Associations of actigraphy measures of sleep duration and continuity with executive function, vigilance, and fine motor control in children with snoring and mild sleep-disordered breathing

Abstract

Study Objectives:

Children with snoring and mild sleep-disordered breathing may be at increased risk for neurocognitive deficits despite few obstructive events. We hypothesized that actigraphy-based sleep duration and continuity associate with neurobehavioral functioning and explored whether these associations vary by demographic and socioeconomic factors.

Methods:

298 children enrolled in the Pediatric Adenotonsillectomy Trial, ages 3 to 12.9 years, 47.3% from racial or ethnic minority groups, with habitual snoring and an apnea-hypopnea index < 3 were studied with actigraphy (mean 7.5 ± 1.4 days) and completed a computerized vigilance task (Go-No-Go) and a test of fine motor control (9-Hole Pegboard). Caregivers completed the Behavior Rating Inventory of Executive Function. Regression analyses evaluated associations between sleep exposures (24-hour and nocturnal sleep duration, sleep fragmentation index, sleep efficiency) with the Behavior Rating Inventory of Executive Function Global Executive Composite index, pegboard completion time (fine motor control), and vigilance (d prime on the Go-No-Go), adjusting for demographic factors and study design measures.

Results:

Longer sleep duration, higher sleep efficiency, and lower sleep fragmentation were associated with better executive function; each additional hour of sleep over 24 hours associated with more than a 3-point improvement in executive function (P = .002). Longer nocturnal sleep (P = .02) and less sleep fragmentation (P = .001) were associated with better fine motor control. Stronger associations were observed for boys and children less than 6 years old.

Conclusions:

Sleep quantity and continuity are associated with neurocognitive functioning in children with mild sleep-disordered breathing, supporting efforts to target these sleep health parameters as part of interventions for reducing neurobehavioral morbidity.

Clinical Trial Registration: Registry: ClinicalTrials.gov; Name: Pediatric Adenotonsillectomy for Snoring (PATS); URL: https://clinicaltrials.gov/ct2/show/NCT02562040; Identifier: NCT02562040.

Citation:

Robinson KA, Wei Z, Radcliffe J, et al. Associations of actigraphy measures of sleep duration and continuity with executive function, vigilance, and fine motor control in children with snoring and mild sleep-disordered breathing. J Clin Sleep Med. 2023;19(9):1595–1603.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Snoring without frequent obstructive or hypoxemic episodes (“mild” sleep-disordered breathing) occurs commonly in children and has the potential to contribute to neurobehavioral morbidity and low academic performance. It is unclear what mechanisms underlie such deficits and specifically whether short sleep duration or fragmented sleep contribute to neurobehavioral morbidity in these children.

Study Impact: In children with mild sleep-disordered breathing, shorter sleep duration and poor sleep continuity were associated with deficits in executive functioning and fine motor control, with associations most evident in boys, young children, and children from low socioeconomic status households or who were Black. Given that executive functioning is fundamental for learning and academic development, our results support the need to consider both sleep duration and continuity when evaluating children with mild sleep-disordered breathing and consider their roles in sleep disparities.

INTRODUCTION

Pediatric sleep disorders are common, often undiagnosed, and frequently associated with neurobehavioral morbidity. Among the most common sleep disorders is sleep-disordered breathing (SDB), which affects 6% to 17% of children.¹ SDB may contribute to neurobehavioral impairment through multiple pathways. Hypoventilation and hypoxemia may impact neuronal apoptosis and necrosis, which can influence gray matter characteristics and cortical structure networks.^2,3 In addition, repetitive airway obstruction may alter sleep macro- and microstructure, negatively impacting sleep sensitive brain areas such as the fronto-parietal attentional network, which is important for sustained attention, executive functioning, and working memory.^4–6 Children may be especially susceptible to sleep loss due to the role of sleep in synaptic connectivity and brain maturation.⁷

Numerous studies have examined neurobehavioral outcomes in children with SDB^8–13 but have not consistently shown improvements in objective measurements of neurocognition after surgical treatment of SDB despite improvements in breathing disturbances. Most studies of children with SDB have relied on parent-reported snoring or single-night polysomnography measures rather than objective measures of sleep duration and continuity made in the child’s typical home environment. In the Childhood Adenotonsillectomy Trial, baseline polysomnography measures of SDB severity did not associate with baseline neurobehavioral functioning or predict neurobehavioral outcomes after adenotonsillectomy.^14,15 While adenotonsillectomy improved the apnea-hypopnea index (AHI), parent-reported measures of executive function, and objective measures of fine motor control,¹⁴ treatment did not lead to statistically significant improvement in the study’s primary outcome of objectively assessed executive function.¹⁵ However, the Childhood Adenotonsillectomy Trial did not objectively assess sleep duration and continuity in the child’s home. It focused on children ages 5 to 9 years with AHI levels of 2 to 30. Thus, it did not include a potentially vulnerable sample of very young children or those with milder forms of SDB. These groups were previously reported to have poorer neurobehavioral functioning than control samples and possibly more impairment compared to children with evident obstructive sleep apnea.^8,16–18

A subsequent trial, the Pediatric Adenotonsillectomy Trial for Snoring (PATS), was designed to evaluate the health and behavior outcomes in children ages 3 to 12.9 years with habitual snoring who were considered clinical candidates for adenotonsillectomy.¹⁹ Eligibility was restricted to children with an AHI less than 3 (“mild SDB”). The PATS study included wrist actigraphy to quantify sleep characteristics measured on multiple nights in the child’s home environment.

In this study, we analyzed baseline data from PATS with the aim to characterize the relationship between actigraphic measures of sleep quantity and continuity with neuropsychological and behavioral outcomes in children with mild SDB. Our outcomes included caregiver assessment of behavior and executive function and objective measures of attention and fine motor coordination. Since sleep quantity and continuity (or fragmentation) are complementary aspects of sleep health, we focused on actigraphic estimates of sleep duration and two measures of sleep continuity (fragmentation index and sleep efficiency). We hypothesized that children with longer or more continuous sleep would have better neurobehavioral functioning than children with shorter or more fragmented sleep. Since associations between sleep and neurocognition can vary by age, sex, and socioeconomic status (SES),^9,20 we also explored potential effect moderation by demographic factors and SES.

METHODS

Study design

PATS is a six-site multicenter, single-blind randomized controlled trial.¹⁹ Enrollment occurred from June 2016 to January 2021. After eligibility criteria were confirmed, including polysomnography criteria and clinical assessment by an otolaryngologist, children were scheduled for a baseline assessment at a pediatric research center at a time when free of acute illness. Assessments included clinical evaluation, anthropometry, collection of urine and blood samples, and neurobehavioral testing. Testing was performed by trained and certified research staff during a morning exam (start between 8:00 am and 9:00 am) in an area suitable for pediatric neurocognitive assessments. The research assistants were trained by two highly experienced licensed and board-certified psychologists familiar with the procedures to ensure integrity of test administration. The psychologists reviewed procedures, provided training, addressed questions regarding test administration, and monitored scoring and data checking for accuracy. Families were informed that the data were collected for research purposes only and did not constitute clinical assessments. At the conclusion of the assessment, children and their families were instructed on use of a wrist actigraph (for the subsequent 7 days), provided information on healthy sleep for children, and then randomized to one of the two treatment arms (watchful waiting and supportive care vs adenotonsillectomy). Written informed consent was obtained from caregivers and assent obtained from children over 7 years old at the start of the study. The study was approved by a single institutional review board at Children’s Hospital of Philadelphia and at the local boards at each participating site.

Eligibility criteria

PATS inclusion criteria were: (1) age 3 to 12.9 years old, (2) tonsillar hypertrophy (Brodsky ≥ 2+ or obstructing at least 25% oropharyngeal obstruction), (3) caregiver-reported snoring at least 3 nights per week, (4) adenotonsillectomy surgical candidacy, and (5) an obstructive AHI under 3 events per hour and obstructive apnea index under 1 event per hour on polysomnography. Children were excluded from the study if they had recurrent tonsillitis, a body mass index z-score ≥ 3, any arterial oxyhemoglobin desaturations under 90% in conjunction with an obstructive event, or a severe comorbidity (eg, cardiopulmonary diseases, bleeding disorders, epilepsy, severe diabetes, poorly controlled asthma), Down syndrome, or developmental delay.

Outcome assessments

Behavior and neuropsychological assessments were performed with a standardized protocol overseen by the study’s Neurobehavioral Quality Control core. Research assistants were trained by board-certified psychologists using videotaped sessions and individual feedback to facilitate accurate testing and scoring. Quality control was maintained throughout the study through regular monitoring of data and frequent communication with research assistants. In this analysis, we focused on the trial’s coprimary outcomes, the Behavior Rating Inventory of Executive Function (BRIEF), and an objective test of vigilance, the Go-No-Go (GNG) test. We also included a second objective measure of neuropsychological functioning, the 9-Hole Pegboard Test, a measure of fine motor control that is similar to one reported in the Childhood Adenotonsillectomy Trial as sensitive to the effects of adenotonsillectomy.¹⁴

Caregivers completed age-appropriate versions of the BRIEF (BRIEF–Preschool Edition for ages 2 to 4 and age 5 years if in preschool, or the BRIEF Second Edition for ages 6 to 18 and age 5 years in kindergarten; PAR Inc, Lutz, Florida).^21,22 These instruments survey behaviors associated with executive functioning (ability to self-regulate, pay attention, and organize in “real-world” situations) and have been shown to provide clinically relevant information on a wide range of behaviors and have good convergent and divergent validity, high test–retest reliability including in preschool children (total score r = .90), and high internal consistency (alpha values .80–.98).²³ We focused on the BRIEF Global Composite Executive (GEC) score, which provides summary information on meta-cognition and self-regulation (higher score equates to lower function).

The Go-No-Go Continuous Performance Task (GNG/CPT) is a computer-based attention test developed for longitudinal studies of heterogenous samples of children ages 3 to 12 years. After a practice session, children were presented with stimuli that consisted of different-colored cartoon fish and gray-colored sharks and asked to “catch” the fish by pressing a key on a single-button response pad as quickly as possible. The primary test outcome from this test is d prime (d’) for the Continuous Performance Task trial block, a signal detectability parameter that assesses the child’s ability to correctly identify targets corrected for their response bias (higher is better). Trial display time and intertrial variability was adjusted in test versions designed for children below 5 years, 5 through 6 years inclusive, and older than 7 years of age.²⁴ Details of the psychometric properties of this test are reported in Clark et al²⁵ and indicate that the GNG/CPT task has limited floor or ceiling effects, was sensitive to development, and correlates with parent-reported executive function and externalizing behavior. The internal reliability of the GNG/CPT was high (coefficient omega = .85), and internal validity of the GNG as an attention test is supported by a correlation of r = .72 between the task’s two latent factors (sustained attention and response inhibition).

The National Institutes of Health Toolbox 9-Hole Pegboard Dexterity Test was used to assess fine motor control.²⁶ Scores for this test are the times needed for the child to put each of nine pegs into the pegboard and then take them out using each hand, converted to age-adjusted scale scores (higher is worse). Scores for each hand were averaged to obtain a total score. In children this measure is reported to have a high test–retest reliability (rs = .81 for the dominant hand and .79 for the nondominant hand) and interrater agreement (rs > .99).²⁷

Actigraphy

Actigraphy was collected as part of the baseline examinations from study start (June 2016) through approximately March 2020 when the onset of the COVID-19 pandemic required protocol simplifications to accommodate both limitations of staff from regularly accessing clinical space and preferences of participants to minimize interactions with research procedures. To allow the study to focus resources on its primary outcomes, the study’s Data and Safety Monitoring Board approved a protocol modification that stopped actigraphy data collection during the pandemic.

Children were asked to wear a wrist actigraph on their nondominant arm for 7 days except when bathing or playing contact sports. Caregivers logged times in and out of bed and device removal during the same 7-day period using a sleep diary. The majority of studies (n = 208) utilized an Actiwatch 2 or Actiwatch Spectrum (Philips Respironics, Bend, OR) actigraph, although the GT3×+ (ActiGraph, Pensacola, FL) was used when other devices were unavailable (n = 90). The actigraphy data for each participant was annotated at the Sleep Reading Center at Brigham and Women’s Hospital Sleep Medicine Epidemiology Program using a hierarchical approach for identifying the main sleep periods and naps based on the sleep diary, light markers, and activity counts, similar to the protocol described in Patel et al.²⁸ After annotations, sleep–wake epochs were classified by Philips Actiware 6 (Philips) using 30-second epoch periods with a device-specific algorithm at medium sensitivity (40 counts for wake identification, as supported by prior research in young children and adolescents²⁹) or by using the ActiLife v6.13 (ActiGraph Inc; scored in 60-second epochs using the Sadeh algorithm).³⁰ The key exposures considered were average (over all days): sleep duration (24 hour and nocturnal), sleep fragmentation index, and sleep efficiency. Average sleep duration during each 24-hour period was calculated as the duration of nocturnal sleep plus daytime naps. The sleep fragmentation index was calculated as the sum of two proportions: the proportion of all epochs during the sleep period that were mobile (ie, the activity count was 2 or greater [Philips] or y-axis counts greater than zero) and the proportion of all immobile bouts (ie, consecutive epochs where the activity count was less than 2 during sleep that was 1 minute or less in duration [Philips] or when sleep was 1 minute or less [greater than zero]). Sleep efficiency is the ratio of the time asleep to the time in bed. As noted in the statistical analysis, device type was used as a covariate to account for differences in devices and software.

Other data

Other data for descriptive purposes or for use as covariates were extracted from standardized questionnaires administered to caregivers or by direct measurement (eg, body mass index). Two measures of SES were used: maternal level of education (high school or higher) and the Neighborhood Socioeconomic Status Index (NSES). The NSES characterizes neighborhood-level SES using the child’s home address matched to census tract data, based on: median household income, percent of households with income below the federal poverty line, adults with high school and bachelor’s educational attainment, unemployment rate, and female-headed households; values vary from 0 to 100, with 50 reported as the average national value in 2010.³¹ Race was reported by parents and dichotomized for analysis as “Black” vs other given that Black children in this study and others³² had the shortest sleep duration across all race and ethnic groups and because Black children have been reported to be at increased risk for higher SDB severity.³³ Sensitivity analyses incorporated both race and ethnicity, dichotomizing the groups as non-Hispanic White vs other.

Analysis

Generalized linear regression was used to test the associations between exposures and outcomes, after testing assumptions of normality of outcomes and linearity in associations. Minimally adjusted models adjusted for actigraph device type (Philips Actiwatch 2 or Actiwatch vs Actigraph GTX), while fully adjusted models adjusted for device type plus age, sex, race (Black vs other), maternal education (high school or more), NSES, and study site. The analyses of the GNG continuous performance test also included adjustment for test version (which varied by age) and an interaction term between test version and age. Exploratory analyses tested for interactions by age (< vs ≥ 6 years), sex, maternal education (high school or more), NSES (≥ 50 vs < 50), and race (Black vs other) for 24-hour and nocturnal sleep, sleep efficiency, and sleep fragmentation index with the outcomes BRIEF GEC and GNG d’. Exploratory analyses tested baseline AHI and percentage of time in N3 and rapid eye movement sleep in regression models as potential confounders or explanatory variables. Analyses were performed with the statistical package R version 4.2.0 (R Core Team, 2022).

RESULTS

Sample characteristics

The sample included 298 participants with at least 3 nights of actigraphy data (mean days of actigraphy 7.5 ± 1.4 days) (Table 1). The sample was on average 6 years old, half female, and more than one-third was overweight or obese. The sample included 26.8% non-Hispanic Black children and 13.4% Hispanic children; 16.8% of the children had a mother with a high school or lower education attainment, and 4.0% were reported as taking medications for attention deficit hyperactivity disorders. Average AHI was .76 (±.72). Of the sample, 37.9% had an average nocturnal sleep duration of < 8 hours and 88.3% < 9 hours, with an average nocturnal sleep duration of 488.5 (± 46.9 SD) minutes (8.1 ± .8 hours). Average sleep efficiency was 83.7% ± 6.1 (Table S1 ^{(378.4KB, pdf)} in the supplemental material). The BRIEF GEC ranged from 33 to 102 and showed mean values somewhat higher (worse) than the national average (55.0 ± 12.2); however, T-scores at or below 59 are considered to be in the typical range. The 9 Hole Pegboard score ranged from 15.5 to 118, and the GNG d’ varied from .61 to 4.35. The characteristics of the analytical sample were similar to randomized participants who did not undergo actigraphy, and no substantive differences were observed according to which actigraphy device was used. (Table 1)

Table 1.

Sample characteristics: PATS participants, by availability of actigraphy data.

Characteristics	Analytic Sample			Sample Without Actigraphy (n = 161)
	Overall (n = 298)	Philips (n = 208)	GTX (n = 90)
Age (years)
Mean (SD)	6.01 (2.31)	5.88 (2.26)	6.32 (2.4)	6.35 (2.27)
Sex, n (%)
Male	152 (51)	102 (49)	50 (55.6)	77 (47.8)
Race/Ethnicity, n (%)
Non-Hispanic White	157 (52.7)	106 (51)	51 (56.7)	79 (49.1)
Non-Hispanic Black	80 (26.8)	61 (29.3)	19 (21.1)	41 (25.5)
Hispanic	40 (13.4)	25 (12)	15 (16.7)	35 (21.7)
Other	21 (7)	16 (7.7)	5 (5.6)	6 (3.7)
BMI, n (%)
Underweight	13 (4.4)	3 (1.4)	10 (11.1)	6 (3.7)
Normal	177 (59.4)	125 (60.1)	52 (57.8)	93 (57.8)
Overweight	58 (19.5)	45 (21.6)	13 (14.4)	24 (14.9)
Obese	50 (16.8)	35 (16.8)	15 (16.7)	38 (23.6)
Maternal education, n (%)
High school diploma or less	50 (16.8)	40 (19.2)	10 (11.1)	36 (22.4)
Some college	119 (39.9)	90 (43.3)	29 (32.2)	65 (40.4)
4-year college degree or more	128 (43)	78 (37.5)	50 (55.6)	59 (36.6)
Missing	1 (.3)	0 (0)	1 (1.1)	1 (.6)
Neighborhood socioeconomic status
Mean (SD)	52.65 (12.74)	52.33 (12.88)	53.4 (12.45)	52.17 (13.72)
≥ 50 (%)	162 (54.4)	113 (54.3)	49 (54.4)	85 (52.8)
BRIEF GEC parent T score
Mean (SD)	55.0 (12.2)	55.51 (12.74)	53.84 (10.82)	56.81 (12.48)
Go-No-Go Continuous Performance Task d’ CPT
Mean (SD)	2.15 (1.01)b	2.15 (1.01)d	2.17 (1.01)e	1.92 (1.14)f
9-Hole Pegboard Test (average of both hands)
Mean (SD)	33.24 (13.52)a	34.47 (14.79)c	30.41 (9.47)	31.23 (13.07)g

^an = 297; ^bn = 294; ^cn = 207; ^dn = 206; ^en = 88; ^fn = 158; ^gn = 160. BMI = body mass index, BRIEF GEC = Behavior Rating Inventory of Executive Function Growth Executive Composite, CPT = Continuous Performance Task, GTX = greater than zero, PATS = Pediatric Adenotonsillectomy for Snoring.

Table S1 ^{(378.4KB, pdf)} further shows the distribution of participant characteristics by actigraphy-based nocturnal sleep duration, sleep fragmentation, and sleep efficiency. Sleep duration tended to be lower among children who were older than 6 years, obese, Black, or prescribed attention deficit hyperactivity disorder medications and higher among children from families with mothers with at least 4 years of college. Sleep fragmentation tended to be higher among White children compared to other race or ethnic backgrounds.

Primary outcomes

The parent BRIEF GEC T-score varied significantly with average sleep duration (both 24 hour and nocturnal), sleep efficiency, and the sleep fragmentation index in fully adjusted models (Table 2). Specifically, longer sleep duration, higher sleep efficiency, and lower sleep fragmentation index were associated with better executive function (ie, lower BRIEF GEC). The fully adjusted model suggested that each additional hour of sleep over 24 hours was associated with a nearly 3-point improvement in the BRIEF GEC score (approximately .3 standard deviations, equivalent to a small change).

Table 2.

Generalized linear regression model coefficients for parent BRIEF GEC T scores, adjusted for device type and covariates.

	BRIEF GEC Parent T Score Generalized Linear Regression (n = 293)
	Device Adjusted				Device and Covariate Adjusted
	beta	SE	P	Std beta	beta	SE	P	Std beta
Average sleep duration (hours)—24-hour period	−2.330	.921	.012	−.156	−3.238	1.037	.002	−0.217
Average nocturnal sleep duration (hours)	−1.998	.959	.038	−.128	−2.618	1.017	.011	−0.168
Sleep fragmentation index	0.271	.114	.018	.177	.278	.125	.026	.182
Average sleep efficiency	−.328	.155	.035	−.163	−.351	.160	.029	−.174

Covariate adjusted: for age, sex, race, maternal education, Neighborhood Socioeconomic Status Index, and study site. All models were adjusted for “device” (Philips vs Actigraph GTX actigraph). Standardized betas are calculated using standardized sleep and outcome values (showing the effect of one standard deviation change of the sleep exposure on the outcome, in standard deviation units). BRIEF GEC = Behavior Rating Inventory of Executive Function Growth Executive Composite.

Shorter nocturnal sleep and higher sleep fragmentation index were associated with worse fine motor control in fully adjusted analyses (Table 3). Analyses showed similar directional trends with greater 24-hour sleep and poorer sleep efficiency as for nocturnal sleep and sleep fragmentation, respectively, but were not statistically significant.

Table 3.

Generalized linear regression model coefficients for 9-Hole Pegboard Test, adjusted for device type and covariates.

	9-Hole Pegboard Test Average of Both Hands (n = 293)
	Device Adjusted				Device and Covariate Adjusted
	beta	SE	P	Std beta	beta	SE	P	Std beta
Average sleep duration (hours)—24-hour period	3.824	1.007	< .001	.230	−1.135	.841	.178	−.068
Average nocturnal sleep duration (hours)	1.489	1.065	.163	.086	−1.888	.815	.021	−.109
Sleep fragmentation index	0.856	.117	< .001	.501	.341	.098	.001	.200
Average sleep efficiency	−.538	.169	.002	−.240	−.131	.129	.311	−.058

Covariate adjusted: for age, sex, race, maternal education, Neighborhood Socioeconomic Status Index, and study site. All models were adjusted for “device” (Philips vs Actigraph GTX actigraph). Standardized betas are calculated using standardized sleep and outcome values (showing the effect of one standard deviation change of the sleep exposure on the outcome, in standard deviation units).

The d’ from the GNG-CPT task, a measure of psychomotor vigilance, did not significantly associate with any of the sleep exposures (Table 4).

Table 4.

Generalized linear regression model coefficients for Go-No-Go Continuous Performance Task (d’ CPT), adjusted for device type and covariates.

	Go-No-Go Continuous Performance Task d’ CPT (n = 290)
	Device Adjusted				Device and Covariate Adjusted
	beta	SE	P	Std beta	beta	SE	P	Std beta
Average sleep duration (hours)—24-hour period	.059	.077	.445	.048	.062	.073	.401	.050
Average nocturnal sleep duration (hours)	.095	.080	.238	.073	.095	.072	.190	.074
Sleep fragmentation index	−.010	.010	.284	−.081	−.012	.009	.188	−.092
Average sleep efficiency	.008	.013	.541	.048	.014	.011	.211	.084

Covariate adjusted: for age, sex, race, maternal education, Neighborhood Socioeconomic Status Index, and study site. All models were adjusted for “device” (Philips vs Actigraph GTX actigraph). Standardized betas are calculated using standardized sleep and outcome values (showing the effect of one standard deviation change of the sleep exposure on the outcome, in standard deviation units).

Tests of interaction by demographic factors

Tests of interactions further identified subgroup differences by sex and age with the BRIEF GEC T-score and pegboard fine hand control time (Table S2 ^{(378.4KB, pdf)} , Table S3 ^{(378.4KB, pdf)} , Table S4 ^{(378.4KB, pdf)} , and Table S5 ^{(378.4KB, pdf)} ). Each 1 hour increase in 24-hour sleep was associated with a 5.78-point improvement in boys but only a .86-point improvement in girls (P interaction = .004). A similar significant interaction for sex and nocturnal sleep was observed, with more improvement in BRIEF scores in boys with longer nocturnal sleep compared to girls with longer sleep (P interaction = .002). Each 1 hour increase in 24-hour sleep was associated with a 4.68-point improvement in the BRIEF GEC score for children younger than 6 years while only associated with a 1.24-point improvement in older children (P interaction = .054). No significant interactions were observed for sleep duration or continuity for any neurobehavioral outcome by maternal education, NSES, or child’s race, although children of mothers with less than a high school education compared to other children appeared to have more improvement in executive function with increasing nocturnal sleep duration (P interaction = .087) or decreasing sleep fragmentation (P interaction = .078), while Black children compared to other children tended to have more improvement in executive function with increasing 24-hour sleep duration compared to other children (P interaction = .088).

Exploratory analyses: associations with polysomnography variables

In exploratory analyses, we evaluated whether the AHI or percentage time in polysomnography-assessed stage N3 or rapid eye movement sleep confounded the associations between actigraphy variables and neurobehavioral outcomes. These polysomnography variables were not associated with the study outcomes, and their inclusion in regression analyses did not change the results (data not shown).

DISCUSSION

Snoring without frequent obstructive or hypoxemic episodes (“mild” SDB) occurs commonly in children and has the potential to contribute to neurobehavioral morbidity and low academic performance.^8,16–18 However, the literature is unclear as to the mechanisms that underlie such deficits. Data from routine polysomnography have not yet identified indices that effectively predict cognitive impairment in children with mild SDB. In this multisite study of a diverse sample of children with snoring and AHI levels less than 3, we found that most children (88%) had nocturnal sleep duration below recommended levels (< 9 hours). We identified significant associations between actigraphy-based measures of sleep duration and continuity with a validated measure of parent-reported executive function (BRIEF GEC). Longer nocturnal sleep was also associated with better performance on an objective measure of fine motor control (9-Hole Pegboard Test). Moreover, we found evidence for stronger effects of sleep on executive functioning among boys compared to girls and children ages 3 to 5 compared to children 6 to 12 years. In addition, we found suggestive evidence that longer nocturnal sleep and less fragmented sleep had a more beneficial effect on executive function in children from low SES households (maternal education lower than high school). Longer 24-hour sleep also was suggested to have greater benefit for executive function for Black children compared to children of other race and ethnic backgrounds. These findings support the importance of sleep quantity and continuity for neurobehavioral function in children with mild SDB and suggest that sleep assessments in children with snoring should include consideration of sleep duration and continuity.

Our findings are consistent with meta-analyses of studies from the general population (excluding children with sleep disorders) that found effects of sleep duration and continuity on academic performance reported by the parent or child or as determined by grades rather than by objective testing, with the strongest findings in young boys.³⁴ Our cross-sectional findings based on objective sleep measures are also consistent with the results of a longitudinal study of children followed from ages 2.5 to 10 years that showed that persistently short nocturnal sleep duration as reported by parents during annual assessments predicted poor receptive vocabulary at age 10 years.³⁵ The promising results of several short-term experimental sleep manipulation protocols (extension and restriction) on cognitive outcomes in healthy children^36,37 supports the need to further evaluate sleep health interventions in children both with and without mild SDB. Notably, even a 1-hour restriction in sleep duration in healthy children results in significant differences in tests of vigilance and multiple measures of behavior.³⁷

We examined both average 24-hour sleep duration and nocturnal sleep duration to capture metrics applicable to children across a wide developmental period with different napping patterns. Similar patterns for each sleep duration measure were observed in relationship to both the BRIEF and pegboard outcomes, although we observed a stronger association between fine motor control and nocturnal sleep than 24-hour sleep. Little is known about the developmental impact of sleep placement across the 24-hour period, though recent studies suggest age-related bidirectional associations between napping and brain development.³⁸ While daytime naps are positively associated with cognitive function in infants and young children, daytime naps predict poorer cognitive function in older children (4–6 years), possibly related to delayed brain maturation or shorter nighttime sleep.³⁸ A recent study also found that nocturnal, but not total, sleep duration of > 10 hours during the transition to kindergarten was associated with better neurocognitive and academic outcomes,³⁹ suggesting that overnight sleep continuity may be especially important as children develop. Unfortunately, actigraphy is not an optimal tool for identifying daytime naps due to problems distinguishing quiet resting from daytime sleep. Further research is needed to understand the contributions of daytime and nighttime sleep to cognitive function in children across the developmental spectrum.

We also found that sleep fragmentation, measured with an actigraphy index that captured frequent transitions from rest to activity, was associated with both executive function and fine motor control but not with d’ from the GNG continuous performance test, developed specifically to evaluate vigilance across a wide developmental age. Our findings are partially consistent with data from an Israeli study of 135 healthy children ages 7 to 12 years that reported significant associations with two measures of sleep fragmentation (the frequency of awakenings and duration of wakefulness after sleep onset) and measures of vigilance (a continuous performance task) and executive function (a symbol-digit substitution test).⁴⁰ Results of our analyses are also consistent with those of Gruber and colleagues who found that better math grades were associated with higher sleep efficiency, but not sleep duration, in children ages 7 to 11 years.⁴¹ Differences in findings between studies may be due to differences in the samples or methods of measurement. Still, together these studies support the importance of sleep duration as well as sleep continuity to neurocognitive functioning in children.

Several factors such as age, sex, and the social and physical environment may influence a child’s sensitivity to the positive effects of healthy sleep or, conversely, to the negative effects of disturbed or curtailed sleep. A meta-analysis of 24 studies of children ages 8 to 18 years old reported modest associations between short sleep duration or poor sleep quality with poor school performance.³⁴ Stronger associations were observed for boys compared to girls, which is consistent with research demonstrating sex differences in sleep as well as in neurobehavioral outcomes,³² and with previous data on the intersectional vulnerability of young boys in particular.⁴² Together, these findings underscore the particular importance of sleep during periods of cortical maturation, particularly for functions governed by the prefrontal cortex, and a need to further understand the roles of sex on vulnerability to sleep disturbances.

Children who live in low socioeconomic environments and those exposed to early childhood adversity may be at risk for poor sleep and more susceptible to the effects of insufficient sleep on development. Their elevated risk may be due to various factors, such as access to fewer resources and support and exposures to multiple physical and psychosocial stressors that reduce sleep duration and quality as well as resilience to poor sleep, among other factors.⁴³ However, prior research suggests a complex relationship between the moderating effects of socioeconomic factors on sleep and neurobehavioral outcomes, with effects potentially varying according to baseline levels of sleep health.²⁰ In addition to increased vulnerability to sleep loss, it is plausible that children from disadvantaged backgrounds may also especially benefit from healthy sleep. In this regard, we observed suggestive evidence of interactions between low maternal education or Black race with sleep metrics on executive functioning. Our exploratory analyses suggested that the impact of longer nocturnal sleep or less sleep fragmentation was associated with more improvement in executive functioning in children from low SES households (as measured by maternal education) compared with children with similarly better sleep who were from higher SES households. Longer 24-hour sleep duration also appeared to have a more positive effect on executive function in Black children compared to children from other racial and ethnic backgrounds. It is important to note that the tests for interactions with SES or race did not meet statistical significance at 5%. It is possible that the lack of significance is due to low power for interaction associated with the study sample size and the need for more precise measures of the social and physical environments. Additional research is needed to confirm whether there is a greater benefit of healthy sleep on neurobehavioral outcomes in children from low SES or minoritized backgrounds than in other children.

While our analyses suggest that sleep duration, efficiency, and fragmentation contribute to neurobehavioral impairment in children with mild SDB, the data available did not allow us to identify whether the variations in sleep reflected subtle effects of sleep-related breathing disturbances. Exploratory analyses did not identify confounding or mediation by the AHI level or by durations of N3 or rapid eye movement sleep. However, analyses of this cohort are planned to extract quantitative metrics of snoring and microarchitecture that may provide further insight into cognitive function in children with mild SDB and to understand whether sleep duration and continuity are related to SDB or rather are more general phenomena.

This study has several strengths, including its use of both objective measures of sleep–wake patterns over multiple 24-hour periods and validated measures of neurobehavior in children from geographically diverse sites collected following stringent quality control procedures. A key outcome was parent-reported executive function, which is sensitive to deficits in the prefrontal cortex (an area sensitive to sleep) and strongly associated with academic performance.^44,45 While many studies have evaluated insufficient sleep duration or poor sleep continuity in healthy samples or children with severe sleep apnea or neurodevelopmental disorders, few studies have focused on children with mild SDB. Furthermore, the effects of sleep duration and continuity on neurobehavioral outcomes in children are poorly studied compared to older adolescents and adults.

Several study limitations merit consideration. We adjusted for several confounders including child age, sex, race, maternal education, neighborhood SES, and the study site, but it is possible that other potential confounders may have influenced results. Due to the COVID-19 pandemic and other logistical problems, we did not have actigraphy data on the entire sample and needed to use devices made by two manufacturers, which we addressed by adjusting for device type in analyses. Actigraphy, while used commonly in epidemiological studies due to its low burden, ability to capture sleep in the child’s typical environment over multiple nights, and high overall accuracy, has limited specificity for detecting wakefulness.⁴⁶ Measurements made over multiple nights in unattended settings often do not correlate well with measurements made during single-night laboratory recordings, emphasizing the need to cautiously compare information from these distinct modalities. Our sample was defined as having mild SDB based on a history of habitual snoring and an AHI < 3. However, it is important to note that there is not a widely accepted definition of “mild SDB” in children, which underscores the need to be explicit in the criteria that define a study population and how results may generalize to other groups. Children with overnight hypoxia were excluded, so our findings do not address the potential effects of hypoxia on cognitive function. All analyses were secondary to the primary aims of the PATS trial and the chance of a false-positive finding is inflated due to multiple comparisons made. Therefore, findings should be confirmed in independent studies. Finally, this analysis was cross-sectional, and identified associations cannot prove cause-and-effect relationships between sleep duration or fragmentation and neurobehavioral measures.

In summary, children with habitual snoring with mild SDB who are considered to be candidates for adenotonsillectomy may also have a high prevalence of curtailed sleep and both shorter sleep duration and poor sleep continuity measured by multiple days of in-home actigraphy. These sleep disturbances may be associated with deficits in executive functioning and fine motor control. A positive effect of longer and less fragmented sleep appeared greatest for boys and for young children, with some evidence suggestive of greater benefit for children from low SES households and Black children. Given that executive functioning is fundamental for learning and academic development, our results support the need for consideration of both sleep duration and continuity when evaluating children with mild SDB.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Supported by National Heart, Lung, and Blood Institute grants 1U01HL125307 and 1U01HL125295. Data are available upon request. After publication of the primary papers, data will be made publicly available via the National Sleep Research Resource (sleepdata.org). R. Chervin reports financial relationship that may be pertinent: Consulting for Eli Lilly and Company, through contract between Eli Lilly and Michigan Medicine; Editor and author for UpToDate; Nonfinancial interests: Serve as member, board of directors, and treasurer, International Pediatric Sleep Association; Serve as member, advisory board, for the nonprofit Pajama Program. F. Hassan reports consulting with Eli Lilly and Company. S. Redline reports consulting fees from Jazz Pharmaceuticals, Eli Lilly, and ApniMed Inc unrelated to this manuscript, and grants from the National Institutes of Health that supported this work. The other authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

BRIEF

Behavior Rating Inventory of Executive Function

GEC

Global Executive Composite

GNG

Go-No-Go

PATS

Pediatric Adenotonsillectomy Trial for Snoring

SDB

sleep-disordered breathing

SES

socioeconomic status