Factors Associated with School Attendance Plans and Support for COVID-19 Risk Mitigation Measures Among Parents and Guardians

Abstract

OBJECTIVE:

To determine which factors are associated with plans for in-person school attendance during the 2020–2021 school year and with support for 15 school-based COVID-19 risk mitigation measures among parents and guardians.

METHODS:

In June 2020, we conducted an online survey of parents and guardians of public school children in Illinois, Michigan, and Ohio. In a child-level analysis, we used linear regression to assess which demographic factors, health-related concerns, and parent/guardian views were associated with plans for in-person school attendance. In a respondent-level analysis, we used linear regression to assess factors associated with the number of risk mitigation measures supported.

RESULTS:

Among 2,202 children in the child-level analysis, in-person school attendance was planned for 71.0%. Such plans were less likely among children of Black respondents (−14.1 percentage points, 95% CI: −25.7, −2.6) and Asian respondents (−16.8, 95% CI: −31.3, −2.2), and among children with perceived high-risk health conditions (−9.7, 95% CI: −15.8, −3.6). Among 1,126 respondents in the respondent-level analysis, the mean number of measures supported was 8.0 (SD 4.4). Several factors were associated with support, but the magnitude of associations was generally modest.

CONCLUSIONS:

During the COVID-19 pandemic, families of children with health conditions or who are of Black or Asian race/ethnicity may be less likely to opt for in-person learning. For these families, addressing barriers to remote education is critical. As schools plan for the 2020–2021 school year and beyond, they should respond to the desire among parents and guardians to implement substantial numbers of risk mitigation measures.

INTRODUCTION

U.S. public school districts have taken widely varying approaches to education in response to coronavirus disease 2019 (COVID-19). Some have begun the 2020–2021 school year with only in-person learning, others with only remote learning, and still others with both in-person and remote learning.¹ Districts offering in-person learning have also taken widely varying approaches to mitigate the spread of COVID-19 within schools. In some states, for example, districts must mandate face coverings for staff and students, while in other states, face coverings are only encouraged.^2,3

Many educators and policymakers are concerned about disruptions in educational achievement among children during the 2020–2021 school year. Disruptions may be particularly great for children who must engage in remote learning but lack stable Internet connections, necessary computing equipment, or adequate family supervision.^4–6 Educators and policymakers are also concerned that remote learning could also exacerbate food insecurity among low-income children who receive school meals, impede social development, and worsen mental health.^7–9 Despite these concerns, full-time remote learning may be necessary in districts located in areas with widespread community transmission of COVID-19. Furthermore, parents and guardians may opt for remote learning even if in-person learning is offered, for example owing to concerns regarding a family member’s risk of severe COVID-19 illness.

School districts may shift between in-person, remote, and hybrid learning during the 2020–2021 school year and may have to plan for COVID-19 during the 2021–2022 school year. To inform COVID-19 planning both now and later, it may be useful to identify which children are less likely to attend school in-person if this option is available. Such information could help educational policymakers identify the children for whom it is most important to address barriers to high-quality remote education. Additionally, it may also be useful to identify which factors are associated with support for school-based COVID-19 risk mitigation measures among parents and guardians. Such information could help districts implement measures by identifying which parents and guardians may require additional outreach to address concerns.

Several studies have reported demographic variation in parents’ plans for in-person school attendance during the 2020–2021 school year and in support for risk mitigation measures.^10–14 However, outside a few exceptions¹⁴, these studies only reported unadjusted differences. In mid-June 2020, we conducted an online survey of parents and guardians of public school children in Illinois, Michigan, and Ohio to assess school attendance plans and support for 15 school-based COVID-19 risk mitigation measures. In a policy brief, we published unadjusted analyses describing how plans and support varied by demographic characteristics, health-related concerns, views on COVID-19, and experiences with severe COVID-19 illness.¹² In this study, we conducted adjusted analyses to identify which factors were independently associated with school attendance plans and support for risk mitigation measures.

METHODS

Study sample.

We recruited respondents from a panel of individuals who volunteer to take surveys from Qualtrics, an experience management company whose services are frequently used in market research across sectors, including health services. Owing to our use of an opt-in non-probability panel, we do not report a response rate, following American Association for Public Opinion Research guidelines.¹⁵ Eligible respondents were parents and guardians who lived in Illinois, Michigan, and Ohio and had at least one child who typically would attend K-12 public school during the 2020–2021 school year. We focused on Illinois, Michigan, and Ohio because they are three populous states with differing political and demographic compositions.^16,17 We imposed state-specific quotas to ensure a minimum amount of representation by sex, age, race/ethnicity, political affiliation, and annual household income. Quotas were based on 2018 American Community Survey data and Gallup data on political affiliation.^17,18 Race/ethnicity was measured because factors such as the disproportionate impact of COVID-19 on minority communities might contribute to greater reluctance to attend in-person classes, which in turn might exacerbate existing systemic educational disparities by race/ethnicity.^4–6,19,20

We created two analytic samples, one for a child-level analysis focused on school attendance plans, and one for a respondent-level analysis focused on support for risk mitigation measures. The sample for the former included all public school children of respondents, except for those with missing data for school attendance plans or for independent variables (described below). The sample for the respondent-level analysis included all respondents, except for those with missing data for support for any of the 15 risk mitigation measures or for independent variables.

Survey development.

To develop survey questions, we examined prior surveys of parents’ views on school re-opening^21,22, demographic questions in federal surveys¹⁸, and lists of recommended school-based risk mitigation measures.²³ We reviewed the survey with a school board president and assistant principal of a middle school in Michigan. Additionally, we pre-tested the survey with eight parents of school-aged children in the Midwest; pre-testing stopped once the survey was stable.

Survey content.

The survey assessed respondents’ demographic characteristics, views on COVID-19, and experiences with severe COVID-19 illness. Respondents reported whether they planned to send their children to school for in-person classes during the 2020–2021 school year, assuming in-person classes were offered. Respondents reported whether each child had an individualized education plan (IEP) and any health conditions. For children with health conditions, respondents reported whether they believed the conditions increased the risk of severe COVID-19 illness.

Respondents reported their support for 15 risk mitigation measures. These include 6 measures focused on limiting contact between students (decreasing the number of students allowed on buses, alternating between in-person and online classes, staggering arrival and pick-up times, eating meals in classrooms instead of cafeterias, closing playground structures, and stopping all extracurricular school programs); 4 measures focused on testing and screening for COVID-19 (daily temperature checks for students, requiring COVID-19 testing for all students in a classroom if a classmate tests positive, randomly testing school staff once per week for COVID-19, and randomly testing students once per week for COVID-19); and 5 measures focused on mandatory face coverings for school staff, students in 6^th grade and above, students in 3^rd-5^th grade, students in 1^st-2^nd grade, and students in kindergarten. Appendix 1 includes the survey instrument.

Survey administration.

The survey was administered between June 12 and June 21, 2020. At the time, none of the three states had mandated school staff and students to use face coverings, although all three subsequently implemented such mandates over the next two months.^2,24,25 The survey could be taken online or via smartphone. Because data were de-identified, this study was exempted from review by the Institutional Review Board of the University of Michigan.

Dependent variables.

In the child-level analysis, the dependent variable was an indicator that equaled 1 if respondents reported that in-person school attendance was likely, and 0 if in-person school attendance was unlikely or if respondents were unsure of plans. In the respondent-level analysis, the dependent variable was the number of measures supported. To generate this variable, we created an indicator that equaled 1 if respondents supported or strongly supported a measure and 0 if they were neutral, opposed, or strongly opposed, then summed indicators across the 15 measures.

Independent variables.

All independent variables were categorical. In the child-level analysis, independent variables included state, child grade, respondent age, respondent sex, respondent race/ethnicity, annual pre-tax household income, respondent political affiliation, residence in an urban or rural zip code, respondent education, respondent employment status, an indicator for experience with severe COVID-19 illness (set to 1 if respondents reported that they, a family member, or a close friend have been hospitalized for COVID-19, or if respondents reported that a family member or close friend has died from COVID-19), respondents’ concern about the impact of COVID-19 on the family’s financial well-being, respondents’ perception of the risk that they or someone in their household will contract COVID-19 within the next 2 months, whether respondents believe the child has a health condition that increases the risk of severe COVID-19 illness (yes/no/unsure), and presence of an IEP.

In the respondent-level analysis, independent variables were identical, except that the child variables for grade, IEP, and perceived high-risk health condition were replaced with indicators for 1) having ≥1 child in kindergarten-2^nd grade, 3^rd-5^th grade, 6^th-8^th grade, and 9^th-12^th grade; 2) having ≥1 child with an IEP; 3) having ≥1 child with a perceived high-risk health condition; and 4) having ≥1 child that respondents were unsure about regarding presence of a high-risk health condition.

Statistical analysis.

We used descriptive statistics to assess sample characteristics. We compared the demographic characteristics of parents and guardians in the respondent-level analysis with weighted 2018 American Community Survey data from parents and guardians of children aged 5–17 years in Illinois, Michigan, and Ohio.¹⁸

In the child-level analysis, we fitted linear regression models with standard errors clustered at the level of respondents. We used linear regression to facilitate interpretation of coefficients as absolute percentage-point changes; results were similar when calculating average marginal effects from logistic regression models (Appendix 2). In the respondent-level analysis, we fitted linear regression models with robust standard errors.

Analyses were conducted using SAS 9.4 (SAS Institute) and Stata 15.1 SE (StataCorp). Significance was set at α = 0.05; hypothesis tests were two-sided.

Additional analyses.

We repeated the child-level analysis but replaced the variable for whether the child had a high-risk health condition with a variable for whether the respondent or any household member had a high-risk condition (Appendix 3). We assessed factors associated with the number of risk mitigation measures opposed and with support for each measure (Appendix 4–5). Finally, for children with selected high-prevalence health conditions, we calculated the proportion for whom in-person school attendance was likely (Appendix 6). We conducted these analyses for interested readers but do not discuss them owing to space limitations.

RESULTS

Sample characteristics.

We received data from 1,193 respondents. Of the 2,282 children of these respondents, we excluded 80 (3.5%) with missing data for school attendance plans or independent variables, leaving 2,202 children in the child-level analysis. Of the 1,193 eligible respondents, we excluded 66 (5.5%) with missing data for support for risk mitigation measures or independent variables, leaving 1,126 respondents in the respondent-level analysis.

Table 1 and Table 2 display characteristics of the samples for the child-level and respondent-level analysis, respectively. Overall, 19.3% of the 2,202 children in the child-level analysis had health conditions perceived to increase the risk of severe COVID-19 illness; 22.1% had an IEP. Compared with parents and guardians of school-aged children in Illinois, Michigan, and Ohio in the 2018 American Community Survey, the 1,126 respondents in the respondent-level analysis were slightly more likely to be aged 18–44 years (4.0 percentage points), female (4.9 percentage points), and white/non-Hispanic (2.1 percentage points); and more likely to have annual household incomes between $50,000-$99,999 (9.0 percentage points) (Appendix 7).

Table 1.

Factors associated with plans for in-person school attendance

Factor	Sample size	% of sample	In-person school attendance is likely (%)	Coefficient (95% CI)
State
Illinois	766	34.8	71.4	Reference
Michigan	677	30.7	72.5	−0.04 (−7.1, 7.0)
Ohio	759	34.5	69.2	−0.6 (−8.0, 6.7)
Child grade
Kindergarten-2nd grade	579	26.3	70.1	Reference
3rd−5th grade	448	20.3	67.6	−3.4 (−9.1, 2.3)
6th−8th grade	461	20.9	72.0	0.2 (−5.5, 5.9)
9th−12th grade	714	32.4	73.1	0.1 (−5.6, 5.8)
Respondent age
18–34 years	556	25.2	68.5	Reference
35–44 years	890	40.4	71.2	−2.4 (−10.0, 5.1)
45–54 years	637	28.9	76.6	1.3 (−7.2, 9.9)
55 years and above	119	5.4	50.4	−22.8 (−35.6, −10.1)
Respondent sex
Male	907	41.2	73.8	Reference
Female	1294	58.8	69.0	0.4 (−5.8, 6.6)
Other	1	0.0	100.0	Not shown
Respondent race/ethnicity
White/non-Hispanic	1567	71.2	75.5	Reference
Black/Non-Hispanic	314	14.3	59.6	−14.1 (−25.7, −2.6)
Hispanic, any race	165	7.5	65.5	−5.6 (−17.2, 6.1)
Asian, non-Hispanic	95	4.3	54.7	−16.8 (−31.3, −2.2)
Other, non-Hispanic (including multiracial)	61	2.8	54.1	−10.1 (−27.3, 6.9)
Household income
$0-$49,999	645	29.3	65.7	Reference
$50,000-$99,999	882	40.1	68.5	−2.1 (−9.6, 5.4)
$100,000 or more	675	30.7	79.3	2.6 (−6.0, 11.1)
Respondent political affiliation
Republican	647	29.4	79.8	Reference
Democrat	939	42.6	67.5	−7.6 (−14.7, −0.5)
Independent, other party, unaffiliated	616	28.0	67.0	−9.0 (−16.7, −1.2)
Urban/rural residence
Urban	2049	93.1	70.7	Reference
Rural	153	6.9	75.2	1.6 (−9.4, 12.7)
Respondent education
High school diploma/GED or less	418	19.0	65.3	Reference
Some college, associate’s or bachelor’s degree	1316	59.8	69.7	1.4 (−7.1, 9.9)
Master’s, professional, or doctorate degree	468	21.3	79.7	6.8 (−3.9, 17.5)
Respondent employment status
Employed full-time or part-time	1601	72.7	74.5	Reference
Stay-at-home parent/guardian	339	15.4	59.0	−13.1 (−22.2, −4.0)
Not working, retired, furloughed, student	262	11.9	65.3	−2.9 (−12.1, 6.3)
Experience with severe COVID-19 illness
No	1860	84.5	71.8	Reference
Yes	342	15.5	66.7	−0.2 (−8.4, 8.0)
Perceived chance that respondent or a family member in household will contract COVID-19 within the next 2 months
No or low chance	1600	72.7	74.8	Reference
Moderate or high chance	602	27.3	61.0	−10.1 (−16.7, −3.5)
Concern about impact of COVID-19 pandemic on family’s financial well-being
Not concerned	465	21.1	81.5	Reference
Somewhat concerned	1043	47.4	74.3	−3.9 (−10.9, 3.2)
Very concerned	694	31.5	58.9	−16.1 (−24.4, −7.9)
Child has a health condition perceived to increase the risk of severe COVID-19 illness
No	1704	77.4	73.8	Reference
Unsure	72	3.3	65.3	−3.4 (−15.0, 8.2)
Yes	426	19.3	60.8	−9.7 (−15.8, −3.6)
Child has individualized education plan
No	1716	77.9	70.4	Reference
Yes	486	22.1	73.0	6.8 (1.4, 12.3)

Coefficients represent absolute percentage-point differences in the probability of reporting that in-person school attendance for a child was likely (versus unlikely or being unsure of school attendance plans). Values for factors with cell sizes of 10 or lower are not shown.

Table 2.

Factors associated with number of COVID-19 risk mitigation measures supported

Factor	Sample size	% of sample	Mean number of measures supported	Coefficient (95% CI)
State
Illinois	399	35.4	8.7	Reference
Michigan	364	32.3	7.7	−0.6 (−1.2, −0.05)
Ohio	363	32.2	7.4	−0.7 (−1.3, −0.1)
≥1 child in kindergarten-2nd grade
No	676	60.0	8.0	Reference
Yes	450	40.0	7.9	−0.7 (−1.3, −0.1)
≥1 child in 3rd–5th grade
No	775	68.8	8.0	Reference
Yes	351	31.2	8.0	−0.4 (−0.9, 0.2)
≥1 child in 6th–8th grade
No	771	68.5	8.0	Reference
Yes	355	31.5	7.8	−0.7 (−1.2, −0.1)
≥1 child in 9th–12th grade
No	632	56.1	8.1	Reference
Yes	494	43.9	7.8	−0.6 (−1.2, −0.005)
Respondent age
18–34 years	285	25.3	7.8	Reference
35–44 years	442	39.3	8.0	0.2 (−0.4, 0.8)
45–54 years	327	29.0	7.9	0.2 (−0.5, 0.9)
55 years and above	72	6.4	8.5	0.9 (−0.1, 1.9)
Respondent sex
Male	449	39.9	8.6	Reference
Female	676	60.0	7.6	−0.9 (−1.4, −0.4)
Other	1	0.1	0.0	Not shown
Respondent race/ethnicity
White/non-Hispanic	816	72.5	7.5	Reference
Black/Non-Hispanic	146	13.0	8.9	0.9 (0.2, 1.6)
Hispanic, any race	84	7.5	9.3	1.1 (0.2, 2.0)
Asian, non-Hispanic	57	5.1	9.9	1.6 (0.4, 2.8)
Other, non-Hispanic	23	2.0	8.6	0.8 (−1.0, 2.6)
Household income
$0–$49,999	340	30.2	7.7	Reference
$50,000–$99,999	447	39.7	7.9	0.3 (−0.3, 1.0)
$100,000 or more	339	30.1	8.4	1.0 (0.2, 1.7)
Respondent political affiliation
Republican	333	29.6	6.7	Reference
Democrat	474	42.1	9.3	2.1 (1.5, 2.7)
Independent, other party, unaffiliated	319	28.3	7.4	0.5 (−0.1, 1.2)
Urban/rural residence
Urban	1058	94.0	8.1	Reference
Rural	68	6.0	6.0	−0.9 (−1.9, 0.1)
Respondent education
High school diploma/GED or less	196	17.4	7.2	Reference
Some college, associate’s or bachelor’s degree	711	63.1	7.9	0.6 (−0.1, 1.2)
Master’s, professional, or doctorate degree	219	19.4	8.9	1.1 (0.2, 2.0)
Respondent employment status
Employed full-time or part-time	820	72.8	8.1	Reference
Stay-at-home parent/guardian	167	14.8	7.5	0.3 (−0.4, 1.0)
Not working, retired, furloughed, student	139	12.3	7.8	0.1 (−0.6, 0.9)
Experience with severe COVID-19 illness
No	981	87.1	7.7	Reference
Yes	145	12.9	9.5	0.9 (0.2, 1.6)
Perceived chance that respondent or a family member in household will contract COVID-19 within the next 2 months
No or low chance	827	73.4	7.6	Reference
Moderate or high chance	299	26.6	8.9	0.7 (0.1, 1.2)
Concern about impact of COVID-19 pandemic on family’s financial well-being
Not concerned	242	21.5	6.0	Reference
Somewhat concerned	527	46.8	8.1	1.6 (1.0, 2.2)
Very concerned	357	31.7	9.1	2.1 (1.5, 2.8)
≥1 child for whom respondent is unsure about regarding presence of a high-risk health condition
No	1070	95.0	8.0	Reference
Yes	56	5.0	7.8	−0.2 (−1.3, 1.0)
≥1 child with a health condition perceived to increase risk of severe COVID-19 illness
No	843	74.9	7.4	Reference
Yes	283	25.1	9.6	1.4 (0.8, 2.0)
≥1 child with individualized education plan
No	824	73.2	7.7	Reference
Yes	302	26.8	8.7	0.6 (0.01, 1.2)

Coefficients represent absolute differences in the number of measures supported. Values for factors with cell sizes of 10 or lower are not shown.

Child-level analysis.

Of the 2,202 children, in-person school attendance was planned for 1,563 (71.0%). This proportion was higher for children of White/non-Hispanic respondents (75.5%) than for children of non-White respondents (54.1%−65.5%) (Table 1). For children in households with annual income of $0-$49,999 and ≥$100,000, in-person school attendance was planned for 65.7% and 79.3%, respectively.

Factors associated with a lower probability of plans for in-person school attendance included Black (−14.1 percentage points, 95% CI: −25.7, −2.6) and Asian race/ethnicity (−16.8, 95% CI: −31.3, −2.2) compared with White/non-Hispanic, presence of a perceived high-risk health condition (−9.7, 95% CI: −15.8, −3.6) compared with no high-risk condition, and employment status of stay-at-home parent/guardian compared with full-time/part-time employment (−13.1, 95% CI: −22.2, −4.0). Other factors associated with a lower probability included respondent age ≥ 55 years (vs 18–34 years), political affiliation other than Republican, belief that the chance of someone in the household contracting COVID-19 in the next 2 months was moderate or high (compared with no or low chance), and being very concerned about the impact of COVID-19 on the family’s financial well-being (compared with not concerned). Factors associated with a higher probability of plans for in-person school attendance included presence of an IEP (6.8 percentage points, 95% CI: 1.4, 12.3). Coefficients for selected factors are displayed in Figure 1. Notable factors unassociated with plans for in-person school attendance included urban/rural residence, household income, respondent education, and experience with severe COVID-19 illness.

Figure 1.

Selected factors associated with a lower probability of plans for in-person school attendance.

Coefficients represent absolute percentage point changes in the probability of in-person school attendance. Boxes represent point estimates and bars represent 95% confidence intervals.

Respondent-level analysis.

Unadjusted support for the 15 measures ranged from 26.6% to 76.9% with a median of 52.2% (Table 3). The measures with the lowest support and highest opposition were stopping all extracurricular school programs and closing playground structures. The measure with the highest support and lowest opposition was daily temperature screens of students. Support for mandatory face coverings for school staff and older students was higher than support for mandatory face coverings for younger students.

Table 3.

Unadjusted support for 15 school-based COVID-19 risk mitigation measures

	Support or strongly support(%)	Oppose or strongly oppose(%)	Neutral(%)
MEASURES TO LIMIT CONTACT BETWEEN STUDENTS
Decreasing the number of students allowed on a school bus	64.4	10.7	25.0
Having groups of students alternate between in-person and online classes to decrease the number of students in school at once	62.3	15.8	21.8
Staggering arrival and pick-up times for students	60.2	13.5	26.3
Requiring students to eat meals in classrooms instead of cafeterias	47.4	21.3	31.3
Closing playground structures	32.2	44.5	23.3
Stopping all extracurricular school programs, such as sports and music	26.6	49.6	23.8
MEASURES FOCUSED ON TESTING AND SCREENING
Conducting daily temperature screens of all students upon arrival at school	76.9	6.8	16.3
Requiring all students in a classroom to be tested for COVID-19 if a classmate tests positive for COVID-19	74.3	11.6	14.0
Testing a randomly selected group of school staff once per week for COVID-19	66.4	14.6	19.0
Testing a randomly selected group of students once per week for COVID-19	48.9	26.4	24.7
MEASURES FOCUSED ON FACE COVERINGS
Requiring school staff to wear a face covering	60.5	21.4	18.1
Requiring students in 6th grade and above to wear a face covering	52.2	27.5	20.2
Requiring students in 3rd−5th grade to wear a face covering	48.0	31.7	20.3
Requiring students in 1st−2nd grade to wear a face covering	40.2	35.2	24.6
Requiring students in kindergarten to wear a face covering	36.4	38.6	25.0

Rows may not sum to 100% due to rounding

Respondents supported a mean of 8.0 measures (SD 4.4) and a median of 8.0 (25^th-75^th percentile: 4–12). Factors associated with higher number of measures supported included Black, Hispanic, and Asian race/ethnicity (compared with White/Non-Hispanic), annual household income ≥ $100,000 (compared with $0-$49,999), Democratic political affiliation (compared with Republican), having a master’s/professional/doctorate degree (compared with high school diploma/GED or less), experience with severe COVID-19 illness, belief that the chance of someone in the household contracting COVID-19 in the next 2 months was moderate or high (compared with no or low chance), being somewhat or very concerned about the impact of COVID-19 on the family’s financial well-being (compared with not concerned), having ≥1 child with a perceived high-risk health condition, and having ≥1 child with an IEP (Table 2). Factors associated with lower numbers of measures supported included residence in Michigan or Ohio (compared with Illinois) and female sex. Among factors associated with the number of measures supported, the absolute magnitude of coefficients ranged from 0.6 to 2.1 measures. Coefficients for selected factors are displayed in Figure 2. Factors unassociated with the number of measures supported included respondent age, urban/rural residence, and employment status.

Figure 2.

Selected factors associated with support for a higher number of school-based COVID-19 risk mitigation measures.

Coefficients represent differences in the number of measures supported. Boxes represent point estimates and bars represent 95% confidence intervals.

DISCUSSION

In this analysis of a June 2020 survey of parents and guardians of public school children in Illinois, Michigan, and Ohio, we identified factors associated with plans for in-person school attendance during the 2020–2021 school year and with support for 15 school-based COVID-19 risk mitigation measures. While much has changed since the survey was fielded, our study provides some of the most detailed information on factors that may influence decision-making by parents and guardians when weighing the benefits and risks of in-person school attendance during pandemics, as well as on factors that may influence support for specific risk mitigation measures. This information may be useful in the short-term as schools plan transitions between remote, hybrid, and in-person learning during the remainder of the 2020–2021 school year, and as schools plan for COVID-19 during the upcoming 2021–2022 school year. Moreover, this information could be useful in the long-term when schools plan for other pandemics.

In our study, in-person school attendance was planned for 71.0% of respondents’ children. In adjusted analyses, such plans were 14.1–16.8 percentage points less likely among children of Black and Asian respondents compared with children of White/non-Hispanic respondents. For Black children, potential reasons for the lower probability of plans for in-person school attendance might include the high prevalence of health conditions that increase the risk of severe COVID-19 illness in Black families, such as diabetes.²⁶ Black and Asian families are also more likely to live in multi-generational households compared with White families.²⁷ Such households may include elderly family members who are at increased risk for severe COVID-19 illness. The presence of additional family members available to supervise children could also make parents and guardians more comfortable with remote learning. Our findings suggest that for families of Black and Asian children, it may be especially important for schools, state policymakers, and federal policymakers to overcome barriers to high-quality remote education and to implement risk mitigation strategies that meet families’ expectations for safe in-person education. Additionally, for children in these families, it may also be especially important for pediatric clinicians to closely monitor academic performance, optimize any physical or mental health conditions that may impede this performance, and screen for mental health or developmental problems that may arise from being kept away from school.²⁸

In unadjusted analyses, plans for in-person school attendance were more likely among children in households with annual incomes of $100,000 or greater compared with children in households with annual incomes between $0-$49,999 (79.7% vs 65.3%). In adjusted analyses, however, this difference did not persist. Findings suggest that household income may be a marker for other factors that drive school attendance plans, such as race/ethnicity.

Our results for income and race/ethnicity contrast with those from a national survey of 730 U.S. parents in early June 2020. In that survey, lower household income was associated with a lower probability of plans for in-person school attendance compared with higher household income, while race/ethnicity was not associated with plans.¹⁴ Potential reasons for this discrepancy include the regional nature of our survey and differences in independent variables. For example, our analyses accounted for political affiliation and urban/rural residence. Importantly, unadjusted differences in school attendance plans by household income were large in our study. Thus, findings still support policymakers’ concerns that existing systemic educational disparities by income and race/ethnicity could be exacerbated if remote learning is not optimally delivered to vulnerable children.^4,6

The presence of health conditions perceived to increase the risk of severe COVID-19 illness was associated with a 9.7 percentage point lower probability of plans for in-person school attendance. Notably, it is unclear which conditions increase the risk of severe COVID-19 illness in children, including common conditions such as asthma.²⁹ Pediatric clinicians caring for children with health conditions should assess families’ concerns regarding these conditions and ensure appropriate perceptions of risk..

Having an IEP was associated with a 6.8 percentage point higher adjusted probability of plans for in-person school attendance. This finding is not unexpected, as children with IEPs often receive school-based services that could be difficult to deliver remotely.³⁰ Additionally, employment status of stay-at-home parent or guardian was associated with a lower probability of plans for in-person school attendance compared with full-time or part-time employment, consistent with a prior study.¹⁴

We identified several factors associated with higher numbers of risk mitigation measures supported among parents and guardians, including Black, Hispanic, and Asian race/ethnicity, high household income, advanced education, Democratic political affiliation, experience with severe COVID-19 illness, and having children with perceived high-risk health conditions. However, these associations were modest in magnitude compared with the number of measures supported by the average parent or guardian in our analysis (8.0). Thus, while variation exists, our findings suggest that most parents and guardians want schools to take a substantial number of steps to mitigate COVID-19 risk.

This study has limitations. First, events since the survey was fielded – such as the nationwide increase in COVID-19 cases³¹, outbreaks of COVID-19 at camps and schools^32,33, and the release of state and school plans on re-opening² – may have changed plans for in-person school attendance and support for risk mitigation measures among parents and guardians. Moreover, many districts in Illinois, Michigan, and Ohio have offered in-person learning^34–36, and early experiences with this learning may have shaped the perceptions of parents and guardians regarding the need for risk mitigation measures. However, many of the associations we demonstrate have been borne out so far. For example, 41% of Black students in Denver Public Schools have chosen remote learning, compared with 22% of White students, findings that are consistent with patterns in other locations.^37–39 Researchers should continue to assess the views of parents and guardians throughout the COVID-19 pandemic; we hope the methods used in this study can facilitate these assessments.

Second, we analyzed a convenience sample of respondents paid to take online surveys. Although our sample was similar to all parents and guardians in Illinois, Michigan, and Ohio on several key demographic characteristics, it slightly over-represented middle-income families. Notably, household income was not associated with school attendance plans in adjusted analyses, although it was associated with the number of risk mitigation measures supported.

Third, generalizability of results to other states is unclear. For example, the proportion of children with an IEP in our study (22.1%) was higher than the proportion nationally (14%).⁴⁰ As having an IEP was associated with a higher likelihood of plans for in-person school attendance, the unadjusted proportion of children for whom in-person school attendance was planned in this survey may be higher than the corresponding proportion for all U.S. children.

Finally, we did not collect information on all factors that might be associated with school attendance plans and support for risk mitigation measures, including average household income in respondents’ school districts, which may be correlated with the ability to implement COVID-19 risk mitigation measures. Additionally, we did not assess the degree to which parents and guardians trusted their school districts and other families to take appropriate precautions to mitigate COVID-19 risk.

CONCLUSIONS

During the COVID-19 pandemic, pediatric clinicians should advocate to ensure that districts have the resources to make in-person learning as safe as possible. Additionally, they should screen for academic challenges among children, including those undergoing remote learning. Importantly, in-person pediatric visits have already decreased during the pandemic and may decrease further if COVID-19 cases continue to rise.^31,41 Consequently, pediatric clinicians should consider routinely screening for academic challenges during all visits, including virtual visits, for example by incorporating this screening during the check-in process.