Source Code for Household COVID-19 risk and In-person schooling

Justin Lessler, M. Kate Grabowski, Kyra H. Grantz, Elena Badillo-Goicoechea, C. Jessica E. Metcalf, Carly Lupton-Smith, Andrew S. Azman, Elizabeth A. Stuart

Code is provided to reproduce the primary in-person schooling (1-Primary-Schooling-Analysis.Rmd) and occupation/work outside home (2-Educator-Analysis.Rmd) analyses using a synthetic dataset. Hence, results produced when knitting these documents will not match those presented in the accompanying paper.

Data to reproduce paper figures/tables are freely available from the CMU Delphi Research Group to researchers at universities and non-profits as detailed at Getting Data Access - Delphi Epidata API (cmu-delphi.github.io).

Preprint: Household COVID-19 risk and in-person schooling

Please reach out to justin[at]jhu.edu if you have any questions.

Main text figures

Figure 1

Spatial distribution of survey responses. (A) Number of survey respondents reporting a school age student in the household by county. (B) Percentage of households with school age children reporting any in-person schooling by county, excluding counties with fewer than 10 responses (excluded counties in dark grey). (C) Percentage of households with a child in in-person schooling reporting full-time in-person schooling, excluding counties with fewer than 10 reporting in-person schooling, (D) Average number of school-based mitigation measures reported for children with in-person schooling, excluding counties with fewer than 10 reporting in-person schooling.

Spatial distribution of survey responses. (A) Number of survey respondents reporting a school age student in the household by county. (B) Percentage of households with school age children reporting any in-person schooling by county, excluding counties with fewer than 10 responses (excluded counties in dark grey). (C) Percentage of households with a child in in-person schooling reporting full-time in-person schooling, excluding counties with fewer than 10 reporting in-person schooling, (D) Average number of school-based mitigation measures reported for children with in-person schooling, excluding counties with fewer than 10 reporting in-person schooling.

Figure 2

Risk from in-person schooling and distribution of mitigation measures by grade. (A) Odds ratio of COVID-19-related outcomes associated with full- and part-time in-person schooling by outcome and grade level, compared to individuals with children in their household not attending in-person schooling and adjusted for individual- and county-level covariates (but not number of mitigation measures) indicating that the strength of the association increases with grade level. (B) Distribution of mitigation measures by grade level and full- versus part-time in-person status across all grades.

Risk from in-person schooling and distribution of mitigation measures by grade. (A) Odds ratio of COVID-19-related outcomes associated with full- and part-time in-person schooling by outcome and grade level, compared to individuals with children in their household not attending in-person schooling and adjusted for individual- and county-level covariates (but not number of mitigation measures) indicating that the strength of the association increases with grade level. (B) Distribution of mitigation measures by grade level and full- versus part-time in-person status across all grades.

Table S4:

Adjusted and unadjusted odds ratio (OR) of COVID-19 among those engaged in full- and part-time in-person schooling versus those engaged in virtual or homeschooling outcomes. All analyses adjust for survey weights. See methods for adjustment factors.

Schooling Loss of taste/smell COVID like illness Test+
adj. OR 95% CI OR 95% CI adj OR 95% CI OR 95% CI adj OR 95% CI OR 95% CI
Overall
Full time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_full_binTRUE 1.21 1.16–1.27 1.46 1.41–1.52 1.38 1.30–1.47 2.21 2.10–2.32 1.30 1.24–1.35 1.07 1.03–1.11
Part time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_part_binTRUE 1.18 1.13–1.24 1.37 1.32–1.43 1.21 1.13–1.29 1.91 1.81–2.01 1.09 1.03–1.14 0.96 0.92–1.00
Grades K or under
Full time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_full_binTRUE 1.11 0.97–1.27 0.93 0.83–1.04 1.16 0.95–1.42 1.15 0.96–1.38 1.21 1.06–1.38 0.87 0.78–0.98
Part time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_part_binTRUE 0.95 0.80–1.12 0.84 0.72–0.98 0.90 0.70–1.16 0.84 0.66–1.07 1.06 0.90–1.25 0.80 0.69–0.93
Grades 1 to 5
Full time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_full_binTRUE 1.19 1.07–1.33 1.10 1.00–1.21 1.12 0.93–1.35 1.13 0.97–1.31 1.31 1.17–1.46 0.95 0.86–1.05
Part time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_part_binTRUE 1.15 1.01–1.32 1.06 0.93–1.20 1.34 1.10–1.62 1.33 1.11–1.59 1.01 0.88–1.16 0.83 0.72–0.94
Grades 6 to 8
Full time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_full_binTRUE 1.30 1.12–1.52 1.20 1.05–1.37 1.42 1.14–1.78 1.41 1.16–1.72 1.38 1.19–1.61 1.14 1.00–1.30
Part time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_part_binTRUE 1.24 1.04–1.48 1.09 0.93–1.28 1.43 1.10–1.87 1.30 1.03–1.65 1.05 0.88–1.27 0.88 0.74–1.03
Grades 9 to 12
Full time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_full_binTRUE 1.36 1.20–1.53 1.31 1.18–1.46 1.64 1.39–1.93 1.85 1.60–2.15 1.53 1.37–1.72 1.19 1.07–1.31
Part time in person NA NA NA NA NA NA NA NA NA NA NA NA
Child_IP_part_binTRUE 1.12 0.98–1.27 0.96 0.86–1.08 1.00 0.84–1.19 1.01 0.86–1.19 1.10 0.98–1.24 0.87 0.79–0.97

Figure 3

Impact of individual mitigation measures. (A) Relationship between number of mitigation measures and percent reporting COVID-19-related outcomes using a log-linear (solid) and spline (dashed) model. (B) Odds ratio of COVID-19-related outcomes by mitigation measure in multivariable model including all measures, versus the reduction due to a generic mitigation measure (dotted line).

Impact of individual mitigation measures. (A) Relationship between number of mitigation measures and percent reporting COVID-19-related outcomes using a log-linear (solid) and spline (dashed) model. (B) Odds ratio of COVID-19-related outcomes by mitigation measure in multivariable model including all measures, versus the reduction due to a generic mitigation measure (dotted line).

Table S9

The association between mitigation measures and the odds-ratio of COVID-19-related outcomes in those engaged in in-person schooling compared to those engaged in virtual- or home-schooling, adjusted for county and individual level covariates.

Mitigatoin Loss of taste/smell COVID like illness Test+
adj. OR 95% CI adj OR 95% CI adj OR 95% CI
student mask mandate NA NA NA NA NA NA
sch_maskMand_stTRUE 0.91 0.84–0.99 0.91 0.81–1.01 0.89 0.81–0.97
teacher mask mandate NA NA NA NA NA NA
sch_maskMand_tchTRUE 0.82 0.76–0.89 0.66 0.59–0.74 0.91 0.83–0.99
same teacher all day NA NA NA NA NA NA
sch_sameTchTRUE 1.05 0.97–1.13 0.94 0.84–1.06 1.00 0.93–1.08
same students all day NA NA NA NA NA NA
sch_sameStTRUE 0.88 0.81–0.95 0.87 0.77–0.98 0.92 0.85–1.00
outdoor instruction NA NA NA NA NA NA
sch_outdoorInstrTRUE 0.89 0.80–0.98 1.02 0.89–1.18 0.88 0.79–0.97
restricted entry NA NA NA NA NA NA
sch_restEntryTRUE 0.94 0.88–1.02 0.92 0.82–1.03 0.88 0.81–0.95
reduced class size NA NA NA NA NA NA
sch_redClassSizeTRUE 0.95 0.89–1.02 0.94 0.85–1.04 1.01 0.94–1.09
closed cafeteria NA NA NA NA NA NA
sch_closedCafeTRUE 1.00 0.93–1.08 1.03 0.92–1.16 1.03 0.96–1.11
closed playground NA NA NA NA NA NA
sch_closedPlayTRUE 1.01 0.92–1.10 1.12 0.99–1.27 1.01 0.92–1.10
desk shields NA NA NA NA NA NA
sch_deskShieldsTRUE 1.07 0.99–1.15 1.29 1.15–1.44 1.13 1.04–1.22
extra desk space NA NA NA NA NA NA
sch_xdeskSpaceTRUE 0.92 0.86–1.00 0.93 0.84–1.04 0.97 0.89–1.05
no extracurriculars NA NA NA NA NA NA
sch_noExtraCurrTRUE 0.84 0.78–0.91 0.86 0.77–0.96 0.73 0.68–0.79
no sharing supplies NA NA NA NA NA NA
sch_noShareSuppTRUE 0.89 0.83–0.96 0.90 0.81–1.01 0.93 0.86–1.00
daily symptom screen NA NA NA NA NA NA
sch_dailySymptScrTRUE 0.82 0.76–0.87 0.75 0.68–0.83 0.78 0.73–0.84
partime in-person NA NA NA NA NA NA
Child_IP_part_binTRUE 1.12 1.05–1.19 1.08 0.99–1.18 0.97 0.91–1.03

Figure 4

Risk of in-person schooling by strata of number of reported mitigation measures. (A) Estimated risk associated with full- and part-time in-person schooling by outcome and number of mitigation measures implemented, adjusted for individual and county-level covariates. (B) Distribution of mitigation measures by total number of measures implemented.

Risk of in-person schooling by strata of number of reported mitigation measures. (A) Estimated risk associated with full- and part-time in-person schooling by outcome and number of mitigation measures implemented, adjusted for individual and county-level covariates. (B) Distribution of mitigation measures by total number of measures implemented.

Table S8:

Association between risk of full-time schooling and number of mitigation measures implemented after accounting for survey design and adjusting for county and individual level covariates.

Schooling Loss of taste/smell COVID like illness Test+
adj. OR 95% CI adj OR 95% CI adj OR 95% CI
Overall
N mitigation measures 0.92 0.91–0.93 0.91 0.89–0.92 0.93 0.92–0.94
Grades K or under
N mitigation measures 0.96 0.93–0.98 0.91 0.88–0.95 0.96 0.94–0.98
Grades 1 to 5
N mitigation measures 0.94 0.93–0.96 0.94 0.91–0.96 0.97 0.95–0.99
Grades 6 to 8
N mitigation measures 0.92 0.90–0.95 0.92 0.88–0.96 0.93 0.90–0.95
Grades 9 to 12
N mitigation measures 0.93 0.91–0.95 0.91 0.88–0.94 0.94 0.92–0.96

Figure 5

Sub-group analysis of association between in-person schooling and COVID-19-related outcomes. Estimated odds ratio (versus those in strata not reporting in-person schooling) of COVID-19-related outcomes from full-time (circles, dashed lines) and part-time (triangles, dotted line) in-person schooling when data is stratified by (A) county population size and relation to metropolitan areas (metropolitan area, non-metropolitan area, adjacent to metropolitan area), (B) quintile of incidence (Q1 is lowest, Q5 is highest) and (C) propensity to report in-person schooling (Q5 most likely to have in-person schooling, Q1 least likely). Horizontal dashed and dotted lines show overall point estimates for full-time and part-time in-person instruction, respectively.

Sub-group analysis of association between in-person schooling and COVID-19-related outcomes. Estimated odds ratio (versus those in strata not reporting in-person schooling) of COVID-19-related outcomes from full-time (circles, dashed lines) and part-time (triangles, dotted line) in-person schooling when data is stratified by (A) county population size and relation to metropolitan areas (metropolitan area, non-metropolitan area, adjacent to metropolitan area), (B) quintile of incidence (Q1 is lowest, Q5 is highest) and (C) propensity to report in-person schooling (Q5 most likely to have in-person schooling, Q1 least likely). Horizontal dashed and dotted lines show overall point estimates for full-time and part-time in-person instruction, respectively.

Supplementary Figures

Figure S1

Distribution of outcomes in first period (left column), second period (center column) and change over time (right column). Results are shown for (A) number of survey respondents reporting ≧1 school-aged child in the household, (B) percent reporting in-person schooling, (C) percent reporting full-time in-person schooling, and (D) average number of in-school mitigation measures.

Distribution of outcomes in first period (left column), second period (center column) and change over time (right column). Results are shown for (A) number of survey respondents reporting ≧1 school-aged child in the household, (B) percent reporting in-person schooling, (C) percent reporting full-time in-person schooling, and (D) average number of in-school mitigation measures.

Figure S2

Survey respondents reporting school age children in the household per 100,000 respondents by state.

Survey respondents reporting school age children in the household per 100,000 respondents by state.

Figure S3

Distribution of selected county-level factors among participants reporting ≧1 school-aged child in the household by schooling type: no in-person schooling (none), part-time in-person schooling (Part), and full-time in-person schooling (Full); outlier values are excluded. Results are shown for county population size (A), percentage of county population that is white (B) percentage with income level under the appropriate poverty threshold (C) Gini index of income inequality (D) and average biweekly SARS-CoV-2 attack rate per 1000 persons (E).

Distribution of selected county-level factors among participants reporting ≧1 school-aged child in the household by schooling type: no in-person schooling (none), part-time in-person schooling (Part), and full-time in-person schooling (Full); outlier values are excluded. Results are shown for county population size (A), percentage of county population that is white (B) percentage with income level under the appropriate poverty threshold (C) Gini index of income inequality (D) and average biweekly SARS-CoV-2 attack rate per 1000 persons (E).

Figure S4

Rates of COVID-19 related outcomes reported by study respondents versus average 2-week county level attack rate over response period calculated from reported confirmed cases among respondents with Pre K-12 children. Limited to counties with 50 or more respondents. Blue lines indicate fit smooth spline relationships. Correlations are Pearson correlation coefficients, and slopes come from a simple linear regression.

Rates of COVID-19 related outcomes reported by study respondents versus average 2-week county level attack rate over response period calculated from reported confirmed cases among respondents with Pre K-12 children. Limited to counties with 50 or more respondents. Blue lines indicate fit smooth spline relationships. Correlations are Pearson correlation coefficients, and slopes come from a simple linear regression.

Figure S5

Rates of COVID-19 related outcomes reported by study respondents versus average 2-week county level attack rate over response period calculated from reported confirmed cases among all respondents. Limited to counties with 50 or more respondents. Blue lines indicate fit smooth spline relationships. Correlations are Pearson correlation coefficients, and slopes come from a simple linear regression.

Rates of COVID-19 related outcomes reported by study respondents versus average 2-week county level attack rate over response period calculated from reported confirmed cases among all respondents. Limited to counties with 50 or more respondents. Blue lines indicate fit smooth spline relationships. Correlations are Pearson correlation coefficients, and slopes come from a simple linear regression.

Fig S6

Correlation and hierarchical clustering between school mitigation measures among participants with ≥1 school-aged child in the household attending any in-person schooling (part-time or full-time). Shading (z-axis) indicates correlation (range 0-1) between reported mitigation measures.

Correlation and hierarchical clustering between school mitigation measures among participants with ≥1 school-aged child in the household attending any in-person schooling (part-time or full-time). Shading (z-axis) indicates correlation (range 0-1) between reported mitigation measures.

Figure S7

Percent of households with ≧1 child attending in-person school reporting each mitigation measure. Counties with less than 10 in-person respondents are excluded (gray).

Percent of households with ≧1 child attending in-person school reporting each mitigation measure. Counties with less than 10 in-person respondents are excluded (gray).

Figure S8

Predicted risk of COVID-19 related outcomes among those living with a child engaged in in-person schooling from model with only number of mitigation measures versus model with effects calculated for each individual mitigation measure, for reporting a positive COVID test (A,D), COVID-like illness (B,E) or loss of taste or smell (C,F). The top row (A-C) shows estimated logit probability of each outcome from a model with just number of mitigation measures versus a more complete model. The bottom row (D-F) shows boxplots of the relative odds of COVID-19 illness from the number of mitigations model versus the more complete model, by number of mitigation measures implemented.

Predicted risk of COVID-19 related outcomes among those living with a child engaged in in-person schooling from model with only number of mitigation measures versus model with effects calculated for each individual mitigation measure, for reporting a positive COVID test (A,D), COVID-like illness (B,E) or loss of taste or smell (C,F). The top row (A-C) shows estimated logit probability of each outcome from a model with just number of mitigation measures versus a more complete model. The bottom row (D-F) shows boxplots of the relative odds of COVID-19 illness from the number of mitigations model versus the more complete model, by number of mitigation measures implemented.

Figure S9

Distribution of selected county-level factors among participants reporting ≧1 school-aged child in the household attending in-person school (part-time or full-time) by the number of reported mitigation measures in the school; outlier values are excluded. Results are shown for county population size (A), percentage of county population that is white (B), percentage with income level under the appropriate poverty threshold (C) Gini index of income inequality (D) and average biweekly SARS-CoV-2 attack rate per 1000 persons (E).

Distribution of selected county-level factors among participants reporting ≧1 school-aged child in the household attending in-person school (part-time or full-time) by the number of reported mitigation measures in the school; outlier values are excluded. Results are shown for county population size (A), percentage of county population that is white (B), percentage with income level under the appropriate poverty threshold (C) Gini index of income inequality (D) and average biweekly SARS-CoV-2 attack rate per 1000 persons (E).

Figure S10

## Error: vector memory exhausted (limit reached?)
## Error: vector memory exhausted (limit reached?)

Figure S11

Sub-group analysis of association between in-person schooling and COVID-19-related outcomes. Estimated odds ratio (versus those in strata not reporting in-person schooling) of COVID-19-related outcomes from full-time (circles, dashed lines) and part-time (triangles, dotted line) in-person schooling when data is stratified by (A) percent white, (B) percent poverty percent of households with access to broad band internet. Horizontal dashed and dotted lines show overall point estimates for full-time and part-time in-person instruction, respectively.

Sub-group analysis of association between in-person schooling and COVID-19-related outcomes. Estimated odds ratio (versus those in strata not reporting in-person schooling) of COVID-19-related outcomes from full-time (circles, dashed lines) and part-time (triangles, dotted line) in-person schooling when data is stratified by (A) percent white, (B) percent poverty percent of households with access to broad band internet. Horizontal dashed and dotted lines show overall point estimates for full-time and part-time in-person instruction, respectively.

Figure S12

## tst2 - MI 
## tst2 - MA 
## tst2 - WA 
## tst2 - AZ 
## tst2 - OR 
## tst2 - IL 
## tst2 - OK 
## tst2 - FL 
## tst2 - VA 
## tst2 - CA 
## tst2 - AK 
## tst2 - SC 
## tst2 - TN 
## tst2 - OH 
## tst2 - NC 
## tst2 - PA 
## tst2 - WV 
## tst2 - AR 
## tst2 - TX 
## tst2 - UT 
## tst2 - MS 
## tst2 - KY 
## tst2 - MT 
## tst2 - NJ 
## tst2 - NY 
## tst2 - GA 
## tst2 - NV 
## tst2 - IA 
## tst2 - NH 
## tst2 - IN 
## tst2 - CO 
## tst2 - LA 
## tst2 - CT 
## tst2 - MN 
## tst2 - VT 
## tst2 - HI 
## tst2 - KS 
## tst2 - MO 
## tst2 - MD 
## tst2 - RI 
## tst2 - NM 
## tst2 - ID 
## tst2 - DE 
## tst2 - ME 
## tst2 - NE 
## tst2 - WI 
## tst2 - AL 
## tst2 - DC 
## tst2 - ND 
## tst2 - SD 
## tst2 - WY 
## cli - MI 
## cli - MA 
## cli - WA 
## cli - AZ 
## cli - OR 
## cli - IL 
## cli - OK 
## cli - FL 
## cli - VA 
## cli - CA 
## cli - AK 
## cli - SC 
## cli - TN 
## cli - OH 
## cli - NC 
## cli - PA 
## cli - WV 
## cli - AR 
## cli - TX 
## cli - UT 
## cli - MS 
## cli - KY 
## cli - MT 
## cli - NJ 
## cli - NY 
## cli - GA 
## cli - NV 
## cli - IA 
## cli - NH 
## cli - IN 
## cli - CO 
## cli - LA 
## cli - CT 
## cli - MN 
## cli - VT 
## cli - HI 
## cli - KS 
## cli - MO 
## cli - MD 
## cli - RI 
## cli - NM 
## cli - ID 
## cli - DE 
## cli - ME 
## cli - NE 
## cli - WI 
## cli - AL 
## cli - DC 
## cli - ND 
## cli - SD 
## cli - WY 
## cli2 - MI 
## cli2 - MA 
## cli2 - WA 
## cli2 - AZ 
## cli2 - OR 
## cli2 - IL 
## cli2 - OK 
## cli2 - FL 
## cli2 - VA 
## cli2 - CA 
## cli2 - AK 
## cli2 - SC 
## cli2 - TN 
## cli2 - OH 
## cli2 - NC 
## cli2 - PA 
## cli2 - WV 
## cli2 - AR 
## cli2 - TX 
## cli2 - UT 
## cli2 - MS 
## cli2 - KY 
## cli2 - MT 
## cli2 - NJ 
## cli2 - NY 
## cli2 - GA 
## cli2 - NV 
## cli2 - IA 
## cli2 - NH 
## cli2 - IN 
## cli2 - CO 
## cli2 - LA 
## cli2 - CT 
## cli2 - MN 
## cli2 - VT 
## cli2 - HI 
## cli2 - KS 
## cli2 - MO 
## cli2 - MD 
## cli2 - RI 
## cli2 - NM 
## cli2 - ID 
## cli2 - DE 
## cli2 - ME 
## cli2 - NE 
## cli2 - WI 
## cli2 - AL 
## cli2 - DC 
## cli2 - ND 
## cli2 - SD 
## cli2 - WY
The relationship between full- and part-time in-person schooling and household COVID-19 outcomes stratified by state and adjusted for individual level factors. Vermont and Washington DC excluded due to model non-convergence. Results are summarized in Table S12.

The relationship between full- and part-time in-person schooling and household COVID-19 outcomes stratified by state and adjusted for individual level factors. Vermont and Washington DC excluded due to model non-convergence. Results are summarized in Table S12.

Figure S13

Association between full- and part-time in-person schooling and COVID-19 related outcomes by quintile of average reported number of interventions in county (restricted to counties with 50 or more responses).

Association between full- and part-time in-person schooling and COVID-19 related outcomes by quintile of average reported number of interventions in county (restricted to counties with 50 or more responses).

Figure S14

Association between full- and part-time in-person schooling and COVID-19 related outcomes by generalized propensity score (i.e., predicted number of reported mitigations if in in person school) strata.

Association between full- and part-time in-person schooling and COVID-19 related outcomes by generalized propensity score (i.e., predicted number of reported mitigations if in in person school) strata.

Figure S15: See 2-EducatorAnalysis.Rmd

Figure S16

Average, period specific, bi-weekly county level attack rate from reported confirmed cases versus average number of interventions reported (limited to county-periods where at least 10 individuals reporting interventions)

Average, period specific, bi-weekly county level attack rate from reported confirmed cases versus average number of interventions reported (limited to county-periods where at least 10 individuals reporting interventions)

Figure S17

## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
## [1] "Pre_K"
## [1] "Grades_15"
## [1] "Grades_68"
## [1] "Grades_912"
Association between full- and part-time in person schooling and COVID-19 related outcomes overall and by grade, stratified by survey period.

Association between full- and part-time in person schooling and COVID-19 related outcomes overall and by grade, stratified by survey period.

Figure S18

Estimated effect of mitigation-measures versus the “average” mitigation measure (dotted line), stratified by time period.

Estimated effect of mitigation-measures versus the “average” mitigation measure (dotted line), stratified by time period.

Figure S19

## [1] "nov-dec"
## [1] "post_dec"
Association between full- and part-time in-person schooling and COVID-19 related outcomes, stratified by number of mitigation measures and period.

Association between full- and part-time in-person schooling and COVID-19 related outcomes, stratified by number of mitigation measures and period.

Figure S20

Note: due to small sample size in the randomly-generate sample data, results are not shown here for non-indicated test positive outcomes. Uncomment lines 1371, 1407, 1506-1535, 1549-1552 in the Rmd to run with all outcomes with full data.

Odds ratio of secondary COVID-19-related outcomes associated with full- and part-time in-person schooling by outcome and grade level, adjusted for individual and county level covariates (but not number of mitigation measures). Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Odds ratio of secondary COVID-19-related outcomes associated with full- and part-time in-person schooling by outcome and grade level, adjusted for individual and county level covariates (but not number of mitigation measures). Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Figure S21

Relationship between number of mitigation measures and proportion reporting secondary COVID-19-related outcomes using a log-linear (solid) and spline (dashed) model. (B) Adjusted odds ratio of COVID-19-related outcomes by mitigation measure in multivariate model including all measures. Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Relationship between number of mitigation measures and proportion reporting secondary COVID-19-related outcomes using a log-linear (solid) and spline (dashed) model. (B) Adjusted odds ratio of COVID-19-related outcomes by mitigation measure in multivariate model including all measures. Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Figure S22

Odds ratio of secondary COVID-19-related outcomes associated with in-person full-time and part-time schooling by number of mitigation measures implemented, adjusted for individual and county-level covariates. An outlier value for the odds ratio of non-indicated positive test result among part-time in-person schooling with zero mitigation measures, for which the sample size was <30, is excluded. Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Odds ratio of secondary COVID-19-related outcomes associated with in-person full-time and part-time schooling by number of mitigation measures implemented, adjusted for individual and county-level covariates. An outlier value for the odds ratio of non-indicated positive test result among part-time in-person schooling with zero mitigation measures, for which the sample size was <30, is excluded. Note that wide confidence intervals for non-indicated tests are due to low numbers (see Table S2).

Figure S23

Comparison of main study related outcomes with coefficients estimated for the relationship between in-person schooling and hypertension.

Comparison of main study related outcomes with coefficients estimated for the relationship between in-person schooling and hypertension.

Figure S24

Association between full- and part-time schooling with a given number of mitigation measures with Blood Pressure included.

Association between full- and part-time schooling with a given number of mitigation measures with Blood Pressure included.

Figure S25

Variable importance, as mean decrease in impurity, for random forests of propensity for in-person schooling. Variables with greater influence on the propensity score have higher values.

Variable importance, as mean decrease in impurity, for random forests of propensity for in-person schooling. Variables with greater influence on the propensity score have higher values.

Figure S26

##  [1] "State"                   "OccupationRed"          
##  [3] "IsMale"                  "Age"                    
##  [5] "Employed"                "num_kid_recode"         
##  [7] "HH_sz"                   "Educational_Level"      
##  [9] "mask_level"              "trav_out_state"         
## [11] "bar_rest_cafe_activ"     "large_event_activ"      
## [13] "pub_transit_activ"       "ext_person_activ"       
## [15] "mkt_grocery_pharm_activ" "work_outside_activ"     
## [17] "Pre_K"                   "Grades_15"              
## [19] "Grades_68"               "Grades_912"             
## [21] "rel_avg_biwk_AR"         "population"             
## [23] "White"                   "Black"                  
## [25] "GINI"                    "BBInternet.HH"          
## [27] "PopPerSqMile"            "Poverty"                
## [29] "NoHealthIns"             "Employed.Essential"     
## [31] "Computer.HH"             "Description"            
## [33] "UrbanRural"
Variable importance, as mean decrease in impurity, for random forests of propensity for in-person schooling. Variables with greater influence on the propensity score have higher values.

Variable importance, as mean decrease in impurity, for random forests of propensity for in-person schooling. Variables with greater influence on the propensity score have higher values.

Supplementary Tables

Note that supplementary tables are not printed here, but can be found in the working directory as tables/TableSX.csv.

Table S1

Selected sociodemographics among participants with ≥1 school-aged child in the household comparing those reporting no in-person schooling to those reporting any part-time and full-time in-person schooling; observed and survey-weighted percentages reported.

Table S2

Selected behaviors relevant to COVID-19 acquisition/transmission and COVID-19-related outcomes among participants with ≥1 school-aged child in the household comparing those reporting no in-person schooling to those reporting any part-time and full-time in-person schooling; observed and survey-weighted percentages reported.

Table S3

Survey-weighted percentage of participants with ≥1 school-aged child in the household by reported schooling type and state.

Table S4: see above

Table S5

Survey-weighted percentages of participants reporting school mitigation measures among those with ≥1 school-aged child in the household attending any in-person school (part-time or full-time) by state.

Table S6

Loadings of school intervention/mitigation measure variables onto principal components (first 9 components explaining ~80% of total variance shown only). Analysis restricted to participants with ≥1 school-aged child in the household attending any in-person schooling (part-time or full-time).

Table S7

Survey-weighted percentage of participants with ≥1 school-aged child attending any in-person school (part-time or full-time) by the reported number of school mitigation measures and state.

Table S8: see above

Table S9: see above

Table S10:

Selected demographics among participants with ≥1 school-aged child in the household attending any in-person schooling (part-time or full-time) by number of reported school mitigation measures; observed and survey-weighted percentages reported.

Table S11

Selected behaviors relevant to COVID-19 acquisition/transmission among participants with ≥1 school-aged child in the household attending any in-person schooling (part-time or full-time) by number of reported school mitigation measures; observed and survey-weighted percentages reported.

Table S12

Summary statistics for estimates of the effect of full- and part-time schooling stratified to the state level.

## # A tibble: 6 x 5
## # Groups:   variable [2]
##   variable          outcome             pct_pos pct_sig_pos ci_cov
##   <chr>             <chr>                 <dbl>       <dbl>  <dbl>
## 1 Child_IP_full_bin COVID like illness    0.837       0.286  0.857
## 2 Child_IP_full_bin Loss of taste/smell   0.776       0.224  0.898
## 3 Child_IP_full_bin Overall Test+         0.755       0.184  0.816
## 4 Child_IP_part_bin COVID like illness    0.673       0.163  0.959
## 5 Child_IP_part_bin Loss of taste/smell   0.714       0.286  0.918
## 6 Child_IP_part_bin Overall Test+         0.796       0.102  0.939

Table S13: see 2-EducatorAnalysis.Rmd