The Nature, Frequency, and Timing of Pediatric Sedation Adverse Events

After institutional board review approval, we examined prospectively collected data of sedation encounters of children 3 months to 18 years of age, characterized by at least 1 AE, from January 1, 2013 to December 31, 2020. Patient characteristics, primary diagnosis, procedure type, nature, frequency, and onset of AEs in each sedation phase were reported from a single center database.

The Pediatric Sedation Program is a pediatric critical care and pediatric hospital medicine-based program at a quaternary-care freestanding academic children’s hospital in an urban environment. Our program is primarily propofol based and follows sedation practice guidelines for monitoring AEs endorsed by the American Academy of Pediatrics.⁴  Procedural sedation was performed in accordance with our standard institutional procedures. This included preprocedure fasting, preemptive supplemental oxygen administration at 6 L/min via a flow-inflating bag and mask during the 3 phases of sedation, routine airway positioning, visual monitoring, tracheal auscultation with a pretracheal stethoscope, continuous pulse oximetry (GE Patient Data Module with Masimo technology), automated blood pressure measurement, and continuous nasal microstream capnography (GE E-sCO). We used a modified (in reverse order) Children’s Hospital of Wisconsin Sedation Scale to assess the sedation levels.⁵ 

We are an active member of the Pediatric Sedation Research Consortium (PSRC). Consequently, our program is used to observe data collection methodology and definitions of sedation-related AEs used by the PSRC.²  Our sedation program cares mostly of children with American Society of Anesthesiologists (ASA) physical status 1 or 2. Procedures are elective and less than 3 hours’ duration. Anxiolysis with oral midazolam or nitrous oxide is used in our program, in selected patients, before peripheral intravenous catheter placement. The goal of the sedation provider is to maintain a natural airway. Airway instrumentation is performed rarely and only in response to AEs. Pediatric sedation providers screen children who require procedural sedation before scheduling to ascertain their likelihood of success with a natural airway. Children with ASA scores ≥3, cardiorespiratory instability including supplemental oxygen requirement, anticipated need for airway instrumentation, age younger than 3 months, cyanotic congenital heart disease, inability to handle oral secretions, and severe gastroesophageal reflux or vomiting are generally referred to the anesthesia service.

Respiratory AEs reported included partial upper airway obstruction (UAO), defined as audible snoring without a stethoscope, complete UAO defined as complete lack of air movement with maintained respiratory effort, apnea defined as no respiratory effort for longer than 15 seconds, and laryngospasm defined as complete or nearly complete lack of air movement with respiratory effort and with or without stridor, not relieved by airway repositioning or oral/nasal airway. Coughing was considered an AE when it delayed or interrupted the procedure. Hypoxemia was defined as pulse oximetry saturation <90% for more than 30 seconds. Other respiratory AEs reported included wheezing, hiccups, and aspiration. Cardiovascular events included blood pressure or heart rate changes >30% from baseline vital signs obtained before induction. Neurologic events included seizures, myoclonus, and agitation/combative behavior interfering with patient care. Gastrointestinal events included vomiting. Finally, miscellaneous events included prolonged sedation (defined as a recovery phase >4 hours), allergic reaction, and other adverse events not defined here.

The sedation encounter was divided in 3 periods: induction, procedure, and recovery. We defined induction (ie, preprocedure phase) as the time from the beginning of sedative administration until the patient was adequately sedated, positioned, and ready for the procedure. We defined procedure phase as the time from the beginning to the end of the procedure. If the procedure required transport to an offsite department, that time was included in the procedure phase. Finally, we defined recovery (ie, postprocedure phase) as the time from the end of the procedure until the patient was awake and ready to be discharged from the clinic.

For each sedation encounter, a sedation nurse recorded demographic characteristics and occurrence of AEs, including type and onset, in a data collection form. Administrative personnel entered the information into an electronic database. An individual sedation encounter could have multiple AEs but for any given AE, the time of onset is unique. Therefore, if there were 2 or more AEs in a sedation encounter (eg, UAO, blood pressure change >30%), we reviewed the electronic medical record to find out the phase of sedation in which each specific AE started (eg, UAO started during induction, blood pressure change >30% started during procedure). We analyzed the phase of sedation that characterized the onset of a particular AE but did not analyze the phase at which the AE finished. Information was only available on induction, procedure, and recovery events but not on AEs that occurred after discharge from the sedation clinic. If an encounter involved 2 or more respiratory AEs, then the earliest phase of sedation among multiple AEs was used to characterize the onset. For example, if a single encounter involved partial UAO during the induction phase whereas both apnea and complete UAO occurred during the procedure phase, then the onset of a respiratory AE for that encounter was set to the induction phase. Finally, if a sedation encounter was “aborted,” defined as ending a sedation encounter after sedative administration but before completion of the procedure, we analyzed the phase of sedation that characterized the onset of the particular AE and included the aborted procedure in the total number of encounters with at least one AE.

Characteristics of the sample were described using the range, median, and interquartile range, mean, and SD, or with frequencies and percentages. Continuous characteristics were compared between groups (at least 1 AE vs no AEs) using the Wilcoxon rank-sum and χ² tests used to compare categorical factors. Each encounter could involve multiple procedures or multiple sedatives and both characteristics were compared between groups using generalized estimating equations.⁶  Analyses were performed using R (v. 4.2.0) with the associated geepack.⁷ 

A total of 12 012 sedation encounters were recorded between January 1, 2013 and December 31, 2020. For all encounters, age ranged from 0.25 to 18.9 years (mean ± SD of 7.6 years ± 4.9) with 48.2% of the encounters involving male patients. Just fewer than 90% of encounters involved patients with ASA II risk classification. The most common diagnosis was hematology/oncology (27.3%) and the most common procedure was radiologic (47.8%). Midazolam was the most common sedative used in 52.3% of sedation encounters. A single sedative was administered in 33.9% of sedation encounters, 2 sedatives were administered in 46.4%, and 3 sedatives or more were administered in 19.6% of sedation encounters.

Of the 12 012 sedation encounters, 765 encounters involving 620 distinct patients had at least 1 AE reported, with 636 (83.1%) having only 1 AE, 110 (14.4%) having exactly 2 AEs, 15 (2.0%) with 3 AEs, and 4 encounters (0.5%) resulting in 4 AEs. Demographic and clinical characteristics of sedation encounters with and without AEs are summarized in Table 1. Table 2 depicts the phase of the sedation encounter during which the AE began. Respiratory AEs were most common (n = 645, 5.4% of all encounters), followed by cardiovascular (n = 67, 0.6%), neurologic (n = 32, 0.3%), vomiting (n = 26, 0.2%), and other events (n = 19, 0.2%). There were no reported major AEs, including aspiration, respiratory arrest, cardiac arrest, or death. The most common respiratory AE was partial UAO (n = 335, 2.8% of all encounters), followed by apnea (n = 144, 1.2%), coughing (n = 90, 0.8%), complete UAO (n = 66, 0.6%), laryngospasm (n = 62, 0.5%), hypoxemia (n = 59, 0.5%), hiccups (n = 9, 0.1%), and wheezing (n = 1, <0.01%).

Most respiratory AEs started during the induction phase of sedation (Fig 1 and Table 2). Of 645 encounters involving respiratory AEs, 416 (64.5%) started during induction phase, 207 (32.1%) during procedure phase, and 22 (3.4%) during the recovery phase of sedation. Partial UAO started more often during induction, 59.4% vs 36.4% procedure vs 4.2% recovery. Complete UAO also started more often during induction, 77.3% vs 22.7% during the procedure. No complete UAO events were reported during recovery. Apnea started more often during induction, 84% vs 15.3% during procedure vs 0.7% during recovery. Coughing started more often during procedure, 54.4% vs 33.3% induction and 12.2% recovery. The onset of laryngospasm events was similarly distributed between induction (48.4%) and procedure phase (46.8%) but rare during recovery (4.8%). The onset of hypoxemia events was similarly distributed between induction (59.3%) and procedure phase (39%) but rare during recovery (1.7%).

Cardiovascular events, primarily hypotension, started more often during the procedure phase, 61.2% vs 29.8% induction, and 9% recovery (Fig 1 and Table 2). Vomiting, neurologic complications, and other events appeared equally distributed among phases, though limited sample size (n < 60) limits our certainty concerning these AEs.

As has been reported in other studies, respiratory AEs were the most common sedation-related AEs reported in this study.^2,3  Most respiratory AEs in our cohort started during the induction phase of sedation. However, the onset of laryngospasm and hypoxemia was similar between induction and procedure phases of sedation. Our findings highlight the need for the sedation team to be especially vigilant during induction and procedure phases of sedation.

The risk of respiratory events during general anesthesia is highest during the induction phase. However, the emergence or recovery phase of anesthesia is also a high-risk period for respiratory AEs. The incidence density per 100 000 person-minutes of respiratory events during general anesthesia is 61.3 during the induction period, compared with 13.7 during maintenance, and 16.5 during emergence/recovery.⁸  Although we did not measure the amount of time a patient spent in each phase of sedation, the frequencies reported in our study suggest similarities and differences between general anesthesia and sedation. Induction is a high-risk phase for both. By contrast, our study found the recovery phase of sedation to be low risk for respiratory AEs. Of 645 respiratory AEs reported, only 22 (3.4%) started during recovery, compared with 207 (32.1%) during the procedure and 416 (64.5%) during the induction phase of sedation.

The differing goals in airway management during procedural sedation compared with general anesthesia may help explain the differences between rates of respiratory AEs that were noted in this study compared with other studies. General anesthesia usually entails planned airway instrumentation with placement of an artificial airway, which happens during induction and is discontinued during recovery or emergence from anesthesia. However, the goal during procedural sedation is essentially always to maintain a natural airway. Therefore, the AEs that occur during the surgical phase under general anesthesia are inherently different from those that occur during the procedure phase under sedation. Airway instrumentation during sedation is rare. A recent report from the PSRC described a rate of unplanned endotracheal intubation or laryngeal mask airway placement of 0.11% and 0.1%, respectively, in sedations with propofol outside the operating room.³  Artificial airways are placed mostly in response to AEs.⁹  This may explain the difference in risk and timing of respiratory AEs between sedation and general anesthesia.

One specific respiratory AE that was evaluated in this study was laryngospasm. Volatile anesthetics have been associated with increased risk of laryngospasm in pediatric anesthesia, but induction with propofol is associated with less risk of laryngospasm.¹⁰  Laryngospasm is felt to be more common during induction and recovery of sedation,¹¹  though our study found the laryngospasm onset to be similar between induction and procedure phases and comparatively rare during recovery phase of sedation. By contrast, during general anesthesia, the risk of laryngospasm is higher during induction and emergence at the end of the operation when the patient is extubated, especially during light anesthesia.¹²  The incidence density per 100 000 person-minutes of laryngospasm during general anesthesia is 13.5 at the induction period, compared with 3.3 during maintenance and 4.8 during emergence/recovery.⁶ 

Our study found both partial and complete UAOs were more common during induction and procedure phases of sedation. During induction, bolus administration of sedative agents relaxes the pharyngeal muscles leading to posterior movement of the tongue and the potential for obstruction of the oropharynx.¹³  Most episodes of UAO respond to simple airway maneuvers (ie, head tilt, chin lift, jaw thrust). At the end of the procedure and at the beginning of recovery phase, sedatives are stopped, with consequent improvement in pharyngeal tone and less risk of airway obstruction.

We found apnea to be more common during the induction phase of sedation. During induction, bolus administration of sedative agents suppresses the central respiratory center and reduces the ventilatory response to a given level of carbon dioxide, which may lead to hypoventilation or apnea.¹⁴  Although this is the case for all sedative agents, it is particularly true of propofol, which is the sedative agent most often used at our sedation program. It is a common practice in our unit for patients to receive a continuous infusion of propofol for procedures longer than 15 minutes. Because the patient is more likely to remain within a therapeutic window of a given drug during a continuous infusion compared with repeated boluses, this may explain the lower frequency of apnea during the procedure compared with the induction phase. In this study, the incidence of apnea significantly decreased during recovery phase. However, premature infants, especially if <60 weeks’ postconceptual age, may be at increased risk of apnea during recovery from general anesthesia.^15,16  Our low incidence of apnea during recovery may be explained by our patients receiving sedation and not general anesthesia, and our practice of referring to anesthesia premature infants who are <60 weeks’ postconceptual age at the time of sedation.

We found the frequency of hypoxemia to be similar during induction and procedure phases of sedation. In contrast, the risk of hypoxia is greater during anesthesia induction and airway instrumentation.⁶  During procedural sedation, most episodes of hypoxemia are secondary to UAO, hypoventilation, or apnea. Half of the procedures reported in this cohort were radiologic and were mostly magnetic resonance imaging procedures. This means the sedation provider did not have immediate access to the child’s natural airway during the procedure. Even with prompt recognition of a respiratory AE via noninvasive monitoring, there is a brief but unavoidable delay as the sedation provider moves into the scanner, regains access to the patient’s airway, and performs the necessary rescue maneuvers. This delay could increase the risk of hypoxemia during the procedure phase and highlights the need for effective patient screening for procedural sedation and for increased vigilance especially during radiology procedures.

Our study has several limitations inherent to database research. A major limitation is the potential for bias in the consistency or accuracy of reporting. Neurologic complications, vomiting and other events appeared equally distributed among phases, though limited sample size (n < 60) limits our certainty concerning these AEs. In addition, we could not determine the timing of major AEs (aspiration, cardiopulmonary arrest, and death) because these events are extremely rare in organized sedation programs.¹⁷  We analyzed the phase of sedation in which a particular AE started but not when it was finished because this information was not available. In addition, information was only available on induction, procedure, and recovery AEs but not on events that occurred after discharge from the sedation clinic. We did not report specific airway interventions in each phase of sedation; however, most sedation respiratory AEs improve with basic airway maneuvers.¹⁰  We did not report the level of sedation achieved but most encounters in which at least 1 AE occurred received propofol (84%) so we can expect deep sedation was achieved. Finally, this is a single-center study, and our practice may not reflect that of other sedation programs.

Most respiratory AEs occurring in this cohort started during the induction or procedure phases of sedation. The sedation team should be especially vigilant and prepared to intervene on the patient’s behalf during these phases of sedation. Our findings may have implications for allocation of staffing/resources during various phases of sedation.