In the present study, we estimated the risk of COVID-19 transmission from extra-household exposures and from a single infected household member in a highly vulnerable setting, a complex of favelas (Complexo de Manguinhos) in Rio de Janeiro city. Our analyses incorporated a complex and intertwining set of demographics, economic, behavioral, and structural factors, as well as COVID-19 vaccine receipt to investigate factors associated with COVID-19 transmission risk. We found that the risk of infection from extra-household exposures, estimated at 74.2%, was considerably higher than the risk of infection from a household member, estimated at 11.4%. We also found that receiving at least one dose of COVID-19 vaccine reduced the risk of infection from within-household and extra-household exposures. The risk of infection from a household member was higher among children and adolescents (10–19 years old), individuals residing in overcrowded households, with low family income, and employed individuals that reported remote work status. In contrast, young adults (20–29 years old), unemployed individuals, and those reporting the use of public transportation as their main means of transportation had a higher risk of infection from extra-household exposures. Interestingly, a higher income increased the risk of infection from extra-household exposures, whereas Bolsa Familia reduced the risk of infection from extra-household transmission.
To fully understand our findings, it is essential to situate our study in space (Complexo de Manguinhos) and time (from November 2020 to December 2021) and be mindful of the epidemiological context in which the study occurred. First, our study was undertaken in a deeply vulnerable neighborhood in Rio de Janeiro city comprised of 16 favelas. This area holds the 5th worst human development index of the city and is characterized by concentrated poverty, high rates of violence, inadequate housing conditions, and lack of access to essential services, such as clean water and sanitation. Prior work from our group conducted in this geographical location revealed a high prevalence of anti-SARS-CoV-2 antibodies (59%) from September 2020 to February 202118. Second, three consecutive SARS-CoV-2 variants (zeta, gamma and delta) emerged during the study period and became the dominant variant in Rio de Janeiro and Brazil. From early 2020 through the end of 2021, the COVID-19 epidemic in the country was characterized by two large epidemic waves19. The first wave (March through September 2020) resulted from multiple introduction events and was characterized by the circulation of several SARS-CoV-2 lineages (including B.1.1.28 and B.1.1.33). The onset of the second wave occurred in October 2020 and resulted from the emergence and circulation of novel consecutive variants such as zeta and gamma. The gamma variant remained dominant from January 2021 until July 2021, when delta replaced it. During gamma dominance, reported COVID-19 deaths and cases increased to a historical maximum, likely due to a large susceptible population and low vaccination coverage. Following gamma, the emergence of delta was not associated with an increase in the number of cases or deaths, and that likely resulted from a reduced pool of susceptible individuals in the community, either due to natural immunity (induced by prior gamma infection) or to the increasing vaccination coverage19. Of particular relevance, our study period coincides largely with the second wave, specifically with the gamma dominance period, and may describe transmission dynamics resulting from dominant variants transitions. Third, approximately halfway through our study period, COVID-19 vaccines became available and were distributed using an age-targeted strategy that initially included elders and progressively included younger individuals. By the end of our study period, vaccine coverage (two doses) reached almost 90% of the Rio de Janeiro city population aged 12 years old or older20.
Effective SARS-CoV-2 transmission is influenced by pathogen features (i.e., variant transmissibility and resistance to neutralizing antibodies), host susceptibility and behavior, household features and other structural factors, as well as community factors such as the population-level burden of infection, population density, and vaccination coverage (direct and indirect effects)21,22. As mentioned briefly in the Introduction, several studies have estimated SARS-CoV-2 household infection risk and SARs in different countries, however, most did not discriminate the source of infection for the secondary cases (i.e., within-household transmission versus extra-household transmission)1,2,6,7. In a study conducted in Guangzhou, China (Jan-Feb, 2020), the risk of infection was estimated at 17.2% within the household and 2.6% from extra-household contacts23. In another study conducted in Geneva, Switzerland (Apr-Jun, 2020), the risk of infection within the household was three-times greater than the risk of infection from extra-household exposures (17.3% versus 5.1%)24. Contrasting with these two studies, we found that the risk of infection was 6.5 times greater from extra-household exposures than from a single infected household member (74.2% and 11.4%, respectively). Adherence to social distancing measures and other non-pharmaceutical interventions was poor in Brazil, particularly among the most vulnerable populations25. The role of Federal leadership in antagonizing the protective benefits of lockdown and non-pharmacological interventions had deleterious effects on individuals' behavior. Additionally, even among mask users, the use of cloth masks was highly prevalent (relative to surgical or N95 masks), and the protective effect of such masks remains unknown and may vary according to the material and layers used26,27. Altogether, we speculate that even during the peak of the SARS-CoV-2 transmission, including the second wave and gamma variant dominance period, adherence to lockdown and non-pharmaceutical preventive measures was poor and extra-household transmission exposures were frequent. The use of public transportation (bus, train, subway), where the limited number of passengers or minimal distancing requirements were not observed, potentially increased extra-household exposure events. Indeed, our results showed that the risk of infection from extra-household exposures was higher among those who reported using public transportation.
Our findings on the risk of infection according to proxies of socioeconomic status, namely employment, family income and receipt of Bolsa Família benefit, highlight the challenges of adhering to social distancing among vulnerable populations. First, our results indicated that the risk of infection from extra-household exposures was higher among those who reported unemployment and low income. To adequately meet survival needs, individuals who were unemployed or with low-paid jobs likely increased their exposure to infection by seeking informal or temporary jobs. In line with this reasoning, we found that receipt of Bolsa Família reduced the risk of infection from extra-household exposures probably because it provided minimum financial security, allowing individuals to adhere to social distancing measures. A study conducted during the COVID-19 pandemic in two favelas in São Paulo found that families who received the Bolsa Familia benefit had fewer experiences of severe or moderate food insecurity28. Contrastingly, the risk of infection from extra-household exposures was also higher among individuals in the highest income category, and this finding may indicate that those individuals with formal "well"-paid jobs, particularly those working in the service sector, did not, or, more realistically, could not adhere to social distancing measures.
As for the socioeconomic factors related to increased risk of infection from household exposure, we found that higher household density (measured as persons per bedroom) and low income were associated with increased risk. Similar findings had been reported previously, mainly in vulnerable settings and minority communities, and may result from the precarious housing conditions (i.e., poor ventilation, small household size, and lack of adequate sanitation and clean water) coupled with the impossibility of isolating an infected household member21,29. Finally, though the prevalence of remote work was very low in our study population, we nonetheless found that individuals who reported working remotely had a higher risk of infection from a single household contact, which may result from intense cohabiting.
Lastly, our results showed that the COVID-19 vaccine was associated with a lower risk of infection from both within-household and extra-household exposures. This finding corroborates those from clinical trials and observational studies showing the efficacy and effectiveness of multiple COVID-19 vaccines30,31. Nonetheless, vaccine coverage was low in the study population, and the proportion of vaccinated individuals was higher in older age groups, as expected given the age-targeted roll-out of vaccination. Vaccination in Rio de Janeiro city began on January 20th, 2021, and up to May 2021, vaccination was restricted to priority groups and elders. In June 2021, vaccination was extended to the general population aged 12 or older, and 90% of the city population had received two doses by December 202120. COVID-19 vaccines were extended to children below 12 years old in January 2022, when this study's data collection had already ended.
Our study has several limitations. First, we gathered our data through a longitudinal study, which provides valuable insights over time. While there are statistical models designed to handle longitudinal data structures32, it has proven challenging to accurately reconstruct each participant's incubation and infectious periods. This information is crucial for informing an appropriate chain binomial model. In our analytical approach, we had to simplify the data by collapsing it over time, essentially mimicking a cross-sectional design. We utilized cumulative time-related information as the input for our model. Regrettably, this simplification leads to less precise estimates compared to what could be achieved with a more detailed longitudinal analysis23. Second, our cohort study was subject to some selection bias in the form of non-random losses-to-follow-up: females and older participants were more likely to be retained in the study. Third, we have used a combination of available proxies/biomarkers of COVID-19 infection to define the outcome of interest, following a hierarchical order as described in the methods. Therefore, we were not able to precisely define the timing of infection or the duration of infectiousness. Fourth, we found a lack of protective association between adherence to social distancing measures and risk of extra-household transmission that may have resulted from reporting bias given the face-to-face format of our interviews, with individuals likely overestimating their adherence33.
A strength of our analysis was the methodological approach that allowed the estimation of the relative contribution of within-household and extra-household exposures in COVID-19 transmission. In addition, we combined the results of consecutive molecular and immunological tests (regardless of symptoms) and major clinical symptoms to ensure high sensitivity and specificity in detecting the outcome.
In conclusion, our study provides important insights into COVID-19 household and community transmission in a highly vulnerable population residing in precarious and overcrowded households in a community that struggled to adhere to lockdown policies and social distancing measures. Compared to prior studies, the much higher extra-household infection risk highlights the extreme social vulnerability of these populations and the need for tailored strategies to mitigate and assist these communities during the emergency of a new transmissible infectious disease. Cash transfer programs can help by providing some level of financial security and, as such, permitting social distancing, whereas prioritizing vaccination of the most socially vulnerable could also protect these individuals and reduce widespread community transmission.