Study population and setting
The study included all publicly reported COVID-19 cases in Mexico, as published by the Directorate General of Epidemiology, which includes daily updates, sociodemographic information common comorbidities, case location, and whether the case died. The analysis was limited to adults 20 years and older who were tested based on symptoms of COVID, resulting in a population of 1,378,002 individuals tested. The authors used multivariate mixed effects logistic regression that allowed for clustering at the municipality level to account for potential random effects based on county. They also established the probability of hospitalization, intubation, and death.
Summary of Main Findings
Of the 1,378,002 people tested, 654,858 tested positive (47.5%). On average, more men tested positive than negative (52.2% of positive tests were among men vs. 46.1% of negative tests were among men), and people testing positive were more likely to be older (46.1 years vs. 42.3 years of age). Additionally, individuals testing positive were more likely to speak an indigenous language (1.01% vs. 0.76%) or go to a social security facility for care (38.5% vs. 25.9%). People speaking indigenous languages were more likely to be hospitalized ( OR 1.64, 95% CI: 1.52 – 1.76), and being in the highest quintile of poverty where 80% or more of their municipality is considered poor increased the odds of hospitalization by 3.25 (95% CI: 2.40 – 4.41), and by 1.95 (95% CI: 1.56 – 2.43) for death. Being older and having comorbidities, in general, also increased the odds of hospitalization and death. Intubation was not clearly associated with the risk factors under investigation.
The study used all reported tests in Mexico to estimate risk factors for COVID-19, hospitalization, intubation, and death. It used multiple outcomes which allowed it to assess varying severity. It also had a very large dataset, making it well-powered, and reducing potential selection bias, given everyone was effectively selected. It also had a number of comorbidities available for analysis, and assessed not only poverty but also facility individuals attended, which addresses socioeconomic risks from multiple angles.
The study only included data from people who tested if they screened as potentially having COVID-19, and as such cannot estimate the population prevalence as asymptomatic cases are not included. Additionally, only individuals who sought care were included, and individuals who did not seek care may have other sociodemographic risks that this study could not capture—for example, attending a Ministry of Health facility vs. social security facility may be related to socioeconomic factors, or due to disease severity and the equipment available at a given facility. Using the municipality-level economic indicator could also misclassify individuals based on their location, rather than their actual income level, which could bias estimates in either direction. Finally, there were a number of study design details that remained unclear, such as the dates when the study was conducted, and whether the comparisons of hospitalization, intubation, and death were among those who tested positive for COVID-19, or among the whole sample. These details would impact our assessment of the validity of estimated effects.
This is one of the largest studies from Mexico investigating COVID-19 risk factors, which is a country not often described in the literature.
This review was posted on: 7 December 2020