Study population and setting
The study sought to describe the association with socioeconomic status and COVID-19 infection and death in Santiago, Chile. The study used the Social Priority Index from 2019, and defined socioeconomic status based on three components: income and poverty, access and quality of education, and access to healthcare and life expectancy. Beginning in January 2020, the EPIVIGILA platform was developed by the Health Epidemiology Unit of the Ministry of Health. Active testing was done for both suspected asymptomatic and pre-symptomatic cases, particularly those living in vulnerable areas or who live in institutions including jails, nursing homes, and the National Service for Minors. The EVIGILA platform contained population projections, testing rates, test positivity rates, and similar data. The reports track cases for each municipality and are published twice per week. The Vital Statistics System in Chile records deaths and standardized clinical terms from the International Statistical Classification of Diseases (ICD-10) for the cause of death. Additionally, COVID-19-resultant deaths are noted based on WHO coding guidelines. The study also used Facebook’s Data for Good Geoinsights portal which has anonymized location data among Facebook users with a smartphone, binned in 8-hour increments. The study calculated the percentage change in individuals gathered in an area compared to baseline for each 8-hour transition period. The study estimated SARS-CoV-2 infections over time using a Poisson deconvolution model based on observed COVID-19 attributable deaths. The study also calculated excess deaths using Gaussian Processes regression and infection fatality rates using hierarchical Bayesian joint model per age group and municipality.
Summary of Main Findings
Across 34 municipalities and 7 million people living in Santiago during 2020, the maximum SARS-CoV-2 incidence in the highest SES municipality Vitacura was 22.6 weekly cases per 10,000 individuals, while La Pintana (the lowest SES municipality) was 76.4 weekly cases per 10,000 individuals. There were also notable changes in human mobility during the lockdown periods, with the highest SES municipalities reducing mobility by 61% in lockdown, while low SES municipalities had 40% reduced mobility. In the researchers’ models, they estimated that the number of infected individuals was 5 to 10 times larger than the reported values, and that between May and July 2020, there were 1.73 [95% Credible Interval: 1.68, 1.79] times the expected number of deaths, with the peak at 2,110 death counts in the first week in June 2020. The study also found a higher testing capacity in wealthy areas, with a strong negative association between positivity and SES, and an association between increased wait times for receiving test results and lower SES municipalities. Finally, the study also found an increased case fatality rate by age and SES, with the fatality rate 1.7 times higher among 40-60 year olds in low SES areas compared to high SES areas, and 1.4 times higher among 60-80 year olds in low vs. high SES areas.
The study made use of a wide range of data sources to inform their estimates. Particularly, researchers had access to Facebook data in 8-hour bins in order to get an accurate measurement of mobility using smartphones. Researchers also used a validated SES index that reflects a number of different SES-related factors, as opposed to only using income or education in order to define this complex and multi-faceted positionality. They also were able to examine testing capacity in order to estimate potential underreporting of prior estimates.
One limitation is potential selection bias in the mobility data, as it is predicated on Facebook use and smartphone ownership, which may not reflect all sociodemographic groups. It is not clear how this may bias estimates, however. Additionally, their SES index examined 3 factors (healthcare access, education, and income), but there may be additional nuances in socioeconomic position that is not represented in these three categories (such as race/ethnicity, access to healthy foods, etc.). It is also not clear how this potential misidentification of SES may bias results.
This is one of the largest studies examining associations with socioeconomic status and COVID-19 infection and mortality in South America.