Abstract
While effects of COVID-19 on physical health are the subject of much research, it is also important to understand risk factors for negative psychosocial and behavioral outcomes. Undergraduates (N = 490, Mage = 20.4) completed measures regarding prior trauma, COVID-19 infection indicators, stressors and trauma, coping, loneliness, social support, sleep behaviors, and negative emotionality. Results demonstrate that pre-existing trauma, COVID-19 stressors, loneliness, and avoidant coping exhibit independent and synergistic associations with poor sleep quality and negative emotionality. Associations between both COVID-19 stressors and avoidant coping with sleep quality were the strongest among those with higher levels of cumulative trauma. Avoidant coping was most strongly associated with higher levels of negative emotions among those reporting COVID infection indicators. Findings suggest a comprehensive set of specific pandemic and general life factors associated with worse outcomes, contributing to the development of a conceptual model of pandemic behavioral and emotional risk for emerging adults.
In December of 2019, China reported an outbreak of the coronavirus disease (COVID-19), a highly contagious virus that mimics pneumonia, in the city of Wuhun (Nishiura et al., 2020). The disease began to spread throughout China and worldwide at an alarming rate, prompting the World Health Organization to declare an international emergency (Mahase, 2020). Countries imposed travel restrictions and social distancing, while businesses and schools closed. The pandemic upended economies worldwide and increased uncertainty and fear (Wang et al., 2020). As the pandemic continues to affect daily life, the emotional, physical, and psychosocial effects of this stressor must be examined (e.g., Holmes et al., 2020; Pfefferbaum & North, 2020), particularly for emerging adults who may be at increased risk (Germani et al., 2020). Given the potentially severe and life threatening impacts of COVID-19, it is also important to address the impacts of this stressor on health behaviors that may already be poor in emerging adults, such as sleep.
While research on the impact of COVID-19 continues to develop, documented effects of previous disease outbreaks and disasters on individual, community, national, and international levels are informative. For example, during the 1995 Ebola outbreak, individuals reported feelings of helplessness and concerns about becoming sick or dying (Hall & Chapman, 2008) that were exacerbated amongst school and business closures (Van Bortel, 2016). During the 2003 SARS epidemic, participants, especially women, living in the affected areas reported post-traumatic stress symptoms (Lau et al., 2005). Similarly, during a 2016 major flood in China, trauma exposure was associated with negative emotions and sleep problems (Zhen et al., 2018).
Initial reports indicate that the COVID-19 pandemic is likewise affecting individuals. In a study examining participants from China 2 weeks after the COVID-19 outbreak, a third of respondents reported moderate to severe anxiety symptoms and half reported moderate to severe depressive symptoms (Wang et al., 2020). Another study examining adults from the United Kingdom reported increased depression, stress, and negative feelings related to social isolation and social distancing. In fact, psychological and social concerns were higher than physical health concerns (Holmes et al., 2020). Researchers examining medical staff treating patients with COVID-19 and individuals in isolation in China found that participants who reported higher levels of anxiety also reported more sleep difficulties (Cao et al., 2020; Liu et al., 2020; Xiao et al., 2020a, 2020b). Women and students were more likely to report higher levels of stress, anxiety, depression, and sleep disturbances (Liu et al., 2020; Wang et al., 2020). This parallels earlier research examining the psychological impact of a quarantine in Australia due to the equine influenza, wherein women and younger individuals were more negatively affected (Taylor et al., 2008). Finally, a recent review regarding the COVID-19 pandemic demonstrated that depression, anxiety, stress, and sleep difficulties were often reported (Rajkumar, 2020). Given the severity of the outbreak, the continued uncertainty regarding the spread of the virus, and the length or possible reinstatement of stay at home orders, it is important to continue to examine the emotional, psychosocial, and behavioral impacts of this pandemic. Such impacts tend to be underestimated and can have long-lasting effects (Ornell et al., 2020). Additionally, findings may inform future psychological interventions that target vulnerable groups (Pfefferbaum & North, 2020; Wang et al., 2020).
One vulnerable group is undergraduate students, who are in the developmental phase of emerging adulthood, particularly those who have already encountered significant stressors and/or trauma (Lucier-Greer et al., 2018). Even outside of pandemic contexts, college students, especially women, report high levels of stress and sleep difficulties (American College Health Association, 2020; Pierceall & Keim, 2007) related to academic issues, uncertainly regarding the future, financial pressures, and social pressures (Hurst et al., 2013; Lee & Dik, 2016). Given the instability associated with emerging adulthood, college students may experience anxiety about the potential effects that COVID-19 can have on their academics (Corine, 2020), future employment opportunities (Wang et al., 2020), or opportunities for social interaction (Cao et al., 2020). In early 2020, students in international programs across the globe were given the news that they needed to return to their countries of origin and complete the remainder of their semester coursework either on main campuses or online. Students from international programs in locations with severe outbreaks, such as China, experienced ostracism and discrimination upon their return. Soon thereafter, many students were told that they needed to leave their campuses within days and complete their coursework via distance learning. This disruption in education and the resulting uncertainties regarding housing, graduation, and coursework were likely quite stressful for students, who may be at increased risk for psychosocial consequences following a public health emergency (Wang et al., 2020). Further, the Southern California students studied in this research were also recently affected by a mass shooting at a local bar’s college night, which took the life of one of their peers, and destructive fires in the fall of 2018, which started the night after the mass shooting, in which education was disrupted and many students lost property and were forced to evacuate or shelter in place.
Students are likely affected to various degrees by these collective stressors (i.e., COVID-19, mass shootings, natural disasters), with most likely experiencing some degree of stress and some experiencing trauma as a result of potentially traumatic aspects of these collective stressors. For example, students who experienced life-threatening complications of COVID-19 either personally or through a loved one, were present at or lost a loved one during the college bar shooting, or felt that the mass fires seriously threatened the health of themselves or loved ones were exposed to traumatic stressors.
Such traumatic stressors differ from other stressors in that they involve experiencing imminent threat to one’s life or bodily integrity, witnessing or hearing about threats to the life or bodily integrity of a loved one, or witnessing anyone dead or seriously injured. Research indicates that exposure to traumatic stressors is associated with negative physical and psychological effects (Joseph et al., 2014; Loeb, et al., 2018; Spertus et al., 2003) and sleep issues among students (Pickett et al., 2016); further, multiple traumas (often referred to as cumulative trauma) are associated with lower psychosocial functioning and emotional distress compared to a single trauma (Delany-Brumsey et al., 2013; Follette et al., 1996; Green et al., 2000; Loeb et al., 2018; Suliman et al., 2009). This student population is especially at risk for cumulative trauma considering the fact that, by the time emerging adults enter college, many have already been exposed to traumas (Merrick et al., 2018). These early traumas may, through biological embedding in the brain, impact emotions and sleep behaviors throughout adulthood (Finch & Crimmins, 2004) as well as impact emotional and behavioral responses to pandemics.
Non-traumatic stressors that are likely heightened during this pandemic context (i.e., financial stress, discrimination due to ethnic background, low daily living resources, general perceived stress) are also associated with impaired sleep in college students (Furman et al., 2018), negative emotions in college-aged emerging adults (Joseph et al., 2020), and impaired sleep and negative emotion in the general adult population (Joseph et al., 2014; Lewis et al., 2013).
Not only is this vulnerable student population likely experiencing a larger stress and cumulative trauma burden than usual, but this burden might be amplified by reduced opportunities to connect with others or enact preferred coping strategies. Negative coping strategies (Lazarus & Folkman, 1984), lack of social support (Cohen & Wills, 1985), and loneliness (Russell, 1996) increase risk for impaired sleep in students (Furman et al., 2018) and negative emotions. Social support is a measure of individuals’ relationships and social interactions and can include support from family, friends, peers, colleagues, and medical professionals (Brugha, 1990). Limited social support is associated with worsened psychological health and poorer sleep quality (Adamczyk & Segrin, 2015; Kent de Grey et al., 2018). Social support may serve as a buffer against the negative stressors associated with COVID-19; a study of medical staff treating patients with COVID-19 in China found that medical staff with adequate social support reported higher quality of sleep and lower levels of anxiety and stress compared to those with lower social support (Xiao et al., 2020a). Another recent study found that undergraduate students in China with higher social support reported lower levels of anxiety than students with lower social support (Cao et al., 2020). Overall, social support is also linked to an increase in help-seeking behaviors, a reduction in stress, improved sleep quality, and likely plays an essential role during public health emergencies such as COVID-19 (Cao et al., 2020; Kent De Grey et al., 2018).
While loneliness is related to social support, experts also report important distinctions between these two variables. Social support is considered to be a more objective measure of relationships and social interactions, while loneliness is considered to be a measure of distress related to perceptions of social isolation (Cacioppo et al., 2015; Hawkley & Cacioppo, 2010). Loneliness has been linked to an increase in mortality, mental-health issues, and cardiovascular problems (Leigh-Hunt et al., 2017), and individuals with two of more lifetime traumatic events report more loneliness (Palgi et al., 2012). Researchers hypothesize that loneliness may exacerbate negative psychological and physical symptoms related to trauma and may also inhibit recovery or growth from traumatic events (Zeligman et al., 2017). Additionally, loneliness has been linked to levels rates of negative emotion (Hawkley & Cacioppo, 2010) and worse sleep quality (Masi et al., 2011; Wakefield et al., 2020). Individuals in China who reported lower levels of loneliness also reported better sleep quality during the COVID-19 outbreak (Xiao et al., 2020b).
How one responds to trauma, or the particular coping strategies that one utilizes, is another important predictor of the severity of negative psychological outcomes. Action-oriented coping strategies, such as problem solving, planning, and support seeking, have been associated with increased well-being compared to avoidant coping strategies, such as denial, behavioral disengagement, substance use, self-distraction, and self-blame (Lo Buono et al., 2017; Meyer, 2001). One review of the literature examining coping strategies during infectious disease outbreaks found that individuals engage in many action-oriented coping strategies such as attempting to solve or prevent problems (e.g., cleaning, buying supplies, seeking alternative medical treatments); however avoidant coping strategies such as distraction and denial are also prevalent (Chew et al., 2020). A recent study examining the impact of COVID-19 on participants with a disability or chronic condition found that the most common strategies used to cope with the disease outbreak were acceptance or distraction (Umucu & Lee, 2020). Additionally, they found that participants that reported higher levels of perceived stress were more likely to practice avoidant coping (Umucu & Lee, 2020), which has been associated with worse sleep outcomes in students (Furman et al., 2018). There is some evidence that avoidant coping may increase the association between various forms of stress and physical and psychological impairment (e.g., Blalock & Joiner, 2000; Chao, 2011), but this has not been examined in the context of the COVID-19 pandemic.
This research aims to examine the relationships and interactions between an extensive set of risk factors, including exposure to multiple traumas and stressors, avoidant coping, and limited social resources, and sleep and negative emotions in a sample of undergraduate students during the first peak of COVID-19 in the United States. Our hypotheses are as follows:
Method
Participants
Participants were 490 undergraduate students (74% female, Mage = 20.4, SD = 1.5). More than half of the participants were White (56%) and 9% were international students. Approximately 42% were seniors whereas 31%, 20% and 7% were juniors, sophomores, and freshmen, respectively. Approximately half (52.4%) were from a household with income at least $100,000. Approximately 11% of the participants were veterans. See Table 1 for demographic information.
Select Sample Characteristics.
Note. N = 490.
Procedure
This research was approved by the Institutional Review Board of Pepperdine University, and data was collected in May, 2020. All undergraduate students who were enrolled in the Spring 2020 semester (N = 3,295) received an email inviting them to complete a survey measuring their beliefs and experiences related to the COVID-19 pandemic. Those who were interested in participating clicked on a link to a Qualtrics survey, where they first provided their consent to participate. Participants were allowed to discontinue the survey at any time and to skip questions they were uncomfortable with answering. Upon completion of the survey, participants were given the opportunity to receive a $5 gift card to Amazon and be entered in a drawing to receive one of five $50 gift cards to Amazon. Participants were then debriefed, provided with information about local and national resources for dealing with stress and trauma, and thanked for their participation. Once we reached our goal of sampling approximately 15% (490 participants equated to 14.8%) of the general student population, data collection was discontinued.
Measures
Demographic questions
Participants provided their year in school (freshman, sophomore, junior, senior), household income, and COVID-19-related pre-existing health-conditions (asthma, cancer, diabetes, high blood pressure, other immuno-compromising conditions), among other demographic variables not relevant to the current paper.
Cumulative trauma
Cumulative trauma was assessed as the number of types of traumas to which participants were exposed, with three possible trauma types (COVID-19 trauma in the form of serious complications, trauma from the mass shooting or fire that impacted the university that these students attended the year prior to the COVID-19 pandemic, and other trauma prior to COVID-19). 1 Trauma history prior to COVID-19 was assessed using the Trauma History Questionnaire (THQ; Hooper et al., 2011) which queries whether or not participants were exposed to each of thirteen traumatic events. Those who reported at least one of the traumas on the THQ were given 1 point on the cumulative trauma variable. COVID-19 trauma was assessed using four items from the Infection History Subscale of the Epidemic–Pandemic Impacts Inventory (EPII; Grasso et al., 2020) that queried whether or not participants experienced serious COVID-19 symptomatology, hospitalization, or a loved one’s death. Those who endorsed at least one of the COVID-19 traumas were given 1 point on the cumulative trauma variable. Mass shooting / fire trauma was assessed using investigator-created questions modifying key stems or phrases from the THQ, e.g., “During the Woolsey fires, feared that your health or life was in danger” and “Present at Borderline the night of the shooting and feared you might be seriously injured” and using a yes or no response format. Those who endorsed at least one of these traumas were given one point on the cumulative trauma variable. Total scores on the cumulative trauma scale ranged from 0 to 3.
COVID-19 stressors
We used the COVID Impacts Questionnaire (Social Psychological Measurements of COVID-19; Conway et al., 2020) to assess financial and resource-related impacts of COVID-19 on participants, e.g., “The Coronavirus (COVID 19) has impacted me negatively from a financial point of view,” “It has been difficult for me to get the things I need due to the Coronavirus (COVID-19).” We added a COVID-19-related discrimination stressor question by modifying the key stem from pre-existing discrimination questions, e.g., “Discriminated against regarding COVID-19 because of ethnicity or country of origin.” The response options for each of these items were yes or no. Total scores were computed as the number of stressors endorsed and ranged from 0 to 9.
COVID-19 infection indicators
Four items of the Infection History Subscale of EPII (Grasso et al., 2020) were used to assess participant’s COVID-19 actual and possible infection history, e.g., “Tested and currently have this disease,” “Had symptoms of this disease but never tested.” Response options were yes or no. Participants that endorsed any of these items received a 1 on this dichotomous variable to indicate the presence of possible or confirmed COVID-19 symptoms; those that did not received a 0.
Coping strategies
Coping strategies were measured via Brief COPE (Carver, 1997). Participants are asked to rate the extent to which they typically use each of the strategies described to manage stressful situations since learning of the COVID-19 pandemic (i.e., “I’ve been getting emotional support from others” or “I’ve been refusing to believe that it has happened”) on a 4-point Likert scale ranging from 1 (not at all) to 4 (usually). As done in other methodological and empirical studies (e.g., Carver et al., 1989, Chao, 2011), we computed an avoidant coping subscale focused on only items describing avoidant strategies, e.g., “I’ve been turning to work or other activities to take my mind off things”. Cronbach’s α was .80 and .84 in these studies and .71 in the current sample.
Social support
Social support was measured by the Social Provisions Scale (Cutrona & Russell, 1987). The scale contains 24 items that measure six dimensions of social support (attachment, social integration, guidance, reliable alliance, opportunity for nurturance, and reassurance of worth). Participants respond to items (e.g., “there are people who like the same social activities I do,” “I do not have a feeling of closeness with anyone”) on a Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). A sum score is calculated for each dimension with scores ranging between 4 and 16; higher scores indicate stronger social support. Adequate test–retest reliability and construct validity of the scale has been demonstrated, and Cronbach’s α coefficients of the six factors range from .67 to .76 (Cutrona & Russell, 1987). Cronbach’s α in the current sample was .94.
Loneliness
Loneliness was measured via the UCLA Loneliness Scale (Version 3; Russell, 1996). This scale includes 20 items such as “How often do you feel that no one knows you well?” and “How often do you feel there are people you can turn to?” Responses range from 1 (never) to 4 (often), and all items are summed and divided by 20 to create a total score that ranges from 20 to 80 with higher scores indicating stronger perception of loneliness. Adequate reliability and validity of the scale has been reported (Russell, 1996). Cronbach’s α in the current sample was .94.
Negative emotionality
Emotionality was measured via the short-form of the Positive and Negative Affect Schedule Short Form (PANAS-SF; Thompson, 2007). The PANAS-SF provides participants with a list of 20 words: 10 measuring positive affect (e.g., enthusiastic) and 10 measuring negative affect (e.g., upset) Participants rate how strongly they have experienced each feeling over the past 7 days on a Likert-scale ranging from 1 (Very Slightly or Not at All) to 5 (Always). The scale has demonstrated acceptable reliability and validity (Thompson, 2007). Cronbach’s α in the current sample was .84. for negative emotions.
Sleep quality
Sleep quality was measured by the Pittsburgh Sleep Quality Index (PSQI), a 19-item self-report questionnaire that assesses sleep quality and disturbances over the past month (Buysse et al., 1989). The questionnaire consists of a 4-point Likert-type scale ranging from 0 (not during the past month) to 3 (three or more times a week). There are seven “component” scores (subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication, and daytime dysfunction) and one global score. Sleep quality levels are categorized by scores: 0–5 (good sleep score); 6–10 (mild sleep difficulty); 11–15 (moderate sleep difficulty); and 16–21 (severe sleep difficulty). The PSQI has previously demonstrated high internal consistency (α = .83; Buysse et al., 1989) and good content and concurrent validity (Buysse et al., 1989; Gelaye et al., 2014). Cronbach’s alpha in the current sample was .71.
Statistical Analyses
Multiple regressions were used to test the main effects hypotheses and moderated effects hypotheses separately for sleep quality and negative emotion. Each main effects regression included demographic variables (year in school, household income, COVID-19-related pre-existing health conditions) in addition to the independent variables of interest (gender, cumulative trauma, COVID-19 stressors, avoidant coping, social support, and loneliness). Each interaction was tested separately to avoid multicollinearity. Thus, moderated effects regressions included the main effects and alternated in interactions between cumulative trauma and each of the psychosocial factors (avoidant coping, lack of social support, loneliness) and COVID-19 stressors; COVID-19 stressors and each of the psychosocial factors; and COVID-19 infection indicators and each of the psychosocial factors. Specifically, for each outcome (negative emotions and sleep quality), the following 10 interactions were tested: cumulative trauma * COVID-19 stressors, cumulative trauma * avoidant coping, cumulative trauma * social support, cumulative trauma * loneliness, COVID-19 stressors * avoidant coping, COVID-19 stressors * social support, COVID-19 stressors * loneliness, COVID-19 infection indicators * avoidant coping, COVID-19 infection * social support, and COVID-19 infection * loneliness. Interactions were tested using multiplicative terms, e.g., the interaction between cumulative trauma and avoidant coping was tested by entering a multiplicative term (cumulative trauma centered around the mean * avoidant coping centered around the mean) into the main effects regression. Each significant interaction was explored using simple slopes analyses. Given the number of exploratory interactions tested, we also used R to examine whether significant interactions for each outcome remained significant when corrected for multiple testing using the Benjamini and Hochberg false discovery rate procedures (Benjamini & Hochberg, 1995).
For each regression, we excluded participants listwise if they were missing data from any of the variables included in that regression. Nine percentage of participants were missing data for the sleep regression and 8.5% were missing data for the negative emotion regression, primarily due to not finishing the survey. Those who did not complete the full survey were slightly younger (p = .02) and had slightly lower levels of social support (p = .001) than those that did, but otherwise did not differ from those that completed the entire survey, e.g., they were similar on demographic variables, stressor and trauma exposure, COVID-19 experiences, psychosocial factors, negative emotions, and sleep quality.
Results
Approximately 25% of the sample reported having a pre-existing health diagnoses associated with elevated COVID-19 complications, while 17% of the sample reported having had COVID-19 symptoms, diagnosis, or serious complications. The majority of the sample (91.6%) reported experiencing at least one trauma type assessed in our cumulative trauma measure, with 3% experiencing all three trauma types. The most commonly reported trauma type was trauma history (87.2%), with trauma from the mass shooting or fire being the second most reported (44.7%), and COVID-19 trauma being the least reported (4.8%). Please see Table 1 for additional description of the characteristics of the study sample.
Sleep Quality
Multiple regression found that the overall main effects model was significant and accounted for 29% of the variance in sleep quality, Adjusted R2, F(10,434) = 19.33, p < .001. Please see Table 2 for main effects results of the multiple regression. In partial support of H1, more cumulative trauma and more COVID-19 stressors were associated with worse sleep quality (b = .84, p = .001 and b = .24, p = .01, respectively) 2 whereas COVID-19 infection indicators was not, p = .29. In partial support of H2, we found that avoidant coping 3 and loneliness were associated with worse sleep quality (b = .31, p < .001 and b = .05, p = .03, respectively) whereas social support was not, p = .18. Of note, social support and loneliness are highly correlated in this sample (r = −.68). Social support was significantly associated with better sleep quality (p < .001) before loneliness was entered into the regression but was no longer associated in the final model that included loneliness. In support of H3, female gender was associated with worse sleep quality (b = .93, p = .01).
Main Effects Multiple Regression Results: Sleep Impairment Dependent Variable.
Separate interaction analyses to further test H2 found a significant interaction between cumulative trauma and avoidant coping such that those that had higher levels of both cumulative trauma and avoidant coping had the worse sleep quality (b = 11, p = .04) and a significant interaction between cumulative trauma and COVID-19 stressors such that those with higher levels of both cumulative trauma and COVID-19 stressors had the worse sleep quality (b = .29, p = .04). Please see Figure 1 for these interactions; no other interactions were significant. Simple slopes analyses found that COVID-19 stressors were most strongly associated with impaired sleep among those with higher levels of cumulative trauma, i.e., b = −.54, p = .22 for those with no cumulative trauma, b = .20, p = .08 for those with exposure to one cumulative trauma type, and b = .40, p = .01 for those with exposure to two cumulative trauma types. Given the low number of participants (N = 15) that endorsed three cumulative trauma types, we were not powered to conduct a simple slopes analyses at that level. Simple slopes analyses also found that avoidant coping was most strongly associated with impaired sleep among those with higher levels of cumulative trauma, i.e., b = −.07, p = .71 for those with no cumulative trauma, b = .28, p < .001 for those with exposure to one cumulative trauma type, and b = .35, p < .001 for those with exposure to two cumulative trauma types. Given the low number of participants (N = 15) that endorsed three cumulative trauma types, we were not powered to conduct a simple slopes analyses at that level.
Cumulative trauma as a moderator of the associations between various psychosocial factors and sleep impairment. Note. The above figures are based on centered values of COVID-19 stressors and avoidant coping as these variables were centered around the mean before being entered into the respective interaction terms. **p < .01. ***p < .001.
It is important to note that, when significance values of interactions were corrected for multiple testing, the significance of the interaction between cumulative trauma and avoidant coping was reduced to non-significance, p = .18, and the significance of the interaction between cumulative trauma and COVID-19 stressors was reduced to non-significance, p = .18. As would be expected, all previously non-significant interactions remained non-significant. Although the corrections reduced the significance of the interactions, we chose to include the interactions in our discussion, given that significant interactions are particularly difficult to detect in non-experimental designs (e.g., McClelland & Judd, 1993) and following the precedent in other research (e.g., Rose et al., 2007) that similarly includes such interactions.
Negative Emotion
Multiple regression found that the overall main effects model was significant and accounted for 34% of the variance in negative emotion, Adjusted R2, F(10,437) = 24.23, p < .001. Please see Table 3 for main effects results of the multiple regression. In partial support of H1, increased amounts of COVID-19 stressors were associated with higher negative emotion (b = .59, p < .001) whereas cumulative trauma and COVID-19 infection indicators were not, p = .14 and .99, respectively. In partial support of H2, avoidant coping and loneliness were associated with higher negative emotion (b = .76, p <.001 and b = .10, p = .02, respectively) whereas social support was not, p = .82. Again, social support and loneliness are highly correlated in this sample (r = −.68). Social support was significantly associated with less negative emotion (p = .01) before loneliness was entered into the regression at the final step but was no longer associated once loneliness was entered. Female gender was not associated with negative emotion, b = .66, p = .33, which does not support H3.
Main Effects Multiple Regression Results: Negative Emotions Dependent Variable.
Separate interaction analyses to further test H2 found significant interactions between COVID-19 infection indicators and avoidant coping such that those that had higher levels of both COVID-19 infection indicators and avoidant coping had the highest levels of negative emotion (b = .45, p = .01). No other interactions were significant. Simple slopes analyses found that the association between avoidant coping and negative emotion was stronger among those that had experienced COVID-19 infection indicators (b = 1.11, p < .001). than those that had not (b = .69, p < .001)).
It is important to note that, when significance values of interactions were corrected for multiple testing, the significance of the interaction between COVID-19 infection indicators and avoidant coping was reduced to a very strong trend, p = .07. As would be expected, all previously non-significant interactions remained non-significant. Again, given the difficulty detecting interactions in non-experimental research (McClelland & Judd, 1993), we made the decision to include these findings in our discussion.
Discussion
The results of the current study support most of our hypotheses by demonstrating that pre-existing trauma, pandemic stressors, restricted social connections, and poor (avoidant) coping repertoires have independent and synergistic associations with sleep health and emotional experience during a pandemic for emerging adults. Results align with and extend similar research on COVID-19 implications, moving beyond simple prevalence estimates and bivariate associations by examining the incremental predictive power among a comprehensive set of risk factors as well as interactions among these variables. It is important to note that our interactions were reduced to trends or to non-significance after accounting for the corrections, therefore our results should be interpreted with caution. Existing COVID-19 work also does not comprehensively assess pre-COVID 19 trauma history, despite robust evidence that pre-existing trauma is known to increase emotional and behavioral vulnerability to new trauma and stress as well as impact coping repertoires and social connections (Loeb et al., 2018; Suliman et al., 2009). As expected, interactions suggest that historical and cumulative trauma may be vulnerability factors to the stressors of a pandemic and that avoidant coping behaviors are associated with the emotional and behavioral impacts of cumulative traumas and COVID-19 infection indicators. Given that corrections reduced the significance of these findings, we recommend that our results be interpreted with caution and recommend that future research examine these variables in a larger sample.
Avoidant coping and loneliness were the psychosocial vulnerability factors with the most robust associations with impaired sleep and negative emotions, and avoidant coping increased the associations between trauma and sleep and between COVID-19 infection indicators and negative emotions. These findings add to the previous literature regarding avoidant coping in the context of a pandemic (Chew et al., 2020; Umucu & Lee, 2020) by demonstrating that this coping style may be particularly problematic for undergraduate students with larger trauma burdens or dealing with having been exposed to a virus. Interestingly, higher COVID-19 infection indicators were associated with lower negative emotion among those exhibiting low avoidant coping. It is possible that participants with lower avoidance coping may have approached and overcome their symptoms or diagnosis in a manner which led to higher perceived control and optimism in their capabilities to cope with the pandemic and lower fear of COVID-19, all of which may be associated with lower negative emotion. Since the interaction was reduced to marginal significance after accounting for corrections, additional research including a larger sample as well as a sample with a higher prevalence of COVID-19 infection indicators may provide additional clarification regarding this finding. Nonetheless, given that avoidant coping is associated with perceived stress (Umucu & Lee, 2020) and poor sleep quality (Furman et al., 2018), it is important to consider these findings as Universities prepare to resume courses either online or in person. It is also noteworthy that loneliness was more strongly related to outcomes for this student population compared to social support. Despite the strong connection between loneliness and social support, it may be that loneliness is more strongly affected by the quarantine and social distancing restrictions imposed by federal and local governments. While the amount of social support an emerging adult possesses may be less likely to change following a pandemic, feelings of loneliness may become exacerbated, at least in the short term. Given that loneliness may inhibit recovery or growth from traumatic events (Zeligman et al., 2017) and has been linked to higher rates of negative emotions (Hawkley & Cacioppo, 2010) and lower sleep quality (Masi et al., 2011; Wakefield et al., 2020) in past research as well as the current study, it is important to also consider the effects of loneliness on the undergraduate student body.
Our hypothesis that women would report worse sleep and negative emotionality than men was partially supported; females reported worse sleep, which is consistent with past literature (Liu et al, 2020; Wang et al., 2020), however no gender differences emerged in negative emotion. The emotion measure utilized in this study may have affected our ability to detect low arousal negative emotions (e.g., sadness), which may be more salient for women. Of note, the outcome variables (sleep and negative emotion) are associated with each other. Research suggests that negative emotions and sleep have a reciprocal influence on each other as negative emotions impact sleep via cognitive, behavioral, and physiological processes and sleep is associated with worse emotion regulation (Beattie et al., 2015; Buckley & Schatzberg, 2005; Buysse, 2014; Deliens et al., 2014; Gunnar & Adam, 2012; Kahn et al., 2013; Miglis, 2017; Vandekerckhove et al., 2011). Therefore, it is conceptually logical that many of the same risk factors are associated with both outcomes. Finally it is also interesting to note that year in school was a significant predictor of both outcome variables. It is possible that students who are farther along in their University experience may be better adjusted to the college experience or more competent in dealing with stressors compared to students only beginning their college experience. We recommend further research in this area.
Implications
Understanding and reducing vulnerability factors prior to pandemics is paramount. By demonstrating that a comprehensive set of pandemic-specific and general life factors are associated with sleep health and emotional experience during a pandemic for emerging adults, the current study supports the development of a conceptual model of pandemic risk centered on these factors. The current findings should be replicated before such a model is advanced. Practically, this conceptual model and knowledge can inform prevention efforts centered on helping emerging adults maintain social health (e.g., diverse social networks and meaningful interactions), diversify coping tools, and address pre-existing trauma long before pandemics occur so that these vulnerabilities do not hinder their ability to be healthy during a pandemic. For example, researchers have long advocated for trauma screeners to be more widely used in healthcare institutions (Liu et al., 2015) in order to identify, refer, and treat vulnerable traumatized individuals given the associations between trauma and many long-term mental and physical health outcomes. Researchers are more recently also advocating for psychosocial factors like social isolation, negative emotion, and sleep to be assessed routinely in the healthcare system and included in the electronic health records (EHRs) as “psychosocial vital signs” (Matthews et al., 2016). Tracking and addressing these traumas and psychosocial issues before a pandemic places strain on resources and priorities is important.
Limitations
This sample of university students may not represent all university students given the socioeconomic and racial/ethnic composition. While we obtained our goal of sampling 15% of the student population at our institution, it can be argued that a larger sample would have been more representative of that population. However, our sample was overall a good representation of the general student community at our institution, therefore it is unlikely that a larger sample would have impacted the results. Specifically, like our sample, the student population under study is predominantly female, with White students being the largest racial/ethnic group and Black or African American students being the smallest racial/ethnic group. Regardless, it is difficult to ascertain whether sampled participants differed on the variables we measured (for example, trauma) than those who were not sampled. The correlational nature of the study also precludes causal conclusions. In addition, methodology could have been improved by ensuring that the trauma history questionnaire assessed traumas prior to the age of 18 and traumas at or after the age of 18 separately in order to better capture the concept of Adverse Childhood Experiences (ACEs). Overall, our study relied on self-report for all variables and retrospective report of early traumatic stressors, and so may be limited by self-report and memory biases. experiences. Further, negative emotions could be assessed in a way that more comprehensively captures the full range of negative emotion, such as sadness or other low arousal negative emotions. Also of note, we were limited in the ability to comprehensively assess the impacts of COVID-19 diagnosis given the low incidence of COVID-19 diagnosis in the sample. Additionally, given the rapidly changing quarantine guidelines or social distancing restrictions it is possible that variables, such as loneliness or social support, may have differed if assessed at various times during the pandemic. Finally, given that we tested for multiple interactions, the likelihood of finding a significant interaction by chance may have been increased. Indeed, after correction for multiple interactions, our results were reduced to non-significance or to trends. Thus, discussions related to our interactions should be interpreted with caution until they are replicated in similar studies.
Future Directions
Future research should examine resilience factors that help emerging adults flourish and thrive in the midst of the negative implications of the trauma and stress that occur in the midst of a pandemic. It is also important to examine specific groups of students (i.e., veterans, racial-ethnic minorities) who may be uniquely affected by the pandemic (Trammell et al., 2021). Further, as time since the beginning of an epidemic increases, emotional and behavioral responses to the pandemic context may change; therefore, future research should also examine longitudinal emotion and sleep trajectories in a university student cohort as well as determine whether the traumas and stressors are associated with long-term biomarkers in this population, i.e., cortisol and inflammation. Longitudinal work may also shed light on whether loneliness perceptions are more vulnerable than social support perceptions to being changed during a pandemic that involves quarantines and social distancing.
Concluding Remarks
COVID-19 and pandemics like it present unparalleled challenges to individuals, institutions, and researchers alike. The current study suggests that for emerging adults, sleep health and emotional stability are at risk during this time and may be related to pandemic-specific factors and individual differences existing long before the pandemic. With further research, conceptual models and translational science could reinforce our defenses against the current pandemic as well as future inevitable collective crises.
Footnotes
Author Contributions
Jennifer Harriger has contributed to conception, design, acquisition, and interpretation; drafted the manuscript; critically revised the manuscript; gave the final approval; and agreed to be accountable for all aspects of work ensuring integrity and accuracy. Nataria Joseph has contributed to conception, design, acquisition, analysis, and interpretation; drafted the manuscript; critically revised the manuscript; gave the final approval; and agreed to be accountable for all aspects of work ensuring integrity and accuracy. Janet Trammell has contributed to conception, design, acquisition, and interpretation; drafted the manuscript; critically revised the manuscript; gave the final approval; and agreed to be accountable for all aspects of work ensuring integrity and accuracy.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial supportfor the research, authorship, and/or publication of this article: funding was provided by internal grants from Pepperdine University.
Notes
Open Practices
No aspects of the study were pre-registered. The raw data, analysis code, and materials used in this study are not openly available but are available upon request to the corresponding author. Data and materials for this study have not been made publicly available. The design and analysis plans were not preregistered.