XY has developed a COVID-19 Community Risk Score using disease prevalence and demographic data from approximately 27,000 communities across the United States. Here we show how to map and explore this risk score interactively.
In previous posts, we discussed preliminary work towards developing a COVID-19 Community Risk Score that leverages surveillance information about disease occurrence and the demographic make-up of communities across the United States. We have since formalized the risk score and placed it on firm statistical footing. We have also mapped the score at the census tract, city, county, and state level across the United States.
Briefly, XY's COVID-19 Community Risk Score aggregates disease prevalence and demographic data at the census tract level (or higher) from the Centers for Disease Control and Prevention (CDC) and the United States Census. These diseases include conditions that have been repeatedly linked with increased morbidity and mortality for COVID-19 patients. These conditions include cancer, hypertension, diabetes, obesity, coronary artery disease, smoking, asthma, COPD, high cholesterol, and arthritis. Age is among the strongest risk factors and is included in our COVID-19 community risk score. We further incorporated the proportion of male and female 65 years olds in a community separately to include in our risk score.
Our statistical approach identifies combinations of these variables that explain the maximal possible variation in the entire dataset. We then correlate this score to indicators of testing data, resource availability (e.g. ICU beds), and socioeconomic status. Look out for a manuscript from XY soon that will be posted on medRxiv with details on how we estimate the score. The score is on a scale of 0-100 where low scores indicate low community risk and higher scores reflect higher community risk across the United States. For a given city, we can rank the city within a state (or a census tract within a city) to get a sense of where a community lies compared to its surroundings.
We have mapped this score across the United States. Below, we zoom in on New York City, the current center of the American COVID-19 epidemic:
For one comparison, here is a map of the total count of COVID-19 cases based on patient address from NYC as of April 3:
XY is currently correlating cases and deaths with the COVID-19 community risk score across the United States, along with other relevant factors including hospital resources, indicators of the built environment (inferred from satellite imagery), and social determinants of health.
You can explore the risk score across the United States interactively here (try zooming in on a city like Detroit):