Effectiveness of the COVID-19 Community Vulnerability Index in explaining COVID-19 deaths

Abstract

Objectives: To explore the effectiveness of a COVID-19 specific social vulnerability index, we examined the relative importance of four COVID-19 specific themes and three general themes of the COVID-19 Community Vulnerability Index (CCVI) in explaining COVID-19 mortality rates in Cook County, Illinois.

Methods: We counted COVID-19 death records from the Cook County Medical Examiner's Office, geocoded incident addresses by census tracts, and appended census tracts' CCVI scores. Negative binomial regression and Random Forest were used to examine the relative importance of CCVI themes in explaining COVID-19 mortality rates.

Results: COVID-19 specific Themes 6 (High risk environments) and 4 (Epidemiological factors) were the most important in explaining COVID-19 mortality (incidence rate ratio (IRR) = 6.80 and 6.44, respectively), followed by a general Theme 2 (Minority status & language, IRR = 3.26).

Conclusion: The addition of disaster-specific indicators may improve the accuracy of social vulnerability indices. However, variance for Theme 6 was entirely from the long-term care resident indicator, as the other two indicators were constant at the census tract level. Thus, CCVI should be further refined to improve its effectiveness in identifying vulnerable communities. Also, building a more robust local data infrastructure is critical to understanding the vulnerabilities of local places.

Keywords: COVID-19 mortality; community vulnerability; negative binomial regression; random forest; social vulnerability index.