Mental health deteriorated on global scale during the COVID-19 pandemic ^{1, 2} . In addition to the concerns and fears associated with becoming infected, the pandemic's economic, political, and social consequences have impacted populations, in particular the most socially and economically vulnerable ^{2, 3} . Many public health campaigns focused on behavioral strategies to control the spread of the virus including physical distancing policies ⁴ . Although these strategies were effective to reduce the virus transmission, some of these recommendations (in particular social distancing) added an additional burden to mental health ^{5– 7} .

Campaigns focusing on mental health started in parallel with the public health messages directed to control viral transmissions. Major health associations, such as the World Health Organization (WHO) and the Center of Addiction and Mental Health (CAMH) created guidelines instructing people on activities to minimize adverse mental health consequences during the pandemic ^{8, 9} . These guidelines emphasized self-care strategies such as eating healthy, keeping active, setting daily routines, meditation and staying in touch with family and friends

A number of reviews and meta-analyses also pointed towards simple and safe strategies that can be adopted by individuals during social distancing and have been shown to improve mental health, reduce anxiety and depression ¹⁰ . Previous studies have identified effective strategies to maintain and improve mental health during stressful periods by correlating self-reported behaviors with mental health scores ^{11, 12} , yet little is known about what individuals believe is effective in protecting their mental health and consequently, what activities people engage in to keep themselves healthy. There is an evident disconnection between scientific information on valid or productive habits and behaviors, and what individuals are actually doing. Therefore, our first focus is an explicit exploration of what individuals in socially vulnerable communities reported as having done to maintain or improve their mental health during the COVID-19 pandemic.

Specifically, we studied these beliefs and behaviors in a large urban slum area. Marginalized communities suffer from added mental health burdens, yet their experiences and voices are being largely ignored in the literature. This omission and bias in current research needs to be addressed in order to help public health campaigns more effectively communicate relevant strategies that can be adopted by these vulnerable populations. In the following, we briefly outline the study context and then present the research question.

The final coding included four main themes: Work, Leisure, Religion, and Others (which included any answer that did not fit into the three other themes). Each of these main themes included sub-themes ( Table 1; Figure 2). Individual responses that could be coded in multiple themes because individuals often reported multiple activities (example: "I read, watch movies and tv shows, and talk to my friends", cis woman, mixed-heritage, 60–64 y/o).

The most frequently mentioned theme "Work" (33.89%, of responses) included responses that focused on active engagement with physical or mental activities. Three sub themes differentiated this broad category further: "Trade (paid work)", "Domestic work and Handiwork", and "Study". The most common responses in the "Work" theme were related to the "Trade (paid work)" subtheme, which included all the responses associated with making a living through paid work, such as "I work" (cis man, mixed-heritage, 50–54 y/o). Domestic work, in which a person reported pursuing various activities at home, such as "cleans the house" (cis woman, mixed-heritage, 60–64 y/o) was the second most frequent response with the larger Work theme. Other responses in the Domestic work subcategory included creative activities and crafts (e.g., "I do crochet", cis woman, white, 45–49 y/o). The last sub-theme captured active engagement with mental activities, either for an ultimate purpose (e.g., study) or for leisure (playful activities that engage the mind). Examples were responses such as "I study" (cis woman, African descendent, 20–24 y/o), "I read" (cis man, white, 60–64 y/o), or "I do crossword puzzles … “, (cis woman, african descendent, 50–54 y/o).

The “Leisure” theme was the second most frequent theme (37.36%) and included entertainment activities related to escapism and relaxation. It contained five subthemes: “Entertainment”, “Physical activities”, ”Social interactions”, “Outings & travel” and “Games & drinking”. The most frequent subtheme within this larger theme was “Entertainment”. It captured activities in which an individual would consume radio, TV or other entertainment content. Example responses included “I watch TV” (cis woman, mixed-heritage, 55–59 y/o) and “Listen to music” (cis woman, african descendent, 45–49 y/o). The second most common response involved physical exercise with an example response "I go to the gym" (cis man, mixed-heritage, 30 – 34 y/o). The third most common subtheme was labelled "Social interactions" and involved face-to-face interactions with friends and family (e.g., “I talk to my friend”, cis man, african descendent, 60–64 y/o). A fourth category was called “Outings & Travel”. This relatively broad theme included activities that involved leaving the place of residence (“I get out of the house”, cis woman, mixed-heritage, 25–29 y/o) as well as travel outside the metropolic area (“travel to Saquarema [rural town]”, cis woman, mixed-heritage, 45–49 y/o). Finally, we included a subtheme which we labelled “Games and Drinking”. This subtheme combined social contact with playing games (typically cards) and drinking, which is a favourite pastime in these communities (e.g.,“I play cards with my friends and drink”, cis man, african descendent, 35–39).

The third theme in the responses involved being in contact with god and religion as a form of improving mental health (10.42%). This main theme was divided into two subthemes: "Faith and prayer" and “Church”. The first subtheme included responses referring to personal or private religious practice (e.g., "reads the bible", cis man, mixed-heritage, more than 70 y/o; "prays", cis woman, African descendent 45–49 y/o). The second subtheme involved activities in church as an institutionalized form of religious practice (e.g., "I go to church", cis woman, african descendent, 55–59 y/o); "I help with social activities in church", cis man, african descendent, 35–39 y/o).

A miscellaneous theme "Others" (18.23%), contained a number of sub-themes such as "Doing nothing", “Avoiding the pandemic”, “Protection” and “Other activities”. The subtheme “Doing nothing” typically meant that individuals responded that they were not actively doing anything to maintain or protect their mental health. In contrast, "protection" included responses that were focused on protecting oneself from being infected with the virus, but not specifically geared towards maintaining one's mental health. We observed that the majority of answers in this category were related to avoiding people or staying at home, which is not recommended from the point of view of mental health. The sub-theme "Avoiding the pandemic" included answers that indicated that individuals tried not to think or seek information about the pandemic such as avoiding television and other forms of communication that related to the COVID-19 pandemic

The “other activities” subtheme did included some responses that contained both mental and physical health relevant responses ("sunbathing and drinking liquids", cis woman, white, 30-34 y/o). What was particularly noteworthy were responses referring to taking care of pets (3 responses) and seeking professional help (primarily in therapy, 4 responses).

We investigated what individuals in a vulnerable community have done to maintain their mental health during the COVID-19 pandemic.These insights advance our understanding of mental health activities in marginalized populations in a number of directions with direct implications for communication strategies about mental health. As a point of departure, many of the responses included local adaptations and implementations of selected activities that have positive mental health effects. On the other hand, what is equally interesting to observers is what was NOT reported, in that highly effective and low cost activities such as meditation or mindfulness that can improve mental health were not reported by the participants. In the following, we selectively discuss some of these insights gained.

The most frequently reported activity was related to active engagement, working, domestic work or studying. In accordance with the WHO and CAMH guidelines, keeping a routine or adapting it to the current context is essential to maintain a healthy attitude ²³ . It is noteworthy that the most frequent responses in our interview were associated with the subthemes "Trade" followed by "Entertainment". Engaging in work means that individuals are keeping a routine, which has been shown to be beneficial for mental health ²⁴ . More importantly, paid work in this marginalized community implies that individuals and their families can have access to food and basic items. In the pandemic context, social and economic adversities have increased and have a direct effect on mental health. Other authors have reported employment as a protective factor against mental health threats in the context of COVID pandemic in socially vulnerable communities in São Paulo, Brasil. On the other hand, food insecurity and lower monthly income are associated with a higher risk of psychological distress ²⁵ . This financial aspect is especially significant in vulnerable communities, where people have little possibility of not working or working from home, given the limited life savings and the characteristics of their work, which typically does not allow telework (e.g., cleaning & other services, security, manual labor).

Focusing on the effectiveness of health campaigns, it is worth elaborating on the responses within the "Protection" subtheme. Many responses related to self isolation, which, although positive in the physical health aspect, are not recommended as a form of maintaining mental health ⁷ . The official guidelines point to social interactions as a vital part of improving mental health. This highlights that public health communications focused on physical health were quite successful in raising the awareness of social distancing for containing the virus. An example of this is the widely used #FicaEmCasa (#Stayathome) hashtag launched by the most popular TV station in Brazil. However, this focus was initially not accompanied by an equally strong message on mental health. This may be the reason why a considerable number of people answered the questions about mental health as if it was the same as physical protection against the virus. Alternatively, participants may have seen health holistically and did not differentiate between physical and mental recommendations. Regardless of the interpretation, it is important for future interventions to emphasize the difference between physical and mental health interventions and highlight both aspects.

Highlighting some of the complexities of mental health prevention in a pandemic context, the dichotomy between self-protection and the wider social and cultural context can also be seen in the subtheme "Games and drinking", which is directly opposed to the guidelines’ recommendations, but a central part of social life in these communities. Given the tropical climate and the cramped living conditions, social life in the “favela” is mainly set in the streets. The identified sub theme included activities considered harmful for mental health, such as the use of tobacco, alcohol and other drugs. Yet, these activities were performed in social context, which has obvious positive mental health consequences. We would like to highlight the complexity of social practices which may combine conflicting activities which both improve and undermine mental health. At the same time, it is important to contextualize the number of responses and emphasize that this was the least frequent of all 14 subthemes.

Finally, when considering the broader mental health literature, it is noteworthy that activities identified as helpful in improving mental health were rarely mentioned by the respondents. Responses indicating seeking support from mental health professionals or taking care of pets were noticeably rare (therapy was mentioned 4 times, while pets only 3 times). A considerable number of respondents reported not taking any measures to maintain their mental health and no responses were related to the practice of meditation or mindfulness. This last point is particularly important in the context of our study. Even though these practices are considered efficient forms for improving mental health by the scientific community ^{10, 24, 26} they are not widely disseminated in the community. The low number of responses regarding using therapy or other institutional mental health services were most likely due to the limited access to psychological care in the public health system in these communities. Getting appointments with mental health services is difficult and mostly limited to treating major psychiatric disorders. Although meditation and mindfulness were not perceived as helpful by respondents, these strategies could be especially impactful in vulnerable communities since they are accessible and cheap ways of improving mental health.

The findings of this study have to be seen in light of some limitations: first the survey was applied to a convenience sample. In this way it is not a representative sample of the larger population and no claims about representativeness of our findings can be made. This may have led to a selection bias since the study was conducted in public spaces and therefore did not include individuals who were socially distancing during this period. The coding by researchers with no lived experience in these communities may have also impacted some of the interpretations that led to specific coding decisions. To overcome these limitations, we consulted community members, but some subtle nuances may have been lost in the coding process. Furthermore, because of how the responses were recorded by interviewers, more elaborated responses were at times summarized as single words and therefore may miss important context. However the investigators tried to reach a consensus through group discussions and liaising with community members.

In conclusion, putting these findings in context, the responses reinforce the need to improve the communication between scientific communities and vulnerable populations. This can be done by creating effective and direct channels of communication between the population and the scientific communities, using accessible language and methods of communication which are widely used by the population, such as locally used social media apps, podcasts and radio. Besides informing the people about mental health strategies, it is also evident from our results that access to mental health services, including telemedicine and online therapy needs urgent attention.