Effect of adenotonsillectomy in children with obstructive sleep apnea and major psychiatric disorders on obstructive apnea-hypopnea index and Epworth Sleepiness Scale scores

Abstract

Study Objectives:

This study evaluates the effectiveness of adenotonsillectomy in the treatment of obstructive sleep apnea in children with major psychiatric disorders as measured by polysomnography and the Epworth Sleepiness Scale (ESS) at a tertiary children’s hospital. Adults with major psychiatric disorders often have higher rates of obstructive sleep apnea and decreased response to treatment. The goal was to determine if children with serious mental illness had outcomes similar to their adult counterparts.

Methods:

A retrospective chart review was undertaken to identify children with obstructive sleep apnea and major psychiatric disorders who underwent adenotonsillectomy as part of their treatment for obstructive sleep apnea and had undergone preoperative and postoperative polysomnography as well as ESS. A multivariable model was run for each of the postoperative outcomes (ESS, obstructive apnea-hypopnea index, or body mass index percentile), adjusting for their respective preoperative value, age, and group.

Results:

There were 34 patients who qualified for this study and who were matched with 66 controls. There was no significant difference between the 2 groups in terms of change in ESS, obstructive apnea-hypopnea index, or body mass index percentile, both before and after adjusting for age. The only significant findings were that preoperative ESS and body mass index percentile were predictive of postoperative ESS and body mass index percentile for both groups.

Conclusions:

Children with psychiatric disorders in our institution respond to surgical management of obstructive sleep apnea similar to pediatric controls without mental illness despite comorbidities and central-acting medications that may alter sleep.

Citation:

Scholes M, Jensen E, Meier M, Friedman N. Effect of adenotonsillectomy in children with obstructive sleep apnea and major psychiatric disorders on obstructive apnea-hypopnea index and Epworth Sleepiness Scale scores. J Clin Sleep Med. 2021;17(4):685–689.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Children with psychiatric disorders are at increased risk of obesity and can have abnormal sleep cycles, which predispose them to the development of sleep disorders. While there are many studies on the effects of major psychiatric disorders on sleep in adults, there are few publications in the pediatric literature. This study aims to evaluate the effectiveness of adenotonsillectomy in the treatment of obstructive sleep apnea in children with psychiatric disorders as measured by polysomnography and the Epworth Sleepiness Scale compared with children with obstructive sleep apnea without psychiatric diagnoses undergoing similar evaluation and treatment.

Study Impact: Our results show that children with obstructive sleep apnea and psychiatric disorders had similar outcomes to children with obstructive sleep apnea without psychiatric disorders.

INTRODUCTION

Obstructive sleep apnea (OSA) in children has been increasingly recognized as a major health problem. In otherwise healthy children adenotonsillectomy (T&A) can be curative in up to approximately 79% of patients and is often the first-line treatment.^1–4 However, there are populations of children with comorbidities in which surgical intervention is less successful. Children with craniofacial disorders, syndromes, obesity, and multiple medical problems with OSA may need adjunctive therapies such as continuous positive airway pressure therapy and further surgery to help alleviate the symptoms and sequelae of OSA.^5,6 Children with psychiatric disorders are at increased risk of weight gain and obesity, which is an independent risk factor for OSA.^7,8 There are multiple studies in the adult literature investigating the relationship between OSA and psychiatric disorders, including the role of obesity, medications, and outcomes of psychiatric disease after treatment of OSA. Conversely, there are limited data on outcomes in children with psychiatric disorders and OSA. Some studies have investigated the relationship of OSA and depression in children and have shown both a link between increased incidence of OSA in children with depressive symptoms as well as risk of developing depressive orders in children with OSA.^9,10 However, it is unclear how effective our routine OSA surgical treatment strategies are for these children. In children, OSA is often treated surgically while adults are more likely to be treated with positive airway pressure therapy in the form of continuous positive airway pressure or bilevel positive airway pressure. This study aims to evaluate the effectiveness of T&A in the treatment of OSA in children with psychiatric disorders as measured by polysomnography (PSG) and the Epworth Sleepiness Scale (ESS). Tiredness, lack of energy, concentration problems, sleepiness or hypersomnolence, and insomnia are common complaints of both OSA and psychiatric disease.^11,12 Our hypothesis is that this population has less improvement compared with controls without psychiatric diagnoses secondary to comorbidities and mood disorders.

METHODS

After institutional review board approval, an electronic medical record search was performed for patients under the age of 18 between 1 January 2010 and 31 December 2015 with diagnosis of OSA and a major psychiatric disorder as defined by the Diagnostic and Statistical Manual of Mental Disorders, fifth edition (DSM-5), who underwent T&A for OSA. Patients also needed to have preoperative and postoperative PSG and ESS data. Major psychiatric disorders queried include major depressive disorder, bipolar disorder, schizophrenia, disruptive mood disorder, mood disorder, major anxiety, disruptive impulse control, and conduct disorders. This study group will be referred to as “SMI” for serious mental illness. Children with autism spectrum disorder, intellectual disability, and mental retardation were not included as we utilized the ESS to measure subjective sleepiness before and after the intervention and this is hard to assess in these populations. Our control group included patients under the age of 18 with a diagnosis of OSA who underwent T&A for OSA who had preoperative and postoperative PSG and ESS data. The charts of these patients were examined for any previous or current history of psychiatric disorders, and if so, were excluded. Children in both groups with attention-deficit/hyperactivity disorder were excluded as there have been studies suggesting that OSA may be linked to attention-deficit/hyperactivity disorder and we did not want this as a confounder.¹³ Subjects from both groups were also excluded if they had any previous surgery related to OSA.

Demographic data were obtained including sex, race, and age (Table 1 and Table 2). Outcome information we analyzed included body mass index percentile (BMI%), obstructive apnea-hypopnea index (OAHI), and ESS (Table 1). Basic demographic and outcome information is summarized as frequencies with percentages for categorical data and means with standard deviations for continuous data. Three multivariate analyses were done separately for the postoperative outcomes: OAHI, ESS, and BMI% for age. Each model was adjusted for the respective preoperative variable (BMI%, ESS, or OAHI), age, and whether they had an SMI diagnosis (SMI vs control). This analysis was felt to be necessary as we were comparing the effect of the presence or absence of an SMI and we did not want BMI% and age to be confounders. Preoperative variables for each category were included in the model to reduce variability within patients and age was included in the model because of clinical relevance. Initial summaries, cleaning, and analysis were conducted using R version 3.6.1 software (http://www.R-project.org/; R Foundation for Statistical Computing, Vienna, Austria).

Table 1.

Demographics stratified by controls and study participants.

	Controls (n = 66)	SMI (n = 34)	Overall (n = 100)
Sex, n (%)
Male	35 (53.0)	23 (67.6)	58 (58.0)
Female	29 (43.9)	11 (32.4)	40 (40.0)
Missing	2 (3.0)	0 (0)	2 (2.0)
BMI percentile
Pre	0.817 (0.270)	0.743 (0.316)	0.790 (0.288)
Post	0.849 (0.240)	0.800 (0.292)	0.831 (0.260)
OAHI
Pre	19.1 (25.6)	16.9 (23.9)	18.4 (24.9)
Post	4.62 (7.05)	5.09 (4.74)	4.79 (6.28)
Epworth score
Pre	8.94 (4.62)	8.84 (5.43)	8.90 (4.93)
Post	6.23 (4.50)	7.32 (4.28)	6.66 (4.41)

BMI = body mass index; OAHI=obstructive apnea-hypopnea index; SMI = serious mental illness.

Table 2.

Additional demographic information.

	Controls (n = 66)	SMI (n = 34)	Overall (n = 100)
Age, y	10.7 (4.20)	13.0 (4.39)	11.5 (4.38)
Race, n (%)
Black/African American	8 (12.1)	4 (11.8)	12 (12)
Hispanic or Latino	26 (39.4)	6 (17.6)	32 (32)
White	19 (28.8)	13 (38.2)	32 (32)
Asian	2 (3.0)	2 (5.9)	4 (4)
More than 1 race	0 (0)	1 (2.9)	1 (1)
Unknown	11 (16.7)	8 (23.5)	19 (19)

SMI = serious mental illness.

RESULTS

A total of 34 patients with SMI met the inclusion criteria along with 66 controls. There was no significant difference between the 2 groups in terms of change in ESS, OAHI, or BMI%, both before and after adjusting for age and group. Table 1 shows the demographic data stratified by group, SMI versus control. A majority of participants in both the SMI and control groups were male (53% and 67.6%, respectively). The OAHI and BMI% were higher for controls both pre- and post-PSG when compared with the SMI group. Table 2 shows the breakdown of the study and control group by age and race. The average age of the study group was 13 years, while the control group’s average age was 10.7 years. Psychiatric diagnosis and medications are summarized in Table 3.

Table 3.

Psychiatry and medications of study participants.

	Overall (n = 34)
Psychiatric diagnosis, n (%)
Depression	16 (48.5)
Anxiety	16 (48.5)
Bipolar disorder	8 (24.2)
Schizophrenia	1 (3.0)
Pharmaceuticals (pre), n (%)
Antianxiety	2 (6.1)
Antidepressants	17 (51.5)
Antipsychotics	10 (30.3)
Mood stabilizer*	9 (27.3)
Pharmaceuticals (post), n (%)
Antianxiety	1 (3.0)
Antidepressants	15 (45.5)
Antipsychotics	10 (30.3)
Mood stabilizer*	8 (24.2)

*Mood stabilizers include lithium and anticonvulsants.

The postoperative OAHI after adjusting for the preoperative OAHI and participant age was, on average, 1.06 points higher for the SMI group compared with the control group. The postoperative ESS score after adjusting for the preoperative ESS score and participant age was, on average, 0.27 points lower for the SMI group compared with the control group. The postoperative BMI% after adjusting for the preoperative BMI% and participant age was, on average, the same for the SMI group compared with the control group. Thus, there was no significant difference between the 2 groups in terms of change in ESS, OAHI, or BMI%, both before and after adjusting for age and group. Full models are summarized in Table 4.

Table 4.

Multivariate models for ESS, BMI%, and OAHI.

Predictors	ESS		BMI%		OAHI
	Estimates (CI)	P	Estimates (CI)	P	Estimates (CI)	P
Control	0.38 (0.18, 0.58)	<.001	0.79 (0.70, 0.89)	<.001	0.02 (−0.02, 0.06)	.406
Group (Study Population)	−0.27 (−2.24, 1.70)	.730	−0.00 (−0.06, 0.06)	.969	1.06 (−1.09, 3.21)	.337
Age in years	0.12 (−0.12, 0.36)	.288	−0.00 (−0.01, 0.00)	.451	0.13 (−0.11, 0.37)	.297

BMI% = body mass index percentile; CI = confidence interval; ESS = Epworth Sleepiness Scale; OAHI = obstructive apnea-hypopnea index.

DISCUSSION

Pediatric OSA is a disease different from adult OSA and T&A has a relatively high cure rate, especially compared with adults.^1,2,4 Our question was if children with OSA as well as a diagnosis of an SMI had outcomes similar to age-matched controls in terms of ESS and reduction in apnea-hypopnea index. We routinely obtain ESS during PSG in our institution. ESS is a measurement of daytime sleepiness and the assumption was that patients with SMI may have increased scores prior to and after intervention as sleepiness and hypersomnolence are often features of psychiatric disorders.^11,14 We also analyzed the outcomes in apnea-hypopnea index in the 2 groups to see if a difference existed. In our investigation, the OAHI and ESS outcomes for the SMI group were similar to those in pediatric controls. These findings suggest that the upper airway anatomy is responsible for the obstructive breathing patterns seen in our study group and that comorbidities of SMI and associated medications did not have an effect.

With regard to degree of daytime sleepiness, there was no significant difference in the ESS score preoperatively or postoperatively between the 2 groups, even when the scores were adjusted for the BMI%. In the adult literature, obesity and psychiatric disease are predictive of higher ESS scores.^11,15 Potentially, our sample size was too small as well as heterogenous to detect any significant difference or the ESS is a poor predictor of daytime sleepiness for our study population. Hypersomnia (prolonged sleep) is a common feature in psychiatric disorders. Recent research has been published on the development of a scale to better understand hypersomnia in relation to sleep disorders as the ESS and other sleep measures may not be sufficient.¹⁶ While the ESS does evaluate daytime sleepiness, it is just a single measurement of sleep-associated symptoms that could be related to psychiatric conditions. Attention should be paid to other sleep instruments as they develop to see if they are applicable to pediatric and adolescent populations.

OSA has been found at a higher prevalence in certain psychiatric disorders, including major depressive disorder, bipolar disorder, schizophrenia, and post-traumatic stress disorder.^11,12,14 Obesity has also been linked to depression, which is a risk factor for OSA.¹⁷ Adolescents with psychiatric disorders are also at a higher risk of obesity.¹⁸ These relationships lead to increased risk of OSA in pediatric patients with SMI and vice versa.

Serotonin is known to affect sleep in several ways and is often affected by psychiatric illness as well as the medications to treat these disorders. A recent systematic review showed a higher rate of OSA in adult patients with SMI, with the highest prevalence in people with major depressive disorder.¹⁴ Another study found that, in adults with bipolar disorder, the prevalence of OSA may be as high as 47.5%.¹⁹ It has been suggested that this is related to sleep fragmentation associated with hypoxemia, altered serotonin metabolism, as well as common shared risk factors such as obesity.²⁰ The interplay between these factors still is not clear. Serotonin plays a role in upper airway dilation through the hypoglossal nucleus, which can improve or decrease hypopharyngeal tone. Serotonin can also reduce rapid eye movement sleep. Thus, alterations in serotonin, which is common in psychiatric disorders as well as in the medications used in treatment, has significant effects on sleep.^21,22 Our study population included children with differing diagnoses and taking different classes of psychiatric medications (Table 3) and is too heterogenous to make any conclusions based on the small number. However, it is important to realize the effects that serotonin and medications that alter serotonin levels may play in adolescents with OSA whom we may encounter in a pediatric setting. The interrelationships of obesity, OSA, and depression in adolescence in terms of effects of serotonin should be explored further as some studies examining these comorbidities in adults suggest that treating patients with OSA and depression with serotonin-modulating medications may improve OSA.^23,24 Additionally, patients with SMI and untreated OSA have exacerbation of psychiatric symptoms.¹²

The Childhood Adenotonsillectomy Trial, which is a study in 453 children with OSA randomized to adenotonsillectomy versus observation, looked at a subsample of children who completed the Children’s Depression Inventory. The Children’s Depression Inventory is a self-reported measure of depression in children ages 7 and above. This study found that there was an increased risk of depression and withdrawal symptoms in children with OSA. Over time, there was improvement in depressed symptoms; however, there was no difference between the treatment groups (surgery vs observation) or in the change of severity of OSA, including resolution.²⁵ Similar to our population, the presence of depressive symptoms did not portend a difference in the outcomes after T&A for OSA.

Our study has several limitations. As a retrospective review the data obtained may be subject to reporting error. However, we did have the PSG and ESS results for all participants, which limits the subjective nature of these data. Since children with SMI make up only 3–7% of the pediatric population according to the Centers for Disease Control and Prevention data on children’s mental health, our study population is small and heterogenous.²⁶ Several different diagnoses were included in our study group to increase the power of the study. Studies of single psychiatric diagnoses may show a difference between the groups if there is a larger number of participants.

We do not routinely administer any assessments of mental health such as the Children’s Depression Inventory, so we may miss patients who have depressive symptoms who do not have a formal diagnosis. The average age of our study group was 13 years, and the control group’s average age was 10.7 years. During our chart review, many of the older children in our dataset did not undergo postoperative sleep studies as a result of family preference or poor follow-up. We did adjust for this limitation within our multivariate analyses. Ideally, it would be beneficial to see if there was benefit of T&A and decreased OAHI on mood and mental health as evidenced by a decrease in symptoms or medication adjustments. However, information on psychiatric health is not able to be accessed through retrospective chart review in our institution without special permission; and since the number of study participants was small, we did not pursue this course. A multi-institutional study would aid in providing more participants for a comprehensive analysis.

Although children and adolescents with SMI make up a small number of our pediatric ear-nose-and-throat population, we must keep in mind their comorbidities, possible effects of central-acting medications, and potential for continued sequelae of OSA after routine treatment. Despite these treatment challenges, the OSA in this population should be treated like in any other child. Our study population had a meaningful reduction in OAHI and ESS similar to our control group. This is an important finding as it may be related to improvement in psychiatric symptoms. Additional studies looking at long-term outcomes of this population as they transition into adulthood would be beneficial.

DISCLOSURE STATEMENT

All authors of this publication have seen and approved this manuscript. Work for this study was performed at Children’s Hospital Colorado. The study was funded by the Children's Hospital Center for Research in Outcomes in Children’s Surgery. The authors report no conflicts of interest.

ABBREVIATIONS

BMI%

body mass index percentile

ESS

Epworth Sleepiness Scale

OAHI

obstructive apnea-hypopnea index

OSA

obstructive sleep apnea

PSG

polysomnography

SMI

serious mental illness

T&A

adenotonsillectomy