Opioid dose and postoperative respiratory adverse events after adenotonsillectomy in medically complex children

Abstract

Study Objectives:

Obstructive sleep-disordered breathing is commonly treated with adenotonsillectomy. Our study objective was to describe perioperative opioid dosing in children with a range of medical complexity evaluated for obstructive sleep-disordered breathing undergoing adenotonsillectomy and to investigate its association with postoperative respiratory adverse events (PRAEs).

Methods:

A retrospective chart review of children who underwent adenotonsillectomy and had preoperative polysomnography performed was conducted. PRAEs included requiring oxygen, jaw thrust, positive airway pressure, or mechanical ventilation. Multivariable logistic regression was performed to examine for associations between covariates and PRAEs.

Results:

The cohort included 374 children with obstructive sleep-disordered breathing, median (interquartile range) age 6.1 (3.9, 9.3) years; 344 (92%) had obstructive sleep apnea (apnea-hypopnea index > 1 events/h) while 30 (8%) had a normal polysomnogram (apnea-hypopnea index < 1 events/h). The median (interquartile range) postoperative morphine-equivalent dose administered was 0.17 (0.09, 0.25) mg/kg. Sixty-six (17.6%) experienced at least 1 PRAE. Multivariable modeling identified the following predictors of PRAE: younger age at surgery (odds ratio 0.90, 95% confidence interval 0.83, 0.98), presence of cardiac comorbidity (odds ratio 2.07, 95% confidence interval 1.09, 3.89), and presence of airway anomaly (odds ratio 3.48, 95% confidence interval 1.30, 8.94). Higher total apnea-hypopnea index and morphine-equivalent dose were associated with PRAE risk, and an interaction between these variables was detected (P = .01).

Conclusions:

This study identified opioid dose in morphine equivalents to be a strong predictor of PRAE. Additionally, severity of obstructive sleep apnea and postoperative morphine-equivalent dose contributed together and independently to the occurrence of PRAEs. Attention to opioid dosing, particularly among medically complex children with obstructive sleep-disordered breathing, is required to mitigate risk of PRAEs.

Citation:

Tsampalieros A, Murto K, Barrowman N, et al. Opioid dose and postoperative respiratory adverse events after adenotonsillectomy in medically complex children. J Clin Sleep Med. 2022;18(10):2405–2413.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Children with obstructive sleep-disordered breathing who are treated with adenotonsillectomy are at risk of postoperative respiratory events and may also be more sensitive to the effects of perioperative opioid dosing. Our aim was to describe perioperative opioid dosing and better understand its role in postoperative respiratory events in a medically complex cohort.

Study Impact: This is one of the largest studies that examined the association of postoperative opioid dosing as well as clinical predictors with postoperative respiratory events postadenotonsillectomy. In this cohort of medically complex children, both severity of obstructive sleep apnea and postoperative morphine dose contributed together and independently to the occurrence of postoperative respiratory events prior to hospital discharge.

INTRODUCTION

Obstructive sleep-disordered breathing (SDB), of which obstructive sleep apnea (OSA) is an extreme form, is present in 1–5% of children¹ and is associated with neurocognitive impairment, poor growth, and cardiovascular and metabolic disease.^1–8 In children, OSA is often caused by upper airway obstruction from adenotonsillar hypertrophy and is commonly treated with adenotonsillectomy, which can result in postoperative respiratory adverse events (PRAEs).^1,2,9–13 Major postoperative respiratory complications (defined as events requiring significant intervention) have been reported to occur in 1–5% of all children undergoing adenotonsillectomy but are approximately 5 times more frequent (20%) in children with OSA.^3–5,8

Postsurgical pain associated with adenotonsillectomy, rated moderate to severe, is traditionally treated with a multimodal approach of acetaminophen, nonsteroidal anti-inflammatory agents, and opioids as needed.^6,7,14 Opioids remain a cornerstone of therapy for pain management in pediatric hospitals.¹⁵ Their analgesic and adverse drug effects are determined by serum drug concentrations.¹⁶ Opioid administration after adenotonsillectomy in children has been reported to be associated with respiratory complications,^17–21 the majority of which occur within the first 12–24 hours after surgery.^3,5,18,19,22 There have also been rare reports of death or anoxic brain injury²³ related to a genetic predisposition to codeine ultra-rapid metabolism to morphine; as a result codeine is no longer used in this population.

The association between SDB and postsurgical respiratory complications has been thought to be caused by an effect of recurrent oxygen desaturations on endogenous opioid mechanisms that increase one’s sensitivity to both analgesic and respiratory depressive responses to exogenous opiates.^21,24,25 In fact, there is growing concern of both heightened sensitivity to the analgesic and respiratory effects of opioids, as well as evidence for increased pain experienced after adenotonsillectomy that may coexist in the same individual with OSA, increasing their risk of an opioid-related PRAE.²⁶ To date, 2 pediatric studies^27,28 have examined the association between opioids and PRAEs; however, they looked exclusively at intraoperative dosing and PRAEs occurring in the postanesthesia care unit (PACU).

The main objective of our study was to investigate the association between all opioids (converted into morphine equivalents) administered during the perioperative period with recorded postoperative respiratory complications prior to hospital discharge in medically complex children evaluated preoperatively for obstructive SDB, who underwent elective adenotonsillectomy. Additional objectives were to determine whether polysomnography (PSG) parameters, including the total apnea-hypopnea index (AHI) and/or the presence of a comorbidity (specifically congenital heart disease and/or the presence of an airway anomaly), increased the likelihood of a PRAE with increased opioid dose. We hypothesized that children who received greater doses of morphine equivalents would be at a higher risk of experiencing a PRAE and that those with more severe OSA would be at higher risk of PRAEs with a given dose of morphine than those without OSA. Clinical predictors of PRAEs have been evaluated and reported in this study population, although perioperative medications have not been considered in our previous analyses.^10,29,30

METHODS

Study population

The methodology for this study was previously reported.¹⁰ This retrospective chart review was conducted at the Children’s Hospital of Eastern Ontario, Ottawa, Canada. Children aged 0 to 18 years who underwent a PSG and then adenotonsillectomy between January 2004 and December 2016 were included. There were no exclusion criteria. Ethics approval was obtained from the hospital research ethics board. Data collection was managed using Research Electronic Data Capture Database.³¹

All children included in this study had previously undergone overnight in-laboratory PSGs, which were performed according to American Academy of Sleep Medicine guidelines³² and included 6-lead electroencephalogram, 2-lead electrooculogram, submental and leg electromyogram, electrocardiogram, nasal pressure and/or thermistor monitoring, chest and abdominal movement monitoring by impedance, and continuous end-tidal and/or transcutaneous carbon dioxide monitoring (Natus Xltek, Oakville, Ontario, Canada). Oxygen saturation was recorded by pulse oximetry with motion resistant technology (Massimo, Irvine, CA). All PSGs were interpreted by 1 of 2 pediatric sleep medicine physicians according to American Academy of Sleep Medicine criteria. Preoperative PSG results collected included the total AHI (events/h), obstructive AHI (events/h), and the lowest oxygen saturation nadir (%). The total AHI was used to categorize OSA as follows: total AHI > 1–5 events/h as mild OSA, total AHI > 5–10 events/h as moderate OSA, and total AHI > 10 events/h as severe OSA.

Demographics and medical history

Medical charts were reviewed for demographic data (sex and age at the time of surgery as well as height and weight at time of PSG and surgical procedure). Preoperative comorbidities recorded included the presence of congenital heart disease, presence of airway anomalies, genetic syndromes, lower respiratory problems, neurological problems, endocrine problems, gastrointestinal problems, and prematurity and obesity (class 2 or 3). Obesity was defined as a body mass index (BMI) > 95% for age and sex.³³ The %BMIp95 was calculated as a percentage of the specific 95th percentile for sex and age. Class 1 obesity was defined as %BMIp95 of 100–120, class 2 as 120–150, and class 3 as %BMIp95 > 150.³⁴

Information collected on PRAEs included 1) the location where they occurred and 2) the type of intervention they required. Information on opioid dosing intraoperatively and postoperatively were also collected. Information collected on opioids included the type of opioid administered (ie, morphine, fentanyl, hydromorphone, remifentanil, codeine, and sufentanil), the dose administered, the frequency, and the route (oral and intravenous). All opioids were converted into intravenous morphine equivalents per kilogram with the help of a pharmacist and expressed as milligrams per kilogram for each participant (the primary exposure measure) (see Table S1 ^{(369.8KB, pdf)} in the supplemental material). Length of stay for those admitted to hospital included time spent in either the PACU or postoperative daycare unit (hours converted into days) as well as the time in days admitted to the ward or pediatric intensive care unit (PICU). To answer a hypothesis that children who stayed in hospital longer received higher doses of opioids, the dose of morphine per day, which is a novel way of considering morphine dose, was calculated and was used only for descriptive purposes and was not incorporated into any of the models. Median dose of morphine while admitted was calculated by taking the total dose of morphine administered and dividing by the length of stay in days (exploratory analysis).

Statistical analyses

All statistical analyses were performed using R statistical software.³⁵ Continuous variables were expressed as mean ± standard deviation or median (IQR [interquartile range]) for nonnormal distributions. A Kruskal–Wallis test was performed to test for differences in total perioperative morphine-equivalent dose by OSA severity. The outcome of interest was the occurrence of any major type of PRAE (including requiring oxygen, jaw thrust, positive airway pressure/bag-mask ventilation or endotracheal intubation and ventilation) at any time during the admission. At our institution, supplemental oxygen was generally initiated in the PACU if the patients came to PACU unconscious (ie, deep residual general anesthetic) or if they were awake and unable to maintain O₂ saturations on room air > 92%. For the purposes of this study, oxygen desaturation requiring oxygen therapy referred to a sustained need for oxygen administration after the patient was awake and/or arousable for more than 30 minutes, as our intent was to identify more severe PRAEs. On the floor, order sets typically stated to provide oxygen to keep saturations > 92%.

The primary analysis included examining the relationship between covariates of interest, which were determined a priori based on the findings from our prior manuscript and PRAEs.¹⁰ The final multivariable logistic regression model thus included the presence of cardiac disease and airway anomaly, age at the time of surgery as a continuous variable, total AHI expressed as events per hour and oxygen saturation nadir from the preoperative PSG, and postoperative opioid dose expressed as total morphine-equivalent dose in milligrams per kilogram. An interaction term between postoperative morphine-equivalent dose and total AHI was also included to test the hypothesis that both may influence respiratory depression. A sensitivity analysis was performed in which the same analysis was carried out separately for the subgroup of participants without any comorbidities and participants with comorbidities. Secondary analyses included examining for (1) an association between intraoperative dosing of morphine equivalents, PACU dosing of morphine equivalents, and whether preoperative midazolam was administered with PRAEs occurring in the PACU and (2) PACU morphine-equivalent dosing and floor/PICU dosing in association with PRAEs occurring while admitted to the floor or PICU.

RESULTS

Participant disease characteristics

There were 374 participants included in this study. The characteristics of the cohort have been previously described and are summarized in Table 1.¹⁰ The median (IQR) age at the time of surgery was 6.8 (4.7, 10.4) years and mean (standard deviation) BMI Z-score was 0.8 (1.6). There were 113 of the 334 participants with an available BMI Z-score (33.4%) with obesity, of whom 53/113 (46.9%) had class 1 obesity, 30 (26.5%) had class 2 obesity, and 30 (26.5%) had class 3 based on %BMIp95. There were 259/374 (69.3%) participants with at least 1 medical comorbidity and 165 (44.1%) with 2. Seventy-seven (20.6%) had cardiac disease (defined as either a resolved issue or ongoing), of whom 1 required bilevel positive airway pressure and supplemental oxygen. Twenty-three (6.1%) had an airway anomaly, which included laryngotracheomalacia (n = 11), subglottic stenosis (n = 4), craniofacial syndromes (n = 3), laryngeal cleft (n = 4), vocal cord paresis (n = 5), cleft palate repair (n = 1), and vascular ring (n = 2). Additional comorbidities included Down syndrome (n = 66), asthma (n = 91), bronchopulmonary dysplasia (n = 3), seizures (n = 22), and cerebral palsy (n = 4). All participants in the study had a PSG and 344/374 (92.0%) children had OSA, of which 112 (32.6%) were mild (total AHI > 1–5 events/h), 84 (24.4%) were moderate (total AHI > 5–10 events/h) and 148 (43.0%) were severe (total AHI > 10 events/h). For those not meeting PSG criteria for OSA, adenotonsillectomy was indicated due to clinical observation of obstructive SDB manifested as increased work of breathing during sleep or clinical impression of upper airway obstruction.

Table 1.

Baseline demographics (n = 374).

Variable
Male, n (%)	215 (57.5)
Age (years) at Surgery, median (IQR)	6.8 (4.7, 10.4)
Total AHI (events/h), median (IQR)	7.6 (3.1, 16.1)
Obstructive AHI (events/h), median (IQR)	4.0 (0.5, 11.4)
Lowest oxygen saturation (%), median (IQR)	88.0 (81.2, 91.0)

AHI = apnea/hyponea index, IQR = interquartile range.

PRAEs

Over the study period, 66 (17.6%) children experienced at least 1 PRAE: 42 had an event in the PACU, and 35 patients had an event while admitted to an in-patient unit (the floor or PICU). Of the 35 children who had a PRAE in the in-patient unit, 1 had a pre-existing oxygen requirement, defined as supplemental oxygen with bilevel positive airway pressure during sleep, prior to the surgery. Respiratory interventions required included endotracheal intubation (n = 1, 0.3%), bag and mask ventilation (n = 1, 0.3%), upper airway support (n = 18, 4.8%, which included jaw thrust, repositioning, oxygen), and oxygen desaturation requiring oxygen therapy (n = 58, 15.5%). There were 11 participants that had events occur in both the PACU and while admitted, and there were 11 participants who had more than 1 intervention for PRAEs. Of the 77 children with cardiac disease, 22/77 (29%) had a PRAE: 14 had a desaturation requiring oxygen therapy, 4 had a desaturation requiring oxygen therapy and airway support, and 4 required airway support.

Opioid dosage by location

The average morphine-equivalent (mg/kg) dose received in hospital by location is summarized in Table 2. Median (IQR) morphine-equivalent dose received intraoperatively was 0.08 (0.05, 0.10) mg/kg, and total postoperative dose was 0.17 (0.09, 0.25) mg/kg. The type of opioid, as well as frequency and dose administered in the perioperative period, is summarized in Table 2. The most common type of opioid administered intraoperatively was a longer acting agent, morphine (71.4%), and the most common opioids administered postoperatively were the long-acting agent morphine (59.1%) and short-acting agent fentanyl (52.4%). Postoperative morphine-equivalent dosing among those who received opioids varied by severity of OSA: median (IQR) for mild 0.15 (0.08, 0.23), moderate 0.17 (0.13, 0.24), and severe 0.20 (0.12, 0.27) mg/kg (P = .009). Children who received remifentanil intraoperatively received a lower postoperative morphine-equivalent dose 0.13 (0.06, 0.23) mg/kg vs 0.16 (0.05, 0.25) mg/kg.

Table 2.

Opioids given, their frequency and dose in IV morphine equivalents per kilogram by location.

Intraoperative (n = 374)
Opioid Given	Frequency; n (%)	Dose per kg; median (IQR)	Dose in IV morphine equivalents per kg (mg/kg); median (IQR)
Short acting
Fentanyl	114 (30.5)	0.90 (0.62, 1.11) mcg/kg	0.09 (0.06, 0.11)
Remifentanil	156 (41.7)	1.60 (1.00, 2.18) mcg/kg	-
Medium acting
Sufentanil	1 (0.3)	0.36 (0.36, 0.36) mcg/kg	0.36 (0.36, 0.36)
Long acting
Morphine	267 (71.4)	0.06 (0.04, 0.09) mg/kg	0.06 (0.04, 0.09)
Hydromorphone	3 (0.8)	0.01 (0.01, 0.02) mg/kg	0.08 (0.08, 0.12)
Codeine	8 (2.1)	0.94 (0.80, 1.25) mg/kg	0.08 (0.07, 0.10)
Total Morphine Equivalent*	366 (97.9)		0.08 (0.05, 0.10)
Among Those Who had Information from the PACU (n = 342)
Short acting
Fentanyl	194 (56.7)	0.63 (0.44, 0.93) mcg/kg	0.06 (0.04, 0.09)
Long acting
Morphine	61 (17.8)	0.04 (0.02, 0.07) mg/kg	0.04 (0.02, 0.07)
Total Morphine Equivalent	237 (69.3)		0.06 (0.04, 0.09)
Among Those Who Went to Daycare (n = 59)
Long acting
Morphine*	10 (16.9)	0.09 (0.07, 0.12) mg/kg	0.03 (0.02, 0.04)
Codeine	9 (15.3)	0.92 (0.83, 0.96) mg/kg	0.05 (0.04, 0.05)
Total Morphine Equivalent	19 (32.2)		0.04 (0.03, 0.05)
Among Those Who Were Admitted to the Floor or PICU (n = 319)
Short acting
Fentanyl	1 (0.3)	0.27 (0.27, 0.27) mcg/kg	0.03 (0.03, 0.03)
Long acting
Morphine	179 (56.3)	0.37 (0.22, 0.58) mg/kg	0.37 (0.22, 0.58)
Codeine	84 (26.4)	3.42 (2.14, 4.63) mg/kg	0.17 (0.11, 0.23)
Total Morphine Equivalent	260 (81.8)		0.14 (0.09, 0.21)
Total Postoperative (n = 374)
Short acting
Fentanyl	196 (52.4)	0.62 (0.43, 0.93) mg/kg	0.06 (0.04, 0.09)
Long acting
Morphine	221 (59.1)	0.32 (0.11, 0.53) mg/kg	0.32 (0.11, 0.53)
Codeine	93 (24.9)	0.15 (0.09, 0.22) mg/kg	0.15 (0.09, 0.22)
Total Morphine Equivalent	323 (86.4)		0.17 (0.09, 0.25)

*Total morphine equivalent excludes remifentanil. IQR = interquartile range, IV = intravenous, PACU = postanesthesia care unit, PICU = pediatric intensive care unit.

Length of stay, morphine dosing and PRAEs

The majority of children in this cohort (n = 319 [85.3%]) were admitted to hospital postoperatively, either to the ward (n = 278) or the PICU (n = 41) (Figure S1 ^{(369.8KB, pdf)} ). There were 41 participants admitted to the PICU, of which 38/41 (95%) were planned, 2 were unplanned, and 1 was not documented. Planned admissions were arranged at the discretion of the treating physician and considered the patient’s medical conditions, including comorbidities, monitoring, OSA, and adenotonsillectomy. For the unplanned admission, admission was due to the occurrence of a PRAE.

The median (IQR) length of stay for those admitted was 1.05 (1.04, 1.08) days, and the median (IQR) postoperative morphine-equivalent dose divided by length of stay did not vary (Table S2 ^{(369.8KB, pdf)} ). The proportion of PRAEs that occurred per number of children admitted per length of stay category was as follows: 16.0% (4/25) for those who stayed less than or 1 day, 14.3% (35/245) for those who stayed 1–2 days, and 49.0% (24/49) for those who stayed more than 2 days.

Primary analysis: multivariable model

Multivariable logistic regression modeling was carried out including clinical factors that were determined a priori based on the results of our prior publication (age at surgery, cardiac disease, airway anomaly, oxygen saturation nadir on PSG, and total AHI). Additionally, postoperative morphine-equivalent dose was considered, along with a possible interaction between this dose and total AHI (Table 3). In the multivariable model, evidence of an association with PRAEs was found for each variable except for oxygen saturation nadir on PSG. Notably, both total AHI and postoperative morphine dose equivalent per weight were associated with PRAE risk. Additionally, an interaction between these variables was detected (P = .01; Figure 1). For total AHI of 1 (minimal OSA) and 10 events/h (severe OSA), the model indicates a positive relationship between the opioid dose received and the estimated marginal mean probability of PRAE, whereas this relationship does not hold for those with a total AHI of 30 events/h (which would represent extremely severe OSA; note that 38 participants in this study had total AHI above 30 events/h) (Table 4). In a sensitivity analysis, we found an association between morphine-equivalent dosing and PRAEs in children both with and without comorbidities; however, an interaction between AHI and morphine-equivalent dosing was detected only in those with a comorbidity.

Table 3.

Multivariable modeling for PRAEs in any location among the whole cohort (n = 374).

Variable	PRAE occurred n/N	Unadjusted OR (95% CI)	P	Adjusted OR (95% CI)	P
Age at surgery (years)		0.87 (0.80, 0.94)	<.001	0.90 (0.83, 0.98)	.01
Morphine dose equivalent post op per 0.1 mg/kg		1.30 (1.09, 1.55)	.004	1.51 (1.17, 1.96)	.002
Total AHI on PSG (events/h)		1.03 (1.01, 1.05)	<.001	1.05 (1.02, 1.08)	<.001
Morphine dose equivalent × total AHI		1.01 (1.00, 1.01)	.04	0.99 (0.98, 1.00)	.01
O2 saturation nadir on PSG		0.98 (0.96, 1.00)	.07	1.00 (0.98, 1.03)	.82
Cardiac comorbidity			.01		.03
No	44/297 (14.8%)	1.0		1.0
Yes	22/77 (28.6%)	2.30 (1.26, 4.12)		2.07 (1.09, 3.89)
Airway anomaly			.01		.01
No	57/351 (16.2%)	1.0		1.0
Yes	9/23 (39.1%)	3.32 (1.32, 7.94)		3.48 (1.30, 8.94)

AHI = apnea-hypopnea index, CI = confidence interval, OR = odds ratio, PRAE = postoperative respiratory event, PSG = polysomnography.

Figure 1. Association been morphine-equivalent dose and probability of PRAEs.

Estimated odds of PRAEs by morphine-equivalent dose per kilogram and selected values of total AHI with other variables set to their observed averages. Shaded bands represent 95% confidence limits. AHI = apnea-hypopnea index, PRAE = postoperative respiratory adverse event.

Table 4.

Risk (95% CI) of PRAE by AHI and morphine dose equivalent.

Morphine Dose Equivalent per Weight*	Total AHI = 1 events/h	Total AHI = 10 events/h	Total AHI = 30 events/h	Total AHI = 50 events/h
25th %ile (0.09)	17 (95% CI 10,26)	22 (95% CI 15,32)	38 (95% CI 25, 53)	57 (95% CI 35, 77)
50th %ile (0.17)	22 (95% CI 15, 32)	27 (95% CI 19, 37)	40 (95% CI 27, 54)	54 (95% CI 34, 74)
75th %ile (0.25)	28 (95% CI 19, 40)	32 (95% CI 22, 43)	41 (95% CI 28, 56)	54 (95% CI 34, 74)

*Percentiles of morphine dose equivalent per weight were calculated from children who received postoperative opioids. AHI = apnea-hypopnea index, CI = confidence interval, PRAE = postoperative respiratory event.

Secondary analysis: multivariable modeling examining PRAEs by location

In a multivariable logistic regression model for PRAEs in the PACU, there was an association with opioids received in the PACU (odds ratio 2.05 per 0.1 mg/kg increase in morphine-equivalent dosing, 95% confidence interval 1.16–3.63, P = .02) and having received midazolam preoperatively (odds ratio 2.2, 95% confidence interval 1.09–4.42, P = .03) (Table S3 ^{(369.8KB, pdf)} ). In a separate multivariable logistic regression model for PRAEs on the floor and in the PICU in relation to opioids received on the floor and in the PICU, there was an association with morphine-equivalent dosing (odds ratio 1.42 per 0.1 mg/kg increase in morphine-equivalent dosing, 95% confidence interval 1.10–1.84, P = .01) (Table S4 ^{(369.8KB, pdf)} ).

DISCUSSION

Our study, which included medically complex children undergoing adenotonsillectomy at a tertiary care center, found a PRAE prevalence of 17.6%, along with evidence from multivariable modeling that both severity of OSA (greater AHI preoperatively) and postoperative morphine dose contribute together and independently to the occurrence of PRAEs in both the PACU and while admitted to the inpatient unit, prior to hospital discharge. The exception to this was among the children with severe OSA (total AHI > 30 events/h), where a greater morphine-equivalent dose did not increase the risk of PRAEs. These findings suggest that children with worsening OSA have increased risk of postoperative opioid-related complications, except those with severe OSA who are already known to have a high risk of PRAE.

Our findings differ from 2 prior studies which have examined for a relationship between opioids and respiratory depression. A large study recently performed by Hamilton et al²⁸ did not detect an association between intraoperative opioids and PRAEs postadenotonsillectomy. This may be explained by the following: a lower occurrence of PRAEs (2%) in the study cohort, the study did not include postoperative opioid doses and the study assessed PRAEs that occurred in the PACU only. Additionally, this cohort was less medically complex, as only 9% had an American Society of Anesthesiologists status ≥ 3. Dalesio et al²⁷ also did not find an association between intraoperative opioid use and postoperative desaturation among 319 children who underwent adenotonsillectomy. This study, however, assessed only for desaturations (defined as an oxygen saturation < 90%) within 2 hours postoperative as their outcome and did not include dosing of opioids but rather used a yes/no categorical variable. Neither of the above-mentioned studies assessed for the preoperative use of benzodiazepines.

While it has been reported that over 60% of respiratory events that occur do so in the immediate postoperative period, they may also occur the day after surgery. Factors thought to contribute to those events include blood or edema in the area of surgery,^11,14,36 increased airway compliance, and decreased neuromuscular function in children with OSA who are receiving prescribed opioids.^37,38 For this reason, it is important to consider the entire time period up until first hospital discharge in evaluating for PRAE.

Children with OSA may be more sensitive to the respiratory effects of opioids.³⁹ A study by Waters et al⁴⁰ found children with severe OSA had 10-fold higher incidence of apneas compared to controls after the same dose of fentanyl. This association between OSA and postsurgical respiratory complications may be explained by the effects of recurrent oxygen desaturations from OSA, which in developing animals have been shown to alter µ-opioid receptors in the brainstem and thus alter their central response to opioids.^25,41 Raghavendran et al⁴² were able to decrease the number of events requiring major medical intervention by 50% when both the intraoperative (from 0.13 mg to 0.10 mg) and postoperative (from 0.04 to 0.02 mg) morphine doses were decreased and intraoperative dexamethasone was administered in children with recurrent hypoxia. The reduction in postoperative complication rate was thought to be related to both the use of dexamethasone and a reduction in opioid dosing.

Despite the above findings, children with OSA may require greater doses of opioids to achieve pain control. Studies in adults have shown that those with OSA have a heightened sensitivity to pain.^43–45 The hypothesized mechanism includes hypoxia-induced systemic inflammation as well as a negative effect of sleep fragmentation on pain perception.^46,47 This finding was supported in a pediatric cohort as well. Yang et al⁴⁷ documented that while children with preoperative OSA symptoms (n = 325) received equivalent operative opioid dosing (0.09 vs 0.08 morphine sulfate equivalent/kg) to children without OSA symptoms (n = 660), those with OSA symptoms required higher rates of PACU analgesia (IV opioids) to control their pain and had higher mean pain arousal scores postoperatively. Thus, while in theory children with OSA should receive lower opioid dosing as they are more sensitive to its respiratory depressant effects, these children may require and receive greater opioid doses to control their post operative pain due to hyperalgesia.

A challenge for clinicians lies in identification of children who require extended monitoring in the postoperative period. Existing clinical care guidelines do not include opioid dose received in their recommendations, however, the Anesthesiology Closed Claims project suggests relaying this information to the involved caregivers beyond PACU.²² The American Academy of Otolaryngology–Head and Neck Surgery Foundation 2019 guidelines⁴⁸ recommend that children < 3 years of age or those with an AHI > 10 obstructive events/h and/or an oxygen saturation nadir < 80% on PSG should be kept for overnight inpatient monitoring post adenotonsillectomy given their higher risk of postoperative respiratory events. Additionally, children with certain medical comorbidities (ie, cardiac complication of OSA, Down syndrome, craniofacial anomalies) may require admission as well. The guidelines also suggest the duration, type, and location of postoperative monitoring should be determined by local practice and therefore the postoperative monitoring duration may range between 2 and 6 hours, assuming no major PRAEs or recurrent minor PRAEs occur.⁴⁹ The findings of our study suggest that opioid dose received may also need to be considered in information relayed and determining disposition planning for children with OSA postadenotonsillectomy.

To our knowledge, our study is one of the few pediatric studies to examine for associations between opioid use and PRAEs in children undergoing adenotonsillectomy who have obstructive SDB. Strengths of our study include the consideration of adverse events up until the time of discharge from hospital and the conversion of all opioid doses into morphine equivalents. The limitations of our study include that it was a single center study that included a cohort of children with medical comorbidities and thus may not be generalizable to other institutions or to the community. The retrospective nature of the study meant that our analysis was limited to the data recorded in medical records; while detailed descriptions of events were available, it is likely that not all PRAEs were documented. For example, if a participant had multiple desaturations requiring intervention in the same location this was only recorded as 1 PRAE. Furthermore, defining severe PRAEs vs transient O₂ desaturations is challenging, especially if details were missing in the medical record (for example whether a patient was awake). We have attempted to mitigate this by applying a PRAE definition that includes prolonged oxygen administration at a time when the child was arousable or awake. Additionally, as the exact time for the administration of opioids was not consistently and/or precisely recorded nor was the exact time for the PRAEs, we could not confirm the temporal relationship and appreciate this is a limitation of the study. Lastly, but importantly, CYP2D6 genotyping to assess if any children were ultra-rapid metabolizers of codeine was not available for our cohort; however, only 25% of patients in the cohort received codeine, with the most recent adenotonsillectomy being in 2011, as codeine has been discontinued at our institute. An ultra-rapid metabolizer with multiple copies of the gene will produce large quantities of active morphine after codeine administration, which may increase the risk of opioid induced respiratory depression.⁵⁰ Khetani et al¹⁴ found among 21 children who underwent adenotonsillectomy and had CYP2D6 testing, 2 were ultra-rapid metabolizers, 15 were extensive, and 4 were intermediate metabolizers. The genotype, however, did not predict postsurgical changes in respiratory parameters. Finally, although opioids were equated in this study by converting them to morphine equivalents, the duration of action of the opioids received was not considered in this study and may impact PRAEs.

CONCLUSIONS

In summary, our study found an association between opioid dosing, OSA, and PRAEs. Our findings highlight the need for close monitoring not only in the immediate postoperative period but potentially for a 24-hour period. Increased vigilance in opioid dosing is needed in children with moderate to severe OSA. The findings from our study should be further validated in a prospective cohort study which also considers the use of preoperative oral sedation (eg, midazolam), intravenous dexamethasone, and duration of action of opioids that could further influence development of PRAEs in this population.

ABBREVIATIONS

AHI

apnea-hypopnea index

BMI

body mass index

IQR

interquartile range

OSA

obstructive sleep apnea

PACU

post anesthesia care unit

PICU

pediatric intensive care unit

PRAE

postoperative respiratory event

PSG

polysomnography

SDB

sleep-disordered breathing

DISCLOSURE STATEMENT

All authors have seen and approved this manuscript. Work for this study was performed at the Children's Hospital of Eastern Ontario. The authors have no conflicts of interest to disclose.