Nearly one-third of individuals in America aged 13–18 years are affected by an anxiety disorder [1]. Adolescent anxiety is increasing, with a 10% rise in anxiety disorders reported between 2012 and 2018 [2]. The coronavirus disease 2019 (COVID-19) pandemic has influenced anxiety levels, the learning environment, and academic performance [3,4]. However, few studies have examined how anxiety symptoms vary among adolescents according to teaching modality.

Sleep problems are also prevalent during adolescence, with common issues including insomnia, delayed sleep phase syndrome, excessive daytime sleepiness, social jetlag, and misalignment between biological and socially imposed sleep schedules [5,6]. Anxiety contributes to sleep disturbances through mechanisms such as heightened physiological arousal and difficulty initiating or maintaining sleep [7,8]. Adolescents with anxiety disorders frequently experience reduced sleep quality, prolonged sleep latency, and increased frequency of nighttime awakenings [9].

Furthermore, previous studies have demonstrated an association between anxiety and sleep-related issues. However, data on adolescents from rural and low-income backgrounds remains limited, often facing greater barriers to mental health services, reduced access to healthcare infrastructure, and elevated levels of unmet needs compared to their urban counterparts [10,11]. These structural challenges may influence both the manifestation of anxiety and sleep disturbances and the capacity to address them during times of crisis, such as the COVID-19 pandemic.

This study was performed in Barrow County, a semi-rural county in Georgia, where 10% of the population lives below the poverty line [12]. The study aimed to elucidate the association between anxiety and sleep behaviors among adolescents during the COVID-19 pandemic using various teaching modalities (in-person, remote, and hybrid).

Three cross-sectional online surveys assessing general health and sleep behaviors were administered to ninth-grade students at two semi-rural high schools in Barrow County, Georgia. Surveys were conducted during different phases of the pandemic, corresponding to varying teaching modalities: in-person instruction pre-pandemic in spring 2020, remote teaching during the fall 2020 semester, and hybrid (in-person and remote) teaching during the spring 2022 semester. Students provided assent acknowledging that their parents could choose to withdraw their children from school. The online survey (Qualtrics, version 2020) was shared with 1,133 students, 54% of whom consented. After excluding non-respondents and those who did not answer the Generalized Anxiety Disorder Criteria for Adolescents (GAD-C) questions, the completed surveys were analyzed at each time point (475/500, 353/500, and 284/500). The Ethics Review Board of Emory University approved the study protocol (IRB00111438). Sleep behaviors were assessed across four domains: duration, social jetlag, daytime sleepiness, and frequency of sleep disturbances. Sleep duration was determined by using self-reported bed and wake times, while social jetlag was calculated as the difference between the average weekend and weekday sleep midpoints [13]. Daytime sleepiness was measured using the Cleveland Adolescent Sleepiness Questionnaire (CASQ) and a self-report question: “Please measure your sleepiness on a typical school day, where 0=not sleepy at all and 10=highest.” Sleep disturbance was assessed using a validated questionnaire that evaluated how frequently students experienced difficulty falling asleep, staying asleep, or sleeping too much. Responses were recorded on a 5-point Likert scale, where 1 indicated “not at all,” 2 “rarely,” 3 “sometimes,” 4 “often,” and 5 “every day.” For analysis, responses were grouped into three categories: not really/weekly (1–2), several days (3), and nearly every day (4–5) [14]. Students self-reported whether they had received a clinical diagnosis of anxiety, responding as yes or no. Anxiety symptoms were scored based on participants’ responses to the GAD-C, with scores averaged across the number of items. Anxiety severity was categorized as 1, indicating “none,” and 2+, indicating “mild or greater.”

Race/ethnicity was self-reported and categorized as White, Black, Hispanic, or other ethnicities. Sex was classified as male or female. Socioeconomic status was assessed using a proxy for eligibility for free or reduced lunch. Descriptive statistics are reported as percentages. Unadjusted and adjusted prevalence ratios (for demographic variables) were calculated for the association between anxiety diagnosis, GAD-C score ≥2, and sleep disturbance using logistic regression. Analyses were conducted using SAS, Version 9.4 (SAS Institute Inc.).

The first pre-pandemic survey, conducted in spring 2020, when teaching was entirely in-person, was completed by 475 ninth-grade students; 52.8% were female, and 48.4% qualified for free and reduced-price lunch (FRL). Among the respondents, 23.4% self-reported having a diagnosis of medical anxiety (Table 1). The second survey, administered during the peak pandemic remote learning phase, was completed by 353 students, whereas the third survey, conducted during the transition hybrid phase, was completed by 284 students. The proportion of female students increased across the phases, reaching 60.6% in the second survey and 66.2% in the third survey. The racial distribution also shifted, with the proportion of white students declining to 42.6% (Table 1). Additionally, chi-square analyses revealed statistically significant differences in several demographic variables across the three pandemic phases, including sex distribution (χ²=13.85, p<0.001), racial composition (χ²=47.28, p<0.001), FRL eligibility (χ²=8.85, p=0.012), and anxiety diagnosis (χ²=38.26, p<0.001). These differences are important to consider when interpreting anxiety and sleep trends, as underlying population characteristics may influence both mental health status and self-reported behaviors. Additionally, a substantial proportion of students scored two or higher on the GAD-C, indicating at least mild anxiety symptoms, with 36.4%, 45.3%, and 34.5% meeting this threshold at each respective time point. Similarly, the proportion of students with a clinical anxiety diagnosis was 23.4% during the pre-pandemic phase, 38.8% at the peak of the pandemic, and 22.2% during the transition phase. In addition to analyzing the average sleep disturbance scores, we examined the frequency distribution of reported sleep disturbances across different time points (Table 1). In the pre-pandemic group, 28.0% of the students reported experiencing sleep disturbances nearly every day. This proportion increased slightly to 30.9% during the peak pandemic remote-learning period and further to 33.5% during the transition phase. However, the percentage of students reporting mild or infrequent sleep issues (not really/weekly) declined from 43.4% pre-pandemic to 34.2% by the transition phase. These patterns underscore a trend toward increasingly frequent sleep disruptions among adolescents as the pandemic progressed. Students experiencing sleep disturbances “every day” were 5.9 times more likely to have a diagnosed anxiety disorder compared to those reporting minimal sleep disturbances (not really) after adjusting for race, sex, and socioeconomic status (adjusted prevalence ratio: pre-pandemic [PR_adj,pre]=5.9; 95% confidence interval [CI], 3.2–10.5) (Table 2). This association was consistent over time (adjusted prevalence ratio during the peak pandemic [PR_adj,peak]= 3.0; 95% CI, 1.8–5.1); adjusted prevalence ratio during the transition phase [PR_adj,trans]=3.2; 95% CI, 1.5–7.1) and for GAD-C scores above two (PR_{adj, pre}=14.5, 95% CI, 8.1–26.0; PR_adj,peak=13.8, 95% CI, 7.5–25.4; PR_adj,trans=19.1, 95% CI, 8.7–42.2). The mean sleep duration, social jetlag, and daytime sleepiness remained unchanged throughout the study period.

Although mean GAD-C scores were relatively stable across all three time points, 1.85 (pre-pandemic), 2.01 (peak pandemic), and 1.84 (transition), a substantial proportion of students reported clinically relevant symptoms. Specifically, 36.4%, 45.3%, and 34.5% of students scored two or higher on the GAD-C (indicating at least mild anxiety) at each time point, respectively. Additionally, 23.4% of the students in the pre-pandemic group reported a clinical diagnosis of anxiety, a rate exceeding the national average for this age group. National data from the Centers for Disease Control and Prevention indicate that 16% of adolescents in the United States aged 12–17 have a diagnosed anxiety disorder [15]. This percentage increased to 38.8% during the peak pandemic remote-learning phase before declining to 22.2% during the hybrid transition phase (Table 1). Similarly, the proportion of students with a GAD-C score ≥2—indicating mild or greater anxiety symptoms—was 36.4% (pre-pandemic), peaked at 45.3% (peak pandemic), and declined to 34.5% (transition). These findings indicate that both clinically diagnosed and self-reported anxiety symptoms peaked during the remote learning phase, potentially reflecting the psychological impact of pandemic-related disruptions and social isolation. Although mean GAD-C scores remained relatively consistent across time points, the proportion of students exhibiting elevated symptoms was highest during the remote learning phase, suggesting that pandemic-related challenges and instructional modalities may have contributed to increased anxiety symptom severity.

In addition, Table 2 presents the association between both clinically diagnosed and self-reported anxiety symptoms and the frequency of sleep disturbances across different phases of the pandemic. Notably, elevated anxiety levels were strongly associated with sleep disturbances. Students who reported experiencing sleep disturbances “every day” were 5.9 times more likely to have a clinical anxiety diagnosis compared to those reporting minimal or no sleep disturbance, even after adjusting for race, sex, and socioeconomic status. This reinforces prior findings that anxiety and poor sleep frequently co-occur, emphasizing the importance of integrated approaches that address both concerns in efforts to promote adolescent mental health.

These findings address a critical gap in the literature by examining the association between sleep disturbances and anxiety among low-income and semi-rural adolescents during the COVID-19 pandemic. Several prior studies have reported that adolescent anxiety worsened or remained elevated throughout the course of the pandemic [16-20]. During the remote learning phase, anxiety symptoms surged among students, with 38.8% reporting a clinical diagnosis and 45.3% scoring ≥2 on the GAD-C. In comparison, the pre-pandemic figures were 23.4% and 36.4%, respectively, while during the transition phase, the corresponding rates were 22.2% and 34.5%. This increase aligns with findings that adolescents experience heightened stress and disrupted routines during the pandemic, particularly in rural areas where limited access to support services, socioeconomic challenges, and increased social isolation exacerbate these effects [21]. Anxiety levels were highest during the remote learning phase, potentially reflecting the abrupt loss of in-person social support, academic uncertainty, and disruption of daily routines experienced during the early stages of the pandemic. These factors have been linked with acute psychological distress among adolescents during periods of quarantine and remote learning [22].

Sleep disturbances also escalated, with 30.9% of students reporting symptoms “nearly every/every day” at the peak of the pandemic, compared to 28.0% pre-pandemic and 33.5% during the transition phase. This trend suggests that sleep problems may have worsened over time, owing to the cumulative effects of prolonged stress, irregular daily schedules, and challenges in readjusting to new hybrid learning formats. Although remote learning initially provided greater flexibility in wake and sleep times, the transition back to hybrid instruction may have disrupted these adjusted routines, resulting in increased irregularities in sleep patterns. Additionally, although anxiety levels declined during the transition period, sleep problems may persist due to the lingering physiological effects of chronic anxiety, as poor sleep is associated with mechanisms such as hyperarousal and dysregulation of the hypothalamic-pituitary-adrenal axis [23]. Even though other studies have reported increased sleep duration and social jetlag during the pandemic [24], our sample likely maintained more consistent routines, which may have mitigated changes in sleep timing despite elevated stress levels.

Notably, the increased proportion of female participants during the peak and transition phases may partially account for the elevated rates of self-reported anxiety, as adolescent females are more likely to report internalizing symptoms, with anxiety generally being more prevalent among females. Even though such demographic shifts do not invalidate the observed associations, they highlight the importance of considering sample composition and potential response bias in voluntary survey studies.

Remote learning during the pandemic may have contributed to better regulation of sleep schedules and preservation of sleep duration. Freed from the demands of early wake-up times and commuting, students may have experienced greater flexibility in aligning their sleep with natural circadian rhythms, reducing discrepancies between weekday and weekend sleep patterns (i.e., social jet lag). However, beyond the increasing frequency of sleep disturbances, other sleep-related indicators, including average sleep duration, social jet lag, and daytime sleepiness, did not exhibit significant differences across the three study periods. This contrasts with findings from several other studies that reported longer sleep duration and greater social jetlag during remote learning phases in adolescents [24]. One possible explanation is that students in our semi-rural population may have experienced fewer fluctuations in external demands or daily routines, despite changes in the learning formats. For instance, limited extracurricular obligations, stable household routines, or more structured remote learning schedules may have helped preserve regular sleep. Alternatively, the measures employed may have been more effective in capturing overt forms of sleep disruption (such as difficulty falling or staying asleep) while being less sensitive to subtle changes in sleep quality or consistency. These differences underscore the importance of contextual factors, including community settings, access to resources, and baseline routines, in shaping adolescent sleep behaviors during periods of disruption.

Our study has several strengths, including a semi-rural, low-income county population and the ability to examine data across three distinct time points, each representing a different learning modality.

A few limitations of this study include reliance on self-reported data for anxiety and sleep behaviors. In addition, variations in response rates by race, sex, and socioeconomic status during the three phases potentially influenced the patterns of anxiety and sleep behaviors. Although non-response is an inherent limitation of survey studies, the survey was not explicitly framed as a study on anxiety or sleep, but rather as a broader assessment of student health, which may have mitigated some response bias related to mental health stigma or interest.

In conclusion, changes in learning modalities during the COVID-19 pandemic significantly impacted adolescent anxiety. Self-reported and clinically diagnosed anxiety were associated with poor sleep behavior across all phases of the pandemic among adolescents from a semi-rural, low-income population. Understanding the trends between sleep and anxiety in diverse populations can inform school district policies regarding interventions for the mental health support of adolescents.