Background

The intranasal route is an effective means of administering sedatives to children who require procedural sedation. Intranasal administration delivers the sedative to the highly vascularized nasal mucosa and the olfactory tissue that is in direct contact with the central nervous system (termed the “nose-brain pathway”), thereby bypassing first-pass metabolism. Although there are commercial products designed specifically for intranasal drug delivery, it is common practice in emergency departments (EDs) to use parenteral formulations of sedatives for intranasal administration. However, commonly available concentrations of these sedatives often necessitate large total volumes to deliver the required weight-based dose in children. In the clinical setting, this total volume is often divided into smaller aliquots, or volumes of administration, that are administered repeatedly by alternating between both nostrils until the total volume is delivered.

Importance

Although the volume of administration is a fundamental aspect of intranasal administration, it is unclear what volume is optimal for achieving timely and effective sedation when intranasal sedatives, such as midazolam, are administered to children. Intranasal medications are typically administered with a mucosal atomization device, with a commonly recommended optimal volume of administration of approximately 0.3 mL and a maximum of 1 mL. This is in contrast to other recommendations stating that a volume of administration should not exceed 0.15 or 0.2 mL because of concerns that any volume in excess of these limits will become runoff, drain out of the nose and not be absorbed, or will be swallowed and subject to first-pass metabolism.

Goals of This Investigation

The aim of this study was to determine the optimal volume of administration of intranasal midazolam in children by comparing clinical outcomes relevant to procedural sedation associated with escalating volumes of administration (0.2, 0.5, and 1 mL) during laceration repair in an ED. Our primary outcome was time to onset of minimal sedation. Our secondary outcomes included procedural distress, time to procedure start, deepest level of sedation achieved, adverse events, and clinician and caregiver satisfaction. We hypothesized that using a volume of administration of 0.2 mL would be associated with a shorter time to onset of minimal sedation compared with using a volume of administration of either 0.5 or 1 mL because we expected that a volume of administration of 0.2 mL would have the least amount of runoff compared with larger volumes.

Study Design

We conducted a randomized, outcome assessor-blinded, 3-arm parallel group (1:1:1), superiority clinical trial in a single urban pediatric ED with an annual census of approximately 55,000 children per year. Our institutional review board approved this study, and written informed consent was obtained from each participant’s legal guardian.

Selection of Participants

Between November 2013 and September 2015, we enrolled a convenience sample of children aged 1 to 7 years who presented to the ED with a simple laceration (defined as length <5 cm and not requiring wound revision) and whose attending physician determined that intranasal midazolam was indicated to facilitate the repair. Patients were enrolled when a research coordinator or study physician was available (9 am to 10 pm on weekdays; variable on overnights and weekends). We excluded children for any of the following: a history of developmental delay, underlying neurologic abnormality, or autism; illness associated with chronic pain; known allergy to midazolam or any other benzodiazepine; weight less than 10 kg; eyelid lacerations (ie, repair would necessitate closed eyes); nasal obstruction that could not be easily cleared; did not speak English or Spanish; or was a foster child or ward of the state.

Interventions

We used an online randomization program, which maintained allocation concealment, to randomize children in blocks of 3 to receive the total dose of midazolam, using 1 of 3 volumes of administration: 0.2, 0.5, and 1 mL. These volumes were selected according to previous recommendations regarding optimal volume of administration, clinical feasibility, and current clinical practice. Clinicians applied lidocaine-epinephrine-tetracaine gel to all lacerations in a standardized fashion, unless contraindicated. Lidocaine injection for local anesthesia was administered at the discretion of the clinician performing the laceration repair (ie, the “proceduralist”). All children received an integrative (nonpharmacologic) intervention, such as a child life specialist, or a developmentally appropriate form of active or passive distraction.

All children received midazolam at 0.5 mg/kg (concentration 5 mg/mL), with a maximum total dose of 10 mg (maximum volume of 2 mL). The medication was administered in aliquots based on the assigned volume of administration by an attending pediatric emergency physician using an LMA MAD Nasal (Teleflex, Morrisville, NC) device attached to a 1-mL syringe with 0.01-mL scale markings. Before every administration, the attending physician completed an “intranasal administration refresher” with a member of the study team ( Figure E1 , available online at http://www.annemergmed.com ). During the refresher, the attending physician was required to demonstrate that he or she could administer the assigned volume of administration accurately. We made the refresher mandatory to ensure that the administration was consistently performed in a standardized fashion and that the assigned volume of administration would be administered correctly.

We administered the intranasal midazolam first into one nostril, using the assigned volume of administration, and then alternated between nostrils until the total volume was delivered. Because the total volume required by each patient was not always divisible by the assigned volume of administration into an equal number of administrations, we administered as many aliquots as possible with the assigned volume of administration and then administered the remaining volume last. Medication could be delivered into the second nostril immediately after the first, but we waited a period of 10 seconds before readministering the medication to the same nostril. We chose the interval of 10 seconds in accordance with both expert consultation with otolaryngologists and pharmacologists and feasibility in a clinical setting. For example, a total dose of 8 mg (and therefore a total volume of 1.6 mL when using the 5-mg/mL concentration of midazolam) would be required for a 16-kg child. If the child were assigned to receive intranasal midazolam with a volume of administration of 0.5 mL, 0.5 mL would be administered to the first nostril, and then another 0.5 mL would be immediately administered to the second nostril. Ten seconds later, another 0.5 mL would be administered to the first nostril, with the remaining 0.1 mL immediately administered to the second nostril, thereby completing administration of the total volume.

It was at the discretion of the proceduralist, who was not blinded, when to begin the procedure; whether restraining the child during the procedure was indicated (and how restraint was implemented); and whether to abort a procedure due to perceived inadequate sedation.

Methods of Measurement and Outcome Measures

A member of the study team videotaped patients for a 1-minute period before intranasal midazolam administration (the baseline period) and then through the entirety of the procedure, starting from just before administration of the intranasal midazolam until 1 minute after procedure completion, defined as time the last suture was cut. Steps were taken during the videotaping and video scoring to create a standardized 1-minute blinding period, starting from the time that the first aliquot was administered, in order to blind outcome assessors to both visual and auditory cues during the administration of the intranasal midazolam.

Three of 4 trained physician outcome assessors (D.S.T., M.I., D.B.F., and J.B.) independently scored each video recording, blinded to the assigned volume of administration and to the results of the other assessors. If one of them was involved in any way with an enrollment, he or she was not eligible to evaluate that video recording. Outcome assessors scored the video recordings in a randomized order that was unique to each of the assessors.

Each outcome assessor evaluated the primary outcome, time to onset of minimal sedation, as defined by the University of Michigan Sedation Scale, which is a 5-point (0 to 4) scale that has been validated and used for children to identify different depths of sedation, where 0=awake and alert, 1=minimally sedated (tired/sleepy, appropriate response to verbal conversation or sound), 2=moderately sedated, 3=deeply sedated, and 4=unarousable. The primary outcome was evaluated by measuring the time starting from the administration of the first aliquot of intranasal midazolam until the onset of minimal sedation.

The outcome assessors evaluated procedural distress, a secondary outcome, with the Observational Scale of Behavioral Distress–Revised. The scale was used to score 4 predetermined clinically relevant phases: baseline, irrigation or cleaning, laceration repair, and postprocedure. The scale is an 8-factor, weighted observational scale used to measure distress associated with medical procedures, which has been validated in children and adults aged 1 to 20 years. The total Observational Scale of Behavioral Distress–Revised score is the sum of the scale scores for each phase, with each phase assigned a score from 0 to 23.5 (0=no distress, 23.5=maximum distress), based on the frequency and types of behaviors observed during a predetermined number of 15-second intervals during each phase. The scale was used to identify statistically significant differences between groups because the change in score representing a clinically meaningful difference and the range of scores representing categories of pain intensity (ie, mild, moderate, and severe) have not yet been clearly defined. Each outcome assessor also determined time to procedure start and deepest level of sedation achieved (as per the University of Michigan Sedation Scale) when scoring the video recording.

The study team member documented the presence of any adverse events during the procedure, including vomiting, apnea (no breath for >20 seconds), paradoxic reaction (ie, agitation precipitated by midazolam), and the need for supplemental oxygen, airway repositioning, or bag-valve-mask ventilation. We also documented whether, per the proceduralist, there was inadequate sedation during the laceration repair, resulting in the inability to complete the procedure. Children who did not reach adequate sedation by 20 minutes after intranasal administration, as per the proceduralist, were also considered to have had inadequate sedation.

After completion of the procedure, a study team member evaluated clinician and caregiver satisfaction by interviewing the attending physician who administered the intranasal midazolam, the proceduralist who performed the laceration repair, and the child’s caregiver. Clinicians and caregivers used a 5-point Likert scale (“strongly agree,” “agree,” “neither agree nor disagree,” “disagree,” or “strongly disagree”) to answer questions related to their satisfaction with ease of intranasal administration, child comfort during intranasal administration, and the adequacy of sedation achieved.

Primary Data Analysis

For our primary outcome of time to onset of minimal sedation, we compared the 3 groups with the Kruskal-Wallis test, with follow-up pairwise comparisons between groups with the Wilcoxon rank-sum test. We chose to use a rank test so that patients who did not have a time to onset of minimal sedation documented (eg, if they not achieve adequate sedation by 20 minutes or if the procedure was started before onset of minimal sedation) could still be included in the analysis. These patients were assigned a maximal score (eg, 999 minutes) in place of a time to onset of minimal sedation to represent a maximal time for the primary outcome, thereby using the most conservative assumption possible. The median of the 3 raters’ scores was used as the time to onset of minimal sedation for each patient, and the median time to onset of minimal sedation for each group was determined and compared. We considered a statistically significant difference for the primary outcome at an experimentwise α=.05. For follow-up pairwise comparisons, we used a Bonferroni adjustment for 3 comparisons, 2-tailed per comparison (α=.0167). We determined our sample size by designing the study to have 0.80 power to detect a clinically meaningful difference in time to onset of minimal sedation of 1.5 minutes, with an estimated SD of 2 minutes for each group. The goal sample size was 108, but enrollment was stopped early because of constraints related to funding.

We analyzed continuous secondary outcomes (ie, Observational Scale of Behavioral Distress–Revised, time to procedure start) with the 1-way ANOVA test, and ordinal and nominal outcomes (ie, deepest level of sedation, adverse events, and satisfaction) with the χ ² test.

An interclass correlation coefficient calculation was performed on the primary outcome of time to onset of minimal sedation and the secondary outcome of procedural distress (Observational Scale of Behavioral Distress–Revised). The Shrout-Fleiss interclass correlation coefficient (1,k) statistic was calculated to show the correlation of raters when each subject was rated by multiple raters, with raters assumed to be randomly assigned to subjects and all subjects have the same number of raters.

Statistical analyses were performed with SPSS (version 22; IBM Corporation, Armonk, NY) and SAS (version 9.4; SAS Institute, Inc., Cary, NC).

Study Design

We conducted a randomized, outcome assessor-blinded, 3-arm parallel group (1:1:1), superiority clinical trial in a single urban pediatric ED with an annual census of approximately 55,000 children per year. Our institutional review board approved this study, and written informed consent was obtained from each participant’s legal guardian.

Selection of Participants

Between November 2013 and September 2015, we enrolled a convenience sample of children aged 1 to 7 years who presented to the ED with a simple laceration (defined as length <5 cm and not requiring wound revision) and whose attending physician determined that intranasal midazolam was indicated to facilitate the repair. Patients were enrolled when a research coordinator or study physician was available (9 am to 10 pm on weekdays; variable on overnights and weekends). We excluded children for any of the following: a history of developmental delay, underlying neurologic abnormality, or autism; illness associated with chronic pain; known allergy to midazolam or any other benzodiazepine; weight less than 10 kg; eyelid lacerations (ie, repair would necessitate closed eyes); nasal obstruction that could not be easily cleared; did not speak English or Spanish; or was a foster child or ward of the state.

Interventions

We used an online randomization program, which maintained allocation concealment, to randomize children in blocks of 3 to receive the total dose of midazolam, using 1 of 3 volumes of administration: 0.2, 0.5, and 1 mL. These volumes were selected according to previous recommendations regarding optimal volume of administration, clinical feasibility, and current clinical practice. Clinicians applied lidocaine-epinephrine-tetracaine gel to all lacerations in a standardized fashion, unless contraindicated. Lidocaine injection for local anesthesia was administered at the discretion of the clinician performing the laceration repair (ie, the “proceduralist”). All children received an integrative (nonpharmacologic) intervention, such as a child life specialist, or a developmentally appropriate form of active or passive distraction.

All children received midazolam at 0.5 mg/kg (concentration 5 mg/mL), with a maximum total dose of 10 mg (maximum volume of 2 mL). The medication was administered in aliquots based on the assigned volume of administration by an attending pediatric emergency physician using an LMA MAD Nasal (Teleflex, Morrisville, NC) device attached to a 1-mL syringe with 0.01-mL scale markings. Before every administration, the attending physician completed an “intranasal administration refresher” with a member of the study team ( Figure E1 , available online at http://www.annemergmed.com ). During the refresher, the attending physician was required to demonstrate that he or she could administer the assigned volume of administration accurately. We made the refresher mandatory to ensure that the administration was consistently performed in a standardized fashion and that the assigned volume of administration would be administered correctly.

We administered the intranasal midazolam first into one nostril, using the assigned volume of administration, and then alternated between nostrils until the total volume was delivered. Because the total volume required by each patient was not always divisible by the assigned volume of administration into an equal number of administrations, we administered as many aliquots as possible with the assigned volume of administration and then administered the remaining volume last. Medication could be delivered into the second nostril immediately after the first, but we waited a period of 10 seconds before readministering the medication to the same nostril. We chose the interval of 10 seconds in accordance with both expert consultation with otolaryngologists and pharmacologists and feasibility in a clinical setting. For example, a total dose of 8 mg (and therefore a total volume of 1.6 mL when using the 5-mg/mL concentration of midazolam) would be required for a 16-kg child. If the child were assigned to receive intranasal midazolam with a volume of administration of 0.5 mL, 0.5 mL would be administered to the first nostril, and then another 0.5 mL would be immediately administered to the second nostril. Ten seconds later, another 0.5 mL would be administered to the first nostril, with the remaining 0.1 mL immediately administered to the second nostril, thereby completing administration of the total volume.

It was at the discretion of the proceduralist, who was not blinded, when to begin the procedure; whether restraining the child during the procedure was indicated (and how restraint was implemented); and whether to abort a procedure due to perceived inadequate sedation.

Methods of Measurement and Outcome Measures

A member of the study team videotaped patients for a 1-minute period before intranasal midazolam administration (the baseline period) and then through the entirety of the procedure, starting from just before administration of the intranasal midazolam until 1 minute after procedure completion, defined as time the last suture was cut. Steps were taken during the videotaping and video scoring to create a standardized 1-minute blinding period, starting from the time that the first aliquot was administered, in order to blind outcome assessors to both visual and auditory cues during the administration of the intranasal midazolam.

Three of 4 trained physician outcome assessors (D.S.T., M.I., D.B.F., and J.B.) independently scored each video recording, blinded to the assigned volume of administration and to the results of the other assessors. If one of them was involved in any way with an enrollment, he or she was not eligible to evaluate that video recording. Outcome assessors scored the video recordings in a randomized order that was unique to each of the assessors.

Each outcome assessor evaluated the primary outcome, time to onset of minimal sedation, as defined by the University of Michigan Sedation Scale, which is a 5-point (0 to 4) scale that has been validated and used for children to identify different depths of sedation, where 0=awake and alert, 1=minimally sedated (tired/sleepy, appropriate response to verbal conversation or sound), 2=moderately sedated, 3=deeply sedated, and 4=unarousable. The primary outcome was evaluated by measuring the time starting from the administration of the first aliquot of intranasal midazolam until the onset of minimal sedation.

The outcome assessors evaluated procedural distress, a secondary outcome, with the Observational Scale of Behavioral Distress–Revised. The scale was used to score 4 predetermined clinically relevant phases: baseline, irrigation or cleaning, laceration repair, and postprocedure. The scale is an 8-factor, weighted observational scale used to measure distress associated with medical procedures, which has been validated in children and adults aged 1 to 20 years. The total Observational Scale of Behavioral Distress–Revised score is the sum of the scale scores for each phase, with each phase assigned a score from 0 to 23.5 (0=no distress, 23.5=maximum distress), based on the frequency and types of behaviors observed during a predetermined number of 15-second intervals during each phase. The scale was used to identify statistically significant differences between groups because the change in score representing a clinically meaningful difference and the range of scores representing categories of pain intensity (ie, mild, moderate, and severe) have not yet been clearly defined. Each outcome assessor also determined time to procedure start and deepest level of sedation achieved (as per the University of Michigan Sedation Scale) when scoring the video recording.

The study team member documented the presence of any adverse events during the procedure, including vomiting, apnea (no breath for >20 seconds), paradoxic reaction (ie, agitation precipitated by midazolam), and the need for supplemental oxygen, airway repositioning, or bag-valve-mask ventilation. We also documented whether, per the proceduralist, there was inadequate sedation during the laceration repair, resulting in the inability to complete the procedure. Children who did not reach adequate sedation by 20 minutes after intranasal administration, as per the proceduralist, were also considered to have had inadequate sedation.

After completion of the procedure, a study team member evaluated clinician and caregiver satisfaction by interviewing the attending physician who administered the intranasal midazolam, the proceduralist who performed the laceration repair, and the child’s caregiver. Clinicians and caregivers used a 5-point Likert scale (“strongly agree,” “agree,” “neither agree nor disagree,” “disagree,” or “strongly disagree”) to answer questions related to their satisfaction with ease of intranasal administration, child comfort during intranasal administration, and the adequacy of sedation achieved.

Primary Data Analysis

For our primary outcome of time to onset of minimal sedation, we compared the 3 groups with the Kruskal-Wallis test, with follow-up pairwise comparisons between groups with the Wilcoxon rank-sum test. We chose to use a rank test so that patients who did not have a time to onset of minimal sedation documented (eg, if they not achieve adequate sedation by 20 minutes or if the procedure was started before onset of minimal sedation) could still be included in the analysis. These patients were assigned a maximal score (eg, 999 minutes) in place of a time to onset of minimal sedation to represent a maximal time for the primary outcome, thereby using the most conservative assumption possible. The median of the 3 raters’ scores was used as the time to onset of minimal sedation for each patient, and the median time to onset of minimal sedation for each group was determined and compared. We considered a statistically significant difference for the primary outcome at an experimentwise α=.05. For follow-up pairwise comparisons, we used a Bonferroni adjustment for 3 comparisons, 2-tailed per comparison (α=.0167). We determined our sample size by designing the study to have 0.80 power to detect a clinically meaningful difference in time to onset of minimal sedation of 1.5 minutes, with an estimated SD of 2 minutes for each group. The goal sample size was 108, but enrollment was stopped early because of constraints related to funding.

We analyzed continuous secondary outcomes (ie, Observational Scale of Behavioral Distress–Revised, time to procedure start) with the 1-way ANOVA test, and ordinal and nominal outcomes (ie, deepest level of sedation, adverse events, and satisfaction) with the χ ² test.

An interclass correlation coefficient calculation was performed on the primary outcome of time to onset of minimal sedation and the secondary outcome of procedural distress (Observational Scale of Behavioral Distress–Revised). The Shrout-Fleiss interclass correlation coefficient (1,k) statistic was calculated to show the correlation of raters when each subject was rated by multiple raters, with raters assumed to be randomly assigned to subjects and all subjects have the same number of raters.

Statistical analyses were performed with SPSS (version 22; IBM Corporation, Armonk, NY) and SAS (version 9.4; SAS Institute, Inc., Cary, NC).