Abstract
Inspired by the POLYMOD study, an epidemiological survey was conducted in June 2021 in one of the most densely populated and socially vulnerable sectors of Belo Horizonte (Brazil). A sample of 1000 individuals allowed us to identify, within a 24-hour period, the rates of social contacts by age groups, the size and frequency of clique in which respondents participated, as well as other associated sociodemographic factors (number of household residents, location of contact, use of public transportation, among others). Data were analyzed in two phases. In the first one, results between two SIR models that simulated an eight-day pandemic process were compared. One included parameters adjusted from observed contact rates, the other operated with parameters adjusted from projected rates for Brazil. In the second phase, by means of a log-lin regression, we modeled the main social determinants of contact rates, using clique density as a proxy variable. The data analysis showed that family size, age, and social circles are the main covariates influencing the formation of cliques. It also demonstrated that compartmentalized epidemiological models, combined with social contact rates, have a better capacity to describe the epidemiological dynamics, providing a better basis for mitigation and control measures for diseases that cause acute respiratory syndromes.
Keywords:
Epidemiological survey; POLYMOD; Social contact rate; Cliques
Thumbnail
Thumbnail
Thumbnail
Source: Research data.Note: The matrix is read in the direction of the line to the column. For example, if we want to know the observed rate of contacts between people aged 0-14 and those aged 35-59, we look for the respective vertex, which indicates it is 0.99 contacts per person per day. In other words, on average, each child or adolescent reported one contact with an adult in the age group in question. We then looked at the line for the 35 to 59 age group to see how many contacts they indicated with a child or adolescent, and found that it was 0.5 on average. The rates do not match because, within the sample, the alteri (indicated) do not necessarily coincide with those indicating (ego). This makes it necessary to symmetrize the matrix, using the arithmetic mean, to include it in the SIR models.
Source: COVID-19: research data.
Source: Research data.