This cross-sectional study uses data gathered between April 2021 and November 2022 as part of the OLA Programme, a study of mental health in urban young people from Latin America (Priebe et al., 2021). The study was conducted in Bogotá (Colombia), Buenos Aires (Argentina), and Lima (Peru), three of the most populated capital cities in Latin America. Argentina, Colombia, and Peru are all upper-middle-income countries characterised by inequality, public insecurity, and weak health and social protection systems.

We recruited a non-randomised sample from educational and community settings and used a self-report questionnaire. We used a checklist of SLEs (Supplementary Data 1) to assess the number and temporality of SLEs; the Patient Health Questionnaire-8 (PHQ-8) to assess symptoms of depression; the General Anxiety Disorder-7 (GAD-7) to assess symptoms of anxiety; and the Manchester Short Assessment of Quality of Life (MANSA) to assess quality of life. To answer our objectives, we used logistic regressions and subgroup analysis.

Participants were adolescents between 15 and 16 years old and young adults between 20 and 24 years old who lived in deprived areas (i.e., economically, socially or environmentally disadvantaged areas) in Bogotá, Buenos Aires, and Lima. The inclusion criteria were: (a) being 15–16 or 20–24 years old; (b) living in the city’s poorest 50% neighbourhoods or districts, according to the United Nations Development Programme’s Human Development Index (United Nations, n.d.) in Bogotá and Lima, and according to the Unsatisfied Basic Needs Index in Buenos Aires (Santos, 2014); and (c) having the ability to give informed consent or assent. In the case of adolescents, informed consent from a parent or tutor was required. Young people with severe mental illness (e.g., psychosis), cognitive impairment, and illiteracy-due to the use of self-report questionnaires-, were excluded. The inclusion and exclusion criteria were self-reported by participants or assessed by the research assistant during the screening stage.

The recruitment of participants varied across cities, reflecting differences in local regulations, restrictions related to COVID-19, and practical options for each local team (Gomez-Restrepo et al., 2023). Participants were recruited in schools, universities, and community settings located in the neighbourhoods or districts eligible for the study. Non-Governmental Organisations and government employment and education schemes facilitated contact with potential participants. Additionally, we used Facebook and Instagram advertisements to reach potential participants.

As previously stated, this study uses the baseline measure from a cohort study. The baseline sample size was calculated to identify variables that predict recovery from symptoms of depression and anxiety with 90% power, 5% significance level, and 25% dropout rate. Detailed information about the sample size calculation can be found in a previous publication (Priebe et al., 2021).

Our aim was to include 2040 participants across the three cities at baseline to build a cohort. We aimed to include 340 young people in each city (1,020 in total) who met the criteria for symptoms of depression and/or anxiety (PHQ-8 and/or GAD-7 score ≥ 10). Additionally, since male participation in research is challenging and we did not want to overrepresent females, we ensured that at least one-third of the participants were male at the screening stage.

Participants who were interested in participating had to undergo at screening process to ensure they met the eligibility criteria. They also needed to provide informed consent or assent. Afterwards, we invited them to complete a paper or online questionnaire with a research assistant on standby to answer any queries. In total, 19 researchers from the three cities were involved in the recruitment and data collection. Assessments were done individually or in groups, and the questionnaire took anywhere between 30 and 60 min to complete. REDCap (Harris et al., 2019), a data collection and management software, was used to record the survey data. In the case of paper questionnaires, a trained research assistant manually entered the data into REDCap.

The questionnaire combined sociodemographic questions (i.e., gender, age, education) and scales. In this study, the following variables and scales were used:

No data imputation methods were used, and each analysis excluded incomplete observations. “I do not know” answers were recoded as missing values.

Descriptive statistics, including frequencies, medians, and interquartile ranges (IQR), were calculated for each variable. Since SLE measures followed a non-normal distribution, the univariate Kruskal Wallis test was used to compare recent and distant SLE levels by categorical groups (e.g., country, gender, and age group). Univariate logistic regressions were performed between gender, age group, and each SLE.

We used crude and adjusted multinomial logistic regressions to explore the relationship between recent and distant SLEs and depression and anxiety symptoms. Odds ratios (OR) and 95% confidence intervals were reported. A multiple linear regression was used to explore the impact of SLEs on quality of life. We assessed the non-multicollinearity of independent variables, the linearity of variables, the normality and independence of residuals, and the homoscedasticity assumptions.

To identify differences by age group and gender in the association between SLEs and the outcomes, we performed the same crude and adjusted multinomial logistic regressions and multiple regressions for each group: adolescents and young adults, and females and males (excluding participants who did not identify as female nor male since they represented about 1% of the sample).

Data were analysed using R, version 4.4.1 (R Core Team, 2023).

The study protocol and instruments were approved by the Institutional Review Boards (IRB) of Universidad de Buenos Aires in Argentina (dated October 2nd, 2020), Pontificia Universidad Javeriana Bogotá in Colombia (ref. FM-CIE-1138-20), Universidad Peruana Cayetano Heredia in Peru (ref. Constancia 581-33-20), and Queen Mary University of London in UK (ref. QMERC2020/02).

Informed consent or assent was obtained for each participant before data collection. For participants under 18 years old, parents or guardians provided informed consent. The consent/assent could be provided by digitally signing the consent/assent document, sending a photo or scan of the signed document, by telephone with an audio recording, adding their signature to a REDCap form, or during an in-person meeting. Each participant was assigned an ID to ensure anonymity, and their identifiable information was stored in a secure location. Respondents who completed the survey received vouchers or cash equivalent to $10 US dollars in each country’s currency.

Participants who had high scores on the depression or anxiety symptoms scales (PHQ-8 ≥ 20 and/or GAD-7 score ≥ 15) received a document with information about depression and anxiety and a list of local mental health services. Additionally, researchers implemented risk management protocols to ensure the participants’ safety and well-being.

The most frequent recent and distant SLEs were related to public safety issues or financial distress, which reflect the context of high delinquency rates and inequality in Latin America and the Caribbean. This region has the highest delinquency rates globally (Machado and Valdés, 2023), and nearly one-third of the region’s population is living in poverty, a percentage that rises to 42.5% in the case of children and adolescents (ECLAC, 2023). Additionally, our sample consisted of young people from deprived neighbourhoods where crime and financial hardship are usually more common.

Other frequent SLEs were the illness or death of family members. Data gathering was during the COVID-19 pandemic when health complications and death of relatives were common globally. South America had the highest rate of children and adolescents who lost their primary or secondary caregivers due to COVID-19 (Hillis et al., 2021). The loss rate was the highest in Peru (14 for every 1,000 children). Therefore, it is likely that this global context influenced the prevalence of death and severe illness of participants’ relatives.

Our results also highlight significant gender disparities in the experience of SLEs. The proportion of young women reporting sexual harassment is between 3.5 (for recent SLEs) to almost 6 times (for distant SLEs) higher than young men. Violence against women affects 1 in 3 women in Latin America and the Caribbean (World Health Organization, 2021b), and Bogotá and Lima are listed as two of the most unsafe capital cities for women in public places (Plan International, 2018). A Colombian study found that 83% of women were harassed while using public transportation at some point in their lives (Veeduría Distrital, 2023). This issue is not as reported in men—specifically heteronormative men—because of the historically sustained gender inequalities that subordinate women.

The findings of this study show that being a victim of bullying had a higher prevalence in women than men. However, the literature results are inconsistent: a study in Latin America also had our finding (Barajas Martínez et al., 2021), but in other regions (Iran and Portugal), men were more likely to be bullied than women (Silva et al., 2013; Mohseny et al., 2019).

Our findings reveal that men had more severe accidents anytime and were more frequently suspended from school in the past. It is well-documented that men have a higher risk for accidents and unintentional injuries than women (Libutzki et al., 2023; World Health Organization, 2023). Since early developmental stages, males are socialised into engaging in more risky behaviour than females and tend to be supervised less by older adults (Udry, 1998; Chou et al., 2022). Additionally, boys are more likely to be suspended or expelled from school than girls (Wallace et al., 2008; Skiba et al., 2014), consistent with our findings.

Data show that an additional recent and distant SLE increases the odds of having mild, moderate, and severe symptoms of depression and anxiety. Moreover, more SLEs were associated with a lower quality of life. Previous research has documented the relationship between the number of SLEs and depression and anxiety symptoms (Heredia-Ancona et al., 2011; Asselmann et al., 2016; Barajas Martínez et al., 2021; Ji et al., 2021; Zhao et al., 2023) and quality of life (Pocnet et al., 2016; Tang et al., 2022). SLEs are perceived as a source of threat or harm that requires the individual to adapt (Cohen et al., 2019). Each additional SLE can add to the overall burden of change (Beards et al., 2020; Cohen et al., 2019). Nevertheless, we must consider that some events are recurrent—such as bullying, financial hardship, and sexual harassment—and can lead to chronic stress conditions and associated health risks.

Recent SLEs’ impact on depression and anxiety symptoms and quality of life was stronger than distant SLEs, which has been previously reported (Riese et al., 2014; Nishikawa et al., 2018). Over time, the influence of distant SLEs may diminish as individuals adapt and the events become less central to their current lives (Ormel et al., 2001; Riese et al., 2014). However, exposure to traumatic events during early developmental stages is associated with depressive symptoms, post-traumatic stress symptoms, and other mental health disorders later in life (Phillips et al., 2015; Asselmann et al., 2016; Dunn et al., 2017; Gårdvik et al., 2021). SLEs, caused by people, that are perceived as intentionally harmful, such as physical abuse, neglect, and school violence, tend to be the most traumatic and can lead to long-term mental health issues (Center for Substance Abuse Treatment (US), 2014; Baldwin et al., 2024).

When considering age as a factor that may influence the relationship between SLEs and mental health outcomes, we found that adolescents had higher odds of experiencing symptoms of depression and anxiety when a recent SLE increases compared with young adults. These results are consistent with previous studies (Mann et al., 2014; Nishikawa et al., 2018). Adolescents are finding their identity, are sensitive to social stimuli, and are going through rapid physical and biological changes: the prefrontal cortex, associated with responsible decision-making and impulse control, continues to develop during this stage (Steinberg, 2005; Blakemore, 2012). External stressors may have a higher psychological burden in this age group because these changes provoke uncertainty and instability (Blakemore, 2012; Mann et al., 2014). On the contrary, young adults may have developed better resources to overcome the stress generated by SLEs because of their accumulated life experiences and prefrontal cortex development (Blakemore, 2012; Mann et al., 2014).

Our study did not find gender as a significant moderator of the relationship between SLEs and mental distress outcomes. Although some studies have found that SLEs have a greater impact on depression, anxiety, post-traumatic stress disorder, and neuroticism among women (Mann et al., 2014; Riese et al., 2014), this finding has not been consistently observed in other studies (Davis et al., 1999; Cohen et al., 2019). Further research is needed to explore the role of gender in SLEs and mental health outcomes.

As Cohen et al. (2019) stated, “stressful events do not fall randomly from the sky”; social and environmental circumstances influence them. For instance, compared to high socioeconomic status neighbourhoods, low socioeconomic status neighbourhoods are marked by more frequent and severe stressor exposures (Evans and Kim, 2010; Cohen et al., 2019).

Many of the most common SLEs experienced by adolescents and young adults, such as being a victim of a robbery, suffering a serious illness or injury, going through financial problems, or being a victim of sexual harassment/bullying, are long-standing, multi-causal, and complex problems. These issues are unlikely to be eradicated and require structural changes, as well as strong financial and political commitment. However, there are measures that national and local governments can implement to prevent them and reduce their incidence. For example, governments should focus on crime prevention and prioritise sustainable and safe public spaces where community and leisure activities can occur (Quintero Cordero, 2020). Regarding sexual harassment and bullying, most suffered by women, addressing and raising awareness about these issues early on can lead to long-term positive change. Previous interventions, such as “Benzies and Batchies”, have been effective in preventing sexual harassment in adolescents in different regions, such as Brazil, the Netherlands, and the US (de Lijster et al., 2016; Cruz et al., 2023). Regarding anti-bullying interventions, although rigorous studies in LMICs are still scarce (Sivaraman et al., 2019), the evidence from high-income countries reveals that anti-bullying programmes can be effective at reducing bullying in schools by taking a whole-school approach, supporting students to develop social and emotional competencies, learning ways to respond to bullying behaviours, providing support and training to school staff, and ensuring systematic implementation and evaluation (Centre for Education Statistics and Evaluation, 2023).

The evidence shows that for many mental health disorders, there is time to identify early symptoms and intervene since the first signs can precede a full diagnosis by up to 3 years. Moreover, since the onset of the first symptoms is usually earlier in children who eventually develop a full diagnosis, implementing early measurements could identify those most likely to go on to a full disorder (Costello, 2016). Performing a universal screening in educational settings to identify those youth at risk or currently experiencing mental disorders is a measure for which there is some available evidence. Still, it must be accompanied by the guarantee of care provision and careful protocols to avoid stigmatization (Mental Health America, 2016).

In addition to reducing the occurrence of some SLEs and early detecting young people at higher risk of developing mental disorders, efforts should be made to strengthen protective factors among children, adolescents, and young adults in deprived communities to enhance the management of stress and reduce its detrimental impact on their mental health and quality of life. There are evidence-based interventions to improve, for example, resilience, social support, and social capital (Ordóñez Barba and Ruiz Ochoa, 2015; Dray et al., 2017; García-Carrión et al., 2019; Tasijawa and Siagian, 2022), that could be implemented in schools, universities, communities, and community-based organisations.

Finally, the larger impact of recent SLEs on the well-being of adolescents compared to distant SLEs and young adults could serve to prioritise actions directed at teenagers and give special attention, in classrooms and community organisations, to recent stressful events experienced by this age group.

The study has some strengths to highlight. First, we assessed, with standardised methods, more than 2000 youth participants from underrepresented deprived communities in Latin America. The SLEs’ relationship with mental health has been vastly explored in high-income countries, and in Latin America, studies tend to have small sample sizes. Second, we explored SLEs such as those related to criminality and aggressive events (i.e., bullying, sexual harassment), which may not be as common in high-income countries. Third, we identified which SLEs were the most common in a sample from Latin America, which to our knowledge, has not been previously explored. Fourth, we have provided new insights in the LAC region on the differential impact of recent versus distant SLEs, and comparisons of their impact on the well-being of different age and gender sub-groups.

Our study also has some limitations. Since this was an observational cross-sectional study, we cannot argue causality between SLEs and our outcomes. However, at least regarding the distant SLEs, we know that their occurrence was prior to assessing the mental health outcomes. A second limitation is that we rely on self-reported questionnaires, which can imply memory and social desirability bias. A third limitation is that the high prevalence of depression and anxiety due to the study design and the COVID-19 context may overestimate the association between the variables. Finally, the total SLE scores suggest that each event has the same weight when predicting mental health outcomes, even though we know that some events are usually more severe and can be experienced differently by each person.