Sleep, Classroom Behavior and Achievement among Children of Color in Historically Disinvested Neighborhoods

Abstract

Children of color are more likely to have poor sleep health than White children, placing them at risk for behavioral problems in the classroom and lower academic performance. Few studies, however, have utilized standardized measures of both classroom behavior and achievement. This study examined whether children’s sleep (parent- and teacher-report) in first grade concurrently related to independent observations of classroom behavior and longitudinally predicted achievement test scores in second grade in a sample of primarily Black (86%) children (n = 572; age = 6.8) living in historically disinvested neighborhoods. Higher teacher-reported child sleepiness was associated with lower adaptive behaviors and higher problem behaviors in the classroom, and predicted lower achievement. Parent-reported bedtime resistance and disordered breathing also predicted lower achievement.

Sleep is an essential component of physical, cognitive, and emotional development among children (Mindell & Owens, 2015). The American Academy of Sleep Medicine recommends 9 to 12 hours of sleep for school-aged children (ages 6 to 12) (Paruthi et al., 2016), but according to reports from parents, 25% of children do not meet these recommendations (Owens et al., 2011). Children of color are at a disproportionately higher risk for poor sleep health (i.e., sufficient sleep duration, absence of daytime sleepiness, consistent sleep schedules, a bedtime routine) and sleep disorders (e.g., obstructive sleep apnea) compared to White children (Nevarez et al., 2010; Owens et al., 2011), and research has demonstrated that poor sleep health and untreated sleep disorders are associated with behavioral problems (Astill et al., 2012; Sakamoto et al., 2017) and impaired school performance (Dewald et al., 2010). Poor sleep health disproportionately harms children of color and children from low socioeconomic status (SES) families (Buckhalt, 2011; Buckhalt et al., 2009), and there is evidence that sleep health plays a role in pathways linking race and SES to academic achievement (Philbrook et al., 2018) and behavior functioning (Bøe et al., 2012; El-Sheikh et al., 2010). We thus examine the cross-sectional relation between children’s sleep (i.e., including sleep health and sleep disorder characteristics) and their behavior in school in first grade as well as the prospective relation between children’s sleep and academic achievement one year later (in second grade) in a sample of primarily Black children growing up in historically disinvested neighborhoods (i.e., neighborhoods in which structural barriers of systemic racism have contributed to a high concentration of families with low incomes).

Sleep among Children of Color in Low-Income Families

Theoretical frameworks including the Social Determinants of Health (Marmot, 2005) and the Integrative Model for the Study of Developmental Competencies in Minority Children (Garcia Coll et al., 1996) posit the important role of social and economic contexts in causing poor health outcomes for children in low-income families. The Social Determinants of Health framework has only recently been applied to children’s sleep health (Hale & Hale, 2010); our understanding of disparities in children’s sleep health remains underdeveloped. The extant literature demonstrates that children in low-income families are at greater risk for poor sleep health and sleep disorders as compared to higher income children (Armstrong et al., 2019; Chervin et al., 2003); children who live in economically and socially disadvantaged neighborhoods are at greater risk than their peers (Bagley et al., 2018; Hale et al., 2019; R. Wang et al., 2017); and these disparities in sleep health begin in infancy (Nevarez et al., 2010). Children of color are at particular risk for poor sleep health as evidenced by a shorter sleep duration than White children (Smith et al., 2019), and for sleep disorders including disordered breathing and sleep apnea (Chervin et al., 2003; Redline et al., 1997). For low-income families of color, large scale social factors likely make it more difficult to obtain high-quality sleep of sufficient duration (Hale & Hale, 2010). To understand the developmental context of sleep health disparities, we extend the Integrative Model for the Study of Developmental Competencies in Minority Children by applying it to sleep health disparities to consider how social position, racism, and discrimination together influence children’s neighborhood environment, family context, and exposure to psychosocial stressors, all of which are likely to be particularly relevant for sleep health among children of color in low-income families (Garcia Coll et al., 1996).

Social position, racism, and discrimination together shape policies and institutional context which influence sleep health by setting constraints on, or opportunities for, conditions that promote positive sleep health (Jackson et al., 2015; Robbins et al., 2019). For example, historical and current interpersonal and institutional racism influence the physical contexts in which low-income families of color live. Families of color are more likely to live in disinvested neighborhoods (Spilsbury et al., 2006; Williams & Jackson, 2005) which, although not homogenous contexts, increases children’s risk of exposure to factors that bear consequence for sleep, including noise, environmental pollutants, access to physical activity amenities, high population density, violence, and safety concerns. These exposures have been shown to increase risk for poor sleep health and sleep disorders among children (Hale et al., 2019; Koinis-Mitchell et al., 2019; Lawrence et al., 2018; Meldrum et al., 2013; Philbrook et al., 2020).

Constraints of the environments in which families live and work further influence family level factors such as routines and parenting styles influencing children’s sleep health. For example, a lack of affordable housing within disinvested neighborhoods leads to more crowded households which may also mean that there are fewer available rooms and beds for sleeping. Children of color are also more likely to have electronics in the room than are White children (Nevarez et al., 2010; Spruyt et al., 2014). Structural features of the work environments of low-income families also impact family routines that support healthy sleep behaviors. Parents working multiple jobs and at variable or evening hours are challenged in setting a stable sleep schedule and bedtime routine for their children. Inconsistent bedtimes and bedtime routines for preschool aged children are more common among Black and Latino families relative to White families and among families with lower income and education (Hale et al., 2009; Patrick et al., 2016). In addition, caregivers in low-income families may have less access to information about healthy sleep practices, a strong predictor of children’s sleep health (Owens et al., 2011).

Psychosocial stressors including racism and acculturation also likely play an important role in the sleep health of children of color. Racism is a pervasive and persistent stressor experienced by families of color, and the experience of racial discrimination is likely an important determinant of sleep health (Jackson et al., 2015). For example, perceived discrimination has been associated with a number of poor sleep health parameters among children and adolescents, including short sleep, poor sleep quality, longer sleep onset, and more time awake after sleep onset (El-Sheikh et al., 2016; Fuller-Rowell et al., 2017; Yip, 2015; Yip et al., 2020). Furthermore, for immigrants of color, acculturation – acquiring the culture of the dominant (i.e., White) society - is another major source of psychosocial stress which has also been shown to have negative impacts on sleep health (Hale et al., 2014; Jackson et al., 2015). Thus, low-income families of color face systemic challenges of poverty and racism which interact with and influence their physical, social, and family contexts to create a developmental environment that may increase their children’s risk for poor sleep health and sleep disorders. This interaction of contexts across multiple levels suggests, however, that despite the multitude of challenges faced by families residing in historically disinvested neighborhoods, there will be individual variation in sleep health.

Consistent with this model of individual variability even in the face of systemic challenges, the immigrant paradox describes a widely-observed pattern in which recent immigrants of color often have more optimal outcomes than non-immigrants of color across multiple domains of health and education. Importantly, findings with regard to potential protective effects of immigrant status are highly nuanced by developmental age, country of origin, language abilities, and socioeconomic factors, underscoring the need for more research (Crosnoe & Turley, 2011). For this reason, we investigate differences in sleep health between children of color from immigrant and US-born families.

Child Sleep and Developmental Competencies at School

The Integrative Model for the Study of Developmental Competencies in Minority Children further posits that child health status is an important predictor of children’s developmental competencies and that health may be of particular importance among children of color given racial disparities in children’s health (Garcia Coll et al., 1996). Building from a health disparities framework, Buckhalt and colleagues (2009; 2011) specifically propose that poor sleep health may disproportionately harm children of color and children from low SES families. Strong evidence shows that poor sleep health among children thwarts healthy development (Spruyt, 2018). In the school context, child behavior and academic achievement are two important developmental competencies. Children who have poor sleep health may thus be at risk for both behavioral challenges (i.e. lower adaptive and greater problem behaviors) and lower academic achievement.

From a neurobiological perspective, sleepiness impairs activation of prefrontal brain areas that support higher-level cognitive skills which are important for higher-order cognitive processes including planning and goal-directed action (Thomas et al., 2000). Lower activation in these areas makes it difficult for children to engage productively in classroom learning. At the same time, lower prefrontal activation decreases children’s ability to regulate their impulses and emotions which can in turn lead to increased problem behaviors (Minkel et al., 2012; Palmer & Alfano, 2017). A meta-analysis demonstrated that insufficient sleep is associated with poor executive functioning, problem behavior (e.g., more externalizing, such as hyperactivity or aggression, and internalizing behaviors, such as fear or anxiety), and worse school performance (Astill et al., 2012). An additional meta-analysis showed that poor sleep health (i.e., insufficient sleep duration, poor sleep quality, and sleepiness) is associated with lower school performance among children and adolescents, especially among school-aged children (Dewald et al., 2010). Conversely, research has shown that sleep health practices, such as bedtime routines that include language-based pre-bedtime activities, are associated with favorable academic outcomes including higher verbal test scores (Hale et al., 2011).

Classroom Behavior

In specifically investigating the relation of sleep to behavior problems, most work has been limited by reliance on parent reports of behavior; relatively fewer studies have specifically investigated behaviors in the classroom setting using teacher report and no studies to our knowledge utilize observation methodologies (Beebe, 2011). Reliance on parent report is problematic for assessing classroom behavior as parents do not observe classroom behavior, and a robust literature finds only small correlations between parent and teacher report of children’s behavioral and emotional problems (Achenbach et al., 1987). Sleep problems, shorter sleep duration, and daytime sleepiness have been associated with teacher reports of greater student-teacher conflict, higher levels of inattention, cognitive problems, and externalizing symptoms, and behavior problems more generally (Gruber et al., 2012; Holdaway & Becker, 2018; Lucas et al., 2017; Paavonen et al., 2009). Sleep disordered breathing has also been related to teacher reports of child behavioral problems (Ali et al., 1993; Beebe et al., 2004). Experimental evidence provides perhaps the strongest evidence that insufficient sleep among children increases teacher-reported problem behaviors (Fallone et al., 2005). Although more recent studies have investigated the relation of sleep to behavior problems among children of color, with one exception, these studies have relied exclusively on parent-reported behavior rather than behavior in the classroom context. Inconsistent bedtimes (measured via actigraphy) were shown to be associated with parent-reported internalizing and externalizing behavior among a sample of African American children (Spruyt et al., 2016). Another study with a diverse sample (approximately 50% children of color) of 5- to 8- year old children found that poorer sleep hygiene (e.g., inconsistent sleep schedule and bedtime routine) was associated with greater parent-reported internalizing and externalizing behavior (Witcher et al., 2012). Among Latino preschool aged children, parent reported sleep disordered breathing and snoring was associated with problem behaviors at school as reported by teachers (Scullin et al., 2011). Relatedly, although not focusing on behavior problems, a small study of preschool aged children (66% White, 29% Black) found relations between sleep duration and socially competent behavior in the classroom (i.e., peer acceptance and social engagement/ motivation) more generally (Vaughn et al., 2015). Thus, more work is needed to understand the relation of sleep health to both adaptive and problem behavior in the classroom among children of color.

Cognitive Functioning and Academic Achievement

Several studies have also examined sleep health and cognitive functioning or academic achievement among racially and ethnically diverse children at various ages. Child race and family SES, however, have been confounded in many studies, although independent effects of each on the relation of sleep to children’s cognitive and academic functioning have been found (ex. Bub et al., 2011, Buckhalt, & El-Sheikh, 2011; Buckhalt et al., 2007; Buckhalt et al., 2009). Buckhalt and colleagues examined sleep health (using actigraphy over the course of one week) and cognitive test scores in a sample of 3^rd grade children (31% Black, 69% White). Their results showed that cognitive performance was similar among White and Black 3^rd grade children who demonstrated good sleep health. However, among children who had poor sleep health, Black children had poorer cognitive test scores than did White children (Buckhalt et al., 2007). Similarly, when examining the role of SES, low family SES was a risk factor for low test scores among children with poor sleep health but not among children with good sleep health (Buckhalt, et al., 2007). A similar pattern of results was found with this sample when examining relations of sleep to academic achievement such that disrupted sleep was more strongly related to achievement for Black children and for children from lower SES families. Specifically, sleep mediated relations of children’s emotional insecurity (about their parents’ marital relationship) to academic achievement for Black children, but not for White children, and for children from lower, but not higher, SES families (El-Sheikh et al., 2007). A follow-up study conducted when these children were in 5^th grade replicated the earlier effect on cognitive test scores and demonstrated significant concurrent positive associations between indicators of sleep health and academic achievement as well as longitudinal relations between sleep health in third grade and academic achievement in fifth grade, for both Black and White children (Buckhalt et al., 2009).

In a similar longitudinal study (65% White and 35% Black), Philbrook and colleagues examined sleep (measured via actigraphy) at age 9 and its association to cognitive test performance at age 11 and found a significant effect of poor sleep (specifically, low sleep efficiency, or low scores on a scale assessing sleep fragmentation where low scores indicate worse sleep health) on cognitive test scores (Philbrook et al., 2017). The association between sleep health and academic outcomes was stronger for Black children than for White children; for Black children, inefficient sleep increased risk for poorer cognitive performance and efficient sleep protected against risk for poorer cognitive performance. A similar pattern of results was found for academic achievement in this sample such that there was an accelerating curvilinear association in which greater sleep efficiency was concurrently associated with higher achievement test scores among Black children, but not among White children (El-Sheikh et al., 2019). Longitudinal analyses demonstrated that 3-year trajectories of child self-reported sleepiness were associated with longitudinal changes in cognitive performance (Bub et al., 2011). Race moderated this association such that sleepiness was more strongly associated with lower cognitive performance among Black children than among White children. Taken together, this research suggests the importance of investigating the relation between sleep and achievement among children of color and among Black children specifically.

The Current Study

The current study advances the literature on sleep, classroom behavior, and academic achievement among school-aged children of color in historically disinvested neighborhoods in New York City. In measuring child adaptive and problem behaviors in the school setting, a key concern is that teacher biases may lead to inaccurate reporting (Downey & Pribesh, 2004; Haller, 1985; Riegle-Crumb & Humphries, 2012). For example, teacher characteristics (such as years of experience, efficacy, burnout, race, ethnicity or culture) may influence ratings of children’s behavior (Mason et al., 2014) and teacher bias in regard to student race and gender can also lead to inaccurate ratings (Chang & Sue, 2003; Hosterman et al., 2008). Because of this concern, we advance past research methodology through our use of standardized observations of children’s adaptive and problem behavior in the classroom. Additionally, in our measurement of sleep, we utilize both parent report of sleep and a measure of teacher report of child sleepiness during the school day. These hypotheses are conceptually grounded in the literature outlined above. However, the RCT and follow-up study which provided data for these analyses, was not designed with the intent of testing these hypotheses. Moreover, several of the measures were created or adapted for this study and thus have not been utilized in prior work.

Preliminary aims of this study include: 1) examining whether reports of children’s sleep differed by child and family sociodemographic characteristics including race/ethnicity, gender, and parental education, and 2) examining relations between teacher and parent reports of sleep. The primary aims of this study are to examine the cross-sectional relation between child sleep health and observed adaptive and problem behaviors in the classroom in first grade, and to examine the longitudinal relation of sleep health to academic achievement one year later (in second grade) in a sample composed primarily of Black children. We hypothesize that poorer child sleep health and sleep disorder symptoms– reported by parents (i.e., bedtime resistance, daytime sleepiness, insufficient sleep duration, disordered breathing, and sleep onset delay) and reported by teachers (i.e., daytime sleepiness at school) -- will be associated with lower levels of observed adaptive behaviors and higher levels of observed problem behaviors in first grade classrooms. We also hypothesize that poorer child sleep health and sleep disorder symptoms as reported by parents and teachers will longitudinally predict poorer academic achievement one year later (in second grade).

Methods

Sample

Participants were children who participated in a longitudinal follow-up study of a cluster (school) randomized controlled trial of a pre-K intervention in 10 elementary schools with a largely Black, low-income population (school selection criteria: 80% Black and 70% eligible for free lunch) in a large urban school district (Brotman et al., 2013, 2016). In large urban school districts in the U.S., the vast majority of students are children of color (i.e. Black, Latino and Asian). Percent Black was included as a school selection criterion to maximize the proportion of families who would meet the only family-level eligibility criterion for the trial of a group-based family-centered intervention (i.e., parent/caregiver English language proficiency). We refer to areas served by these schools as historically disinvested neighborhoods to recognize the structural barriers of systemic racism that have contributed to a high concentration of families with low incomes. To illustrate, the neighborhoods of study participants from these 10 schools were, on average, characterized as 36.5% low-income (under 200% poverty threshold), 9.5% unemployed, 17.6% less than a high school diploma, 46.7% immigrant, 85.4% Black, and 7.9% Latino. These schools are representative of the elementary schools (with and without pre-K programs) in the two geographical zones (within the large urban district) from which they were drawn. The original sample included 1050 students enrolled in pre-K programs in these 10 schools from 2005 to 2008 (4 cohorts of children). Children were followed annually through the end of second grade (three years later) if they remained enrolled in the original study schools (n = 758 at first grade and n = 682 at second grade). Sleep data was not collected from children in cohort 1 (~25% of the sample) in first grade. For primary analyses, we limited the sample to children who had either parent or teacher-reported measures of sleep in first grade (n = 572). This represents 75% of the total number of children (n = 758) who were followed into first grade; children who left these schools before first grade were not followed. Parent or teacher-reported sleep data may be missing due to the informant or child being unavailable or the parent choosing not to complete the survey.

The mean age of children (n = 572) when they were assessed at the end of first grade was 6.8 years and 52% were girls. Parent race/ethnicity at study entry when children were in pre-K was Black for 86% (18% African American; 68% African Caribbean), Latino for 8%, and other races or ethnicities for 6% of the sample. Overall, most children were from immigrant families (68% of the 536 who reported on immigrant status). For the larger study, African American parents were by definition born in the United States. African Caribbean parents were primarily first generation immigrants to the U.S. (12% were second or later generation, born in the U.S.) coming from a number of countries including Jamaica, Trinidad, Guyana, & Haiti. About half (49%) of Latino parents were immigrants to the U.S. coming from outside the mainland U.S. including from Puerto Rico and the Dominican Republic. Although we have made the distinction between racially Black individuals who identify ethnically as African American versus African Caribbean, we note that in the review of the prior literature, the term African American may have been used to encompass any participant who identified as racially Black, regardless of ethnic identity. Of those reporting parent income (n = 373), 40% of families were low-income as defined by an income to need ratio of less than 200% of the federal poverty level. Of those reporting (n =479), 46% of parents had a high school education or less.

Procedures

With the exception of demographic data, which was collected in pre-K (~age 4), this study uses data from assessments conducted when children were in first grade and in second grade (~ age 6 and 7). In first and second grade, teachers provided ratings of children’s daytime sleepiness, and parents (86% biological mothers; 10% biological fathers) provided ratings of children’s sleep behavior and sleepiness. Children’s behavior in their first and second grade classroom was observed and rated by research staff. Children’s academic achievement in math and reading was assessed through direct standardized assessments in second grade, administered individually to students during the regular school day by trained research assistants.

Measures

Child sleep health and sleep disorders symptoms.

Parent reported child sleep health and sleep disorder symptoms were measured using a subset of 5 domain subscales from the validated and widely used Child Sleep Habits Questionnaire (CSHQ) (Owens et al., 2000). We used a subset of domain subscales from the CSHQ and shortened the daytime sleepiness subscale, with a focus on domains and items that were most relevant for the sample (in terms of stage of development and context) due to logistical time constraints of administering a telephone survey to parents who were also asked to report on a series of other measures as part of the longitudinal follow up study. The 5 domain subscales include: bedtime resistance (e.g., “your child went to bed at the same time” and “your child struggled at bedtime like cried or refused to stay in bed”), sleep duration (e.g., “your child slept about the same amount each day” and “your child slept the right amount”), disordered breathing (e.g., “your child snored loudly”), daytime sleepiness (e.g., “your child seemed tired” and “your child took a long time to become alert in the morning”), and sleep onset delay (“your child fell asleep within 20 minutes after going to bed”). Items were rated on a 3-point scale indicating the frequency at which the behavior occurred, where usually (1) = 5 to 7 times per week, sometimes (2) = 2 to 4 times per week, and rarely (3) = 0 to 1 time per week. Items within each subscale were averaged to create the subscale score. Consistent with the original CSHQ, sleep duration consisted of 3 items (Cronbach’s alpha = .627); bedtime resistance consisted of 6 items (Cronbach’s alpha = .642); sleep disordered breathing consisted of 3 items (Cronbach’s alpha = .581), and sleep onset delay consisted of 1 item. The daytime sleepiness subscale was shortened from 8 items to the 2 items listed above (Cronbach’s alpha = .632). Reliabilities were acceptable and largely similar to prior studies using this scale (Owens et al., 2000; Urfer-Maurer, 2017). Higher scores on each CSHQ component indicates more problem sleep behaviors.

Teacher-reported assessment of child sleep was collected using an adapted measure from the CSHQ. Specifically, teachers were asked to respond to three statements: “This child took a long time to become alert in the morning,” “This child seemed tired,” and “This child has fallen asleep or almost fallen asleep” on a scale where usually (1) = 5 or more times in the past week, sometimes (2) = 2 to 4 times in the past week, and rarely (3) = 0 to 1 time in the past week. Items were reverse coded averaged for a single scale of teacher-reported sleep with high reliability (Cronbach’s alpha = .857). Higher scores indicate higher sleepiness.

Classroom behavior: adaptive behaviors and problem behaviors.

The Assessing Behavior in the Classroom (ABC) (Brotman, Kamboukos, Acra, & Dawson-McClure, 2007) is an observational coding system that was developed for the larger study to assess an individual student’s behavior for 15 continuous minutes during structured classroom activity. Full details of the measure are available in Supplementary Material. The ABC included time sampling of behavior for a 15-minute observation per individual student. Students were observed for 15 minutes on two occasions approximately 2 days to 1 week apart. Seven categories of problem behaviors and one category of adaptive behaviors were considered as present or absent during 15 one-minute time-sampling segments. Adaptive behaviors were defined as the student being actively engaged in learning as demonstrated by non-verbal (e.g. listening, nodding, sitting up, working on an assigned task) or verbal (e.g. raising hand to respond, answering questions, participating in discussion) active signs of learning. Problem behaviors included behavioral and emotional problems exhibited by the student that are observable in the classroom.

During each 1-minute segment, the observer put a check mark in the specific time block for each behavior if it was observed. Behaviors across the seven problem behavior and one adaptive behavior domains could be coded simultaneously. Approximately 30% of observations were double-coded to assess inter-rater reliability. Intraclass correlations ranged from .685 to .969, which indicate good to excellent inter-rater reliability (Cicchetti, 1994; Hallgren, 2012). Final scores for each category of behavior were calculated by summing the number of occurrences across the 15-minute observation (ranging from 0 to 15). The sum was then divided by the length of the observation in order to obtain percentage scores. Percentage scores for problem behaviors can be greater than 1 as multiple categories of behavior were coded. Percentage scores were used in all analyses. Percentage scores were averaged over the two observation periods and across the two raters (when two ratings were conducted).

Academic Achievement.

The Kaufman Test of Educational Achievement (K-TEA) Brief Form-Second Edition (Kaufman & Kaufman, 2005) assessed achievement in kindergarten and second grade. The K-TEA is an individually-administered standardized test that provides reliable and valid estimates of reading, writing, and math achievement (Kaufman & Kaufman, 2005). Full details of the measure are available in Supplementary Material. Composite scores combined across these three subtests (math, reading, and writing) and were converted to standard scores, based on a grade- normed sample (M=100, SD=15).

Missing Data

Missing data analyses examined differences between the original sample of 1050 children and the current sample of children for whom either parent or teacher-reported measures of sleep were available (n = 572) in first grade. Based on available data, children who had data on sleep measures were more likely to be African Caribbean (r = .071, p = .023), less likely to be Latino (r = −.081, p = .009), had higher achievement in kindergarten (r = .077, p = .028), and exhibited greater adaptive behavior on the ABC (r = .347, p <.001). The study sample did not differ from the original sample in terms of parental education, child sex, first grade problem behaviors, or academic achievement in 2^nd grade. Within the current study sample, there was missing data on outcome variables (12% missing observations of classroom behavior and 24% missing academic achievement in 2^nd grade) and covariates (see Table 1 and descriptive results). We also asked parents to report income data, but because of high rates of missingness (35% missing), we rely on parent education only in our models. Parent education was coded dichotomously as having a high school education or less or as having completed at least a technical degree beyond high school. In order to deal with potential bias from missing data, full information maximum likelihood (FIML) was used in estimating all models in Mplus version 8 (Enders & Bandalos, 2001).

Table 1.

Descriptive statistics for first grade (G1) (n =572) and second grade (G2) (n = 660) samples.

First grade sample (n = 572)	N	Mean	S.D.	Range
G1 Bedtime resistance (parent)	432	1.33	0.36	(1 –3)
G1 Sleep duration (parent)	431	1.17	0.35	(1 –3)
G1 Disordered breathing (parent)	431	1.11	0.29	(1 –3)
G1 Daytime sleepiness (parent)	432	1.30	0.50	(1 –3)
G1 Sleep onset delay (parent)	432	1.43	0.67	(1 –3)
G1 Daytime sleepiness (teacher)	516	1.18	0.40	(1 –3)
G1 Child age- Spring	501	6.80	0.30	(6.1–7.4)
G1 Adaptive Behaviors (% time)	502	0.84	0.14	(0.1–1.0)
G1 Problem Behaviors (% time)	502	0.67	0.30	(0.0–1.9)
Academic Achievement (second grade)	433	100.85	14.91	(45–144)
Second grade sample (n = 660)
G2 Bedtime resistance (parent)	531	1.29	0.36	(1 –3)
G2 Sleep duration (parent)	530	1.19	0.35	(1 –3)
G2 Disordered breathing (parent)	531	1.12	0.30	(1 –3)
G2 Daytime sleepiness (parent)	531	1.29	0.48	(1 –3)
G2 Sleep onset delay (parent)	531	1.42	0.69	(1 –3)
G2 Daytime sleepiness (teacher)	558	1.20	0.40	(1 – 3)
G2 Child age- Spring	530	8.03	0.32	(7.3 – 8.8)
G2 Adaptive Behaviors (% time)	546	0.86	0.14	(0.15 – 1.00)
G2 Problem Behaviors (% time)	546	0.68	0.29	(0 – 1.70)

Analytic Plan

We first used independent t-tests or ANOVA to examine whether reports of children’s sleep differed by child and family sociodemographic characteristics including race/ethnicity, gender, and parental education. Next, we examined correlational relations between teacher and parent reports of sleep. We then used linear regression with robust standard errors in Mplus 8 to examine whether parent and teacher-reported child sleep is concurrently related to observed classroom behavior in first grade and longitudinally predictive of academic achievement in second grade. Parent and teacher reports of sleep were examined in separate models. All models controlled for child age, sex, race/ethnicity, and parent education. Models predicting academic achievement test scores in second grade also controlled for academic achievement test scores in kindergarten. All models included dummy variables for children’s school at pre-K in order to account for the study design in which children are nested in schools and to control for differences between schools, including school randomized intervention status.

For these primary regression analyses, we limited the sample to children who had either parent or teacher-reported measures of sleep in first grade (n = 572). Analyses examining parent measures of sleep were limited to children who had these measures (n = 432). Analyses examining teacher measures of sleep were limited to children who had these measures (n = 516). There was missing data for both covariates and outcome variables, so we used full information maximum likelihood to mitigate potential bias from missing data. Finally, we replicate these analyses with data from children’s sleep and classroom behaviors in second grade.

Results

Descriptive Statistics

Descriptive statistics for the sleep health and sleep disorder symptoms indicators and behavioral and academic outcomes are shown in Table 1. According to first grade teachers, 22.3% of children exhibited at least some level of daytime sleepiness. According to parents, 68.3% of children had bedtime resistance more than once per week; 33.1% had delay in sleep onset more than once per week; 32.9% had daytime sleepiness more than once per week, 25.8% had insufficient sleep duration more than once per week and 18.3% displayed disordered breathing more than once per week.

Mean differences in children’s sleep by sociodemographic characteristics

When comparing girls and boys, no differences were found in any parent rated measure of sleep (bedtime resistance, sleep duration, disordered breathing, daytime sleepiness, and sleep onset delay). On the teacher-report measure of daytime sleepiness, teachers rated boys (M = 1.24, SD = .44) to be more sleepy than girls (M = 1.12, SD = .34); t(466.93) = −3.42, p = .001). In examining racial and ethnic differences, we compared parent and teacher reports of sleep across children of African-American, African Caribbean, and Latino parents. One-way ANOVA results indicated significant differences in teacher ratings of daytime sleepiness [F(2, 484) = 4.88, p = 0.008] and no significant differences in parent report of sleep health. Post hoc tests with Games-Howell correction for multiple comparisons indicated that African American children (M = 1.29, SD = .53) were on average rated by teachers as .35 SD higher in sleepiness than African Caribbean children (M = 1.15, SD = .36); p = .043). Latino children (M = 1.21, SD = .43) did not differ from either African American or African Caribbean children. There were no significant differences in parent or teacher report of child sleep by parental education status.

Concurrent and longitudinal correlations of parent and teacher reports of sleep

As shown in Table 2, of the four parent-reported sleep constructs, only higher disordered breathing was associated with higher teacher-reported sleepiness (r = .146, p < .01). Parent reports of sleep health were moderately stable from first to second grade (rs = .29 to .53, ps < .05). Teacher reported sleepiness (across different teachers) was also moderately stable from first to second grade r = .29, p < .05).

Table 2.

Correlations between sleep, classroom behaviors, and achievement across first and second grades.

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1. G1 Bedtime Resistance (p)
2. G1 Sleep Duration (p)	.27**
3. G1 Dis. Breathing (p)	.01	.06
4. G1 Day Sleepiness (p)	.28**	.38**	.16**
5. G1 Sleep Onset Delay (p)	.27**	.29**	–0.02	.21**
6. G1 Day Sleepiness (t)	.04	.03	.15**	0.07	–.003
7. G1 Adaptive Behaviors	–.07	–.003	–.01	–.01	–.06	–.23**
8. G1 Problem Behaviors	.05	0.03	–.02	.05	–.03	.16**	–.46**
9. G2 Bedtime Resistance (p)	.53**	.21**	–.04	.28**	.18**	.06	.003	.05
10. G2 Sleep Duration (p)	.23**	.34**	–.004	.26**	.17**	.04	–.07	.10*	.39**
11. G2 Dis. Breathing (p)	.08	.09	.46**	.15**	–.01	.004	–.044	–.02	.10*	.17**
12. G2 Day Sleepiness (p)	.25**	.22**	.11*	.37**	.15**	.06	–.06	–.01	.22**	.21**	.19**
13. G2 Sleep Onset Delay (p)	.17**	.24**	.12*	.13*	.29**	.18**	.03	–.04	.16**	.25**	.11*	.11*
14. G2 Day Sleepiness (t)	0.07	.002	–.01	.05	.08	.29**	–.19**	.09*	0.07	.06	0.04	–.02	.03
15. G2 Adaptive Behaviors	–.10	.02	–.04	.02	–.01	–.11*	.16**	–.17**	–.06	–.04	–.06	–.02	.02	–.11*
16. G2 Problem Behaviors	–.01	–.09	–.03	–.07	–.01	.06	–.09*	.30**	.02	–.04	.05	–.11*	–.05	.11*	–.46**
17. G2 Academic Achievement	–.14*	–.10	–0.07	–0.05	–.08	–.30**	.14**	–.13**	–.14**	–.13**	–.07	–.07	–.10*	–.25**	.18**	–.07

Notes:

^*p < .05

^**p < .01.

P = parent rated. T = teacher rated.

Regression analyses: Associations of sleep to classroom behavior and academic achievement

We first examined the relation of teacher-reported sleepiness in first grade to concurrent observed classroom behavior and later academic achievement in second grade, controlling for child gender, ethnicity, age, parental education, and school (Table 3). Teacher-reported higher sleepiness was associated with lower adaptive behaviors and with higher problem behaviors such that a 1 SD difference in teacher-reported sleepiness was associated with a −.21 SD difference in adaptive behaviors and a .13 SD difference in problem behaviors. Teacher-reported higher sleepiness in first grade longitudinally predicted lower academic achievement in second grade even after controlling for academic achievement in kindergarten such that a 1 SD difference in teacher-reported sleepiness was associated with a −.10 SD difference in academic achievement one year later.

Table 3.

Regression analyses examining relations of G1 teacher-reported sleep to G1 behavior and G2 academic outcomes (n =516).

	Adaptive Behavior			Problem Behavior			Academic Achievement
	Estimate	S.E.	p	Estimate	S.E.	p	Estimate	S.E.	p
Daytime Sleepiness	−0.075	0.021	<.001	0.094	0.033	0.004	−3.578	1.297	0.006
Female	0.039	0.013	0.002	−0.078	0.026	0.002	0.371	0.89	0.677
African Caribbean	−0.024	0.015	0.121	0.012	0.031	0.708	3.437	1.249	0.006
Latino	−0.016	0.021	0.452	−0.017	0.053	0.752	1.793	1.901	0.345
Other Ethnicity	0.029	0.025	0.257	−0.082	0.066	0.219	12.33	2.245	<.001
Age	0.028	0.024	0.246	−0.026	0.045	0.566	−0.233	1.458	0.873
Parental Education	−0.012	0.014	0.414	0.021	0.029	0.455	1.616	0.996	0.105
Kindergarten Achievement							0.713	0.035	<.001

Notes: Analyses also control for school dummy variables. Race/ethnicity reference category is African American. Adaptive and problem behaviors measured at first grade. Academic achievement measured at second grade. Significant predictors are bolded.

We next examined the relation of parent-reported sleep quality in first grade to concurrent observed classroom behavior and later academic achievement in second grade, controlling for child gender, race/ethnicity, age, parental education, and school (Table 4). No aspects of parent-reported sleep quality in first grade were related to adaptive behaviors or problem behaviors in first grade. Higher bedtime resistance and higher disordered breathing reported by parents in first grade longitudinally predicted lower academic achievement in second grade, even after controlling for academic achievement in kindergarten. A 1 SD difference in bedtime resistance was associated with a −.09 SD difference in academic achievement, and a 1 SD difference in disordered breathing was associated with a −.09 SD difference in academic achievement.

Table 4.

Regression analyses examining relations of G1 parent reported sleep to G1 behavior and G2 academic outcomes (n =432).

	Adaptive Behaviors			Problem Behaviors			Academic Achievement
	Estimate	S.E.	p	Estimate	S.E.	p	Estimate	S.E.	p
Bedtime Resistance	−0.02	0.021	0.352	0.053	0.046	0.247	−3.591	1.477	0.015
Sleep Duration	0.014	0.025	0.581	0.003	0.05	0.948	−0.305	2.308	0.895
Disordered Breathing	−0.02	0.028	0.472	−0.012	0.049	0.814	−4.552	1.552	0.003
Daytime Sleepiness	−0.001	0.017	0.936	0.043	0.033	0.196	0.066	1.217	0.957
Sleep Onset Delay	−0.02	0.013	0.14	−0.023	0.026	0.387	−1.007	0.881	0.253
Female	0.06	0.015	<.001	−0.083	0.031	0.007	0.53	1.023	0.605
African Caribbean	−0.005	0.018	0.799	−0.033	0.036	0.369	3.061	1.403	0.029
Latino	−0.021	0.029	0.474	−0.02	0.068	0.77	1.037	2.114	0.624
Other Ethnicity	0.022	0.036	0.535	−0.08	0.09	0.373	10.565	2.685	<.001
Age	0.027	0.027	0.315	−0.004	0.05	0.938	−0.45	1.726	0.794
Parental Education	−0.014	0.016	0.393	0.037	0.031	0.236	1.322	1.131	0.242
Kindergarten Achievement							0.73	0.04	<.001

Notes: Analyses also control for school dummy variables. Race/ethnicity reference category is African American. Adaptive and problem behaviors measured at grade 1. Academic achievement measured at grade 2. Significant predictors are bolded.

As a robustness check on the significant concurrent relation between first grade teacher-reported sleepiness and observed behavior problems, we repeated analyses after excluding somatization/anxiety from the total problems behavior. We did this to provide a more conservative test given overlap between symptoms of somatization/anxiety and sleepiness (i.e. putting head down, sleeping). Teacher-reported sleepiness remained a significant predictor of total problem behavior even when somatization/anxiety symptoms were excluded (b = .069, p = .037).

Replication of results in second grade

We examined whether these associations held in second grade. Specifically, we examined the concurrent relations between second grade teacher-reported sleep quality and observed classroom behavior, controlling for child gender, race/ethnicity, age, parental education, and school (Supplementary Table 1). Largely the same pattern of results was found in second grade. Teacher-reported higher sleepiness was marginally associated with lower adaptive behaviors and higher problem behaviors. For parent-reported sleep quality (Supplementary Table 2), no aspect of parent-reported sleep was related to adaptive behaviors, which replicated first grade findings. Higher parent-reported daytime sleepiness, however, was related to lower problem behavior.

Discussion

Sleep is an essential component of healthy development for children, and children of color are at elevated risk for poor sleep health and sleep disorders, which can set the stage for sleepiness in school, increased problem behavior, decreased adaptive behavior, and lower academic achievement. This is the first study to examine the ways in which sleep health and sleep disorder symptoms are related to independently observed classroom behavior and academic achievement test scores among primarily Black children growing up in historically disinvested neighborhoods.

In line with the Integrative Model proposed by Garcia Coll and colleagues (1996), sleep health was predictive of both behavioral and academic competencies in the classroom among children of color. Results demonstrate that higher teacher-reported sleepiness is concurrently associated with both lower adaptive behaviors and higher problem behaviors observed in the classroom and longitudinally predicts lower academic achievement test scores one year later in second grade. Sleepiness in the classroom may limit children’s ability to engage in adaptive classroom behaviors. From a neurobiological perspective, sleepiness makes it more difficult to activate prefrontal brain regions that support higher-level cognitive skills which are important for planning and goal-directed action (Thomas et al., 2000). Behaviorally, this would make it much more difficult for children to engage in adaptive behaviors to support the goal of learning such as focusing attention, listening, participating in discussion, and working on assigned tasks. At the same time, lower prefrontal activation would also predispose children to difficulties regulating their impulses and emotions which may set the stage for problem behaviors (Minkel et al., 2012; Palmer & Alfano, 2017). While these concurrent relations of sleep to behavior may be directly due to sleepiness, the longitudinal relation between sleep health and academic achievement may in part also result from an accumulation of risk (Buckhalt et al., 2009). If sleepiness puts children at risk for acquiring fundamentals in first grade, then it will be more difficult for them to further build knowledge in second grade because this later learning builds on the attainment of earlier fundamentals (Buckhalt et al., 2009).

Results indicate that first grade teacher ratings of a child’s daytime sleepiness in the classroom are an important predictor of children’s school success. To take advantage of teacher’s direct observations of children during the school day, we adapted the widely used parent reported CSHQ to assess teacher perceptions of child daytime sleepiness, whereas previous research has largely relied on parent-reported child sleep (McDowall et al., 2017). Results demonstrated that teacher report of children’s sleepiness during the school day was more strongly related to children’s independently observed classroom behavior than was parent report of children’s sleep health and was an equally strong predictor of academic achievement. These results are consistent with a recent study which assessed teacher perceptions of child daytime sleepiness using the Teacher’s Daytime Sleepiness Questionnaire (TDSQ; Owens et al., 2000) and found that in a sample of children with ADHD, teacher-rated daytime sleepiness was related to teacher-rated emotional and behavioral problems whereas the more typically used parent-rated sleep problems were not (Lucas et al., 2017). Given that children spend much of their day in school, and that teachers may have a more accurate frame of reference as to the range of manifestations of sleepiness because they observe children over the course of the school day in a range of activities, teachers are potentially highly useful informants about this aspect of sleep health. Prior work with older (fifth grade) children found a low correlation between child self-report and teacher-report of children’s daytime sleepiness and thus concluded that teachers may not be accurate in identifying which of their students have sleep problems (Amschler & McKenzie, 2005). Although it is true that teachers may not have in-depth knowledge of the types of sleep problems that children are experiencing at home, study findings suggest that teacher knowledge of which children appear sleepy in class is related to multiple important school outcomes. Additionally, teacher reports of sleep offer several advantages for population and field-based research with children. The very brief 3-item report used in the current study is feasible to add to other survey measures that teachers are asked to fill out about their students. Although using teacher reports does not give the precise information that can be obtained from polysomnography or other objective measures of sleep, these other measures are often not feasible to include in large population studies where researchers collect data from children in the school setting. Furthermore, unlike the TDSQ which includes items assessing behavior problems, the 3-item report used in this study focused specifically on daytime sleepiness.

Moreover, results also suggest the potential importance of educating teachers about the prediction from their observations of sleepiness to child academic achievement so that teachers understand that sleepiness in the classroom is an important warning sign. This knowledge would help prepare teachers to in turn share their observations of sleepiness with parents, in a collaborative manner, so that parents are aware that sleepiness may be interfering with learning. For example, just as elementary school teachers provide feedback about children’s social skills in the classroom, they could also provide information about sleepiness. Schools could also provide basic sleep recommendations such as the number of hours of sleep recommended by age, information on healthy bedtime routines, and guidance on brief evidence-based interventions for parents to address sleep problems. This information could help make sense of other difficulties children may be having in school and support parents in improving bedtime routines and accessing sleep problem interventions as an important way to help children succeed in school. Receiving this information from teachers may also support parents in identifying the need for further consultation with pediatricians to assess whether any underlying issues such as obstructive sleep apnea are present.

In contrast to teacher reports, parent reports of child sleep were not meaningfully related to children’s classroom behavior. Specifically, there were no associations between parent reports of daytime sleepiness, sleep disordered breathing, sleep duration, bedtime resistance, or sleep onset delay and children’s observed adaptive or problem behaviors in the classroom. This is in contrast with expectations and findings from some prior studies which have demonstrated relations between parent reports of sleep health and children’s behavior problems as rated by parents (Spruyt et al., 2016; Witcher et al., 2012) and teachers (Scullin Ornelas, & Montgomery-Downs, 2011). Of note is a recent study of children with ADHD that also failed to find associations between parent-reported sleep problems and emotional and behavioral problems in the classroom (Lucas et al., 2017).

Although parent reported child sleep was unrelated to concurrent classroom behavior, parent-reported bedtime resistance and sleep disordered breathing in first grade predicted lower academic achievement test scores one year later. These findings are consistent with past studies which have demonstrated links between parent reported sleep health and children’s academic achievement (Astill et al., 2012; Dewald et al., 2010) and sleep disordered breathing and poor outcomes in the classroom (Witcher et al., 2012). Of note, parent reported bedtime resistance was the most common sleep problem reported for this sample, with more than two thirds of parents reporting that bedtime resistance was a problem more than once a week. From a population health perspective, this common problem among first grade children of color from historically disinvested neighborhoods that was associated with a significant decrement (.09 SD) in academic achievement test scores in second grade is a promising candidate for universal intervention. Although sleep disordered breathing was less common, it was still reported to occur more than once a week in nearly one fifth of the sample; disordered breathing is an indicator of sleep apnea, which can interfere with numerous aspects of child health and development. The association with academic achievement one year later was similar in size to that found for bedtime resistance, and suggests further opportunity for health promotion and prevention efforts.

As a check for scientific rigor, we repeated concurrent analyses when children were in second grade. The pattern of results for associations with teacher-reported sleepiness largely replicated what was seen in first grade. Higher teacher-reported sleepiness was concurrently associated with higher observed adaptive behaviors and lower problem behaviors (although these relations were only marginally significant). The pattern of results with parent reported sleep health was also similar but less consistent. Of note, an association between parent-reported daytime sleepiness and problem behaviors that was not seen in first grade did emerge in second grade, but this association was in the unexpected direction such that higher sleepiness was related to fewer problem behaviors. One possible interpretation is that children who are rated by parents as sleepy may exhibit fewer problem behaviors because they are generally less active and somewhat subdued or withdrawn. This finding, however, should be interpreted with caution as it was not hypothesized. Overall, the pattern of results in second grade, particularly with regard to the associations with teacher-reported sleepiness, is largely similar to the pattern seen in first grade.

This study offers a novel methodological contribution in using a standardized assessment of child behavior through systematic observations conducted by trained observers, rather than teacher report. This approach may aid in reducing the bias in teacher reports of observing relatively more problematic behavior among children of color (Downey & Pribesh, 2004; Haller, 1985; Riegle-Crumb & Humphries, 2012). The observation methodology lends itself to greater objectivity in that observers are directed to focus on a specific child for a specific amount of time and observers use a time sampling methodology to record discrete behaviors. In the literature on implicit bias, it has been shown that teachers look longer at Black children, and Black boys in particular, when expecting challenging behaviors (Gilliam et al., 2016). The observational methodology minimizes this aspect of bias by directing observers to concentrate on an individual child for a set period of time. Moreover, if teachers are spending more time looking for challenging behaviors among Black children, they may then actually observe or believe that they have observed more challenging behaviors among these children and thus provide global ratings of behavior that reflect this bias. By utilizing a time sampling methodology, however, these observer ratings focus on actual instances of behavior rather than a global recall of behavior that may be more subject to bias. The high intra-class correlations, indicating similar observations across different observers, is a reassuring indicator of the consistency with which observers were able to apply this methodology. Although this method greatly increases our confidence in the objectivity of the outcome measures, observers are not free of racial bias which may impact their observations. Moreover, implicit racial biases could play a role in how sleepiness is interpreted by teachers and in how it subsequently shapes interactions with students and students’ opportunities to learn in the classroom.

Racism is pervasive and persistent in all institutions in the U.S. including schools; therefore we must consider the ways in which daytime sleepiness may pose an even greater risk for students of color than it does for White students (Feagin & Bennefield, 2014; Garcia Coll et al., 1996; Buckhalt et al., 2009; Buckhalt 2011). When children of color are sleepy in the classroom, they may be more likely to be perceived by teachers or peers as not caring about school. When sleepiness leads to behavioral problems for students of color, teachers may be more likely to make trait based attributions about behavior, categorizing a child as lazy, aggressive, or non-compliant rather than recognizing that the temporary state of sleepiness is making it harder for some children to engage in school (Wang & Hall, 2018). Furthermore, teachers may make negative attributions about parenting and the home environment and the extent to which parents care about their children and schooling and are thus worth engaging. Moreover, Black children and Black boys in particular are more likely to receive harsher punishments including suspensions for behavior problems at school than White children (Townsend, 2000), which suggests that the potential consequences of sleepiness and problem behavior related to sleepiness are much greater for Black children.

Within the study sample of Black children, teachers rated children of African American parents as higher in sleepiness than children of African Caribbean parents. Of note, the vast majority of African Caribbean children were 2^nd generation (their parents had immigrated to the U.S.). In line with the immigrant paradox, the protective factor of immigrant status has been associated with better sleep health in Latino adult populations (Seicean et al., 2011), and evidence suggests that African Caribbean groups report fewer sleep problems or complaints across the lifespan compared to African Americans, European Americans or Eastern Europeans (Adenekan et al., 2013; Jean-Louis et al., 2001). This is, however, the first study to investigate whether the same pattern is present in African Caribbean children’s sleep health. Importantly, however, in the United States, racism experienced among African Caribbean/West Indian groups erodes the protective buffer of the immigrant paradox the longer they reside in the US (more than 20 years), putting them at equal risk for poor health outcomes (Hamilton & Hummer, 2011; Paradies et al., 2015), except in the case of sleep health (Adenekan et al., 2013). Future research should investigate the strength-based factors that preserve the immigrant paradox buffer and protect African Caribbean groups from poor sleep health. Alternatively, there is the possibility that teachers exhibited implicit biases in their rating of Black children, rating African American children more negatively (i.e., sleepier); this is a viable explanation particularly since there was no difference in parent ratings of children’s sleep health. Further work is needed to understand the role of ethnicity, immigrant status, acculturation, and associated protective factors in potentially explaining this difference between African American and African Caribbean children in teacher reports of child sleepiness.

Limitations and Future Research

Several limitations exist. First, we did not have any objective measures of sleep health such as actigraphy or polysomnography which is considered to be the gold standard assessment of sleep measures in children and adults. Moreover, in one study which attempted to validate the CSHQ, none of the CSHQ subscales used in this study were associated with actigraphy or polysomnography (Markovich et al., 2014). Despite this lack of correlation, however, subjective and objective sleep measures are believed to each provide unique contributions to understanding sleep health. Also, the CSHQ subscales had only acceptable or nearly acceptable levels of reliability (Cronbach’s alpha for the various dimensions ranged from 0.58 to 0.64). Although these measures have been used widely, the paper which validated the CSHQ had similar reliability (Cronbach’s alpha ranged from 0.51 to 0.70, Owens et al., 2000) as did a recent study using these measures for assessing child sleep (Urfer-Maurer et al., 2017). The low reliability may suggest, however, that the measure is not adequately capturing sleep health in this population. The current study included measures of sleep from two different sources – parents and teachers-- to provide a broader assessment of children’s sleep health. Nevertheless, it is possible that there may be inherent biases in these measures particularly with regard to Black children generally, or African American children specifically, being incorrectly rated as sleepy. More work is needed to validate and understand biases in teacher-rated daytime sleepiness in the classroom and to utilize objective measures of sleep at home and in school to further understand sleep health among children of color. Our measurement of family socioeconomic characteristics was also limited to a dichotomous measure of parental education which may have reduced our ability to detect differences in children’s sleep health.

Additionally, although we controlled for several important covariates and examined longitudinal relations with academic achievement, we cannot make causal claims about the relations between sleep health and classroom behavior or achievement. As in any cross-sectional analysis, the possibility of reverse causation remains for analyses within first grade and replicated within second grade. For example, students who exhibit “challenging” behavior in the classroom may lose sleep resulting from negative interactions or consequences from both teachers and parents. Moreover, it is possible that a range of social determinants of health place children at risk for poor sleep, challenging behavior in the classroom and underachievement. Stressors such as racism, discrimination, financial strain, shift work, crowding housing and community-level noise may operate through the same or different pathways for these three outcomes of sleep, behavior and academic achievement. Possible mechanisms include parent mental health, parenting stress, limited resources and opportunities and child self-regulation (Blair & Raver, 2012). Controlling for earlier achievement in predicting later achievement helps to move toward understanding the independent contribution of sleep health to academic achievement, but future work is needed to understand the mechanisms by which sleep health influences behavioral health and school success. Based on the literature on child health and development, poverty, parenting stress, parent mental health, racism, discrimination and child self-regulation are all important processes to explore. Furthermore, as this study included relatively few LatinX children, findings cannot be generalized to LatinX or other populations of children of color. Future work should include children from these racial and ethnic backgrounds.

Conclusion

Poor sleep health and sleep disorders among children are pressing public health concerns which disproportionately affect children of color. This study contributes to the literature by examining the associations of parent and teacher-reported child sleep with observed behavior in classroom settings and academic achievement test scores. Teacher report of child daytime sleepiness is meaningfully related to observed classroom behavior (adaptive and problematic). First grade teacher reports of child sleepiness and parent reports of bedtime resistance and disordered breathing predict academic achievement test scores in the subsequent school year. Given the high rates of bedtime resistance and disordered breathing reported by parents (over two thirds of the sample experience resistance and nearly one fifth have disordered breathing more than one night a week), and the simple nature of the teacher rated sleepiness variable, a range of universal strategies or interventions warrant future study. A sleep health curriculum to engage teachers and parents on the ways to promote sleep health hygiene may serve as a stepping-stone towards moving towards a school-based program to address this issue. Home-based solutions for parents, such as audio-based mobile apps (limiting screen exposure) may prompt soothing bedtime routines that may be an effective intervention, reinforcing sleep health curriculum taught in schools. Furthermore, screening for sleepiness may be incorporated into standard practice followed by culturally relevant information and referrals for families.