Open access Original research

CHILD AND ADOLESCENT MENTAL HEALTH

Do coping mechanisms moderate the effect of stressful

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
life events on depression and anxiety in young people?
A case–control study from Latin America
Georgie Hudson ‍ ‍,1 Catherine Fung,2 Diliniya Stanislaus Sureshkumar,2
Carlos Gómez-­Restrepo,3,4 José Miguel Uribe-­Restrepo,3 Karen Ariza-­Salazar,4
Francisco Diez-­Canseco,5 Liliana Hidalgo-­Padilla,5 Mauricio Toyama,5
Luis Ignacio Brusco,6 Natividad Olivar,6 Santiago Lucchetti ‍ ‍,6 Stefan Priebe,7
James B Kirkbride ‍ ‍1

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
► Additional supplemental ABSTRACT
material is published online Background Stressful life events (SLEs) are associated WHAT IS ALREADY KNOWN ON THIS TOPIC
only. To view, please visit the ⇒ Stressful life events can increase the risk of
journal online (https://​doi.​ with increased risk of depression or anxiety. Coping
org/​10.​1136/​bmjment-​2024-​ mechanisms may moderate this relationship but little is experiencing mental health problems. By
301087). known on this topic in young people or in Latin America. using certain coping mechanisms, this risk
1 Aim To investigate whether coping strategies predict may be decreased, but coping mechanisms
Division of Psychiatry, UCL,
London, UK odds of depression and/or anxiety and moderate the are not often looked at in young people, who
2
Unit for Social and Community relationship between SLEs and depression and/or anxiety are exposed to different events and may cope
Psychiatry, Queen Mary in young people in Peru, Lima and Bogotá. differently to adults. In addition, most of this
University of London, London, research has been conducted in the Global
UK
Method Using case–control data from people
3
Department of Psychiatry aged 15–24, we used logistic regression to examine West, so it is unclear if these relationships will
and Mental Health, Pontificia associations between coping mechanism, SLEs and be different here.
Universidad Javeriana, Bogota, caseness for depression or anxiety, adjusting for
Colombia WHAT THIS STUDY ADDS
4 sociodemographic and socioeconomic factors. We
Department of Clinical ⇒ We found that young people with symptoms
Epidemiology and Biostatistics, included interaction terms to model whether this
association varied depending on coping mechanisms of depression and/or anxiety in Latin America
Hospital Universitario San
Ignacio, Bogota, Colombia (positive cognitive restructuring, problem focused, are less likely to use positive, active coping
5
CRONICAS Center of support seeking, distraction, avoidant). strategies and more likely to use avoidance as
Excellence in Chronic Diseases,
Results We included 1437 cases and 965 controls. a coping strategy. Using these active coping
Universidad Peruana Cayetano strategies (support seeking, problem-­focused
Heredia, Lima, Peru Cases reported less use of positive cognitive restructuring
6
Department of Psychiatry and (OR 0.66; 95% CI 0.57 to 0.75) and problem-­focused coping and positive cognitive restructuring)
Mental Health, University of coping (OR 0.82; 95% CI 0.73 to 0.93), and more use appears to reduce the impact of stressful life
Buenos Aires, Buenos Aires, of avoidance than controls (OR 1.33; 95% CI 1.19 to events on mental health.
Argentina
7
Unit for Social and Community 1.50) in adjusted models. They had greater odds of HOW THIS STUDY MIGHT AFFECT RESEARCH,
Psychiatry, East London NHS reporting lifetime (OR 1.07; 95% CI 1.04 to 1.10) and PRACTICE OR POLICY
Foundation Trust, London, UK past-­year (OR 1.05; 95% CI 1.01 to 1.10) SLEs than
⇒ By encouraging and teaching the use of support
controls. We found weak but consistent evidence of
Correspondence to effect modification; the association between lifetime SLEs seeking, problem-­focused coping and positive
Georgie Hudson, Division of and case–control status was stronger in those who used cognitive restructuring coping, we may be
Psychiatry, UCL, London W1T
less support seeking (p=0.09), problem-­focused coping able to help prevent a decline in mental health
7NF, UK; ​georgie.​hudson.​22@​ after exposure to upsetting events. Further
ucl.​ac.​uk (p=0.08) or positive cognitive restructuring (p=0.09).
Conclusions Relationships between SLEs, coping longitudinal research is needed to confirm any
Received 21 March 2024 mechanisms and depression/anxiety appear similar in causal relationships between coping strategy
Accepted 19 December 2024
these Latin American cities to other contexts. Active and depression or anxiety.
coping strategies may ameliorate the impact of SLEs on
mental health of young people.
depressive or anxiety disorders by 2.48 and 1.68
times, respectively.2 Nonetheless, most studies of
© Author(s) (or their childhood trauma and mental health to date have
employer(s)) 2025. Re-­use INTRODUCTION been conducted in the Global North, including
permitted under CC BY. There are strong associations between stressful 89% of the studies in the above meta-­analysis2; only
Published by BMJ Group.
life events (SLEs) in childhood and the chances 5% included people in Latin America. This meta-­
To cite: Hudson G, Fung C, of developing depressive or anxiety symptoms or analysis found that in non-­clinical population-­level
Sureshkumar DS, et al. BMJ disorders.1 For example, a meta-­ analysis found studies, all forms of child maltreatment (physical
Ment Health 2025;28:1–8. that child maltreatment increased the risk of abuse, sexual abuse, emotional abuse, neglect and
Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087    1
 Open access
domestic violence) were associated with depressive disorders, we used the Unsatisfied Basic Needs Index (NBI14) to estimate
and most with anxiety disorders. However, this meta-­analysis the proportion of households in each district experiencing unmet
did not consider SLEs beyond maltreatment, including parental needs. The bottom 50% of districts in HDI or NBI rankings were

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
death or experiencing a serious illness or injury. selected for sample recruitment.
It is possible that the relationship between SLEs and common We recruited participants using convenience sampling. At
mental disorders differs across continents and contexts. For each site, we aimed to recruit 340 young people aged 15–16 and
example, exposure to childhood trauma and adversities has been 20–24 who met threshold criteria for depression or anxiety (see
reported to be higher in Latin America than other contexts.3 below; ie, cases) and 340 people without (ie, controls). For full
Reasons for this may include greater risk of exposure to poverty, details, see Gómez-­Restrepo et al.10
poor education and housing, low employment rates or substance
use.4 Due to more limited financial resources and mental health- Participants
care investment, young people in Latin America rarely receive Participants had to be aged between 15 and 16 or 20 and 24
formal mental healthcare.5 Despite this, most young people years when providing informed consent, be resident in an
appear to recover from depression within a year.6 It is unclear included district and have capacity to provide informed consent/
what traits may reduce mental health symptoms for these young assent. Exclusion criteria were diagnosis of psychosis, bipolar
people. disorder, schizophrenia, or cognitive impairment, or illiteracy.

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
One possibility is that young people in these contexts use Informed consent was obtained from all participants aged 18
different types or levels of coping mechanisms. In a study from or over, and for those under 18, assent was obtained alongside
the Netherlands, self-­ blame, rumination and catastrophising informed consent from their parent/guardian.
were all associated with worse depressive symptoms in adoles-
cents, while positive reappraisal appeared to be protective.7 They Measures
also found that the relationship between stress and depressive All measures used self-­reported questionnaires.
symptoms was stronger in those who employed coping strategies
of self-­blame or rumination.7 Other evidence suggests that active
Outcome
and problem-­solving coping strategies positively impact mental
All participants completed the Patient Health Questionnaire-­815
health, whereas emotion-­focused and passive (including distrac-
and the General Anxiety Disorder-­716 to screen for depression
tion and avoidance) strategies have negative impacts.8 Despite
and anxiety, respectively. We defined cases as participants who
this, coping strategies in young people have received little atten-
scored 10 or more on either instrument, consistent with estab-
tion,9 and it is unclear to what degree different coping strategies
lished cut-­offs.15 16 Participants who did not meet these thresh-
moderate associations between SLEs and depression or anxiety
olds were defined as controls.
in young people outside of the Global North.
We sought to clarify these issues in a sample of young people
in three major cities in Latin America. Using these data, we have Exposures
previously demonstrated that young people with depression or An adapted version of Heubeck and O’Sullivan’s scale17 was used
anxiety are more likely to experience SLEs.10 11 In this paper, to measure the number of SLEs experienced, which captures
we sought to address whether different forms of coping predict whether participants have experienced any of 30 SLEs in the
the odds of experiencing depression and/or anxiety, and whether past year or more than 1 year ago. This included events such
these moderated the relationship between SLEs and odds of as the participant/someone close to them experiencing a serious
depression or anxiety. illness or injury, changing schools, serious financial problems
We hypothesised that young people with depression and/ and bullying. We assessed two exposure measures: the number
or anxiety would have decreased odds of using active coping of SLEs experienced in the participant’s lifetime and the number
strategies (positive cognitive restructuring, support seeking and of SLEs experienced in the past year.
problem-­focused coping) and increased odds of passive coping
strategies (avoidance and distraction coping). We hypothe- Moderators
sised that active coping strategies would lessen the association We assessed participants’ coping strategies using a version of the
between SLEs (lifetime or past year) and depression and anxiety Children’s Coping Strategy Checklist, modified by Cline et al.18
caseness, while greater use of passive coping strategies would This assessed how frequently participants employed different
increase this association. coping strategies when experiencing problems. The 26 items
mapped onto five different coping strategies: positive cogni-
tive restructuring, problem-­focused coping, distraction, avoid-
METHODS ance and support-­seeking strategies. The scores for each coping
Study design strategy were generated by averaging items on each domain. Due
We used data from the OLA Project10 12; a large, multisite study to an oversight during data collection, one item was not included
based in three study locations in Latin America: Buenos Aires (‘I try to figure out why things like this happen’; problem-­focused
(Argentina), Bogotá (Colombia) and Lima (Peru). Two groups coping), reducing the number of items to 25.
of participants were recruited: those aged 15–16 and those aged
20–24.
Confounders
Confounders were selected based on prior research7 9 and theo-
Sampling strategy and inclusion criteria retical knowledge. We constructed a directed acyclic graph
We recruited participants living in the most deprived half of (online supplemental figure 1) to model the hypothesised causal
administrative districts in each city. In Lima and Bogotá, we used pathways and to identify the following relevant confounding
the United Nations Development Programme’s Human Devel- variables:
opment Index (HDI13) to estimate the proportion of households 1. Gender (male/female/other).
in each district achieving basic living standards. In Buenos Aires, 2. Age group (aged 15–16 or 20–24).
2 Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087
 Open access
3. Socioeconomic status (SES) measured via three variables: between each exposure and covariate and case–control status.
i. Number of people per bedroom. Second, we fitted multivariable regressions, adjusting for
ii. Whether the participant had health insurance (yes/no/I confounders, including other coping strategies. We present ORs

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
don’t know). Types of health insurance differed across and 95% CIs. Third, we included interaction terms in our regres-
the three countries due to availability, but reflected a mix sion models to model the effect modification of the association
of public, social, police/armed forces and private health between SLEs (lifetime and past year) and case–control status
insurance policies. by the level of each coping strategy employed. All confounding
iii. Highest parental education of either caregiver (no formal variables and other coping strategies were controlled for in these
education/primary/secondary/higher education). analyses. Statistical significance of the interactions was deter-
4. Parental history of mental health treatment: The participant mined via inspection of whether Wald p values met the threshold
reported whether their parent(s) had received mental health of p<0.05. We visualised any putatively relevant continuous by
treatment (one parent/both parents/neither parent). continuous interactions graphically using marginal plots.
5. Neighbourhood safety: We estimated mean scores from sev- For all models, we standardised continuous variables (coping
en items related to neighbourhood safety from the Short strategies, neighbourhood safety and perceived social support)
Social Capital Assessment Tool,19 which was specifically de- to have a mean of 0 and SD of 1. ORs represented the change
signed for use in low-­income countries, and validated for use in odds of caseness associated with a one SD change in the

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
in Latin America.20 Items were scored yes/no. One item was exposure.
subsequently reverse coded so for all questions, one repre- All analyses were performed using Stata V.17.
sented higher neighbourhood safety than zero.
6. Drug and alcohol use: Nine items from the Alcohol, Smoking
and Substance Involvement Screening Test,21 designed for RESULTS
adolescents, were used to assess drug and alcohol use. Par- Sample characteristics
ticipants were asked how often they had used various illegal We recruited 2402 participants, comprising 1437 cases (59.8%)
drugs in the last 3 months. The drug most frequently used and 965 controls (40.2%).
was entered in analyses. Drug and alcohol use were included Compared with controls, cases were more likely to have expe-
separately. rienced SLEs in their lifetime and past year; use lower levels
7. Perceived social support: The Multidimensional Scale of of all five coping mechanisms; have higher levels of drug and
Perceived Social Support22 is a 12-­item measure, with high alcohol use; have a parent in receipt of mental health treatment;
scale reliability exceeding 0.90.23 Total perceived social sup- perceive less social support; report lower neighbourhood safety;
port was estimated by averaging scores from all 12 items. have health insurance; and have parents with higher levels of
In this study, age, gender and SES were considered the most education (all p<0.001; table 1).
essential covariates to include in our models; however, all were
important and included in all analyses. We considered including Missing data
other variables as confounders such as parental drug use and Of the 2402 participants, 858 (35.7%) had some missing
attachment type; however, through discussion and a search of covariate or exposure data, although the percentage of item-­level
the literature it was decided that these variables were unlikely to missing data was low (0.92%). Those with missing data were
be confounders between SLEs and anxiety/depression. more likely to be in the younger age group; have lower drug use;
use lower levels of positive cognitive restructuring and support-­
Statistical analysis seeking strategies; and perceive less social support (all p<0.001;
Missing data online supplemental table 2). Those with missing data were no
Participant responses of ‘I don’t know’ were recoded as missing more likely to be cases than controls (χ2(1)=2.2, p=0.14).
data. We used multiple imputation by chained equations to
impute missing covariate, moderator and exposure data. 20
Main effects of coping mechanisms and SLEs on depression
datasets were imputed using logistic, linear, multinomial logistic
and anxiety
and ordinal logistic regressions including auxiliary variables.
In our univariable models, young people with depression or
Our primary analysis investigated the association between
anxiety had decreased odds of using all forms of coping—posi-
coping mechanism use and case–control status using the imputed
tive cognitive restructuring (OR 0.60; 95% CI 0.55 to 0.65),
dataset, combined using Rubin’s rule,24 which we checked
problem-­ focused coping (OR 0.65; 95% CI 0.60 to 0.71),
against our complete case analysis results in a sensitivity analysis.
distraction strategies (OR 0.85; 95% CI 0.78 to 0.92), avoid-
See the online supplemental material for further information.
ance strategies (OR 0.91; 95% CI 0.84 to 0.99) and support-­
seeking strategies (OR 0.70; 95% CI 0.64 to 0.76)—compared
Analysis procedure with controls. After adjustment for covariates, the relationship
We summarised continuous variables by estimating the median between caseness and distraction (OR 1.07; 95% CI 0.96 to
and IQR, given evidence of non-­ normally distributed data 1.18) and support-­seeking strategies (OR 0.93; 95% CI 0.83
using the Shapiro-­Wilk test (online supplemental table 1). We to 1.04) was no longer statistically significant, while the effect
compared cases and controls on their exposure and covariate of avoidance strategies changed direction, whereby cases had
values using the χ2 test for categorical data and Mann-­Whitney increased odds of using this coping mechanism (OR 1.33; 95%
U test for continuous data. We also used these tests to investigate CI 1.19 to 1.50). Cases remained less likely than controls to use
differences between participants with and without missing data positive cognitive restructuring (OR 0.66; 95% CI 0.57 to 0.75)
on exposure and confounding variables. and problem-­focused coping strategies (OR 0.82; 95% CI 0.73
We used logistic regression to establish whether coping strate- to 0.93) in multivariable models. Our adjusted logistic regression
gies and SLEs were associated with case–control status. First, we model following multiple imputation showed each additional
fitted univariable models to examine the unadjusted association lifetime (OR 1.07; 95% CI 1.04 to 1.04) and past-­year (OR
Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087 3
Open access

Table 1 Participant demographics and summary statistics

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
Number of lifetime SLEs experienced, median (IQR) 9 (6–13) 8 (5–11) 10 (7–13) z=−10.7,
p<0.001
Number of SLEs experienced in the past year, median (IQR) 2 (1–4) 2 (1–3) 3 (1–5) z=−9.1,
p<0.001
Frequency of coping strategy used, median (IQR)
 Positive cognitive restructuring 2.3 (2.0–2.8) 2.7 (2.2–3.0) 2.3 (1.8–2.7) z=12.3,
p<0.001
 Problem-­focused coping 2.6 (2.0–3.0) 2.8 (2.2–3.2) 2.4 (2.0–3.0) z=10.0,
p<0.001
 Distraction strategies 2.0 (1.5–2.5) 2.0 (1.5–2.5) 2.0 (1.5–2.5) z=3.9,
p<0.001
 Avoidance strategies 2.7 (2.3–3.0) 2.7 (2.3–3.0) 2.7 (2.3–3.0) z=2.4,
p=0.02
 Support-­seeking strategies 2.0 (1.5–2.5) 2.0 (1.5–2.8) 1.8 (1.3–2.3) z=9.3,

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
p<0.001
Gender, n (%)
 Male 815 (34.0) 428 (44.4) 387 (26.9) χ2(2)=80.6,
 Female 1560 (65.0) 533 (55.2) 1027 (71.5) p<0.001
 Other 24 (1.0) 4 (0.4) 20 (1.4)
 Missing 3 (0.1) – 3 (0.2)
Age group, n (%)
 Young group (15–16 years) 1080 (45.0) 435 (45.1) 645 (44.9) χ2(1)=0.009,
 Older group (20–24 years) 1322 (55.0) 530 (54.9) 7892 (55.0) p=0.93
Age, median (IQR)
 Young group (15–16) 15 (15–16) 15 (15–16) 15 (15–16) z=0.4,
p=0.71
 Older group (20–24) 21 (20–23) 21 (20–23) 21 (20–23) z=−0.6,
p=0.57
Centre, n (%)
 Argentina 621 (25.9) 280 (29.0) 341 (23.7) χ2(2)=15.4,
 Colombia 965 (40.2) 344 (35.7) 621 (43.2) p<0.001
 Peru 816 (34.0) 341 (35.3) 475 (33.1)
Illicit drug use in the last 3 months, n (%)
 Not used 1964 (81.8) 848 (87.9) 1116 (77.7) χ2(4)=40.4,
 Once or twice 182 (7.6) 47 (4.9) 135 (9.4) p<0.001
 Monthly 66 (2.8) 19 (2.0) 47 (3.3)
 Weekly 77 (3.2) 18 (1.9) 59 (4.1)
 Every day/almost every day 99 (4.1) 29 (3.0) 70 (4.9)
 Missing 14 (0.6) 4 (0.4) 10 (0.7)
Alcohol use in the last 3 months, n (%)
 Not used 1082 (45.1) 495 (51.3) 587 (40.9) χ2(4)=37.0,
 Once or twice 756 (31.5) 297 (30.8) 459 (31.9) p<0.001
 Monthly 302 (12.6) 89 (9.2) 213 (14.8)
 Weekly 234 (9.7) 75 (7.8) 159 (11.1)
 Every day/almost every day 23 (1.0) 6 (0.6) 17 (1.2)
 Missing 5 (0.2) 3 (0.3) 2 (0.1)
Parent received mental health treatment, n (%)
 Neither parent 1532 (63.8) 667 (69.1) 865 (60.2) χ2(2)=17.1,
 One parent 256 (10.7) 85 (8.8) 171 (11.9) p<0.001
 Both parents 36 (1.5) 7 (0.7) 29 (2.0)
 Missing 578 (24.1) 206 (21.3) 372 (26.0)
Ever experienced, n (%)
 Low mood 1298 (54.0) 302 (31.3) 996 (69.3) χ2(1)=337.5,
p<0.001
 Anxiety 1115 (46.4) 228 (23.6) 887 (61.7) χ2(1)=338.2,
p<0.001
If ever experienced, currently experiencing, n (%)
 Low mood 769 (32.0) 89 (9.2) 680 (47.3) χ2(1)=154.8,
p<0.001
 Anxiety 789 (32.9) 109 (11.3) 680 (47.3) χ2(1)=115.6,
p<0.001

Continued
4 Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087
 Open access

Table 1 Continued
Characteristic All participants (N=2402) Controls (n=965) Cases (n=1437) Statistic

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
If experienced mental health problems, ever received mental health treatment, n (%)
 Yes 380 (23.7) 82 (20.3) 298 (24.8) χ2(1)=3.1,
 No 1232 (76.1) 321 (79.3) 911 (75.1) p=0.078
 Missing 6 (0.3) 2 (0.5) 4 (0.3)
Perceived social support, median (IQR) 4.9 (4.1–5.8) 5.3 (4.5–6.1) 4.7 (3.8–5.5) z=12.5,
p<0.001
Neighbourhood safety, median (IQR) 0.2 (0–0.3) 0.3 (0.1–0.4) 0.2 (0–0.3) z=10.3,
p<0.001
Highest parental education, n (%)
 No formal education 33 (1.4) 12 (1.2) 21 (1.5) χ2(3)=17.9,
 Some primary education 306 (12.7) 139 (14.4) 167 (11.6) p<0.001
 Some secondary education 1009 (42.0) 436 (45.2) 573 (39.9)

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
 Some higher education 955 (39.8) 335 (34.7) 620 (43.2)
 Missing 99 (4.1) 43 (4.5) 56 (3.9)
Number of people per bedroom, median (IQR) 1.6 (1.3, 2.0) 1.7 (1.3–2.0) 1.5 (1.3–2.0) z=1.1,
p=0.28
Has health insurance, n (%)
 Yes 1676 (69.8) 635 (65.8) 10 341 (72.4) χ2(1)=4.3,
 No 424 (17.7) 184 (19.1) 240 (16.7) p=0.038
 Missing 302 (12.6) 146 (15.2) 156 (10.9)
SLE, stressful life event.

1.05; 95% CI 1.01 to 1.10; table 2) SLE was associated with to report more SLEs. We found weak but consistent evidence
increased odds of depression and/or anxiety. that active coping strategies (positive cognitive restructuring,
We observed only minor differences between the analysis problem-­focused coping and support seeking) buffered against
using the imputed dataset and complete case analysis (online the increased odds of depression and/or anxiety associated
supplemental table 3), suggesting that the missing data mecha- with experiencing more lifetime SLEs. No moderating effect
nisms underlying our dataset were at least missing at random and was observed for more passive coping strategies (avoidance,
unrelated to variables not observed in this study. distraction).

The relationship between SLEs and depression and anxiety Comparison with prior literature
for different coping strategies employed Similar to Kraaij et al,7 we found the relationship between SLEs
We found weak but consistent trends to suggest that the rela- and depression was stronger for young people who used active
tionship between lifetime SLEs and depression/anxiety was coping strategies less frequently. We have shown that this rela-
modified by active coping strategies; positive cognitive restruc- tionship also exists for support-­seeking strategies and when both
turing (p=0.09), problem-­focused coping (p=0.08) and support depression and anxiety were included in our outcome definition.
seeking (p=0.09). We visualised these interactions in marginal Additionally, it appears this relationship may generalise across
plots (figure 1). These suggested that the relationship between settings, as our work in Latin America was consistent with that
increased odds of depression and/or anxiety and more lifetime in the Netherlands.7 Although the interactions we observed
SLEs was weaker for participants with greater use of support between three active coping strategies and lifetime SLEs did
seeking, problem-­focused coping and positive cognitive restruc- not achieve conventional statistical significance (p=0.08), these
turing. There was no evidence of effect modification between consistent results suggest that active coping strategies may help
lifetime SLEs and either passive coping strategy (distraction buffer against the odds of developing anxiety or depression after
p=0.37, avoidance p=0.84) on the odds of depression and/ experiencing SLEs.
or anxiety, and no evidence of effect modification with any In our sample, cases reported statistically significantly
coping style when only SLEs in the last year were considered (all lower levels of perceived social support than controls. This
p>0.35; online supplemental table 4). may mean that it is harder for young people at risk of mental
health problems to achieve the support seeking that may miti-
DISCUSSION gate this risk. One method to increase the availability is via
Principal findings social support interventions, which appear to be beneficial.
There is limited previous research investigating the impact of For example, children showed significantly reduced depres-
coping strategies on depression and/or anxiety, and whether sive symptoms following a 2-­y ear social support interven-
coping can moderate the relationship between SLEs and anxiety tion compared with no symptomatic difference in a control
and/or depression in young people, particularly in the Global group who did not receive the intervention. 25 Additionally,
South. When controlling for confounders, we found young in our study, problem-­f ocused coping appeared to moderate
people with depression and/or anxiety had decreased odds of the relationship between lifetime SLEs and odds of depres-
using positive cognitive restructuring and problem-­ focused sion/anxiety. A randomised controlled trial has shown that
coping, and increased odds of using distraction strategies. a 7-­week mindfulness-­based stress reduction programme for
Young people with depression and/or anxiety were more likely university students was associated with statistically higher
Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087 5
Open access

Table 2 Logistic regression model results for the association between SLEs, coping mechanisms, and depression or anxiety (20 imputations)
Univariable model Multivariable model*

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
95% CI 95% CI
Characteristic OR Low High OR Low High
Number of SLEs experienced in the lifetime 1.11† 1.09 1.14 1.07† 1.04 1.10
Number of SLEs experienced in the past year 1.17† 1.13 1.21 1.05† 1.01 1.10
Positive cognitive restructuring (z-­standardised) 0.60† 0.55 0.65 0.66† 0.57 0.75
Problem-­focused coping (z-­standardised) 0.65† 0.60 0.71 0.82† 0.73 0.93
Distraction strategies (z-­standardised) 0.85† 0.78 0.92 1.07 0.96 1.18
Avoidance strategies (z-­standardised) 0.91† 0.84 0.99 1.33† 1.19 1.50
Support-­seeking strategies (z-­standardised) 0.70† 0.64 0.76 0.93 0.83 1.04
Gender
 Male (Ref) 1 1
 Female 2.13† 1.08 2.54 2.07† 1.69 2.54

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
 Other 5.54† 1.88 16.34 2.57 0.81 8.20
Age group
 Young group (15–16 years) (Ref) 1 1
 Older group (20–24 years) 1.01 0.86 1.19 0.91 0.74 1.12
Illicit drug use in the last 3 months
 Not used (Ref) 1 1
 Once or twice 2.18† 1.55 3.07 1.47 0.99 2.16
 Monthly 1.88† 1.09 3.22 1.61 0.87 2.97
 Weekly 2.51† 1.47 4.28 1.79 0.95 3.35
 Every day/almost every day 1.84† 1.19 2.87 1.59 0.94 2.72
Alcohol use in the last 3 months
 Not used (Ref) 1 1
 Once or twice 1.30† 1.08 1.57 1.13 0.91 1.42
 Monthly 2.02† 1.53 2.65 1.71† 1.24 2.37
 Weekly 1.80† 1.33 2.42 1.75† 1.21 2.52
 Every day/almost every day 2.39 0.94 6.12 2.79 0.95 8.15
Parent received mental health treatment
 No (Ref) 1 1
 One parent 1.55† 1.17 2.05 1.31 0.96 1.80
 Both parents 2.73† 1.28 5.85 2.49† 1.06 5.84
Perceived social support (z-­standardised) 0.57† 0.52 0.62 0.72† 0.64 0.80
Neighbourhood safety (z-­standardised) 0.63† 0.57 0.69 0.75† 0.67 0.83
Highest parental education
 None 0.93 0.45 1.93 0.77 0.33 1.79
 Primary 0.65† 0.50 0.84 0.70† 0.51 0.96
 Secondary 0.71† 0.59 0.85 0.73† 0.59 0.91
 Higher (Ref) 1 1
Number of people per bedroom 1.01 0.93 1.10 1.06 0.96 1.18
Has health insurance
 No (Ref) 1 1
 Yes 1.29† 1.05 1.60 1.28 0.99 1.66
*Adjusted for all variables in the model.
†Indicates significance at the level of p<0.05.
SLE, stressful life event.

levels of problem-­focused coping, compared with a treat- of childhood trauma is not displayed until later in life. 27
ment as usual control group. 26 Schools and colleges may be Young people exposed to more recent SLEs may not have
well placed to identify those at risk and implement inter- had sufficient time to fully develop and implement their
ventions to improve their coping mechanisms or to provide preferred coping strategy. Given the strong relationship
them with the opportunity to seek support from an appro- we observed between SLEs in the past year and depres-
priate person. sion/anxiety, independent of lifetime SLEs, the immediate
We found no evidence of interactions between coping period after exposure to SLEs for young people may be
strategy and the number of SLEs experienced in the past a critical period for bonds to be developed to employ
year; only for the number experienced in the lifetime. support-­s eeking strategies and minimise the risk of devel-
There is evidence of a ‘sleeper effect’, whereby the impact oping anxiety or depression.
6 Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087
 Open access

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
Figure 1 Marginal plot depicting the relationship between stressful life events (SLEs) and odds of depression/anxiety caseness, depending on the
level of (a) support seeking, (b) problem-­focused coping or (c) positive cognitive restructuring used.

Strengths and limitations depression and/or anxiety. It is possible that coping strategies are
Strengths of our study included a well-­controlled case–control related to cognitive biases that are influenced by symptoms of
design, a large sample size, control for important covariates and depression or anxiety; further longitudinal evidence is required
low item-­level missing data. We also employed multiple impu- to tease out the directionality of these relationships. Nonethe-
tation techniques to account for any missing data, and results less, we hypothesised that coping strategies would moderate the
from our imputed sample were consistent with those from association between SLEs and mental health, and found stronger
our complete case analysis. We included exposure, outcome, support for this in relation to lifetime SLEs, consistent with
covariate and several auxiliary variables when performing the longitudinal evidence on this topic.29 It has also been suggested
multiple imputation. The imputation approach we used assumes that coping mechanisms may act as a mediator of the relation-
the data are at least missing at random. It is plausible that reasons ship between SLEs and psychopathology30; however, we could
for participants having missing data on exposure to SLEs were not investigate this given our study design.
related to their unobserved exposure status. For example, partic- Although we recruited a large sample, our study was likely
ipants may not have wanted to reveal they had experienced to have been underpowered to detect statistical interactions
traumatic events if they were unsure this information would between lifetime SLEs and coping strategies at the observed
be treated confidentially. However, since other variables that coefficient effect sizes. Although our observed p values for
predicted both missingness and exposure status were included in interactions between active coping strategies and lifetime SLEs
the imputation model, this increases the plausibility that the data were just outside of conventional statistical significance, these
were missing at random. Additionally, both SLE exposure and trends were consistent, and the buffering effects in the hypothe-
covariates had very low levels of missing data (except for parental sised direction. This lends credence to the possibility that active
mental health treatment), meaning any systematic differences in coping strategies can ameliorate the potential harmful effects of
missingness are unlikely to have impacted the results. Therefore, SLEs on mental health in young people, and warrants replication
the assumption of the data missing at random appears likely to in larger longitudinal studies in diverse settings.
have been met.
We also note several potential limitations which mean that
causality cannot be inferred. First, case–control studies are CONCLUSIONS
subject to the possibility of recall bias. Recall bias may have acted There was some evidence to suggest that active coping strategies
differentially, if depressed or anxious participants were more moderated the odds of depression/anxiety for young people who
or less likely to have recalled negative life events than control had experienced SLEs in three deprived Latin American settings.
participants; there is evidence that people with common mental The evidence was strongest for the impact of problem-­focused
disorders under-­report levels of childhood adversity.28 strategies in buffering the impact of SLEs on mental health, but
A second related issue is that we could not establish the this relationship may exist for positive cognitive restructuring
temporal order between SLEs, coping strategies and caseness. and support seeking as well. Schools and community organisa-
Due to the case–control design, we were not able to infer tions may be well placed to offer interventions to young people
any causal relationships between SLEs, coping strategies, and at risk of experiencing mental health problems after exposure to
Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087 7
 Open access
SLEs to improve their coping mechanism and reduce their risk. 2 Gardner MJ, Thomas HJ, Erskine HE. The association between five forms of child
However, due to the case–control design, causality cannot be maltreatment and depressive and anxiety disorders: A systematic review and meta-­
analysis. Child Abuse Negl 2019;96:104082.
inferred. Further longitudinal research is needed to confirm the

BMJ Mental Health: first published as 10.1136/bmjment-2024-301087 on 9 January 2025. Downloaded from https://mentalhealth.bmj.com on 21 June 2025 by guest.
3 Myers HF, Wyatt GE, Ullman JB, et al. Cumulative burden of lifetime adversities:
existence of a moderating effect of coping mechanisms, partic- Trauma and mental health in low-­SES African Americans and Latino/as. Psychol
ularly for young people living in deprived contexts outside the Trauma 2015;7:243–51.
Global North. 4 Patel V, Flisher AJ, Hetrick S, et al. Mental health of young people: a global public-­
health challenge. The Lancet 2007;369:1302–13.
X Georgie Hudson @Georgie_Hudson5 and James B Kirkbride @ProfKirkbride 5 Alarcón RD. Mental health and mental health care in Latin America. World Psychiatry
2003;2:54–6.
Contributors SP, LIB, CG-­R, FD-­C, Melanie Smuk, JBK, Ricardo Araya, Craig
6 Stapinski LA, Montgomery AA, Heron J, et al. Depression symptom trajectories and
Morgan, Sandra Eldridge, Paul Heritage and Victoria Bird designed the study. GH
associated risk factors among adolescents in Chile. PLoS ONE 2013;8:e78323.
and JBK conducted the data analyses and interpretation and are the guarantors. All
7 Kraaij V, Garnefski N, de Wilde EJ, et al. Negative Life Events and Depressive
authors carried out the literature search and wrote the manuscript.
Symptoms in Late Adolescence: Bonding and Cognitive Coping as Vulnerability
Funding GH is funded by the Wellcome Trust through a PhD Fellowship in Mental Factors? J Youth Adolesc 2003;32:185–93.
Health Science (218497/Z/19/Z). The OLA Project and other authors are supported by 8 Folkman S, Moskowitz JT. Coping: pitfalls and promise. Annu Rev Psychol
the Medical Research Council (grant number: MR/S03580X/1). 2004;55:745–74.
Competing interests None declared. 9 Thapar A, Eyre O, Patel V, et al. Depression in young people. The Lancet
2022;400:617–31.

Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies.
Patient and public involvement statement Young people from Bogotá, Lima, 10 Gomez-­Restrepo C, Diez-­Canseco F, Brusco LI, et al. Risk and protective factors for
and Buenos Aires were highly involved in the design and preparation of this study. depression and anxiety among young people from deprived urban neighbourhoods
Each centre had a Lived Experience Advisory Panel made up of young people aged in south america: a case-­control study in bogotá, buenos aires and lima. SSRN
15–24 with lived experience of depression and/or anxiety. The advisory panels, 2024.
along with professionals, assisted in the decisions of which traits and resources 11 Diez-­Canseco F, Carbonel A, Bernabe-­Ortiz A, et al. Association between stressful
may reduce mental health symptoms, and therefore should be investigated. The life events and depression, anxiety, and quality of life among urban adolescents and
panels additionally piloted the assessment battery to determine its feasibility and young adults in Latin America. Front Psychol 2024;15:1466378.
acceptability. For full details on the patient and public involvement strategy, see 12 Priebe S, Fung C, Brusco LI, et al. Which resources help young people to prevent and
Priebe et al. (2021). overcome mental distress in deprived urban areas in Latin America? A protocol for a
Patient consent for publication Not applicable. prospective cohort study. BMJ Open 2021;11:e052339.
13 United Nations development programme human development reports. Available:
Ethics approval This study involves human participants and was approved by
https://hdr.undp.org/data-center/human-development-index#/indicies/HDI [Accessed
the Institutional Review Boards of Universidad de Buenos Aires (2 October 2020),
08 Dec 2022].
Pontificia Universidad Javeriana (FM-­CIE-­1138-­20), Universidad Peruana Cayetano
Heredia (Constancia 581-­33-­20) and the Research Ethics Committee of Queen 14 Santos ME. Measuring multidimensional poverty in Latin America previous experience
Mary University of London (QMERC2020/02). The authors assert that all procedures and the way forward. 2014.
contributing to this work comply with the ethical standards of the relevant national 15 Kroenke K, Strine TW, Spitzer RL, et al. The PHQ-­8 as a measure of current depression
and institutional committees on human experimentation and with the Helsinki in the general population. J Affect Disord 2009;114:163–73.
Declaration of 1975, as revised in 2008. Participants gave informed consent to 16 Spitzer RL, Kroenke K, Williams JBW, et al. A brief measure for assessing generalized
participate in the study before taking part. anxiety disorder: the GAD-­7. Arch Intern Med 2006;166:1092–7.
17 Heubeck B, O’sullivan C. An exploration into the nature, frequency and impact of
Provenance and peer review Not commissioned; externally peer reviewed. school hassles in the middle school years. Aust Psychol 1998;33:130–7.
Data availability statement Data are available upon reasonable request. 18 Cline JI, Belsky J, Li Z, et al. Take your mind off it: Coping style, serotonin transporter
Individual participant data and the data dictionary will be made available to others. linked polymorphic region genotype (5-­HTTLPR), and children’s internalizing and
This will be individual participant data that underlie the results reported in this externalizing problems. Dev Psychopathol 2015;27:1129–43.
article after deidentification (text, tables, figures and appendices). Additionally, the 19 De Silva MJ, Huttly SR, Harpham T, et al. Social capital and mental health: a
study protocol will be made available. Data will be available beginning 9 months comparative analysis of four low income countries. Soc Sci Med 2007;64:5–20.
and ending 36 months following article publication to researchers who provide a 20 De Silva MJ, Harpham T, Tuan T, et al. Psychometric and cognitive validation of a social
methodologically sound proposal, for analyses conducted to achieve the aims of the capital measurement tool in Peru and Vietnam. Soc Sci Med 2006;62:941–53.
approved proposal. Proposals should be directed to the authors to gain access; data 21 Humeniuk R, Henry-­Edwards S, Ali R, et al. The Alcohol, Smoking and Substance
requestors will need to sign a data access agreement. Involvement Screening Test (ASSIST): manual for use in primary care; Geneva World
Health Organization; 2010. Available: https://apps.who.int/iris/handle/10665/44320
Supplemental material This content has been supplied by the author(s). It
22 Zimet GD, Powell SS, Farley GK, et al. Psychometric characteristics of the
has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have
Multidimensional Scale of Perceived Social Support. J Pers Assess 1990;55:610–7.
been peer-­reviewed. Any opinions or recommendations discussed are solely those
23 Calderón C, Ferrando PJ, Lorenzo-­Seva U, et al. Multidimensional Scale of Perceived
of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
responsibility arising from any reliance placed on the content. Where the content Social Support (MSPSS) in cancer patients: psychometric properties and measurement
includes any translated material, BMJ does not warrant the accuracy and reliability invariance. Psicothema 2021;33:131–8.
of the translations (including but not limited to local regulations, clinical guidelines, 24 Rubin DB. Multiple Imputation for Nonresponse in Surveys. Wiley, 1987.
terminology, drug names and drug dosages), and is not responsible for any error 25 Wang J, Wang X, Li P, et al. Effects of social support and positive psychology
and/or omissions arising from translation and adaptation or otherwise. intervention on depression and subjective well-­being among the left-­behind children:
a two-­year longitudinal study. Curr Psychol 2023;42:30008–19.
Open access This is an open access article distributed in accordance with the 26 Halland E. Mindfulness training improves problem-­focused coping in psychology
Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits and medical students: Results from a randomized controlled trial. Coll Stud J
others to copy, redistribute, remix, transform and build upon this work for any 2015;49:387–98. Available: https://www.ingentaconnect.com/content/prin/csj/2015/​
purpose, provided the original work is properly cited, a link to the licence is given, 00000049/00000003/art00007#expand/collapse
and indication of whether changes were made. See: https://creativecommons.org/​ 27 Holmes MR. The sleeper effect of intimate partner violence exposure: long-­term
licenses/by/4.0/. consequences on young children’s aggressive behavior. J Child Psychol Psychiatry
2013;54:986–95.
ORCID iDs
28 Hardt J, Rutter M. Validity of adult retrospective reports of adverse childhood
Georgie Hudson http://orcid.org/0000-0002-5002-4890
experiences: review of the evidence. J Child Psychol Psychiatry 2004;45:260–73.
Santiago Lucchetti http://orcid.org/0009-0005-9445-1412
29 Kingsbury M, Clayborne Z, Colman I, et al. The protective effect of neighbourhood
James B Kirkbride http://orcid.org/0000-0003-3401-0824
social cohesion on adolescent mental health following stressful life events. Psychol
Med 2020;50:1292–9.
REFERENCES 30 van Dijk FA, Schirmbeck F, Boyette L-­L, et al. Coping styles mediate the association
1 Turner RJ, Lloyd DA. Stress burden and the lifetime incidence of psychiatric disorder in between negative life events and subjective well-­being in patients with non-­affective
young adults: racial and ethnic contrasts. Arch Gen Psychiatry 2004;61:481–8. psychotic disorders and their siblings. Psychiatry Res 2019;272:296–303.

8 Hudson G, et al. BMJ Ment Health 2025;28:1–8. doi:10.1136/bmjment-2024-301087