Association between COVID-19 Risk-Mitigation Behaviors and Specific Mental Disorders in Youth

Abstract Background : Although studies of adults show that pre-existing mental disorders increase risk for COVID-19 infection and severity, there is limited information about this association among youth. Mental disorders in general as well as specific types of disorders may influence their ability to comply with risk-mitigation strategies to reduce COVID-19 infection and transmission. Methods : Youth compliance (rated as “Never,” “Sometimes,” “Often,” or “Very often/Always”) with risk mitigation was reported by parents on the CoRonavIruS Health Impact Survey (CRISIS) in January 2021. Responses were summarized using factor analysis of risk mitigation, and their associations with lifetime mental disorders (assessed via structured diagnostic interviews) were identified with linear regression analyses (adjusted for covariates). All analyses used R Project for Statistical Computing for Mac (v.4.0.5). Results : A two-factor model was the best-fitting solution. Factor 1 (avoidance behaviors) included avoiding groups, indoor settings, and other peoples’ homes; avoidance was more likely among youth with any anxiety disorder (p=.01). Factor 2 (hygiene behaviors) included using hand sanitizer, washing hands, and maintaining social distance; practicing hygiene was less likely among youth with ADHD (combined type) (p=.02). Mask wearing, which did not load on either factor, was not associated with any mental health disorder. Conclusion and Relevance : Findings suggest that education and monitoring of risk-mitigation strategies in certain subgroups of youth may reduce risk of exposure to COVID-19 and other contagious diseases. Additionally, they highlight the need for greater attention to vaccine prioritization for individuals with ADHD.


Introduction
The COVID-19 pandemic constitutes an ongoing global public health threat. SARS-CoV-2 is a highly contagious virus that transmits mainly through inhalation of airborne droplets and transfer from direct contact with surfaces that are contaminated. Public health o cials across the world initially responded to this threat by urging people to follow several risk-mitigation strategies to reduce the chance of infection and transmission. Adults and children alike were urged to avoid close contact, maintain at least 6-feet of physical distance from others, wear face masks in public, engage in frequent and intensive hand washing, adhere to stay-at-home orders, and self-isolate when exhibiting symptoms of infection 1 . Such efforts have been associated with a reduction in COVID-19 There is growing evidence regarding the potential impact of mental health disorders on COVID-19 risk. Studies of adults show that pre-existing mental disorders are associated with increased risk of COVID-19 infection 5 , severity 6-9 , and mortality [7][8][9][10][11][12] . A recent study showed that individuals with substance use disorders are also at increased risk of COVID-19 infection, hospitalization, and death 8, 9 . These ndings have been largely con rmed by several meta-analyses [7][8][9]13 .
Accordingly, the Centers for Disease Control designated mental and behavioral health disorders among the preexisting conditions that increase vulnerability to COVID- 19 illness. An important effect of this recognition is greater priority in receiving vaccines and boosters as they are rolled out, which help to protect individuals and reduce community spread. Included in this list are disorders that typically affect children, such as neurodevelopmental disorders like autism spectrum disorder and ADHD. However, there is scant research on the role of mental disorders in COVID-19 vulnerability among youth. One recent study of electronic health records among patients (aged 2 months to 103 years) in Israel found that COVID-19 infection was associated with ADHD (but no other psychiatric diagnosis examined), male gender, age below 20 years, and low-medium SES group 14 . Interestingly, the association between ADHD and COVID-19 infection in this sample was especially elevated among youth (ages [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] and untreated ADHD cases of any age. Additional research on youth with a range of mental disorders is needed to understand the potential mechanisms associated with COVID-19 vulnerability. In particular, the ability of youth to adhere to risk-mitigating practices may differ by the key phenomena underlying mental disorders such as inattention, anxiety, fear, impulsivity, etc. To our knowledge, no studies have investigated COVID-19 risk-mitigation adherence among youth with mental disorders. The purpose of this study was to examine associations between COVID-19 risk-mitigation practices and speci c mental disorders among female and male youth from the Healthy Brain Network (HBN) in the New York City metropolitan area.

Sample
Participants were recruited from the ongoing Healthy Brain Network (HBN) initiative that seeks to create and share a 10,000-participant biobank of data from children and adolescents ages 5-21 from the New York City area 15 .
Data collected includes psychiatric, cognitive, behavioral, genetic, and lifestyle information as well as MRI and EEG neuroimaging. The HBN collection sites are on Staten Island, in Midtown Manhattan, and in Harlem. As part of the HBN survey battery, participants and their parents/guardians completed a variety of age-based questionnaires assessing basic demographic characteristics, dimensional assessments of domains associated with mental health, substance use, and socioeconomic status. For participants under the age of 11, a trained research assistant read and explained individual items and collected responses from participants. HBN's latest data release includes 4139 participants; data are available to researchers by registering for a data usage agreement (http://fcon_1000.projects.nitrc.org/indi/cmi_healthy_brain_network/Pheno_Access.html#DUA). The study was approved by the Chesapeake Institutional Review Board (https://www.chesapeakeirb.com/). Prior to conducting the research, written informed consent is obtained from participants ages 18 or older. For participants younger than 18, written consent is obtained from their legal guardians and written assent obtained from the participant.
Between April and July 2020 (Wave 1), parents of HBN participants were invited to complete the CoRonavIruS Health Impact Survey (CRISIS) 16 about their child via Research Electronic Data Capture (REDCap). The CRISIS was designed and piloted at the beginning of the COVID-19 pandemic to assess mental and behavioral health, lifestyle behaviors, and sources of stress induced by the COVID-19 epidemic. In total, parents of 1780 HBN participants completed the Wave 1 survey. Parents were then invited to complete a modi ed version of the CRISIS in January 2021 (Wave 2). The Wave 2 modi cations included questions on frequency of compliance with COVID-19 riskmitigation practices. The current study sample included 314 female and 514 male participants (ages 5-21) whose parents completed the CRISIS at Wave 2.

Measures
Measures derived from HBN participation included child age at Wave 2, sex, race/ethnicity, family structure, family socioeconomic status (SES), and consensus diagnosis. Child age, sex, race/ethnicity (Caucasian, African American, Hispanic, Asian), and family structure (indicator of single caregiver household) were reported by parents/caregivers during a structured clinical history interview. Family SES was measured by the Barratt Simpli ed Measure of Social Status, which is based on parent/caregiver reports of parent/caregiver education and occupation 17 . Continuous scores are generated with higher scores indicating higher SES. In single caregiver families, scores were based on that caregiver alone. SES scores were subsequently grouped into tertiles to determine low, middle, and high SES.
Diagnostic interviews were conducted using the computerized Kiddie Schedule for Affective Disorders and Schizophrenia 18 that was administered to parents by an experienced research clinician or social worker. Following Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria, consensus lifetime diagnosis was achieved by two study psychiatrists based on these interviews and other symptomatic information such as standardized rating scales. Up to 10 separate diagnoses were allowed per participant. Diagnoses were grouped into the following categories: attention de cit hyperactivity disorder inattentive/hyperactive type [ADHD-I), ADHD combined type (ADHD-C), autism spectrum disorder, any depressive disorder (major depressive disorder [MDD], persistent depressive disorder [PDD], disruptive mood dysregulation disorder [DMDD], depressive disorder due to another medical condition, unspeci ed depressive disorder, substance/medication-induced depressive disorder, other speci ed depressive disorder, other (or Unknown) substance-induced disorders), any anxiety disorder (unspeci ed anxiety disorder, generalized anxiety disorder [GAD], separation anxiety, social anxiety, speci c phobia, agoraphobia, panic disorder, selective mutism, other speci ed anxiety disorder), and any other behavior disorder (oppositional de ant disorder, conduct disorder, intermittent explosive disorder, or other speci ed disruptive, impulse-control disorder).
Seven separate COVID-19 risk-mitigation practices among youth were measured via parent report at Wave 2 of CRISIS administration. Speci cally, parents were asked with respect to the past two weeks "To what extent has your child been taking the following steps to prevent infection or spread of the virus? Wearing a mask or face covering in public; Wearing gloves in public; Washing hands; Using hand sanitizer; Staying at least six feet away from others; Avoiding visits to other people's homes; Avoiding group in-person activities; Avoiding indoor public places (e.g., stores) when possible." Responses were rated as "Never," "Sometimes," "Often," or "Very often/Always." Analysis Primary analyses were conducted using the parent CRISIS survey conducted in January 2021. This survey was limited to the 1578 past HBN participants (2015-2020; total: 3600) that completed an initial survey in April-June 2020 (Wave 1). 955 parents completed the survey in 2021. Participants that did not complete enough of the HBN study to yield a consensus DSM diagnostic pro le based on the KSADS-COMP and licensed clinical evaluators, were removed. Five participants with missing Barratt data were also removed. The nal analytic sample comprised 314 female and 514 male participants between ages 5-21 years (N=828). Sample characteristics did not differ by study-completion status (see Supplement). All statistical analyses were conducted using The R Project for Statistical Computing for Mac 19 .
Responses to the 7 risk-mitigation practices were summarized using factor analysis, and model tness was evaluated by parallel analysis. Associations between the resulting factors and lifetime mental disorders were identi ed with linear regression analyses. Three models were used in the analysis: unadjusted, adjusted for demographics (age, sex, SES, single caregiver, and race), and additionally adjusted for comorbid mental disorders.
Each factor underwent the three-model analysis to examine the associations between speci c mental disorders and each derived factor of risk-mitigation behaviors. Table 1 presents the number and percentages of youth who "very often/always" engaged in the 7 risk-mitigation behaviors, by demographic characteristics. Overall, the percentage of mask wearing "very often/always" was high (90%), and higher in females, African Americans, and Hispanics. The overall percentage of "very often/always" maintaining social distance, using hand sanitizer, and washing hands was 46%, 46%, and 59%, respectively. These behaviors differed signi cantly by race, as the percentages were elevated among African Americans and Hispanics and lower among Caucasians. Further, maintaining social distancing "very often/always" was positively associated with age and negatively associated with SES, as was washing hands "very often/always." The overall percentages of avoiding other people's homes, avoiding in-person groups, and avoiding indoor public places "very often/always" was 58%, 46%, and 43%, respectively. These were highly similar across demographic factors. Note. 1 De ned as "Very often/Always" according to the original responses of "Never," "Sometimes," "Often," or "Very often/Always." Chi-Square group differences are represented by asterisks; * p < 0.05, ** p < 0.01, *** p < 0.001. 4 Barratt total score was divided into tertiles: Low The correlations among the risk-mitigation items appear in Fig. 1. All correlations were positive and ranged from 0.09 to 0.63. Mask wearing correlated weakly (r < 0.31) with all other items. Maintaining social distance correlated moderately (r = 0.38 to r = 0.44) with using hand sanitizer, avoiding other people's homes, avoiding in-person groups, and avoiding indoor public places, and washing hands. Using hand sanitizer correlated strongly (r > 0.63) with washing hands and weakly (r < 0.14) with avoiding other people's homes, avoiding in-person groups, and avoiding indoor public places. Washing hands correlated weakly (r < 0.21) with avoiding other people's homes, avoiding in-person groups, and avoiding indoor public places. The three avoidance items -avoiding other people's homes, avoiding in-person groups, and avoiding indoor public places -were moderately to strongly intercorrelated (r = 0.53 to r = 0.60).

Results
Results from the parallel analysis (Fig. 2) show that a two-factor model was the best-tting solution. That is, a two-factor solution explained more variance than what would be expected due to chance based on a null distribution of eigenvalues. Results from the factor analyses are shown in Fig. 3. The rst factor (avoidance behaviors) included avoiding groups, indoor settings, and other peoples' homes. This factor accounted for 27.7% of the cumulative variance. The second factor (hygiene behaviors) included using hand sanitizer, washing hands, and maintaining social distance. The addition of this factor increased the cumulative variance explained to 49.8%.
Mask wearing loaded poorly on each factor, a nding consistent with the weak individual correlations with this item. Further, its inclusion to the model did not meaningfully improve the t or the amount of variance explained. For these reasons, mask wearing was analyzed subsequently as a separate item. Table 2 presents the results from the regression analyses examining associations between the two resulting factors (and, separately, mask wearing) and mental disorders. The table presents results from the unadjusted model, the model (a) adjusted for demographics, and the model (b) adjusted for demographic variables and comorbid mental disorders. Focusing on the fully adjusted model (b), several of the demographic variables were signi cant covariates (results not shown). African American race was positively associated with both avoidance behaviors and hygiene behaviors. Age, Hispanic race, other race, or nondisclosed race were positively associated with hygiene behaviors. Male sex was positively associated with mask wearing. As shown in Table 2, the two factors were associated with speci c mental disorders. The avoidance behaviors were more likely among youth with any anxiety disorder, a nding that was signi cant in the unadjusted model (p = .008), the model that adjusted for demographic variables (p = .01), and the model that adjusted for demographics and comorbid disorders (p = .01). No other disorder was signi cantly associated with this factor. The hygiene behaviors were less likely among youth with ADHD-C in the unadjusted model (p = .01), the model that adjusted for demographic variables (p = .02), and the model that adjusted for demographics and comorbid disorders (p = .02). Hygiene behaviors were negatively associated with depression, but only in the unadjusted model (p = .04). Analyzed as a separate item, mask wearing was less likely among youth with ADHD-C in the unadjusted model (p = .03). However, this association was not signi cant after adjusting for demographics and comorbid disorders. No other mental disorder was signi cantly associated with mask wearing. . ADHD-I: attention de cit hyperactivity disorder inattentive/hyperactive type, ADHD-C: ADHD combined type; ASD: autism spectrum disorder; Depression: any depressive disorder including major depressive disorder, persistent depressive disorder, disruptive mood dysregulation disorder, depressive disorder due to another medical condition, unspeci ed depressive disorder, substance/medication-induced depressive disorder, other speci ed depressive disorder, other (or Unknown) substance-induced disorders); Anxiety: any anxiety disorder including generalized anxiety disorder, separation anxiety, social anxiety, speci c phobia, agoraphobia, panic disorder, selective mutism, unspeci ed anxiety disorder, other speci ed anxiety disorder; Behavior: any other behavior disorder including oppositional de ant disorder, conduct disorder, intermittent explosive disorder, or other speci cized disruptive, impulse-control disorder.

Discussion
The importance of risk-mitigation measures in reducing exposure and severe disease have been one of the most important public health measures in response to the prolonged nature of the COVID-19 pandemic, now entering its third year. Physical distancing, avoiding touch, washing hands frequently and intensively, wearing a face mask in public, staying at home, and maintaining quarantine have been particularly challenging for youth with mental disorders. Here we report two important ndings regarding the association between compliance with different riskmitigation factors with different types of disorders. First, the COVID-19 risk-mitigation behaviors tended to cluster into two factors: avoidance behaviors (avoiding groups, indoor settings, and other peoples' homes) and hygiene behaviors (using hand sanitizer, washing hands, and maintaining social distance). Second, the tendency to practice these behaviors differed by speci c mental disorders. Avoidance was more likely among youth with any anxiety disorder, whereas practicing hygiene was less likely among youth with ADHD-C. Mask wearing was largely unrelated to other risk-mitigation behaviors and, when examined as a separate item, was not associated with mental disorders after adjusting for demographic factors or comorbid mental disorders.
Our ndings add to the limited research on the role of mental disorders in COVID-19 risk among youth, despite ample evidence showing that children and adolescents with mental disorders are at increased risk of myriad physical diseases 20 . In particular, ADHD has been shown increase risk of in ammatory and immune-related disorders including asthma, eczema, certain allergies [21][22][23][24] , respiratory infections and in uenza 12 , and several preventable negative outcomes including sexually transmitted infections, accidental injuries, and injury-related mortality [25][26][27] . Psychosocial stress, anxiety, negative affect, and depression are also associated with increased risk of acute respiratory infections as well as poorer clinical prognosis 28 . Depression has been associated with an increased risk of a developing a wide range of infections 29 including sexually transmitted diseases and poorer outcomes thereof 30,31 . Children and adolescents with autism spectrum disorders are at increased risk of several medical conditions including immunological, gastroenterological, neurological, and other medical complaints 22 . This work provides some insight into the increased risk by showing that youth with speci c mental disorders are more (or less) likely to engage in certain COVID-19 risk-mitigation behaviors.
There are several potential mechanisms for our ndings. Youth with anxiety disorders, who were more likely to practice avoidance behaviors, may be so inclined because avoidance is a central characteristic of many anxiety disorders including social phobia, agoraphobia, social anxiety disorder, and generalized anxiety disorder. Avoidance-related manifestations of anxiety disorder may serve to mitigate COVID-19 vulnerability by reducing social contacts with others in groups, indoor settings, and other peoples' homes. The failure to comply with hygiene behaviors among youth with ADHD-C may occur through other mechanisms. Youth with ADHD-C, where both inattention and hyperactivity/impulsivity are present, may be less capable of complying with 6-foot distance rules and hand-cleaning practices. This may help explain the nding reported by Merzon and colleagues 14 that youth with ADHD are at increased risk of COVID-19 infection, and that the association was greater for untreated than treated ADHD. Several non-signi cant ndings are notable as well. Although youth with ASD can sometimes struggle with daily living skills including maintaining personal hygiene 32 that may lead to lower compliance with hand washing and using hand sanitizer, we did not observe this nding. Youth with oppositional de ant disorder or conduct disorder, who tend to disregard social rules, were no more or less likely to consistently engage in harmmitigation behaviors. We similarly found no independent association between depressive disorders and riskmitigation behaviors. In summary, these interpretations are suggestive and warrant further research.
Aside from risk-mitigation compliance, there are other mechanisms by which pre-existing mental and behavioral health disorders may increase COVID-19 vulnerability. These include increased exposure to COVID-19 at home or at school due to greater household or community density, greater susceptibility due to enhanced physical disease vulnerability as described above, or poorer general health status related to obesity, physical inactivity, or familial exposure to smoking. Future research on COVID-19 vulnerability should incorporate the full range of risk factors for COVID-19 infection as well as compliance with COVID-19 risk-mitigation behaviors.
To further reduce COVID-19 incidence and community transmission, it is crucial to identify risk factors for COVID infection in youth. Although some research shows that young children are less susceptible to infection than adults [33][34][35] , other research shows that children may be as likely as adults to become infected with COVID-19 36 . In fact, a recent epidemiologic study 37 of a pediatric sample in Virginia reported a SARS-CoV-2 infection rate (8.5%) that was higher than a sample of adults (2.4%) from a similar region and period 38 . Further, children appear to play a role in community transmission through their social interactions and hygienic habits 33 , a nding that underscores the importance of risk-mitigation strategies among youth especially as the COVID-19 pandemic continues to evolve and become even more transmissible, as in the case of the most recent circulating variant (Omicron). Indeed, a recent experiment found that the Omicron variant survives longer than other variants on plastic and skin, a factor that may have contributed to the rapid community spread of Omicron 39 . Taken together, ndings further underscore the importance of being vigilant about risk-mitigation behaviors to combat vulnerability to COVID-19 infection and transmission.
This work provides novel information on the associations between mental disorders and COVID-19 risk-mitigation behaviors in youth. Reduced practice of prevention measures among those with speci c types of disorders highlights the need to gain greater insight into the speci c di culties underlying the reduced compliance observed here. Risk-mitigation education could then be tailored to the speci c components of these conditions that reduce compliance efforts. It may also be worthwhile to consider prioritizing vaccinations among individuals with mental and neurodevelopmental disorders 40 that may reduce their ability to prevent exposure. The study was approved by the Chesapeake Institutional Review Board.

Abbreviations
Dr. Conway, Dr. Merikangas, and Dr. Milham conceptualized and designed the study. Ms. Bhardwaj, Dr. Paksarian and Ms. Kang carried out the data management, performed the analyses, contributed to the interpretation of the results, and drafted the initial manuscript. Dr. Conway, Dr. Merikangas, and Dr. Milham contributed to the interpretation of the results and supervised the work. All authors reviewed and revised the manuscript. All authors approved the nal manuscript as submitted and agreed to be accountable for all aspects of the work.

Funding
This work was supported in part by the Intramural Research Program of the National Institute of Mental Health (ZIAMH002953) and Morgan Stanley. Additionally, by philanthropic gifts from Phyllis Green, Randolph Cowen, and Joe Healey to the Child Mind Institute. The views and opinions expressed in this article is those of the authors and should not be construed to represent the views of any of the sponsoring organizations, agencies, or the U.S.
Government. The funding sources had no role in the design of this study and will not have any role during its execution, analyses, interpretation of the data, or decision to submit results.

Availability of data and materials
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.