ABSTRACT
The study proposes an index of vulnerability to the spread of COVID-19 using multivariate and geospatial analyses. Vulnerability was considered as a combination of exposure, susceptibility, and responsiveness of the population. The methodology consisted of six steps selection and grouping of variables;definition of indicators;normalization of the variables;weight calculation using principal component analysis;index calculation and normalization;and classification using Jenks Kmeans, Quantile, and Hclust techniques. The exposure of Fortaleza, Brazil, to COVID-19 was assessed by means of population density, number of subnormal and precarious housing, percentage of aged people per residence and proximity to bus terminals. The study integrates factors related to socioeconomic characteristics, public water supply and sanitation for the classification of census tracts in five levels of vulnerability. Most of them present high (Jenks and Quantile) and moderate (K-means and Hclust) vulnerability. The most vulnerable regions are in the South and West Zones of the city, where many subnormal agglomerates are located. The results can assist the development of coping strategies for the groups most exposed to the risks associated with COVID-19, as well as in the preparation for future public health crises. The methodology can be replicated in other cities and can be useful for public managers.