Table 2. Association of Social Vulnerability Index Measures With Change in Weekly Cumulative COVID-19 Incidence and Mortality Rates.
| Variable | COVID-19 incidence ratea,b | COVID-19 mortality ratea,c | ||
|---|---|---|---|---|
| IRR (95%CI) | P value | IRR (95%CI) | P value | |
| Overall SVI | ||||
| Time, per wk | 1.26 (1.25-1.26) | <.001 | 1.20 (1.19-1.20) | <.001 |
| Overall SVI indexd | 1.12 (1.10-1.13) | <.001 | 1.22 (1.19-1.26) | <.001 |
| Interaction of time and overall SVI score | 1.01 (1.01-1.01) | <.001 | 1.01 (1.01-1.01) | <.001 |
| Socioeconomic status subindex | ||||
| Time, wk | 1.26 (1.26-1.27) | <.001 | 1.21 (1.20-1.21) | <.001 |
| Socioeconomic status indexd | 1.09 (1.07-1.11) | <.001 | 1.19 (1.16-1.22) | <.001 |
| Interaction of time and socio-economic status indexd | 1.01 (1.01-1.01) | <.001 | 1.00 (1.00-1.00) | <.001 |
| Household and disability subindex | ||||
| Time, wk | 1.28 (1.27-1.28) | <.001 | 1.22 (1.21-1.22) | <.001 |
| Household characteristics and disability indexd | 1.04 (1.02-1.05) | <.001 | 1.11 (1.08-1.14) | <.001 |
| Interaction of time and household characteristics and disability indexd | 1.01 (1.01-1.01) | <.001 | 1.00 (1.00-1.00) | <.001 |
| Racial/ethnic minority status and language subindex | ||||
| Time, wk | 1.27 (1.26-1.27) | <.001 | 1.16 (1.16-1.17) | <.001 |
| Racial/ethnic minority status and language indexd | 1.15 (1.13-1.17) | <.001 | 1.21 (1.18-1.25) | <.001 |
| Interaction of time and racial/ethnic minority status and language indexd | 1.01 (1.01-1.01) | <.001 | 1.01 (1.01-1.01) | <.001 |
| Housing and transportation sub-index | ||||
| Time, wk | 1.28 (1.276-1.284) | <.001 | 1.20 (1.19-1.21) | <.001 |
| Housing type and transportation indexd | 1.09 (1.08-1.11) | <.001 | 1.16 (1.13-1.19) | <.001 |
| Interaction of time and housing type and transportation indexd | 1.01 (1.01-1.01) | <.001 | 1.01 (1.00-1.01) | <.001 |
Abbreviations: COVID-19, coronavirus disease 2019; IRR, incidence rate ratio.
Each of the regression models was adjusted for population density, urbanicity, and state-wide COVID-19 testing rate. Time (in weeks) was centered at week 19 (May 28, 2020–June 3, 2020). In separate regression models for each measure, time was interacted with the overall SVI score and subindices along with the individual outcomes to study how the SVI index was associated with the weekly change in COVID-19 incidence or mortality. All regression models included an offset for the total population in the county. Analytic sample exclude the counties spanning New York, New York.
Incidence rates were estimated using mixed-effects negative binomial regression with a random intercept for repeated measures within county.
Mortality rates were estimated using mixed-effects zero-inflated negative binomial regression with a random intercept for repeated measures within county. COVID-19 cases per 100 000 population were used to model the logit part in each model estimating excess zero count.
Owing to rescaling of the variables, IRR for each index shows the change in incidence or mortality for a 0.1 unit of the original index measure.