Skip to main content
. 2021 Nov 9;7(11):e29319. doi: 10.2196/29319

Table 4.

Logistic regression model of associations between substance use and any COVID-19 testing, accounting for covariates in the full sample of survey participants (N=1084).a

Variable Unadjusted prevalence and percentage testing within category, n (%) Odds ratio (95% CI)
Age N/Ab 1.01 (1.00-1.02)
Education

Graduate or professional degree 64 (29.8) Refc

College graduate/4-year degree 162 (29.6) 0.70 (0.47-1.05)

High school or less 19 (17.8) 0.45 (0.23-0.86)

Some college/2-year degree 34 (15.8) 0.41 (0.24-0.69)
Hispanic or Latino 118 (42.8) 1.67 (1.18-2.37)
Not Hispanic or Latino 161 (19.9) Ref
Racial identity

White 165 (23.2) Ref

Asian 9 (11.7) 0.66 (0.31-1.44)

Black or African American 89 (37.2) 1.60 (1.12-2.30)

Other or more than one 16 (27.6) 1.26 (0.63-2.52)
Essential worker 164 (42.3) 2.22 (1.60-3.09)
Not essential worker 115 (16.5) Ref
Income N/A 0.85 (0.77-0.94)
Household size N/A 1.33 (1.17-1.50)
Opioid use

Daily versus none 32d (69.6d) 3.35 (1.63-6.86)

1-6 days versus none 93e (51.7e) 2.08 (1.39-3.11)

Daily versus 1-6 days 154f(18.0f) 1.61 (0.78-3.34)

aItalicized text indicates reference groups and additional models for substance use variables where reference groups were switched to allow for additional comparisons. Unadjusted prevalence values are provided for categorical variables only.

bN/A: not applicable.

cRef: reference category.

dDaily.

e1-6 days.

fNone.