Association Between Habitual Snoring and Cognitive Performance Among a Large Sample of Preadolescent Children

Abstract

Importance: Previous studies have identified an association between habitual snoring and lower cognitive performance in children. However, whether and to what extent this association is confounded by pertinent demographic, anthropometric, and socioeconomic characteristics is unknown.

Objective: To assess the extent to which potential confounding factors modify the association between parent-reported habitual snoring and cognitive outcomes among a large and diverse sample of typically developing preadolescent children.

Design, setting, and participants: This cross-sectional analysis used a baseline data set (version 2.0.1) from children enrolled in the ongoing Adolescent Brain Cognitive Development study between September 1, 2016, and October 15, 2018. Children aged 9 to 10 years without serious psychiatric or neurological comorbidities were recruited at 21 research sites in the US. Study recruitment was designed to approximate the racial and socioeconomic diversity of the US population. Data were analyzed from February 1 to March 31, 2020.

Exposures: Parent-reported habitual snoring in children that occurs 3 or more nights per week.

Main outcomes and measures: Associations between habitual snoring and cognitive performance were assessed using the Sleep Disturbance Scale for Children and the National Institutes of Health Toolbox Cognition Battery, which includes 7 domain-specific and 3 composite (total cognitive function, fluid cognition, and crystallized cognition) standard scores that are uncorrected for covariates. Cognitive performance was examined before and after adjustment for covariates, which included age, sex, body mass index percentile, annual household income before taxes, and highest educational level of caregiver. The extent of confounding was assessed by the effect size, represented by Cohen d, before and after inclusion of covariates using linear mixed-effects models.

Results: A total of 11 873 children aged 9 to 10 years (6187 boys [52.1%]; 6174 White [52.0%]) with available data were included in the study. Of those, habitual snoring (≥3 nights per week) was reported in 810 children (6.8%), and nonhabitual snoring (1-2 nights per week) was reported in 4058 children (34.2%). In the unadjusted models, the total cognitive function composite score among children who habitually snored was significantly lower compared with children who never snored (Cohen d, 0.35; 95% CI, 0.28-0.42). Differences were also identified in the crystallized cognition (Cohen d, 0.34; 95% CI, 0.26-0.41) and fluid cognition (Cohen d, 0.28; 95% CI, 0.21-0.35) composite scores. The association between habitual snoring and cognitive performance was substantially attenuated after adjustment for covariates (Cohen d, 0.16 [95% CI, 0.09 to 0.24] for total cognitive function, 0.14 [95% CI, 0.07 to 0.21] for crystallized cognition, and 0.13 [95% CI, 0.06 to 0.21] for fluid cognition). Similar mitigation was also observed for all domain-specific scores.

Conclusions: In this cross-sectional study, when adjusted for baseline demographic, anthropometric, and socioeconomic characteristics, the association between parent-reported habitual snoring and cognitive performance was substantially attenuated among children aged 9 to 10 years.

Figure.. Comparison of Unadjusted and Adjusted Models for the Association Between Parent-Reported Frequency of Snoring and Cognitive Test Scores Among Children Aged 9 to 10 Years

Estimated marginal mean scores of cognitive tests included in the National Institutes of Health Toolbox Cognitive Battery. Habitual snoring represents children who snore 3 or more nights per week, and nonhabitual snoring represents children who snore 1 to 2 nights per week. All outcome measures were derived from linear mixed-effects models. Span of vertical shading represents 95% CIs. Flanker indicates Flanker Inhibitory Control and Attention Test, which measures inhibitory control and attention. A, Unadjusted models. Models include only the principal fixed effect (snoring frequency), with recruitment site as a random effect. B, Adjusted models. Models include age, sex, race, body mass index (calculated as weight in kilograms divided by height in meters squared) percentile, highest educational level of caregiver, and annual household income, with recruitment site as a random effect.