Abstract
We estimated within/extra-household SARS-CoV-2 infection risk and associated factors in a household cohort study in one of the most vulnerable neighborhoods in Rio de Janeiro city. Individuals ≥1 years-old with suspected or confirmed COVID-19 in the past 30 days (index cases) and eligible household contacts were enrolled (Nov/2020-Dec/2021) and followed at 14 and 28 days. RT-PCR testing, COVID-19 symptoms, and SARS-CoV-2 serologies were ascertained in all visits. Chain binomial household transmission models were fitted using data from 2,024 individuals (593 households). Extra-household infection risk was 74.2% (95% credible interval [CrI] 70.3 - 77.8), while within-household infection risk was 11.4% (95%CrI 5.7 - 17.2). Vaccination reduced both within/extra-household infection risks. Within-household infection risk was higher among participants aged 10-19 years, from overcrowded households, and with low family income. Contrastingly, extra-household infection risk was higher among participants aged 20-29 years, unemployed, and public transportation users.
Introduction
While many studies have estimated the Secondary Attack Rate (SAR) of COVID-19 within households, few have successfully disentangled within-household transmission from extra-household (community) transmission. This distinction is vital for public health policy, especially in highly vulnerable urban areas like Brazil’s favelas. In these informal settlements, "stay at home" orders are often a luxury. High population density, precarious housing, and the need for informal work make social distancing nearly impossible. This study sought to quantify the relative risks of these two transmission pathways and identify the specific socioeconomic factors—like income, transport use, and government aid—that influenced them during the height of the Gamma and Delta variant waves.
Methods
The researchers conducted the Comvida 2 study, a household longitudinal cohort study in the Manguinhos neighborhood of Rio de Janeiro. They enrolled 2,024 individuals across 593 households between November 2020 and December 2021.
To maximize sensitivity, infection was defined using a hierarchical approach: RT-PCR positivity, COVID-19 symptoms, anti-N IgG antibodies, and anti-S IgG antibodies (for the unvaccinated). The team employed Bayesian chain binomial transmission models implemented in Stan to estimate the probability of infection. This approach allows for the reconstruction of infection sequences within a household, distinguishing between multiple external introductions and internal spread.
Key Results and Insights
The Dominance of Community Transmission
The most striking finding is the 6.5x disparity between community and household risk. The extra-household infection risk was estimated at 74.2% (95% CrI: 70.3% – 77.8%), while the within-household risk was only 11.4% (95% CrI: 5.7% – 17.2%). In this setting, the "outside world" was far more dangerous than the home, likely due to the collapse of social distancing measures and the economic necessity of public interaction.
Socioeconomic Drivers and Overcrowding
The study found a "poverty paradox": while low income was associated with higher within-household risk due to overcrowding, a slightly higher income within this poor community actually increased extra-household risk. This suggests that those with formal service sector jobs were more exposed than those with no income at all.
As shown in the data above, households with higher density (persons per bedroom) faced distinct challenges. Specifically, households with more than 4 persons per bedroom had the highest within-household transmission risk. Age also played a role: adolescents (10-19) were primary drivers within the home, whereas young adults (20-29) faced the highest risk in the community.
Transport, Vaccination, and Social Safety Nets
Public transportation users (bus, train, subway) had a significantly higher probability of community-acquired infection. Conversely, receipt of the Bolsa Família (a conditional cash transfer) was associated with lower extra-household infection risk, providing empirical evidence that financial safety nets allow vulnerable people to reduce exposure by staying home.
Vaccination proved highly effective; receiving at least one dose significantly reduced the probability of infection from both household and community sources.
Discussion and Limitations
The authors acknowledge several limitations, including the simplification of longitudinal data into a transversal snapshot for modeling and potential selection bias toward female and older participants. Furthermore, adherence to social distancing was self-reported, and the lack of a protective association in the data suggests participants may have overestimated their compliance.
Ultimately, the study suggests that future pandemic responses in vulnerable areas must prioritize tailored strategies—such as targeted vaccination and expanded cash transfers—rather than one-size-fits-all lockdown policies that are economically impossible for the poor to follow.
Summary updated