Abstract
Objectives.
To test whether adolescents’ mental health during the COVID-19 pandemic is associated with the combination of their instructional approach(es) and their sleep patterns.
Design.
Cross-sectional.
Setting.
Adolescents were recruited through social media outlets in October/November 2020 to complete an online survey.
Participants.
Participants were 4,442 geographically and racially diverse, community-dwelling students (grades 6–12, 51% female, 36% non-White, 87% high schoolers).
Measurements.
Participants completed items from the PROMIS Pediatric Depressive Symptoms and Anxiety scales. Participants reported their instructional approach(es), bedtimes, and wake times for each day in the past week. Participants were categorized into five combined instructional approach groups. Average sleep opportunity was calculated as the average time between bedtime and wake time. Social jetlag was calculated as the difference between the average sleep midpoint preceding non-scheduled and scheduled days.
Results.
Emotional distress was elevated in this sample, with a large proportion of adolescents reporting moderate-severe (T-score≥65) levels of depressive symptoms (49%) and anxiety (28%). There were significant differences between instructional approach groups, such that adolescents attending all schooldays in-person reported the lowest depressive symptom and anxiety T-scores (p<.001, ηp2=.012), but also the shortest sleep opportunity (p<.001, ηp2=.077) and greatest social jetlag (p<.001, ηp2 =.037) of all groups. Adolescents attending school in person, with sufficient sleep opportunity (≥8–9 hours/night) and limited social jetlag (<2hrs) had significantly lower depressive (ηp2 =.014) and anxiety (ηp2 =.008) T-scores than other adolescents.
Conclusions.
Prioritizing in-person education and promoting healthy sleep patterns (more sleep opportunity, more consistent sleep schedules) may help bolster adolescent mental health.
Keywords: adolescence, mental health, sleep, social jetlag, education, COVID-19
INTRODUCTION
The US Surgeon General recently declared a national mental health crisis among youth, noting an ongoing rise in depression and anxiety among adolescents that has been exacerbated by the coronavirus (COVID-19) pandemic1. The COVID-19 pandemic led to nationwide changes across all domains of life, including the educational system. In the Fall of 2020, schools across the country implemented a wide array of instructional approaches to adapt to COVID-19 restrictions. As a result, students across the U.S. received differing combinations of instructional approaches that included in-person classes, online instruction, and hybrid or mixed formats. Despite significant changes to students’ instructional format, little is known regarding how these changes relate to adolescents’ mental health.
Preliminary evidence suggests that in-person instruction may be protective of adolescent mental health2 and well-being3. For instance, a recent study found that adolescents receiving in-person instruction during the COVID-19 pandemic reported less stress and depressive symptoms than those who received virtual instruction2. While this work distinguished between the mode of learning (in-person, virtual), however, it did not distinguish between virtual classes that were scheduled and led by live instruction (synchronous) vs. classes that could be completed at any time (asynchronous) or whether students received a combination of these instructional formats.
In a large sample of US adolescents, we have previously shown that sleep schedules and sleep opportunity differ based on the instructional approach that adolescents receive the next day, and the combination of instructional approaches they receive in the week influences the night-to-night consistency of their sleep schedules4. Our main conclusion was that in-person instruction was associated with an insufficient sleep opportunity, which is a known factor that contributes to poorer mental health outcomes5.This stands in contrast with aforementioned studies supporting the benefits of in-person instruction2,3. We propose that instructional approaches not only affect sleep but that the combination of instructional approach and sleep together influence adolescents’ mental health.
Adolescents are vulnerable not only to insufficient sleep opportunity but also social jetlag6, another potential risk factor for depression and anxiety7,8. Calculated as the discrepancy between an individual’s sleep midpoint preceding work/school days and their free (non-work/school) days, social jetlag is a chronic, repeated jet-lag like phenomenon that reflects the misalignment between an individual’s endogenous circadian clock and actual sleep times6,9. Relative to adults, adolescents are more likely to experience social jetlag6 because they have a delayed internal circadian phase and yet are enforced by early school start times to wake early on school days10–12. This circumstance leads adolescents to shift sleep times frequently when transitioning between school days and weekends. The COVID-19 pandemic initially led to a nationwide lockdown, and there is evidence that social jetlag among adolescents decreased during that time13. It is unknown, however, how much social jetlag adolescents experience in the context of different instructional approaches that have been instituted since then and whether the combination of social jetlag and these instructional approaches associate with adolescent mental health.
As previously described, the Nationwide Education and Sleep in TEens During COVID (NESTED) study was developed to examine associations between instructional approaches, school start times, and sleep outcomes in a large, racially diverse sample of adolescents across the U.S4. This paper extends the previous findings from the NESTED study, examining how adolescents’ mental health (depressive symptoms and anxiety), weekly sleep opportunity, and social jetlag differ based on the combination of instructional approaches they experience across their school week. In addition, this study examines whether the instruction received by an adolescent and their sleep characteristics together associate with mental health, such that adolescents who experience both the instructional approach linked with the best mental health outcomes and optimized sleep (sufficient sleep opportunity and limited social jetlag) have better mental health outcomes than other adolescents.
PARTICIPANTS AND METHODS
Study Design
Complete details of the NESTED study design have been described4. In brief, the NESTED study recruited adolescents through Facebook and Instagram social media platforms from October 14 to November 26, 2020. Using the Facebook marketing platform, which markets through both Facebook and Instagram, we targeted adolescents ages 13–18 years and over-marketed to specific demographic subgroups identified as underrepresented amongst initial survey respondents, such as Black males. Through our social media ads, individuals were provided a study description and a link to a REDCap survey. Individuals were eligible to enroll if they reported that they were a U.S. resident, a student in grades 6–12, and had internet access at the time of study enrollment. As described in our previous report4, the resulting sample reflects a broad range of instructional approaches and is consistent with key demographics in the School Enrollment in the United States: 2018 report from the U.S. Census bureau.14 This minimal risk study was approved by the BRANY SBER Institutional Review Board (#20-053-528) with a waiver of informed parental consent. Upon clicking a link to the study, participants were provided study information, informed of the voluntary nature of the study, potential risks, and notified that they had the ability to discontinue at any point. To continue to the survey, participants had to select an option indicating assent to participate. No compensation was given.
Measures
Current Anxiety and Depressive Symptoms.
Four items from the PROMIS Pediatric Depressive Symptoms and three items from the PROMIS Pediatric Anxiety item banks were included in the survey to assess symptoms of depression and anxiety, respectively (see Supplement for items). Both these custom short form scales had Cronbach alpha’s >= .88 in our sample, indicating high internal consistency and reliability. Participants were asked to rate each item based on the past 7 days. Each question had a 5-point Likert response scale (“Almost Always” to “Never”). The Health Measures Scoring Service (https://www.assessmentcenter.net/ac_scoringservice) was used to generate a final T-score for each scale based on data that was centered on representative U.S. samples (T-score mean =50, standard deviation =10). In primary analyses, the depressive symptoms and anxiety T-scores were included in analyses as continuous variables. In secondary analyses with a focus on interpreting sample distribution and symptom severity, participants’ scores were classified using the PROMIS T-Score cut-offs that correspond with both Depressive Symptoms and Anxiety scales: Within-Normal Limits, T-score <50; Mild, 50–54.9; Moderate, 55–64.9; Moderate-Severe, 65+15.
Instructional Approaches.
Participants were asked to report on the format(s) of their school education in the past week. For each weekday (Monday – Friday), they were asked to select one of the following instructional format options: in-person, online/synchronous (live online classes or scheduled interactions with teachers), online/asynchronous (online, but without live classes or scheduled teacher interactions), or did not attend school. Students could report experiencing up to three instructional formats, along with not attending school and non-school days, for the week. Participants were categorized based on the combination of their reported instructional formats across the school week (Monday – Friday) into one of the following instructional approach groups: (1) In-Person (all 5 days), (2) Hybrid (at least one in-person instruction day/week and at least one online day), (3) Online/Synchronous (all 5 days), (4) Online/Mixed Synchronous (no in-person, but at least one online synchronous instruction day/week), and (5) Online Asynchronous/No School (all 5 days).
Sleep Opportunity and Social Jetlag.
As previously described4, for each instructional approach selected, participants were asked what time they tried to fall asleep the night before (bedtime; BT) and what time they woke up on that school morning to start their day (wake time; WT). All participants were also asked to report their BT and WT on weekends/non-school days, with the assumption that sleep schedules were similar for these days. Consistent with our previous report, we defined sleep opportunity as the total number of hours between reported BT and WT, and, as a common method used in pediatric sleep survey studies, we examined sleep opportunity as a proxy for sleep duration4,16. While we previously focused on how sleep opportunity during a given night associates with instructional approach the next day, here we calculated the average sleep opportunity participants obtain across the week (5 weekday and 2 weekend nights) to examine how it relates to students’ education across the school week and their mental health. In secondary analyses, participants were categorized based on whether they experienced sufficient sleep opportunity (y/n). Sufficient sleep was defined as ≥9 hours for middle schoolers and ≥8 hours for high schoolers based on the lower end of recommendations provided by the American Academy of Sleep Medicine16.
Social jetlag was calculated as the absolute difference between the average sleep midpoint on free days versus scheduled days 6,9. We refer to days on which participants attended school in-person or online/synchronous as “scheduled” days given that these days entailed a set school start time. We refer to weekends and weekdays that participants reported having online/asynchronous classes or no school as unscheduled or “free” days. Based on participants’ reported weekly schedule and associated BTs and WTs per instructional format, we derived their average sleep timing (BT, WT, and midpoint between BT and WT) preceding free days, scheduled days, and whole week. Of note, our calculation would yield no social jetlag for participants who endorsed having a free day every day (i.e., 5 days online/asynchronous and/or no school). To avoid potentially biasing the data, we did not include these participants in analyses that involved social jetlag. In secondary analyses, participants were categorized based on whether they experienced limited social jetlag (y/n). Limited social jetlag was defined as <2 hours social jetlag based on evidence linking ≥2 hours social jetlag to poor mental health and health outcomes8,17–20.
Covariates.
Demographic Variables: Participants were asked to self-identify grade, gender and race/ethnicity. From these data, the following categorical variables were included as dummy-coded variables in analyses: school level (middle school, high school), gender (male, female, non-binary/other/prefer not to respond), and race/ethnicity (White, Black, Hispanic/Latino[a], Asian, Mixed and Other).
History/Presence of Anxiety and Depression:
To distinguish whether participants have a history or presence of depression or anxiety, participants were asked to complete two items: “Has a health care provider ever told you that you have any of the following: Depression (yes/no), Anxiety (yes/no).” Each item was included as a dichotomous variable in analytical models.
Chronotype:
Evening chronotype has been linked to poorer mental health. Participant chronotype was calculated and included as a covariate in analyses to distinguish whether instructional approach and sleep characteristics are associated with mental health beyond effects of individual differences in chronotype. Consistent with an established method of assessing chronotype from sleep timing9, we calculated chronotype as the midpoint between self-reported average bedtime and waketime on free days minus (i.e., corrected for) sleep debt. Sleep debt was calculated as the difference between TIB on free days and scheduled days. Since the Online Asynchronous group endorsed having non-scheduled days every day, a chronotype calculation was not possible. We preserved comparison of all five instructional approaches in the primary analyses, and conducted supplementary analyses with chronotype as an additional covariate comparing mental health outcomes across four instructional approach groups (omitting Online Asynchronous).
School Start Time:
School start time information was gathered for in-person days and online/synchronous days. Participants who endorsed having either of these instructional formats were asked to report when their first class began on each of these days. Since other formats (online/asynchronous, no school days) by definition did not involve scheduled classes, no school start time data were gathered for these days. As a result, SST data was available specifically for the In-Person (all 5-days) and Online Synchronous (all 5 days) groups.
Statistical Analyses
Statistical analyses were performed with SPSS® 26.0 (IBM Corporation, Armonk NY, USA) for Windows®/Apple Mac®. Data were first analyzed with one-way ANOVAs to test for differences in mental health outcomes (anxiety and depression T-scores) between participant demographic groups (gender, school-level, race), and between participants with vs. without a self-reported history of anxiety and of depression. To consider the role of school start time, we tested whether school start time correlated with sleep opportunity, sufficient sleep (y/n), and mental health T-scores. To examine differences in mental health between instructional approach groups, an ANCOVA was performed controlling for participant demographics, and mental health history. Supplementary ANCOVAs were conducted with chronotype included as an additional covariate. These analyses were used to identify which instructional approach is linked with the best mental health outcome. Next, participants were categorized based on whether they had the instructional approach associated with the best mental health outcome or not, whether they experienced sufficient sleep, and whether they had limited social jetlag. ANCOVAs were performed to test whether average depressive symptoms and anxiety T-scores differed between these groups, and Pearson chi-square analyses were conducted to test whether the symptom severity classifications differed significantly across groups. For ANCOVA analyses, partial eta-square (ηp2) was used to determine effect size, with .01= small effect, .06= medium, .14= large effect. For the chi-square analyses, Cramer’s V (df=3) was used to determine effect size with .06= small effect, .17= medium effect, and .29= large effect21.
RESULTS
Participant Characteristics
Consent and instructional approach responses were provided by 6,577 adolescents. Of these, 4,442 (67.5%) provided sleep data and PROMIS anxiety items; 4,384 participants also completed the PROMIS depressive symptoms items. Approximately 11% (n= 502) of participants did not indicate having both scheduled and free days, and were excluded from analyses that included social jetlag because they did not report the corresponding sleep data needed to calculate social jetlag. This subset included the 10.3% (n=456) of participants who reported having online/asynchronous learning or no school 5 days a week.
Participant characteristics of the full NESTED sample and the differences in participant characteristics between instructional approach groups have been described elsewhere4. Differences in demographics and sleep characteristics between the original NESTED sample and current subsample are reported in the Supplement. Consistent with the larger sample, the current subsample was predominantly high schoolers (88%), and diverse in self-reported gender, race, and ethnicity (see Table 1). Over one-third of participants reported having ever been told by a health care provider that they had depression (34%) or anxiety (42%). Chi-square analyses showed small differences between instructional approaches, with the highest proportion of adolescents reporting a history of depression or anxiety in the Asynchronous/No School group (small effects, Cramer’s Vs: .06-.07; Supplementary Table 1).
Table 1.
Differences in Mental Health across Participant Demographics and Self-Reported History of Anxiety and Depression
| Sample Distribution | Depressive Symptoms Mean (95% CI) | Depressive Symptoms Test Statistics | Anxiety Mean (95% CI) | Anxiety Test Statistics | |
|---|---|---|---|---|---|
|
| |||||
| Gender | |||||
|
| |||||
| Female | 51.9% | 64.3 (63.9–64.7) |
F(2,4292)=80.97 ηp2 =.036, p<.001 | 60.6 (60.2–61.0) |
F(2,4332)=47.84 ηp2 = .022, p<.001 |
| Male | 39.1% | 61.6 (61.1–62.1) | 58.6 (58.2–59.1) | ||
| Non-Binary/Other | 8.9% | 68.3 (67.4–69.0) | 63.2 (62.4–64.0) | ||
|
| |||||
| School Level | |||||
|
| |||||
| Middle School | 11.9% | 62.5 (61.6–63.5) | F(1,4382)=6.07 ηp2=.001, p=.014 | 58.3 (57.5–59.2) | F(1,4440)=20.67 ηp2=.005, p<.001 |
| High School | 88.1% | 63.7 (63.4–64.0) | 60.3 (60.0–60.6) | ||
|
| |||||
| Race/Ethnicity | |||||
|
| |||||
| White | 66.1% | 63.3 (62.9–63.6) | 60.1 (59.8–60.5) | ||
| Black | 4.2% | 63.7 (62.0–65.3) | F(4,4290)=4.50 ηp2=.004, p=.001 | 58.8 (57.4–60.2) | F(4,4330)=1.19 ηp2=.001, p=.312 |
| Hispanic | 16.7% | 64.7 (64.0–65.5) | 60.0 (59.3–60.7) | ||
| Asian | 3.7% | 62.3 (60.6–63.9) | 59.5 (58.1–60.9) | ||
| Multiracial/Other | 9.3% | 64.6 (63.6–65.5) | 60.4 (59.5–61.2) | ||
|
| |||||
| Self-Reported History of Depression | |||||
|
| |||||
| No | 65.0% | 61.0 (60.6–61.4) | F(1,4292)=596.03, ηp2=.122, p<.001 | 58.1 (57.8–58.4) | F(1,4332)=403.24 ηp2=.085, p<.001 |
| Yes | 35.0% | 68.5 (68.1–68.9) | 63.7 (63.3–64.1) | ||
|
| |||||
| Self-Reported History of Anxiety | |||||
|
| |||||
| No | 58.4% | 61.2 (60.8–61.6) | F(1,4292)=368.03, ηp2=.079, p<.001 | 57.8 (57.5–58.2) | F(1,4332)=35.02 ηp2=.084, p<.001 |
| Yes | 41.6% | 67.0 (66.6–67.4) | 63.2 (62.8–63.6) | ||
Legend: Anxiety Symptoms refer to PROMIS Anxiety T-Scores, Depressive Symptoms refer to PROMIS Depressive Symptoms T-Scores.
School Start Time (SST) and Mental Health
Consistent with our parent study, we found that earlier SST among adolescents attending school 5 days In-Person or 5 days Online Synchronous was correlated with shorter sleep opportunity (τb=.16, p<.001) and obtaining insufficient sleep (τb=.15, p<.001). SST, however, was not correlated with either depressive symptoms (τb=.00, p=.940) or anxiety T-scores (τb=.00, p=.723).
Instructional Approach and Mental Health
On average, reported symptoms of depression (T-score mean: 63.6±10.2, range: 38.1–78.7) and anxiety (T-score mean: 60.0±9.2, range:38.5–72.3) fell in the moderate range (T-score 55–65). Forty-nine percent of participants reported depressive symptoms in the moderate-severe range (T-score > 65), while 28% reported moderate-severe levels of anxiety. As shown in Table 1, depressive symptom and anxiety T-scores differed as a function of participant demographic variables and were significantly higher for individuals who indicated a history of depression or anxiety. After controlling for the effects of these demographics and mental health history, we found that depressive symptom T-score differed as a function of instructional approach, with the lowest average score observed in the In-Person group (F(4,4284)=12.98, p<.001, ηp2=.012, indicating a small effect; Figure 1A). Using the same model, we found that anxiety T-scores similarly differed between instructional approach groups, with In-Person group reporting the lowest score, although the effect size was minimal (F(4,4324)=3.704, p=.005, ηp2=.003; Figure 1B). Separate models covarying for participant chronotype showed consistent results (see Supplement).
Figure 1. Adolescents’ Mental Health (A & B), Sleep Opportunity (C) and Social Jetlag (D) Significantly Differ based on their Combined Instructional Approach.
Error bars represent 95% confidence intervals.
Instructional Approach, Sleep Opportunity, and Social Jetlag
Weekly average sleep opportunity was 8.5 (SD=1.3) hours per night, and 62% of participants had sufficient sleep opportunity based on their average reported sleep opportunity across the week. On average, participants reported later bedtimes, later wake times, and longer sleep opportunity on nights preceding free days compared to nights preceding scheduled days (Supplementary Table 2). Approximately 38% reported ≥2 hours social jetlag. A small percentage of participants (3%) reported an advance in sleep midpoint such that they slept earlier before free days compared to scheduled days. Of note, removal of these participants had minimal influence on our results, thus, presented results include these participants.
In our previous paper, we found that adolescents reported the earliest bedtimes, earliest wake times, and shortest sleep opportunity on nights before attending school in-person relative to other learning formats4. Consistent with this report, we found that after controlling for participant demographics and mental health history, adolescents’ weekly average sleep opportunity differed significantly across instructional approach groups (F(4,4324)=89.93, p<.001, ηp2=.077 indicating a medium effect). As shown in Figure 1C, post-hoc contrasts showed the In-Person group had, on average, less sleep opportunity than other groups. Using the same model, we also found social jetlag significantly differed between instructional groups (F(3,3834)=48.53, p<.001, ηp2 =.037, indicating a small effect). As shown in Figure 1D, adolescents attending school In-Person had, on average, significantly more social jetlag than those in other groups.
In Person Instruction, Sufficient Sleep, and Limited Social Jetlag
As described above, we found that the In-Person instructional approach was associated with the best mental health outcomes, with students attending school in person everyday reporting the lowest levels of anxiety and depressive symptoms. While In-Person instruction was also associated with less sleep opportunity than other combined instructional approaches, average sleep opportunity within this group showed considerable variability in (mean: 8.0± 1.2hrs, range: 4.1–11.5hrs/night). In our secondary analyses, we examined whether the combination of having daily In-Person instruction and sufficient sleep opportunity was associated with differences in depressive symptom and anxiety scores. We categorized participants into 4 groups based on the combination of whether they attended school In-Person daily (yes/no) and/or had Sufficient Sleep opportunity (yes/no). ANCOVA analyses showed small effects of group status on both depressive symptoms (F(3,4285)=20.40, p<.001, ηp2=.014) and anxiety (F(3,4325)=14.13, p<.001, ηp2=.010) T-scores, with the In-Person/Sufficient Sleep group having the lowest symptom scores (Supplementary Figure 1). Supplementary analyses including chronotype as an additional covariate showed consistent results (see Supplement). As reported in Table 2, Pearson Chi-Square analyses showed that the distribution of depressive symptom and anxiety symptom T-scores significantly differed across groups, such that the In-Person/Sufficient Sleep group had the largest portion of adolescents that had T-scores falling within normal limits or in the mild symptoms range and the smallest portion who fell in the severe range. In contrast, the Other Instructional Approach/Insufficient Sleep group had the largest portion of adolescents reporting symptoms in the severe range.
Table 2.
In-Person Instruction and Sufficient Sleep Opportunity are associated with Lower Depressive and Anxiety Symptoms
| In-Person, Sufficient TIB | In-Person, Insufficient TIB | Other Instructional Approach, Sufficient TIB | Other Instructional Approach, Insufficient TIB | Test Statistic | |
|---|---|---|---|---|---|
|
|
|||||
| Depressive Symptoms | (n=464) | (n=486) | (n=2257) | (n=1175) | Pearson Chi Square(9)= 65.3, Cramer’s V(3)= .07, p<.001 |
| % Within Normal Limits | 15.7 | 9.7 | 10.2 | 8.6 | |
| % Mild | 10.8 | 8.6 | 8.4 | 6.4 | |
| % Moderate | 35.8 | 34.2 | 33.6 | 27.9 | |
| % Moderate-Severe | 37.7 | 47.5 | 47.8 | 57.2 | |
| Anxiety Symptoms | (n=467) | (n=493) | (n=2283) | (n=1199) | Pearson Chi Square(9)=50.6, Cramer’s V(3)= .06, p<.001 |
| % Within Normal Limits | 21.2 | 16.8 | 17.3 | 12.8 | |
| % Mild | 14.8 | 12.2 | 10.1 | 9.4 | |
| % Moderate | 44.3 | 45.8 | 45.2 | 45.3 | |
| % Moderate-Severe | 19.7 | 25.2 | 27.3 | 32.5 | |
Note: Sufficient TIB refers to time in bed ≥8 hrs among highschoolers, and ≥9hrs for middle schoolers.
Similar to the above analyses, we examined whether the combination of In-Person and limited social jetlag (<2 hrs.) would associate with lower depressive symptom and anxiety symptom T-scores. We categorized participants based on whether they attended school In-Person (yes/no) and/or had Limited Social jetlag (yes/no). ANCOVA analyses showed small effects of these groups on depressive symptoms (F(3,3798)=8.04, p<.001, ηp2 =.006) and anxiety (F(3,3834)=4.07, p=.007, ηp2 =.003) T-scores with In-Person/Limited Social Jetlag had the lowest scores (Supplementary Figure 1). Pearson Chi-Square analyses showed significant differences between groups in the distribution of T-scores, with In-Person/Limited Social Jetlag group having the lowest proportion of adolescents reporting symptoms in the severe range (Table 3). Finally, we tested whether the combination of In-Person instruction and Optimal Sleep (defined as having both sufficient sleep opportunity and limited social jetlag) would be associated with significant differences in depressive symptom and anxiety T-scores. We found that depressive symptom and anxiety T-scores varied as a function of these groups (F(3,3798)=17.6, p<.001, ηp2 =.014 and F(3,3834)=10.5, p<.001, ηp2 =.008, respectively). Adolescents with all three factors had the lowest average depressive symptom and anxiety T-scores (Figure 2), and lowest proportion of adolescents with T-scores in the severe range compared to other groups (Table 4).
Table 3.
In-Person Instruction and Limited Social Jetlag are associated with Lower Depressive and Anxiety symptoms
| In-Person, SJL<2hrs | In-Person, SJL≥2hrs | Other Instructional Approach, SJL<2hrs | Other Instructional Approach, SJL≥2hrs | Test Statistic | |
|---|---|---|---|---|---|
|
|
|||||
| Depressive Symptoms | (n=465) | (n=485) | (n=1813) | (n=1125) | Pearson Chi Square(9)= 47.26, Cramer’s V(3)=.06, p<.001, |
| % Within Normal Limits | 12.5 | 12.8 | 10.9 | 8.0 | |
| % Mild | 11.4 | 8.0 | 8.4 | 6.8 | |
| % Moderate | 34.6 | 35.3 | 33.8 | 29.1 | |
| % Moderate-Severe | 41.5 | 43.9 | 46.9 | 56.1 | |
| Anxiety Symptoms | (n=471) | (n=489) | (n=1838) | (n=1142) | Pearson Chi Square(9)=37.04, Cramer’s V(3)= .06, p<.001 |
| % Within Normal Limits | 18.5 | 19.4 | 17.2 | 13.2 | |
| % Mild | 15.1 | 11.9 | 10.0 | 10.3 | |
| % Moderate | 42.5 | 47.6 | 45.9 | 45.6 | |
| % Moderate-Severe | 24.0 | 21.1 | 27.0 | 30.8 | |
Figure 2. In-Person Instruction and Optimal Sleep patterns are Associated with Lower Depressive Symptom and Anxiety T-Scores.
Optimal sleep refers to having both sufficient sleep opportunity (≥9 hours for middle schoolers and ≥8 hours for high schoolers) and social jetlag <2hrs. Error bars represent 95% confidence intervals.
Table 4.
In-Person Instruction, Sufficient Sleep Opportunity, and Limited Social Jetlag are associated with Lower Depressive and Anxiety symptoms
| In-Person, Sufficient TIB, SJL<2hrs | In-Person, Insufficient TIB and/or SJL≥2hrs | Other Approach, Sufficient TIB, SJL<2hrs | Other Approach, Insufficient TIB and/or SJL≥2hrs | Test Statistic | |
|---|---|---|---|---|---|
|
|
|||||
| Depressive Symptoms | (n=237) | (n=713) | (n=1181) | (n=1757) | Pearson Chi Square(9)= 73.19, Cramer’s V(3)=.08, p<.001 |
| % Within Normal Limits | 14.8 | 11.9 | 12.2 | 8.2 | |
| % Mild | 13.5 | 8.4 | 9.2 | 6.9 | |
| % Moderate | 33.8 | 35.3 | 35.6 | 29.5 | |
| % Moderate-Severe | 38.0 | 44.3 | 43.0 | 55.4 | |
| Anxiety Symptoms | (n=239) | (n=721) | (n=1192) | (n=1788) | Pearson Chi Square(9)=51.20, Cramer’s V(3)=.07, p<.001 |
| % Within Normal Limits | 20.9 | 18.3 | 19.7 | 13.0 | |
| % Mild | 15.9 | 12.6 | 9.6 | 10.5 | |
| % Moderate | 41.4 | 46.3 | 45.0 | 46.3 | |
| % Moderate-Severe | 21.8 | 22.7 | 25.8 | 30.3 | |
Note: Sufficient TIB refers to time in bed ≥8 hrs among highschoolers, and ≥9hrs for middle schoolers.
In addition to these additive effects, we considered interaction effects of instructional approach and sleep opportunity and of instructional approach and social jetlag, on mental health. Linear regression analyses showed no significant interaction effects (p’s>.05).
DISCUSSION
In this large, nationwide study, we found elevated symptoms of depression and anxiety in adolescents during the COVID-19 pandemic, with average T-scores one standard deviation above the mean of the reference population. Further, a striking 49% and 28% of adolescents reported moderate-severe levels of depressive symptoms and anxiety, respectively. This study was performed in the Fall of 2020, and our findings provide a snapshot of adolescent mental health during a time when school systems were in flux. By surveying the diverse array of instructional formats adolescents experienced and their associated sleep patterns, this study is the first to show that the combination of in-person learning, sufficient sleep, and consistent sleep schedules are together associated with lower levels of anxiety and depressive symptoms.
We found that adolescents attending classes in person all schooldays (In-Person group) reported significantly fewer depressive symptoms and anxiety than those with only virtual forms of instruction. This finding is consistent with findings from a recent study comparing depressive symptoms between students attending in-person versus virtual learning2, which suggests that connection and engagement with the in-school environment appears to support adolescent mental health. We did not assess social isolation or feelings of social connectedness in this study, but a previous study found that greater feelings of school connectedness, including perceived teacher support, school engagement, a sense of belonging, and positive peer relationships, explained part of this association between instructional approach and depressive symptoms. Similarly, another survey study showed students receiving only virtual education reported missing face-to-face interactions, socialization, and engagement in extracurricular activities22. In-person schooling may also provide important access to school counselors and school-based mental health interventions23. Future studies are needed to replicate our findings and additionally evaluate whether social connectedness and lower social isolation, and factors unique to the school environment might contribute to the association between in-person learning and lower anxiety and depressive symptoms.
Self-reported history of depression and anxiety in the current study were each associated with higher current symptom scores. Relative to other combined instructional approach groups, a smaller proportion of adolescents in the In-Person group endorsed a history of depression or anxiety; in contrast, the largest proportion was observed in the Online Asynchronous group. Adolescents with mental health vulnerabilities, when given the option, may have been more likely to select virtual instructional formats without scheduled instruction. We did not collect data on whether participants were assigned or self-selected their instructional formats, nor did we collect data regarding when their endorsed depression or anxiety occurred. We found, however, that the association between instructional approach and mental health outcomes persisted even after controlling for this self-reported anxiety or depression history, suggesting an independent effect of instructional approach.
While the In-Person group had the most positive mental health outcomes, this group also reported on average significantly less sleep opportunity across the week and more social jetlag compared to other groups. In our previous study on the parent NESTED sample from which the current subsample is derived, we found that adolescents experienced the shortest sleep opportunity before in person school days with early school start times, which suggested that both school start time and instructional approach impacts adolescents’ sleep schedules4. Consistent with this initial report, we found in this subsample that earlier school start time among adolescents attending 5 days in-person or 5 days online synchronous was associated with, on average, shorter sleep opportunity. Notably, we found that adolescents attending school 5 days in-person experienced the largest shift in sleep timing when transitioning from scheduled to free-days, while those who had online mixed instruction (both asynchronous and synchronous) reported the least. The adolescent delay in circadian phase and are natural predisposition to sleep and wake at later times is well documented10,24,25. Yet, early school start times enforce early wake times while their sleep regulatory systems lead to difficulty falling asleep10,11. In the traditional school format, adolescents are often required to wake earlier on school days in order to get to school on time, and they sleep later and longer on weekends when unconstrained by school schedules in an attempt to “catch up” on lost sleep 10,11. Thus, students with more opportunity to choose their school start time, as with online asynchronous learning, are likely able to sleep more and maintain more consistent sleep schedules aligned with their intrinsic preference (less social jetlag). Despite this apparent benefit for sleep, asynchronous, online learning was also linked to the highest average levels of depressive symptoms and anxiety, and thus online instruction alone cannot be recommended.
In this study, the lowest levels of depressive symptoms and anxiety were found in adolescents who had the combination of in-person instruction, sufficient sleep opportunity, and consistent sleep/wake times. Our previous report highlighted that early school start times for either in person or online-synchronous instruction were associated with the shortest sleep opportunity4, and the current study highlights the possible benefits associated with preserving both in-person learning and sleep. The combination of in-person learning and healthy sleep patterns appear integral for adolescent wellbeing, as echoed in our other finding that sleep disturbances partially explain the influence of instructional format on adolescents’ school and learning success26. The robust literature on school start times has clearly shown early start times are the primary factor that restricts an adolescent’s sleep opportunity27,28. Thus, healthy secondary school start times are essential to achieve the combination of in-person learning and sufficient sleep opportunity. While we found that school start time alone was not correlated with depressive symptoms or anxiety, delaying school start times may support adolescent mental health by increasing sleep opportunity and decreasing social jetlag while preserving benefits of in-person attendance.
The current study has several limitations. This study was cross-sectional in design with data collected during a 6-week window and relied on adolescents’ subjective reports of their mental health and sleep schedules for each instructional approach. Our study was conducted during a time when school closings and re-openings were ongoing. While this was an opportunity for us to consider mental health and sleep patterns in the context of education changes, abrupt shifts in school schedules prior to the survey may have affected some participants’ sleep and mental health. Since we did not collect data pre-pandemic, we are unable to assess within-person changes in mental health, and with the cross-sectional design of this study, we are unable to conclude causal effects. For instance, adolescents and their families may have chosen specific instructional approaches based on pre-existing mental health or sleep concerns not captured in this study. We also did not collect information on general health status and were unable to examine whether pre-existing health concerns may have contributed to the observed associations in this study. In addition, the use of Facebook and Instagram for recruitment may have limited the generalizability of our findings, given possible differences between users and non-users of these platforms. However, at the time of data collection more than 2/3 of teens reported using Facebook or Instagram at least once per week29. Further, our purposeful sampling technique helped facilitate representativeness of the sample regarding key demographic variables such as age, gender, and race/ethnicity. Finally, future studies are needed to consider how the associations between instructional approach, sleep and mental health may differ by adolescent demographics not examined here, including family socioeconomic status and urban vs rural locations. Overall, the present study provides important insight derived from a time when different instructional approaches were implemented across the nation and highlights the possible link between both in-person instruction and healthy sleep to lower levels of anxiety and depressive symptoms in adolescents.
Supplementary Material
Supplementary Figure 1. In-Person Instruction with Sufficient Sleep or Limited Social Jetlag are Associated with Lower Depressive Symptom and Anxiety T-Scores. Sufficient sleep opportunity refers to ≥9 hours for middle schoolers and ≥8 hours for high schoolers, and limited social jetlag, <2hrs. Error bars represent 95% confidence intervals.
ACKNOWLEDGMENTS
Funding Sources:
The NESTED study did not receive any external funding. Support for MAC and DB was received from the National Institute of General Medical Sciences (P20GM139743). Other individual investigators were also supported by the National Institutes of Health during the time of the study and/or preparation of this article (T32MH019927 to PW, K23HL150299 to SMH, K01MH109854 to JMS, and K01HL135452 and R01HL152453 to AS), as well as the Jacobs Foundation and the Rhode Island Foundation (awards for JMS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Declarations of interest: none
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Contributor Information
Patricia Wong, Alpert Medical School of Brown University/E. P. Bradley Hospital, Sleep for Science Research Lab
Lisa J. Meltzer, National Jewish Health
David Barker, Alpert Medical School of Brown University
Sarah M. Honaker, Indiana University School of Medicine
Judith A. Owens, Harvard Medical School.
Jared M. Saletin, Alpert Medical School of Brown University/E. P. Bradley Hospital, Sleep for Science Research Lab
Azizi Seixas, The University of Miami Miller School of Medicine, Department of Psychiatry and Behavioral Sciences
Kyla L. Wahlstrom, University of Minnesota
Amy R. Wolfson, Loyola University Maryland
Mary A. Carskadon, Alpert Medical School of Brown University/E. P. Bradley Hospital, Sleep for Science Research Lab
REFERENCES
- 1.Protecting Youth Mental Health. US Surgeon General’s Advisory. 2021. [PubMed] [Google Scholar]
- 2.Hertz MF, Kilmer G, Verlenden J, et al. Adolescent mental health, connectedness, and mode of school instruction during COVID-19. Journal of Adolescent Health. 2022;70(1):57–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Duckworth AL, Kautz T, Defnet A, et al. Students attending school remotely suffer socially, emotionally, and academically. Educational Researcher. 2021;50(7):479–482. [Google Scholar]
- 4.Meltzer LJ, Saletin JM, Honaker SM, et al. COVID-19 instructional approaches (in-person, online, hybrid), school start times, and sleep in over 5,000 US adolescents. Sleep. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Short MA, Booth SA, Omar O, Ostlundh L, Arora T. The relationship between sleep duration and mood in adolescents: a systematic review and meta-analysis. Sleep medicine reviews. 2020;52:101311. [DOI] [PubMed] [Google Scholar]
- 6.Wittmann M, Dinich J, Merrow M, Roenneberg T. Social jetlag: misalignment of biological and social time. Chronobiol Int. 2006;23(1–2):497–509. [DOI] [PubMed] [Google Scholar]
- 7.Mathew GM, Hale L, Chang A-M. Sex moderates relationships among school night sleep duration, social jetlag, and depressive symptoms in adolescents. Journal of biological rhythms. 2019;34(2):205–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mathew GM, Li X, Hale L, Chang A-M. Sleep duration and social jetlag are independently associated with anxious symptoms in adolescents. Chronobiology international. 2019;36(4):461–469. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Roenneberg T, Allebrandt KV, Merrow M, Vetter C. Social jetlag and obesity. Curr Biol. 2012;22(10):939–943. [DOI] [PubMed] [Google Scholar]
- 10.Carskadon MA. Sleep in adolescents: the perfect storm. Pediatric Clinics of North America. 2011;58(3):637. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Crowley SJ, Wolfson AR, Tarokh L, Carskadon MA. An update on adolescent sleep: New evidence informing the perfect storm model. Journal of adolescence. 2018;67:55–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Meltzer LJ, Williamson AA, Mindell JA. Pediatric sleep health: It matters, and so does how we define it. Sleep Medicine Reviews. 2021:101425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Genta FD, Rodrigues Neto GB, Sunfeld JPV, et al. COVID-19 pandemic impact on sleep habits, chronotype, and health-related quality of life among high school students: a longitudinal study. Journal of Clinical Sleep Medicine. 2021;17(7):1371–1377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.United States Census Bureau. School Enrollment in the United States: 2018. 2020. [Google Scholar]
- 15.Carle AC, Bevans KB, Tucker CA, Forrest CB. Using nationally representative percentiles to interpret PROMIS pediatric measures. Quality of Life Research. 2021;30(4):997–1004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Paruthi S, Brooks LJ, D’Ambrosio C, et al. Consensus statement of the American Academy of Sleep Medicine on the recommended amount of sleep for healthy children: methodology and discussion. Journal of clinical sleep medicine. 2016;12(11):1549–1561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Levandovski R, Dantas G, Fernandes LC, et al. Depression scores associate with chronotype and social jetlag in a rural population. Chronobiology international. 2011;28(9):771–778. [DOI] [PubMed] [Google Scholar]
- 18.Rutters F, Lemmens SG, Adam TC, et al. Is social jetlag associated with an adverse endocrine, behavioral, and cardiovascular risk profile? Journal of biological rhythms. 2014:0748730414550199. [DOI] [PubMed] [Google Scholar]
- 19.Beauvalet JC, Quiles CL, de Oliveira MAB, Ilgenfritz CAV, Hidalgo MPL, Tonon AC. Social jetlag in health and behavioral research: a systematic review. ChronoPhysiology and Therapy. 2017;7:19–31. [Google Scholar]
- 20.Islam Z, Hu H, Akter S, et al. Social jetlag is associated with an increased likelihood of having depressive symptoms among the Japanese working population: the Furukawa Nutrition and Health Study. Sleep. 2020;43(1):zsz204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kim H-Y. Statistical notes for clinical researchers: Chi-squared test and Fisher’s exact test. Restorative dentistry & endodontics. 2017;42(2):152–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Sofianidis A, Meletiou-Mavrotheris M, Konstantinou P, Stylianidou N, Katzis K. Let students talk about emergency remote teaching experience: Secondary students’ perceptions on their experience during the COVID-19 pandemic. Education Sciences. 2021;11(6):268. [Google Scholar]
- 23.2017–18 School Survey on Crime and Safety (SSOCS). National Center for Education Statistics, U.S. Department of Education;2018. [Google Scholar]
- 24.Carskadon MA, Vieira C, Acebo C. Association between puberty and delayed phase preference. Sleep. 1993;16(3):258–262. [DOI] [PubMed] [Google Scholar]
- 25.Andrade MM, Benedito-Silva AA, Domenice S, Arnhold IJ, Menna-Barreto L. Sleep characteristics of adolescents: a longitudinal study. Journal of Adolescent Health. 1993;14(5):401–406. [DOI] [PubMed] [Google Scholar]
- 26.Saletin J, Wolfson A, Wahlstrom K, et al. School went online during the COVID-19 pandemic: influences on sleep and academic readiness in 4081 adolescents in the NESTED study. in submission. [Google Scholar]
- 27.Wheaton AG, Chapman DP, Croft JB. School start times, sleep, behavioral, health, and academic outcomes: a review of the literature. Journal of School Health. 2016;86(5):363–381. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bowers JM, Moyer A. Effects of school start time on students’ sleep duration, daytime sleepiness, and attendance: a meta-analysis. Sleep health. 2017;3(6):423–431. [DOI] [PubMed] [Google Scholar]
- 29.The Common Sense Census: Media Use by Tweens and Teens. 2021. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Figure 1. In-Person Instruction with Sufficient Sleep or Limited Social Jetlag are Associated with Lower Depressive Symptom and Anxiety T-Scores. Sufficient sleep opportunity refers to ≥9 hours for middle schoolers and ≥8 hours for high schoolers, and limited social jetlag, <2hrs. Error bars represent 95% confidence intervals.