BACKGROUND: Upper respiratory infections (URIs) are associated with airway adverse events (AAEs) during general anesthesia. There have been no large studies examining the relationship between URIs and AAEs during pediatric procedural sedation. We hypothesized that there would be a relationship between URI status and AAEs in pediatric procedural sedation.
METHODS: We examined prospectively collected data from the Pediatric Sedation Research Consortium database. Specific questions regarding URI status were added to the database to facilitate our analysis. Characteristics of patients, procedure types, adjunctive medications, adverse events, and airway interventions (AIs) were reported. We performed bivariate analysis of adverse events and URI status, then used a multivariable logistic regression model to assess the relationship between URI status and adverse events. We examined the secondary outcome of AI similarly.
RESULTS: Of the 105 728 sedations entered into the Pediatric Sedation Research Consortium database during the study period, we were able to use 83 491 for analysis. Controlling for multiple patient, drug, and procedure characteristics, recent and current URI were associated with increased frequency of AAEs. In general, the frequency of AAEs and AIs increased from recent URI, to current URI-clear secretions to current URI-thick secretions. We did not find a relationship between URI status and non-AAEs.
CONCLUSIONS: URI status is associated with a statistically significant increase in frequency of AAEs and AI during pediatric procedural sedation for the population sedated by our consortium. Although URI status merits consideration in determining potential risk for sedation, rates of some AAEs and AIs remained low regardless of URI status.
- AAE —
- airway adverse event
- AI —
- airway intervention
- ASA —
- American Society of Anesthesiology
- URI —
- upper respiratory infection
- PSRC —
- Pediatric Sedation Research Consortium
- NPO —
- nil per os
What’s Known on This Subject:
Several studies suggest that when anesthesia is administered to patients with a URI there are increased AAEs. Until now, there have been no studies that assess the relationship between URI and AAEs in pediatric procedural sedation.
What This Study Adds:
We examine data from 83 491 sedations and demonstrate that URIs are associated with an increased rate of AAEs. Regardless, the rates of AAEs remain low even in children with recent or current URIs.
The identification of potential patient or procedural factors associated with the occurrence of adverse outcomes during pediatric procedural sedation can be used to enhance patient safety. Upper respiratory infections (URIs) are the most common illnesses in children. Estimates of the prevalence of these infections indicate as many as 30% of children have a URI during the winter months in the Northern Hemisphere.1,2 The impact of this disease burden on pediatric procedural sedation is unclear. Numerous studies suggest that when anesthesia is administered to patients with a current or recent URI there is an increased incidence of perioperative adverse events.3–6 These studies suggest that the incidence of airway adverse events (AAEs) such as coughing, bronchospasm, and laryngospasm is higher not only during an active URI, but may also be increased in the 2 to 4 weeks after the infection. On the other hand, there are no studies that specifically address the issue as to whether the presence of a URI at the time of pediatric procedural sedation is associated with an increase in similar adverse events. Because it is impossible to study this subject in a randomized, controlled, or blinded manner, we used an observational methodology in a large, ongoing, multicenter collaborative data sharing effort (the Pediatric Sedation Research Consortium [PSRC] database) to investigate the relationship between URI symptoms and the frequency of AAEs as reported by our member institutions. This database is designed to collect data on sedation demographics, practice characteristics, and adverse events that occur during procedural sedation or immediately afterward. To facilitate this study, we customized the data collection interface for a defined period, adding questions about URI symptoms specifically so that we could investigate this potential association. We hypothesized that we would find an increased incidence of AAEs in patients with current URIs. We also examined airway interventions (AIs) as a secondary outcome, and hypothesized that we would find an increased incidence of AIs in patients with current URI as well.
The Pediatric Sedation Research Consortium
The PSRC was established in 2003 as a collaborative, multidisciplinary group of sedation practitioners dedicated to promoting the delivery of optimal sedation for children undergoing tests and procedures. The PSRC database was initiated in 2004 for the purpose of collecting information on a large number of pediatric procedural sedations to better understand and improve the practice of sedation in children. Participation in the PSRC database is voluntary, but participating institutions are required to obtain approval from their respective institutional review boards, identify a primary investigator, and agree to standardized methodology for data collection and quality oversight. Audits are required every 6 months to include an accounting of numbers of sedation encounters and quality of data entry to ensure data integrity. To avoid selection bias, primary investigators agree to only enter data from sedation areas in which >90% of sedation encounters can be reliably captured. Data collection methodology has been described in a detailed report of the first 30 000 sedations published in 20067 and in numerous subsequent publications.7–12
Data were entered into the PSRC database via a standardized, web-based data collection tool. A secure site, web-based data system, maintained by the Dartmouth Bioinformatics Group was employed for data collection and storage. The data entry tool consisted of ∼25 separate areas of inquiry and collected information on patient demographics, diagnostic categories associated with the procedure, coexisting diagnoses, procedures performed, medications administered, adverse events and interventions. To ensure consistency in data entry, for any field that involves interpretation, pop-up text was available to explain the field. Only deidentified data were collected.
In 2012, questions designed to specifically identify patients with a URI were added to the PSRC data collection tool (Supplemental Figure 2). Parents were asked about symptoms related to the upper respiratory tract. If the child had runny nose, nasal congestion, sneezing, or mild cough (without sputum production), and the parent and sedation provider agreed that these symptoms were because of a current URI, they were eligible for inclusion in the URI cohort. This definition is similar to that previously used in the anesthesiology literature.6,13 The instructions to PSRC members was to make this decision in conjunction with the families. The nature of secretions was based on the parental description. The symptoms and signs of URI were collected before sedation as part of the standard data collection for presedation assessment. Patients were excluded from being categorized as having a URI if they exhibited systemic symptoms such as fever, or if they had physical examination findings suggesting specific lower respiratory tract pathophysiology such as wheezing or rales. The URI variable was coded as either no URI, recent URI, meaning that the patient had URI symptoms within the past 2 weeks, but none currently, or current URI at the time of sedation. Because there may be a spectrum of severity of URI, the data entry tool had the user specify whether a current URI at time of sedation was most accurately characterized by thin and/or clear secretions or thick and/or green secretions by the parent or the provider. Data for this study were collected between November 10, 2012 and July 7, 2015. Forty-five centers contributed data to the PSRC database during this period (Supplemental Table 6).
The PSRC database collects data on multiple potential adverse events. Definitions for the adverse events have been agreed on and are available to all participants as a part of the data entry tool. The data collection tool includes “hover text” that defines each element we are collecting. For example, if the PSRC member places the cursor over “oxygen desaturation”, the definition pops up in text. This definition reads, “oxygen saturation decreased to <90% for more than 30 seconds.” There is also an option for “<80% for more than 30 seconds” and “<70% for more than 30 seconds.” Our definition of laryngospasm requires evidence of stridorous breathing with or without evidence of frank airway obstruction. We do not require that an intervention was taken to reverse the event.
In collaboration with the participating centers, the authors (a priori) defined those events that seemed directly related to the airway as AAEs. This category included airway obstruction, apnea oxygen desaturation, cough, secretions requiring suction, laryngospasm, stridor, wheezing, emergent AI, and snoring.
AIs collected by the PSRC include bag-mask ventilation, continuous positive airway pressure, endotracheal tube placement, jaw thrust or chin lift, nasotracheal tube placement, nasopharyngeal airway placement, oxygen delivery (blowby, nasal cannula, or mask blowby), oral airway placement, airway repositioning, suction, laryngeal mask airway placement, and other.
Descriptive statistics were used to examine patient characteristics, procedures performed, and medications used. For ease of presentation, nil per os (NPO) data were collapsed into a single variable to indicate the presence of one or more NPO violations. A patient was deemed to have had an NPO violation if before sedation they had clear liquids within 2 hours, breast milk within 4 hours, nonfat solids within 6 hours, or a full meal within 8 hours.
χ2 tests were used to assess for differences in variables based on URI status. A multivariable logistic regression model was developed to assess the adjusted association of URI status with AAEs and AIs. Variables shown to be associated with adverse events in our data and previous studies were included in the initial model. Interactions between medication variables were assessed and significant interaction terms were retained in the model. Models were assessed for collinearity and adjusted by removing variables, if necessary. Because of the large number of sedations in our dataset, we could control for a large number of variables in the model. We controlled for demographic factors, comorbidities, procedure types, and locations as well as sedative agents and adjunctive medications. The final model included American Society of Anesthesiology (ASA) classification, age, obesity, sex, NPO status, asthma, developmental delay, metabolic and/or genetic, neurologic, seasonal allergy, procedure type (bone, dental, gastrointestinal, hematology and/or oncology, neurologic, radiology, surgery, painful, other), propofol, ketamine, dexmedetomidine, atropine, glycopyrrolate, fentanyl, and the interaction between propofol and fentayl. Because suction and dental procedures were highly correlated (suction was performed in >80% of dental procedures), we removed dental procedures from the model for suction. Obesity was defined as a weight greater than the 95th percentile for sex and age based on the Centers for Disease Control and Prevention growth curves.
During the study period, information on 105 728 sedations were entered into the PSRC database. We excluded 19 282 (18.2%) for missing URI status information and 2948 (2.8%) for missing other important demographic data, leaving 83 491 sedations in our analysis data set.
Patient demographics are presented in Table 1. Patients with no URI were older than those with recent or current URI. The boy to girl ratio was similar in each group, but trended toward a higher proportion of boys in the current URI categories compared with recent and no URI categories. The majority of each group had ASA classification I or II, but the current URI with thick and/or green secretion group did have a higher proportion of classification III or IV patients, and fewer classification I patients. The proportion of patients who had at least 1 NPO violation was lowest in the current URI groups.
Table 2 lists the 6 most commonly performed procedure categories broken down by URI status. Regardless of URI status, the majority of patients were sedated for radiology procedures, followed by hematology and/or oncology procedures. Eighty percent of radiology procedures were MRIs.
A synopsis of sedative, analgesic, and adjunctive medication use is provided in Table 3. Propofol was the most commonly used sedative in all groups. Ketamine was the second most commonly given sedative. Opioid analgesics were administered in 23.3% of cases, and were more commonly used in patients with no URI (P < .001). Atropine and glycopyrrolate were given to 2.7% and 6.5% of all patients respectively. Administration of these agents did vary by URI status (P < .001).
Rates of adverse events by URI status are provided in Table 4. Most of the AAEs occurred with increasing frequency from no URI to recent URI, to current URI with thin and/or clear secretions, and then to current URI with thick and/or green secretions. Exceptions included stridor (only 2 recorded for current URI with thick and/or green secretions), and emergent AI (only 1 recorded for current URI with thick and/or green secretions). All of the AAEs except apnea demonstrated a statistically significant association with URI status in χ2 analysis. Occurrence of any AAE increased progressively from 6.3% for those with no URI to 22.2% for those with current URI with thick and/or green secretions. Adverse events not directly related to the airway were less likely to demonstrate a statistically significant relationship with URI status.
Rates of AI by URI status are provided in Table 5. Like AAEs, most AIs increased from no URI to recent URI, to current URI with thin and/or clear secretions, and then to current URI with thick and/or green secretions.
Figure 1 provides a forest plot of the odds ratios for adverse events based on URI status by using no URI as the referent from the multivariable logistic regression model that controlled for a multitude of patient and procedure characteristics. The outcomes for which odds ratios are depicted include all individual AAEs, occurrence of any AAE, occurrence of any non-AAE, and procedure not complete because of a problem related to sedation. Recent URI and current URI with thick and/or green secretions demonstrated a statistically significant association with 7 of 10 AAEs when compared with no URI. Current URI with thin and/or clear secretions demonstrated a statistically significant association with 8 of 10 AAEs when compared with no URI. For a number of the AAEs, confidence intervals of odds ratios for URI with thick and/or green secretions and URI with thin and/or clear secretions did not overlap, indicating differences in the odds ratios for various AAEs based on the characteristics and timing of their URI. Because the event "procedure unable to be completed due to problem with sedation" was strongly associated with URI status, and because this outcome could be plausibly related to the occurrence of AAEs, we chose to include it in Fig 1.
Figure 2 provides a forest plot of the odds ratios for the various AIs based on URI status by using no URI as the referent from the multivariable logistic regression model that controlled for the same characteristics as the model used to create Fig 1. As with AAEs, there were statistically significant differences between the odds ratios for various AIs based on the characteristics and timing of their URI.
URIs are ubiquitous in children.14 Therefore, it is not surprising that children with active or recent URIs will present for therapeutic or diagnostic procedures facilitated by procedural sedation. Given the high prevalence of URIs in childhood, a generalized decision to cancel or delay a sedation and procedure because of a URI would have an enormous impact on the ability to accomplish needed tests and procedures in this patient population. The deferral or delay of a procedure could adversely affect outcomes for children undergoing cancer treatment or other time-sensitive procedures. Even if it would not affect patient health, the blanket cancellation of procedures for all children with URIs would lead to costly logistical problems for parents and medical centers alike. Understanding the potential risks and complications of providing sedation to a child with an active or recent URI is necessary to assist practitioners in developing an optimal sedation plan or in the decision to defer a procedure. Appreciation of the rate and nature of AAEs and AIs associated with URIs will allow appropriate counseling of patients and families concerning the risk versus benefit of procedural sedation.
URIs are associated with AAEs during general anesthesia in children. Investigators in the field of anesthesia have defined AAE to include specific events such as oxygen desaturation, laryngospasm, bronchospasm, coughing, and breath-holding.5,6,15–17 Review of this data has led to the conclusion that when the URI is not associated with fever or any lower respiratory tract or systemic symptoms the majority of these AAEs are minor in nature and do not indicate that anesthesia is contraindicated. Whether children are at the same risk (or greater) for AAEs during procedural sedation during an uncomplicated URI is uncertain. There is limited data in this area of practice. Grunwell et al18 performed a retrospective case-control evaluation of children who had failed procedural sedation in a single institution. Eighty-three cases of failed sedation between January 2007 and December 2011 were compared with a convenience sample of 523 successful sedation cases between January 2011 and February 2011. Thirteen patient characteristics and comorbidities were subsequently analyzed for possible association with failed sedation. After regression analysis, the presence of URI, obstructive sleep apnea or snoring, ASA classification III, and older age were likely to predict sedation failure. However, the presence of a URI alone was a poor predictor of whether a child would fail sedation. Furthermore, no correlation between the presence of URI and the nature of any AAE could be made in this study.
We are aware of no prior studies that have examined the relationship between current or recent patient URI and AAEs during pediatric procedural sedation. Our data demonstrate that both current and recent URI symptoms in children are associated with an increased risk of specific AAEs during pediatric procedural sedation. Furthermore, it appears that the character of the URI is an important determinant of risk for AAEs. Children with a current URI were more likely to experience airway obstruction, oxygen desaturation, snoring, cough that interfered with the procedure, secretions requiring suctioning, laryngospasm, stridor, wheezing, any AAE, or an inability to complete the procedure because of a sedation related problem than those children who did not have a URI. Children with URI were also more likely to receive AIs, including intubation. However, they do not appear to have a statistically significant increased risk of apnea or need for emergent AIs when compared with those without URI. This implies that intubation, in the few cases that it was used, was likely an intended part of the sedation plan. And for almost all patients who were intubated, the sedation provider was an intensivist or anesthesiologist.
Furthermore, our data indicate that the character of secretions matters: children with URIs characterized by thick and/or green secretions were more likely to experience airway obstruction, oxygen desaturation, snoring, secretions requiring suction, wheezing, or the occurrence of any AAE than those children with URI characterized by thin and/or clear secretions. Children with a recent URI (within 2 weeks of the procedure, but not with active symptoms at the time of the procedure) were more likely to experience airway obstruction, oxygen desaturation, cough that interfered with the procedure, secretions requiring suctioning, stridor, emergent AI, wheezing, and any AAE than those children who did not have a URI. There was no increased risk of a non-AAE for either children with a current or a recent URI when compared with children without a URI. Our use of multivariable regression analyses to account for variables characterizing demographics, underlying medical problems, procedures, sedative medications, and adjunctive medications, which could independently impact AAEs, failed to show any effect of these factors and strengthens the validity of the association between URI and AAEs as well as AIs during procedural sedation.
It is interesting to note that the odds ratios were higher for patients with a current URI versus a recent URI for all AAEs except for the need for an emergent AI and apnea. Our findings are similar but distinct when compared with those in the anesthesia literature17 in which AAEs are noted to be significantly increased over baseline for the 2 weeks after the resolution of a URI. Subtle differences in these findings are likely because of the more invasive airway management that is associated with general anesthesia versus procedural sedation.17 Airway hyper-reactivity in association with a viral infection diminishes with resolution of the infection, but the increased odds for an emergent AI in the recent URI group highlights the need for ongoing vigilance during procedural sedation in this 2-week period after a URI.
As with any study that includes tens of thousands of patients, we found statistical significance in some AAE subcategories in which the absolute difference in incidence is relatively small. For example, although the difference in laryngospasm between those with a URI (0.7%) and without a URI (0.3%) is statistically significant, the clinical significance may not be of great concern to the sedation provider. On the other hand, the consistency of the increase in risk between the different URI categories is noteworthy, as is the cumulative adverse event rate, which was found to be over 15% in those with a URI and 6.3% in those without.
Notable demographic findings include an older age for patients with no recent or current URI who undergo sedation. It is likely that older patients have already developed an immunity to many of the viruses that cause URI symptoms. Medication utilization also differed for patients with a current URI versus no URI, with current URI patients receiving more propofol and less ketamine and narcotic. This may be because of the increased frequency of URI in patients who were sedated for MRI compared with those who underwent a painful procedure. The ease of titration and rapid offset of propofol would be desirable in patients with an active URI, as would avoiding the sialorrhea associated with ketamine. Anticholinergic medications were also more commonly used in patients with a current URI than those without a URI, although the impact of this practice on decreasing AAEs is uncertain. A meta-analysis of ketamine use for pediatric procedural sedation found an increase in adverse events associated with use of anticholinergics in conjunction with ketamine.19 Our data indicate the need for additional study regarding the need for adding anticholinergics when sedating children with URIs.
The large number of patients available for analysis because of our multicenter approach is an important strength of this study. Our study incorporated specific questions about the presence of a URI to our existing prospectively collected database for the duration of the study period. These focused questions allow for identification of patients with current and recent URI. It is likely that our identification of patients with a current URI is more accurate than those with a recent URI, which is the reason they are present in a greater number. On the other hand, those identified as having had a recent infection are almost certainly accurately identified. Furthermore, complications were specifically defined a priori for consistency in data collection. Our results reflect the effect of URIs on AAEs across various pediatric procedural sedation programs, institutions, provider types, and medication regimens. Thus, our results should be minimally impacted by individual center effect and applicable to many current pediatric procedural sedation practices.
Previous reports from the PSRC have outlined the limitations of our database.7,12 This is a retrospective analysis of prospectively collected data and has all the inherent difficulties associated with observational data. Blinding and randomization are not possible, and although definitions of the various AAEs are provided with the database entry tool, there may be some variation in various providers’ threshold for calling out a particular AAE. Twenty-one percent of database entries during the study period could not be used because of URI status or other data elements not being recorded, which may have caused a selection bias. Unfortunately, subgroup analysis is not possible with this database model. Additionally, we are not able to comment on the number of patients who may have presented for sedation only to have been turned away because of the severity of URI symptoms. It is possible that our selection of patients with URIs represents a group with milder symptoms. We also recognize that the centers involved in the PSRC are willing to devote significant time and effort to this quality improvement project and, as such, may be higher performance systems than average. The data presented here may not reflect data that would be collected from all centers around the United States or internationally where procedural sedation care is provided for children.
URIs are common in children who present for procedural sedation. Based on the results of this large database study, children with current URIs have an increased occurrence of some AAEs and AIs. Among those children with current URI, there is an increased rate of some AAEs and AIs for those with thick and/or green secretions compared with those with thin and/or clear secretions. Despite the statistically significant relationships that we found, the rates of major AAEs such as laryngospasm, aspiration, emergent AI, unplanned admission and emergent call for anesthesia all remained <1% regardless of URI status. In many instances, the increased rate of AAEs and AIs that we document reaches statistical significance, but is of questionable clinical relevance. Similar to the data that has been produced concerning general anesthesia, our data supports the general safety of procedural sedation in children with URIs.
The authors recognize the members of the Society for Pediatric Sedation without whom this project would not have been completed. The PSRC database is maintained as a core function of and with funding from the Society for Pediatric Sedation.
The authors wish to recognize Susan Gallagher for her contributions to this project.
- Accepted April 19, 2017.
- Address correspondence to Michael D. Mallory, MD, MPH, Emergency Medicine, Children’s Healthcare of Atlanta at Scottish Rite, 2133 Kodiak Drive, Atlanta, GA 30345. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Dingle JH,
- Badger GF,
- Jordan WS
- Cravero JP,
- Blike GT,
- Beach M, et al; Pediatric Sedation Research Consortium
- Cravero JP,
- Beach ML,
- Blike GT,
- Gallagher SM,
- Hertzog JH; Pediatric Sedation Research Consortium
- Langhan ML,
- Mallory M,
- Hertzog J,
- Lowrie L,
- Cravero J; Pediatric Sedation Research Consortium
- Scherrer PD,
- Mallory MD,
- Cravero JP,
- Lowrie L,
- Hertzog JH,
- Berkenbosch JW; Pediatric Sedation Research Consortium
- Chonmaitree T,
- Revai K,
- Grady JJ, et al
- Green SM,
- Roback MG,
- Krauss B, et al; Emergency Department Ketamine Meta-Analysis Study Group
- Copyright © 2017 by the American Academy of Pediatrics