Small Area Vulnerability Index for COVID-19

Project Description

Published in the Journal of Epidemiology & Community Health (2021), this study developed the Small Area Vulnerability Index (SAVI), an evidence-based composite measure of community-level COVID-19 vulnerability derived from ecological analysis across all 6,789 Middle Super Output Areas (MSOAs) in England.

The work was motivated by the unequal impact of the pandemic: COVID-19 mortality was twice as high in the most deprived areas, and communities in the North of England and Midlands were disproportionately affected. Existing resource allocation mechanisms had not incorporated specific measures of local population vulnerability, leaving the most exposed communities under-served.

The study assessed five candidate vulnerability measures: care home bed density, prevalence of long-term health conditions (hospitalisation records 2014-2018), ethnic minority population share, overcrowded household rate, and income deprivation, against age-adjusted COVID-19 mortality using a multivariable Poisson regression framework. Income deprivation dropped out of the multivariable model; the remaining four measures were independently associated with mortality:

• Care home beds (per capita): +28% increase in mortality per SD
• Long-term health conditions: +19% per SD
• Overcrowded housing: +11% per SD
• Ethnic minority population: +8% per SD

The final index was constructed by removing wave-specific epidemic dynamics (time-to-first-cases, regional dummies) from the model predictions and applying a shrinkage procedure to reduce small-area uncertainty. The mapped index revealed concentrated vulnerability clusters in the North West, West Midlands, and North East regions where multiple risk factors co-occur.

An interactive map of SAVI was published openly via the PLDR platform, and all underlying data were made publicly available to support local public health planning ahead of subsequent pandemic waves.

Skills Used

• R
• Poisson Regression
• Generalised Estimating Equations (GEE)
• Small Area Analysis
• Spatial Epidemiology
• Index Construction & Shrinkage Estimation
• Hospital Episodes Statistics (HES)
• ONS / Census Data