Space-time Classification Index for Assessing COVID-19 Hotspots
Objectives: To develop new methods to address problems associated with use of traditional measures of disease surveillance, including prevalence and positivity rates. Methods: We use data from the public New York Times Github repository to develop a space-time classification index of COVID-19 hotspots. The Local Indicator of Spatial Association (LISA) statistic is applied to identify daily clusters of COVID-19 cases, from July 4th to July 19th. Results: The classification index is a spatial and temporal assessment tool that seeks to incorporate temporal trends of the clusters that are "high-high" and "high-low". Two classifications support the index: severity and temporal duration. We define severity as the number of times a county is statistically significant and temporal duration captures the number of consecutive days a county is a hotspot. Conclusions: The space-time classification index provides a statistically robust measure of the spatial patterns of COVID-19 hotspots. Spatial information is not captured through measures like the positivity rate, which merely divides the number of cases by tests conducted. The index proposed in this paper can guide intervention efforts by classifying counties with six-levels of importance.