awakening was documented by the anesthesiologist on duty using routinely recorded standardized scores (Modified Aldrete score ≥ 9). Third, the statistical strategy was designed to improve interpretability and robustness: the primary model used Firth penalization to mitigate small-event bias, and dose–response modeling supported an approximately linear association within the studied dose range (P for nonlinearity = 0.451). Additionally, sensitivity analyses restricted to the MRI-only subgroup (comprising 82% of the cohort) yielded consistent results (aOR = 3.96), minimizing confounding by procedure type. Subgroup analyses were restricted to baseline/procedural characteristics to avoid post-exposure stratification.
Nevertheless, this study has several limitations. First, as a single-center retrospective cohort study, it cannot establish causality and may be subject to unmeasured confounding (e.g., severity of underlying conditions, operator experience). Marked baseline/procedural imbalances across dose groups suggest protocol-driven dosing; therefore, residual confounding by indication may remain, and results should be interpreted as associations. Second, a total of 416 patients were excluded due to incomplete documentation of key fields (dose and/or timestamps required to define the recovery interval), which may introduce selection bias if missingness is related to workflow or outcomes. Third, the low outcome incidence (2.6%, 56 events) may limit precision despite penalized regression. Fourth, while delayed awakening is assumed to impact resources and anxiety, these specific outcomes (e.g., costs, parental stress scores) were not directly measured in this retrospective design. Fifth, although outcome assessment used routinely recorded standardized scores, the data came from a single institution's sedation center, whose sedation regimen (fixed combination with midazolam) and operational workflow may limit generalizability to other settings (e.g., emergency departments or dental clinics) [1, 23]. Future multicenter prospective cohort studies or randomized controlled trials incorporating standardized recovery assessment tools and broader sedation regimens are needed to further validate these findings.
Conclusion
This retrospective cohort study indicates that among children receiving the"oral midazolam combined with intranasal dexmedetomidine"sedation regimen, the intranasal dexmedetomidine dose was associated with the risk of delayed awakening, demonstrating an approximately linear positive association within the dose range studied. Younger children appeared to have higher odds of delayed awakening than older children. Given that delayed awakening may prolong observation time and reduce outpatient care efficiency, these findings suggest that when implementing this combined sedation regimen, it is necessary to balance sedative efficacy with recovery time. In particular, higher doses of dexmedetomidine should be used cautiously in younger children, and more individualized dosing strategies should be explored.
Supplementary Information
Supplementary Material 1: Supplementary Table S1. Missingness by variable in the analytical cohort (n=2,116).
Supplementary Material 2: Supplementary Table S2. Sensitivity Analysis: Association Between Dexmedetomidine Dose and Delayed Awakening in MRI-Only Subgroup (Firth Penalized Logistic Regression, n=1,743).
Acknowledgements
Not applicable.
Abbreviations
- ABR
Auditory Brainstem Response
- aOR
Adjusted Odds Ratio
- CI
Confidence Interval
- CT
Computed Tomography
- EMR
Electronic Medical Record
- GAM
Generalized Additive Model
- IQR
Interquartile Range
- IRLS
Iteratively Reweighted Least Squares
- MRI
Magnetic Resonance Imaging
- NE
Not Estimable
- OR
Odds Ratio
- PACU
Post-Anesthesia Care Unit
- SD
Standard Deviation
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
Authors’ contributions
S J and L J contributed equally to this work, including the study design, data collection, and analysis. Y L assisted in drafting the manuscript and conducted the statistical analysis. J L participated in data collection and provided critical revisions to the manuscript. X S contributed to the supervision of the study and discussions regarding the results. Y L served as the corresponding author, overseeing the research process and providing guidance throughout the project. All authors reviewed the manuscript.
Funding
None.
Data availability
The datasets analyzed during the current study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
The study was approved by the Ethics Committee of Guangxi Zhuang Autonomous Region Maternal and Child Health Hospital [Approval No.: (2024–3)166], and the requirement for study-specific informed consent was waived due to the retrospective design and the use of de-identified data. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments and complied with the requirements of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Shunzhong Jing and Liping Jiang contributed equally to this work.
References
- 1.Fantacci C, Fabrizio GC, Ferrara P, Franceschi F, Chiaretti A. Intranasal drug administration for procedural sedation in children admitted to pediatric Emergency Room. Eur Rev Med Pharmacol Sci. 2018;22(1):217–22. 10.26355/eurrev_201801_14120. PMID: 29364490. [DOI] [PubMed] [Google Scholar]
- 2.Tsze DS, Rogers AP, Baier NM, Paquin JR, Majcina R, Phelps JR, et al. Clinical outcomes associated with intranasal dexmedetomidine sedation in children. Hosp Pediatr. 2023;13(3):223–43. 10.1542/hpeds.2022-007007. PMID: 36810939. [DOI] [PubMed] [Google Scholar]
- 3.Zhou X, Zhao J, Tu H, Chen K, Hu Y, Jin Y. The effect of age on outpatient pediatric procedural sedation with intranasal dexmedetomidine and oral midazolam. Eur J Pediatr. 2024;183(1):169–77. 10.1007/s00431-023-05240-5. Epub 2023 Oct 19. PMID: 37855928; PMCID: PMC10858144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Dai C, Zhao X, Li A, Zhang X, Wang P, Zhang Y, Wu Y. Comparative evaluation of intravenous versus intranasal dexmedetomidine on emergence delirium and hemodynamics in pediatric patients undergoing adenotonsillectomy: a randomized controlled trial. Front Pharmacol. 2025;16:1543344. 10.3389/fphar.2025.1543344. PMID: 39950111; PMCID: PMC11821972. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Joshi VS, Kollu SS, Sharma RM. Comparison of dexmedetomidine and ketamine versus propofol and ketamine for procedural sedation in children undergoing minor cardiac procedures in cardiac catheterization laboratory. Ann Card Anaesth. 2017;20(4):422–6. 10.4103/aca.ACA_16_17. PMID: 28994677; PMCID: PMC5661311. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Azimaraghi O, Rudolph MI, Luedeke CM, Ramishvili T, Jaconia GD, Scheffenbichler FT, et al. Association of dexmedetomidine use with haemodynamics, postoperative recovery, and cost in paediatric anaesthesia: a hospital registry study. Br J Anaesth. 2024;133(2):326–33. 10.1016/j.bja.2024.03.016. Epub 2024 Apr 17. PMID: 38631942. [DOI] [PubMed] [Google Scholar]
- 7.Chatrath V, Kumar R, Sachdeva U, Thakur M. Intranasal fentanyl, midazolam and dexmedetomidine as premedication in pediatric patients. Anesth Essays Res. 2018;12(3):748–53. 10.4103/aer.AER_97_18. PMID: 30283188; PMCID: PMC6157221. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ericksen AM, Kane TD, Tubog TD. Intranasal dexmedetomidine compared with intranasal midazolam for pediatric emergence delirium: a systematic review and meta-analysis. J Perianesth Nurs. 2025;40(5):1375-1385.e2. 10.1016/j.jopan.2025.01.011. Epub 2025 Apr 25. PMID: 40278813. [DOI] [PubMed] [Google Scholar]
- 9.Fu SH, Miao MR, Zhang LY, Bian J, Fu YX, Zhang JQ, et al. Prophylactic application of dexmedetomidine reduces the incidence of emergence delirium in children: a systematic review and meta-analysis. Anaesth Crit Care Pain Med. 2024;43(6):101426. 10.1016/j.accpm.2024.101426. Epub 2024 Sep 16. PMID: 39293539. [DOI] [PubMed] [Google Scholar]
- 10.Wu H, Wu P, Xiang L, Huang Q, Xiang Y, Zhang J, Zhao Z, Xu T. Effect of different intranasal dexmedetomidine doses on pediatric postoperative delirium and agitation: network meta-analysis. Pediatr Res. 2025;98(4):1266–75. 10.1038/s41390-025-03851-2. Epub 2025 Mar 13. PMID: 40082658. [DOI] [PubMed] [Google Scholar]
- 11.Manning AN, Bezzo LK, Hobson JK, Zoeller JE, Brown CA, Henderson KJ. Dexmedetomidine dosing to prevent pediatric emergence delirium. AANA J. 2020;88(5):359–64 PMID: 32990204. [PubMed] [Google Scholar]
- 12.Moore AD, Anghelescu DL. Emergence delirium in pediatric anesthesia. Paediatr Drugs. 2017;19(1):11–20. 10.1007/s40272-016-0201-5. Erratum in: Paediatr Drugs. 2017 Jun;19(3):267. 10.1007/s40272-017-0227-3. [DOI] [PubMed] [Google Scholar]
- 13.Yahya CQ, Hariyanto H, Santoso KH, Andriyanto L, Utariani A, Semedi BP, et al. Emergence agitation in pediatrics after dexmedetomidine vs. sevoflurane anesthesia: a randomized controlled trial. PLoS One. 2025;20(10):e0333576. 10.1371/journal.pone.0333576. PMID: 41061010; PMCID: PMC12507210. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Olsen JM, Sumie M, Yang A, Englesakis M, Niimi N, Campisi P, et al. Dexmedetomidine in pediatric tonsillectomy: a systematic review with meta-analysis. Can J Anaesth. 2025;72(7):1066–78. 10.1007/s12630-025-02991-7. English. Epub 2025 Jul 11. PMID: 40646381. [DOI] [PubMed] [Google Scholar]
- 15.Vitraludyono R, Utariani A, Hanindito E. Comparison of dexmedetomidine alone or with other sedatives for paediatric sedation during magnetic resonance imaging: a systematic review. Med Glas (Zenica). 2023;20(1). 10.17392/1532-22. Epub ahead of print. PMID: 36435998. [DOI] [PubMed]
- 16.Li BL, Guan YP, Yuen VM, Wei W, Huang M, Zhang MZ, Li AW, Standing JF, Zhong GP, Song XR. Population pharmacokinetics of intranasal dexmedetomidine in infants and young children. Anesthesiology. 2022;137(2):163–75. 10.1097/ALN.0000000000004258. PMID: 35503980. [DOI] [PubMed] [Google Scholar]
- 17.Li BL, Yuen VM, Zhou JL, Zhang N, Huang JX, Tian H, Song XR. A randomized controlled trial of oral chloral hydrate vs intranasal dexmedetomidine plus buccal midazolam for auditory brainstem response testing in children. Paediatr Anaesth. 2018;28(11):1022–8. 10.1111/pan.13498. Epub 2018 Oct 3. PMID: 30281180. [DOI] [PubMed] [Google Scholar]
- 18.Nie J, Li C, Yang G, Chang H, Ding G. An evaluation of dexmedetomidine in combination with midazolam in pediatric sedation: a systematic review and meta-analysis. BMC Anesthesiol. 2024;24(1):210. 10.1186/s12871-024-02570-1. PMID: 38907338; PMCID: PMC11191149. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Kim JY, Kim KN, Kim DW, Lim HJ, Lee BS. Effects of dexmedetomidine sedation for magnetic resonance imaging in children: a systematic review and meta-analysis. J Anesth. 2021;35(4):525–35. 10.1007/s00540-021-02946-4. Epub 2021 May 18. PMID: 34002258. [DOI] [PubMed] [Google Scholar]
- 20.Squillaci D, Console K, Colussi L, Kiren V, Rabusin M, Stocco G, et al. Dexmedetomidine versus midazolam for propofol sparing in procedural sedation of children with leukemia: a consecutive case series. Acta Anaesthesiol Scand. 2025;69(8):e70107. 10.1111/aas.70107. PMID: 40740101; PMCID: PMC12311750. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Barot G, Patel M, Patel C, Fernandes MC, Patel F, Mehta M. Comparative effectiveness and safety of dexmedetomidine and midazolam in pediatric dental sedation: a systematic review and meta-analysis. J Dent Anesth Pain Med. 2025;25(3):147–59. 10.17245/jdapm.2025.25.3.147. Epub 2025 May 30. PMID: 40521425; PMCID: PMC12160033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Qiu L, Cao L, Lang Z, Li X, Lin H, Fan T. Preoperative sedation in children with congenital heart disease: 50% and 95% effective doses, hemodynamic effects, and safety of intranasal dexmedetomidine. J Clin Anesth. 2022;81:110908. 10.1016/j.jclinane.2022.110908. Epub 2022 Jun 29. Erratum in: J Clin Anesth. 2022 Nov;82:110936. 10.1016/j.jclinane.2022.110936. [DOI] [PubMed] [Google Scholar]
- 23.Mulcrone AE, Camp EA, Frierson E, Koukaz Y, Chumpitazi CE. Minimal sedation use for laceration repair in the pediatric emergency department. J Emerg Med. 2025;75:1–13. 10.1016/j.jemermed.2025.03.007. Epub 2025 Mar 19. PMID: 40561621. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1: Supplementary Table S1. Missingness by variable in the analytical cohort (n=2,116).
Supplementary Material 2: Supplementary Table S2. Sensitivity Analysis: Association Between Dexmedetomidine Dose and Delayed Awakening in MRI-Only Subgroup (Firth Penalized Logistic Regression, n=1,743).
Data Availability Statement
The datasets analyzed during the current study are available from the corresponding author upon reasonable request.