Demographics and Analysis
Analyses included 16,017 observations across 6,874 unique participants (Table 1). Incomplete youth-parent/caregiver dyad data (e.g., parents/caregivers, hereafter referred to as “parents”, but not their children returning Q1 or vice versa) was not an exclusionary criterion. Relative to the entire ABCD cohort, our surveyed sample was more likely to (1) have higher incomes, (2) live in less disadvantaged census tracts, (3) identify the youth’s race as white, and (4) identify the youth’s ethnicity as non-Hispanic (Table 1). Here, ADI covaried with household income, Spearman’s rho (ρ) = 0.50, p < .001. At Q1, youth participants were ~ 12.5 years old (range: 10.6–14.6).
Table 1
Demographics for the Adolescent Brain Cognitive Development (ABCD) Study
|
Release 3.0 (%) [Baseline Data]
|
Sample in this Report (%)
|
Youth Sex
|
|
|
Male
|
6,196 (52.1%)
|
3,612 (52.5%)
|
Female
|
5,682 (47.8%)
|
3,262 (47.5%)
|
Annual Household Income
|
|
|
<$5,000
|
417 (3.5%)
|
168 (2.4%)
|
$5,000-$11,999
|
421 (3.5%)
|
187 (2.7%)
|
$12,000-$15,999
|
274 (2.3%)
|
137 (2.0%)
|
$16,000-$24,999
|
524 (4.4%)
|
254 (3.7%)
|
$25,000-$34,999
|
654 (5.5%)
|
342 (5.0%)
|
$35,000-$49,999
|
934 (7.9%)
|
503 (7.3%)
|
$50,000-$74,999
|
1,499 (12.6%)
|
926 (13.5%)
|
$75,000-$99,999
|
1,572 (13.2%)
|
983 (14.3%)
|
$100,000-$199,999
|
3,315 (27.9%)
|
2,371 (34.5%)
|
≥$200,000
|
1,250 (10.5%)
|
1,003 (14.6%)
|
Missing/Undefined
|
1,018 (8.6%)
|
0 (0.0%)
|
Area Deprivation Index
|
|
|
≤ 33 Percentile (Low)
|
5,392 (45.4%)
|
3,655 (53.2%)
|
34–66 Percentile (Mid)
|
3,499 (29.5%)
|
2,162 (31.5%)
|
≥ 67 Percentile (High)
|
2,055 (17.3%)
|
1,057 (15.4%)
|
Missing/Undefined
|
932 (7.8%)
|
0 (0.0%)
|
Youth Race
|
|
|
American Indian / Alaska Native
|
62 (0.5%)
|
26 (0.4%)
|
Asian
|
276 (2.3%)
|
192 (2.8%)
|
Black
|
1,869 (15.7%)
|
822 (12.0%)
|
Native Hawaiian / Pacific Islander
|
16 (0.1%)
|
8 (0.1%)
|
Other
|
1,959 (16.5%)
|
1,096 (15.9%)
|
White
|
7,525 (61.1%)
|
4,730 (68.8%)
|
Missing/Undefined
|
171 (1.4%)
|
0 (0.0%)
|
Youth Ethnicity
|
|
|
Hispanic
|
2,411 (20.3%)
|
1,258 (18.3%)
|
Not Hispanic
|
9,314 (78.4%)
|
5,616 (81.7%)
|
Missing/Undefined
|
153 (1.3%)
|
0 (0.0%)
|
Total
|
11,878 (100%)
|
6,874 (100%)
|
Primary analyses employed linear mixed-effects models with linear and quadratic terms of household income and ADI as the predictors of interest, controlling for race, ethnicity, parental education, and youth age and sex at birth for analyses of youth data or parent data about the youth participant (see Tables S4-S35). ABCD study site and participant ID were included as random effects. Family risk/exposure analyses incorporated generalized linear mixed-effects models. Statistical significance was assessed by comparing each coefficient divided by its standard error to the t-distribution.
COVID-19 Disease Burden: Family Exposure Risk and Reported Diagnoses
Across Q1-Q3, risk of COVID-19 exposure due to essential-job employment or public-transit use significantly increased with ADI (i.e., greater neighborhood disadvantage), t(15313) = 10.45, p < .001, partial correlation coefficient (rp) = .084, but this pattern plateaued across the highest ADI tracts [(ADI)2], t(15313) = 5.48, p < .001 (Figs. 1A-B; Table S4). While risk of exposure generally increased with household income, t(15313) = 9.14, p < .001, rp = .074, there was a substantial decrease across the largest household incomes [(household income)2], t(15313) = 6.07, p < .001, in that, for household income, intermediate income households were those more likely at increased risk of exposure.
At Q2 (late June to July 2020), 3.4% of parents (178/5223) reported that at least one immediate family member (i.e., same household) had been diagnosed with COVID-19. Families with lower household incomes, t(5210) = -5.52, p < .001, rp = .076, and those living in higher ADI census tracts, t(5210) = 3.74, p < .001, rp = .052, reported more family members having been diagnosed with COVID-19 (Figs. 1C-D; Table S5; also see Table S3 for a detailed breakdown of these data). In the highest ADI decile (most disadvantaged neighborhoods), 10.1% of families reported having at least one family member diagnosed with COVID-19; in the most affluent neighborhoods, 2.7%. Similarly, while 12.5% of the lowest-income households reported that at least one family member had been diagnosed with COVID-19, 2.5% of the highest-income households reported at least one COVID-19 diagnosis; for families with the lowest household incomes who also lived in both the most disadvantaged neighborhoods, 17.6%. Thus, as predicted here and consistent with previous reports,15–17 lower household incomes and residence in greater ADI census tracts were associated with greater familial COVID-19 disease burden and, thus, greater risk of diagnosed COVID-19 exposure for parents and youth.
Perceived Risk
Contrary to our predictions, decreased household income was associated with decreases in believing that the participant his-/her-/themself would get COVID-19, t(10081) = 5.42, p < .001, rp = .054, that someone close to them would get COVID-19, t(10082) = 7.10, p < .001, rp = .071, and that someone close to them would be hospitalized or die from COVID-19, t(10082) = 2.96, p = .003, rp = .030 (Tables S6-S9).
Associations of ADI with perceived risk were considerably weaker than with household income (Tables S6-S9). Nonetheless, similar to findings with household income, greater neighborhood disadvantage (higher ADI) was associated with significant decreases in participants believing he/she/they would get COVID-19, t(10081) = 2.57, p = .010, rp = .026, and that someone close to them would get COVID-19, t(10082) = 2.07, p = .038, rp = .021; the other perceived-risk relationships with ADI were not significant, ps ≥ .491.
As higher ADI and lower income were associated with having one or more family members diagnosed with COVID-19, we conducted sensitivity analyses including only those who had not had immediate family members diagnosed with COVID-19 to examine the possibility that perceived risk may differ based on experiencing positive COVID-19 tests within the household. The relationships with household income were maintained (albeit weaker) for thinking that one’s self, t(7992) = 4.22, p < .001, rp = .047, or someone close to him/her/them would get COVID-19, t(7993) = 5.59, p < .001, rp = .062, and for whether someone close to him/her/them would be hospitalized and/or die from COVID-19, t(7993) = 2.10, p = .036, rp = .023 (Table S10-S13). However, upon accounting for reported rates of diagnosis, ADI was no longer associated with perceived risk of exposure, ps ≥ .092, suggesting that perceived risk may not align with actual disease burden or likelihood of infection.
COVID-Related Worry
Lower household income was related to greater parental worry, t(15357) = 2.87, p = .004, rp = .023. Parental worry also tended to increase with ADI in higher ADI tracts [(ADI)2], t(15357) = 2.05, p = .041] (Fig. 3A-B; Table S14). Although youth worry decreased with increasing household incomes, t(12510) = 2.32, p = .020, rp = .021, with these levels plateauing at greater income levels [(household income)2], t(12510) = 2.90, p = .004, youth worry was neither linearly nor quadratically related to ADI, ps ≥ .072 (Table S15). Parent-reported youth worry levels about the health- and non-health-related consequences (e.g., financial) of COVID-19 were also negatively associated with household income, ps < .001, but not ADI, ps ≥ .342 (Tables S16-S17). Analyses also indicated that greater disease burden was related to increases in parent but not youth worry levels (see Supplemental Information, “COVID-19-Related Worry and Disease Burden”).
Youths’ worry levels were highly correlated with, but noticeably lower than, their parents’ worry levels, Spearman’s rho (ρ) = .28, p < .001 (Fig. 3C) (Parent: M = 3.01, SEM = 0.01; Youth: M = 2.36, SEM = 0.01). While youth were only asked about general COVID-19-related worry, youth’s self-reported worry was more highly correlated with their parents’ report on their health-related, ρ = .26, p < .001, than non-health-related worry, ρ = .14, p < .001 (Fig. 3D), suggesting that youth’s general COVID-19-related worry was more related to their concerns about getting sick from COVID-19 rather than its non-health-related consequences.
Families’ Responses to the COVID-19 Pandemic
Family-level disadvantage (i.e., lower household income) was associated with both increased disease burden (risk/exposure) and increased youth and parent worry, while neighborhood-level disadvantage (i.e., higher ADI) was associated with increased disease burden. However, increased socioeconomic disadvantage across both levels was associated with reduced perceived risk. To determine whether more disadvantaged families were differentially engaging in potential coping or disease-risk reduction strategies given increased COVID-19 risk and disease burden, we analyzed indicators of parent-youth communication about COVID-19 risk and prevention, parental reassurance about COVID-19, parental transparency with their child regarding their own COVID-19-related concerns, and youth’s COVID-19 preventative behaviors.
Parent-Youth Communication about COVID-19 Risk and Prevention
Lower household income was associated with increased communication on all topics queried regarding COVID-19 prevention (Fig. 4A): the importance of handwashing, t(15063) = 3.12, p = .002, rp = .025; the importance of social distancing, t(15062) = 5.03, p < .001, rp = .041; cancellations of school and other events, t(15061) = 5.27, p < .001, rp = .043; avoiding visits with friends/family, t(15061) = 7.11, p < .001, rp = .058; COVID-19 symptoms, t(15057) = 6.08, p < .001, rp = .049; protecting the elderly/vulnerable, t(15061) = 4.55, p < .001, rp = .037; and, wearing masks, t(4804) = 3.34, p = .001, rp = .048 (Tables S22-S28). In other words, families with lower incomes were speaking with their children about COVID-19 prevention more frequently than their higher-income counterparts were. Aside from parents’ talking about COVID-19 symptoms and handwashing, ps ≥ .359, these associations tended to plateau at the highest income levels [(household income)2], ps ≤ .028.
For ADI, while there were small negative associations between ADI and frequency of parent-youth discussions on three of the queried COVID-19 prevention topics [importance of social distancing, t(15062) = 2.81, p = .005, rp = .023; avoiding visits with friends/family, t(15061) = 3.06, p = .002, rp = .025; and, wearing masks, t(4804) = 2.34, p = .019, rp = .034], there were significant positive quadratic terms for ADI for each of these topics, ps ≤ .016 (Fig. 4B; Tables S22-S28). To better understand the quadratic relationships between ADI and parent-youth communication on COVID-19 prevention, we conducted bivariate ADI-by-parent/youth-communication correlational probe analyses (for all prevention topics) separately for those with ADI ≤ 40th percentile (Low ADI; n = 4,414 participants) and for those with ADI > 40th percentile (High ADI; n = 2,460 participants), given the minimal change in parent-youth communication below the 40th percentile (i.e., the 40% least deprived per national percentile; Fig. 4B). For High ADI participants, there were significant positive correlations between ADI and parent-youth communication frequency on all queried topics related to COVID-19 risk/prevention, ρs ≥ .12, ps < .001. In contrast, these relationships were substantially weaker for Low ADI participants (hand washing: ρ = .01, p = .484; social distancing: ρ = .04, p < .001; cancellations: ρ = .01, p = .338; avoiding visits: ρ = .04, p < .001; COVID-19 symptoms: ρ = .02, p = .041; protecting the elderly/vulnerable: ρ = .01, p = .252; wearing masks: ρ = .04, p = .018), further suggesting that parents in more disadvantaged neighborhoods were talking more with their children about COVID-19 prevention (Fig. 4).
Parents were also asked about how often they engaged in parental reassurance (i.e., “everything will be okay”) and parental encouragement to not dwell on COVID-19, as well as items related to transparent parental communication (i.e., parents’ discussing their own feelings about COVID-19, avoiding discussions about COVID-19, and discussing with their child about the safety and life-altering impact of COVID-19).
Lower household income was associated with more parental encouragement, t(9839) = 2.72, p = .007, rp = .027, but there was no relationship between income and parental reassurance, t(9842) = 1.18, p = .239, rp = .012 (Tables S29-S30). Except for talking about their own COVID-19-related feelings, p = .698, parents with lower household incomes were more likely to avoid talking to their child about COVID-19, t(9842) = 5.09, p < .001, rp = .051, more likely to tell their child that they may not be fully safe from COVID-19, t(9841) = 4.40, p < .001, rp = .044, and more likely to prepare their child that their lives may change significantly, t(9842) = 4.28, p < .001, rp = .043 (Fig. 5A; Tables S31-S34). Higher ADI (greater neighborhood disadvantage) was associated with more parental reassurance, t(9842) = 2.58, p = .010, rp = .026, and encouragement, t(9839) = 2.43, p = .015, rp = .024 (Fig. 5B; Tables S29-S30). However, there were no linear or quadratic associations with ADI and parental transparency items, ps ≥ .300 (Fig. 5B; Tables S31-S34).
Therefore, like parent-youth communication on COVID-19 risk/prevention, parents with lower household incomes and/or living in higher ADI tracts may be providing their child with greater preventative and anxiety-reducing emotional support in the wake of increased risk of COVID-19 exposure. Increased frequency of parent-youth discussions on COVID-19 prevention was also associated with less perceived risk, particularly in High ADI participants (see Supplemental Information, “Parent-Youth Communication and Perceived Risk”).
Youths’ Preventative Actions
Given increased disease burden (i.e., exposure, risk) and more frequent parent-youth discussions on COVID-19 prevention in disadvantaged families, we analyzed how often these children endorsed engaging in COVID-19-reducing behaviors (i.e., average frequency across multiple items: wearing a mask, avoiding others inside and outside their house, using hand sanitizer, washing hands, wiping surfaces, and avoiding touching people and things).
In the face of increased COVID-19 disease burden within families, youth of lower-income families and those living in more disadvantaged neighborhoods reported greater engagement in preventative actions. Greater household income was associated with decreased frequency of youths’ preventative actions, t(8071) = 4.05, p < .001, rp = .045, plateauing at the greatest household incomes [(household income)2], t(8071) = 3.48, p = .001 (Fig. 6A, Table S35). An increase in youth preventative actions was also evident in the more disadvantaged neighborhoods [(ADI)2], t(8071) = 3.41, p = .001 (Fig. 6B). As with parent-youth communication frequency, there was a strong positive relationship between ADI and youth preventative actions for High ADI participants, ρ = .16, p < .001, but a weaker, negative relationship for Low ADI participants, ρ = .07, p < .001.
The direct relationships between socioeconomic disadvantage and youths’ engagement in preventative behaviors mirrored the relationships with how often parents reported discussing prevention with their children. This finding was confirmed via a strong association between the average frequencies of youths’ preventative actions and parent-youth discussions on COVID-19 risk and prevention, ρ = .30, p < .001 (Fig. 6C). While youth who were more worried about COVID-19 also engaged more in COVID-19 preventative actions, ρ = .28, p < .001 (Fig. 6D), parents who were more worried about COVID-19 were also more likely to talk to their children about COVID-19 risk and prevention strategies (see Fig. 3), ρ = .37, p < .001 (Fig. 6E). More frequent parent-youth discussions on COVID-19 prevention and greater youth engagement in preventative behaviors were also associated with increased parental support and transparency (see Supplemental Information, “Preventative Actions, Parental Support, and Parental Transparency”). Thus, frequency of parent-youth communication on COVID-19 risk/prevention paralleled how often children endorsed engaging in preventative actions, both occurring more often in families with lower household incomes and/or those living in more disadvantaged census tracts.