Adverse childhood experiences, coping and self-harm among adolescents attending public psychiatry and mental health clinics in Malaysia: a cross-sectional study

Abstract

Background

Adverse childhood experiences (ACEs) are increasingly recognised as key determinants of self-harm among adolescents, particularly within psychiatric and mental health service settings in Malaysia, where coping resources may shape vulnerability and resilience.

Setting

A cross-sectional study was conducted among adolescents referred to psychiatry and mental health services in Malaysia.

Data and participants

Of 1244 eligible patients aged 10–24 years, 1164 completed a clinician-administered, interviewer-guided questionnaire.

Outcome measures

Self-harm status was assessed in clinical care and was compared across ACE burden, individual ACE domains and coping-related protective factors through a clinician-led interview. Multivariable logistic regression was used to examine adjusted associations with self-harm.

Results

Among 1164 participants, 827 (71.0%) reported lifetime self-harm. In adjusted analyses, a greater burden of ACEs was strongly associated with higher odds of self-harm. Compared with participants with no ACE exposure, the odds increased in a dose–response manner, from an adjusted OR (aOR) of 2.54 (95% CI 1.27 to 5.20) among those with 2–3 ACEs to 6.14 (95% CI 3.13 to 12.40) among those with ≥4 ACEs. Several coping-related factors were associated with lower odds of self-harm, including the ability to cope with stress (aOR 0.32, 95% CI 0.23 to 0.45), a sense of responsibility towards children, loved ones or pets (aOR 0.71, 95% CI 0.51 to 0.98) and perceived social support (aOR 0.63, 95% CI 0.40 to 0.97). In secondary analyses of individual ACE components, violence against household members, emotional abuse, sexual abuse, bullying and witnessing community violence were independently associated with higher odds of self-harm.

Conclusions

This study showed high prevalence of self-harm among adolescents attending the psychiatric and mental health services. A greater cumulative burden of ACEs was associated with increased odds of self-harm, whereas protective factors including effective stress coping, a sense of responsibility towards loved ones or pets and social support were associated with reduced odds. These findings emphasised the importance of trauma-informed assessment and interventions that enhance coping capacities and social connectedness among young people presenting to mental health services.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This multisite clinical study provides evidence from Malaysian government psychiatry and mental health services on how specific adverse childhood experience domains and coping resources relate to self-harm among young people.

• The analysis considered both adversity exposures and potentially modifiable protective factors for consideration in future interventions.

• The cross-sectional design limits causal inference, and measurements were retrospective, self-reported and may be affected by recall and reporting bias.

• Findings may not be generalisable to community populations.

Introduction

Self-harm is recognised as a predictor of suicidal behaviour.¹ Among adolescents presenting to mental health services, self-harm is a common clinical issue and a key target for early identification and intervention. Globally, lifetime rates among adolescents and young adults have been estimated at around 15%–20%.^{2 3} Evidence also suggests a higher prevalence among the Asian population, with a meta-analysis reporting 19.5% in Asia compared with 14.7% in other continents.⁴ Self-harm remains a priority for public health as it contributes to substantial morbidity and service utilisation and is often accompanied by multiple psychological and social difficulties affecting the functioning of children and adolescents at home, school and even work.^{5 6} Furthermore, repeated self-harm is associated with increased risk of later suicidal behaviour, particularly among young people.⁷

Adverse childhood experiences (ACEs) are recognised as long-term exposures associated with poorer mental health outcomes in later adulthood.⁸ Experiences of maltreatment, emotional neglect, household mental illness or violence, bullying or violence influence developmental trajectories and can shape the children’s or adolescents’ behaviour and increase vulnerability to adverse outcomes, including self-harm and mental health problems.⁹ Exposure to adversity does not inevitably lead to self-harm, but the risk may increase when adolescents have limited coping resources to manage emotional distress. Coping resources include individual capacities, such as problem-solving, emotion regulation and the ability to tolerate frustration, as well as external supports from family, peers, school and/or services.¹⁰

Current evidence on self-harm, ACEs and coping resources is largely dominated in high-income settings with comparatively limited Asian representation.¹¹ In addition, a recent systematic review reported that only a small proportion of studies examined protective factors such as coping resources.⁵ This matters because cultural and structural contexts can shape both patterns of adversity exposure and individual responses.¹² For instance, in some Asian contexts where stricter, disciplined settings are socially more inclined than Western settings, this could influence the distribution and impact of childhood adversity.¹³ In Malaysia, national data indicate that adolescent suicidal ideation remains a concern. The National Health and Morbidity Survey reported an increase in suicidal ideation among adolescents aged 13–17 years from 7.9% in 2012 to 10.0% in 2017, with local data also reporting an increase in the suicide rate from 2014 to 2019.¹⁴ Despite this, locally generated evidence in Southeast Asia examining self-harm alongside ACEs and coping resources remains limited. This study, therefore, aimed to examine the relationship between ACEs, coping resources and self-harm among adolescents.

Methods

A multicentre cross-sectional study was conducted across 21 government psychiatry and mental health service clinics in Malaysia between October 2023 and October 2024. Data were collected during routine clinic visits using interviewer-administered, standardised data collection forms. Participants were recruited through systematic sampling: at each site, a random starting point was selected from the first two eligible attendees, after which every second eligible participant was approached until the target sample size was achieved.

Eligible participants were patients aged 10–24 years referred to participating clinics during the study period. In line with expanded conceptualisations of adolescence, the age range of 10–24 years was adopted to reflect earlier pubertal onset and the continuation of biological and social transitions into the early 20s.¹⁵ Only participants who provided informed consent were included; for those under 18 years, written parental or guardian consent and participant assent were obtained in accordance with ethical requirements. Patients presenting with acute psychosis, acute medical or surgical conditions or severe neurodevelopmental disorders (including severe autism spectrum disorder and severe intellectual disability) were excluded.

Sociodemographic information was obtained through the interviewer-administered questionnaire during the clinic visit. Participants were asked about their age, gender, ethnicity, religion, highest education level, employment status, household income and marital status. Where participants were unable to provide specific sociodemographic information, such as household income, information was obtained from accompanying parents or caregivers when available. Age was categorised as 10–14, 15–19 and 20–24 years. Ethnicity was grouped as Malay and non-Malay, religion as Muslim and non-Muslim, education level as primary, secondary and tertiary education, employment status as student or employed and marital status as married or not married. Household income was categorised using the Department of Statistics Malaysia household income group thresholds and dichotomised into lower-income households and middle-/high-income households for analysis. Presence of psychiatric disorder was recorded based on clinical assessment of the attending clinician or clinic documentation and categorised as yes or no.

Sample size

The sample size was estimated to ensure adequate precision for estimating the prevalence of self-harm and adequate power to examine associations with key exposures. For the prevalence estimate, we assumed an expected proportion of 0.27,¹⁶ 5% precision and 95% confidence level. For the association analysis, a two-proportion calculation using α=0.05 and 80% power. Violence at home was selected as a key exposure based on published estimates of non-suicidal self-injury among youth who were exposed versus unexposed (16.6% vs 8.2%).¹⁷ The calculated sample size was inflated for an anticipated 20% non-response rate and adjusted for clustering due to the multicentre design. The final minimum required sample size was approximately 700 participants.

Measures

Self-harm history and coping-related protective factors were assessed using items adapted from the Ministry of Health Malaysia’s Self-Harm Hospital-based Accurate Reporting Project (SHHARP) clinical reporting system.¹⁸ Lifetime self-harm was defined as any self-reported history of self-harm using the question, “Have you ever self-harmed?”, with responses of ‘yes’ or ‘no’. Participants who answered ‘yes’ were asked follow-up questions on self-harm behaviour to support clinical clarification, including board method categories and additional intentional self-injurious behaviours by the clinicians. For analytical purposes, the outcome variable was dichotomised, with participants who reported any lifetime self-harm coded as ‘yes’ coded as 1 and those who reported no lifetime self-harm as ‘no’ coded as 0. Coping-related protective factors were assessed using the protective factors section of the SHHARP clinician-guided questionnaire. During the interview, clinicians asked all participants about the presence of six protective factors: ability to cope with stress or tolerate frustration; strongly held religious or cultural beliefs; realistic life goals or future plans; responsibility to loved ones, including children or pets; perceived social support and a positive therapeutic relationship. Based on the participant’s response and clinical clarification during the interview, each item was coded as ‘yes’ if present and ‘no’ if absent. For analysis, each coping-related protective factor was coded as a binary variable, with ‘yes’ coded as 1 and ‘no’ coded as 0.

ACEs were assessed using the WHO Adverse Childhood Experiences International Questionnaire (ACE-IQ),¹⁹ which encompasses 13 domains, including emotional, physical and sexual abuse, violence in the household, household substance misuse, household mental illness/suicidality, imprisonment of a household member, parental separation/absence, emotional and physical neglect, bullying, community violence and collective violence exposure. ACEs were scored using the binary scoring approach described in the ACE-IQ analysis guidance.¹⁹ For each domain, any affirmative response within the relevant item or items was coded as exposure to that domain. Responses indicating ‘once’, ‘a few times’, ‘many times’ or ‘yes’ were coded as present, while ‘never’ or ‘no’ responses were coded as absent. The item-to-domain mapping and dichotomisation rules are provided in online supplemental table S1.

For the cumulative exposure to childhood adversity, a total ACE score was calculated by summing the 13 dichotomised ACE domains, giving a possible score range of 0 to 13. Higher scores indicated greater cumulative ACE exposure. For descriptive and regression analyses, ACE burden was categorised as 0, 1, 2–3 and ≥4 ACEs. The ≥4 ACE threshold has been widely used in research to indicate high cumulative adversity, and similar cumulative ACE groupings have been applied in adolescent studies of self-harm-related outcomes.^{20 21} The Malay-language version of the ACE-IQ has demonstrated acceptable internal consistency (Cronbach’s alpha=0.701) and satisfactory content validity (Item Content Validity Index >0.8) from a previous study by Asyraf et al,²² and in the present sample, the Cronbach’s alpha is 0.82.

Ethical approval

This study was approved by the Malaysian Medical Research and Ethics Committee with reference number NMRR ID-23-02279-HBP. To protect participants’ privacy and confidentiality, all data were anonymised prior to data analysis by the data manager.

Patient and public involvement

Patients and the public were not involved in shaping the research question, study design or recruitment procedures. The findings will be shared with participating clinics and relevant stakeholders through routine service channels and publication.

Statistical analysis

All statistical analyses were conducted using R software V.4.5.1.²³ Descriptive statistics were used to summarise sociodemographic characteristics of study participants and the distribution of study variables. Categorical variables were presented as frequencies and percentages. Total ACE score was summarised using median and IQR because it was treated as an ordinal cumulative exposure.

The primary outcome of interest was lifetime self-harm, which was treated as a binary variable (coded as 1=presence of lifetime self-harm; 0=no lifetime self-harm). Descriptive comparisons between participants with and without lifetime self-harm were conducted using Pearson’s χ² test or Fisher’s exact test for categorical variables and the Wilcoxon rank-sum test for total ACE score. Crude ORs with 95% CIs were estimated for the main exposures of interest, namely ACE burden and coping-related protective factors. Multivariable logistic regression was then used to examine the adjusted associations between ACE burden, coping-related protective factors and lifetime self-harm. The main model included the ACE burden category and the six coping-related protective factors. The model was adjusted for age group, gender, ethnicity, education level and household income based on subject-matter relevance and prior literature. A sensitivity analysis was also conducted by additionally adjusting the primary multivariable model for the presence of psychiatric disorder, given its clinical relevance to self-harm. This sensitivity model retained ACE burden category, all six coping-related protective factors, age group, gender, ethnicity group, education level and household income, with psychiatric disorder added as an additional covariate.

To further explore whether specific ACE domains were associated with lifetime self-harm, a secondary multivariable logistic regression model adjusted for the same sociodemographic covariates was conducted. In this model, all individual ACE items and coping-related protective factors were entered simultaneously and adjusted for the same sociodemographic covariates. Adjusted ORs (aORs) with 95% CIs were reported. Model fit was assessed using appropriate diagnostic measures, including the Hosmer–Lemeshow goodness-of-fit test and inspection of residuals. Multicollinearity among independent variables was evaluated using variance inflation factors, with values greater than 5 indicating potential multicollinearity concerns. All statistical tests were two-tailed, and a p value of <0.05 was considered statistically significant.

Results

A total of 1244 eligible participants were screened, of whom 1164 agreed to participate in the study, yielding a response rate of 93.6%. Among the 1164 participants, 827 (71.0%) reported a history of self-harm, while 337 (29.0%) reported no self-harm (see table 1). The majority of participants were female (77.1%), aged 20–24 years (42.0%), of Malay ethnicity (76.9%) and Muslim (81.3%). Within the self-harm group, there is a significantly higher proportion of females (86.1% vs 54.9%), participants aged 20–24 years (46.7% vs 30.9%), tertiary education (30.7% vs 21.7%) and employment (34.5% vs 25.5%). However, students remained the largest employment status category in both groups. Ethnicity, religion, household income, marital status and the presence of any psychiatric disorder did not differ significantly by self-harm status. The overall presence of any psychiatric disorder did not differ significantly by self-harm status. Some participants were referred for assessment, risk evaluation or psychosocial concerns but did not have a confirmed psychiatric diagnosis recorded at the time of data collection. Among participants with any recorded psychiatric disorder, mood disorder and personality disorder were more frequently reported among participants with self-harm, whereas neurodevelopmental disorder was more common among those without self-harm (online supplemental table S2).

Table 1. Sociodemographic characteristics of adolescents referred to psychiatry and mental health services in Malaysia by self-harm status (n=1164).

Characteristics		Self-harm
Characteristics	Overall n=11 641	No self-harm n=337 (29.0%)	Self-harm, n=827 (71.0%)	P value*
Gender				<0.001
Male	267 (23%)	152 (45%)	115 (14%)
Female	897 (77%)	185 (55%)	712 (86%)
Age group				<0.001
10–14 years	230 (20%)	114 (34%)	116 (14%)
15–19 years	444 (38%)	119 (35%)	325 (39%)
20–24 years	490 (42%)	104 (31%)	386 (47%)
Ethnicity group				0.923
Malay	895 (77%)	258 (77%)	637 (77%)
Non-Malay	269 (23%)	79 (23%)	190 (23%)
Religion group				0.965
Muslim	946 (81%)	273 (81%)	673 (81%)
Non-Muslim	218 (19%)	64 (19%)	154 (19%)
Highest level of education				<0.001
Primary education	146 (13%)	75 (22%)	71 (8.6%)
Secondary education	691 (59%)	189 (56%)	502 (61%)
Tertiary education	327 (28%)	73 (22%)	254 (31%)
Employment status				0.004
Student	793 (68%)	251 (74%)	542 (66%)
Employed	371 (32%)	86 (26%)	285 (34%)
Household income				0.181
Lower-income households	756 (65%)	209 (62%)	547 (66%)
Middle-/high-income households	408 (35%)	128 (38%)	280 (34%)
Marital status				0.320
Not married	1131 (97%)	330 (98%)	801 (97%)
Married	33 (2.8%)	7 (2.1%)	26 (3.1%)
Presence of psychiatric disorder	921 (79%)	255 (76%)	666 (81%)	0.076

^*Pearson’s χ2 test. Education was grouped as primary level (ages 7–12), secondary level (approximately ages 13–17) and tertiary education level (diploma or university). Household income was categorised using Department of Statistics Malaysia household income group thresholds 2019 (released 10 July 2020), representing the bottom 40%, middle 40% and top 20% of households by income. For analysis, income was dichotomised to lower-income households (<RM4850/month) versus middle-income/high-income households (≥RM4851/month). A breakdown of psychiatric disorder categories is provided in the online supplemental table S2.

Table 2 presents the distribution of ACE and coping-related protective factors by self-harm status. Participants with self-harm reported a substantially higher burden of ACE than those without self-harm. Nearly all participants in the self-harm group had at least one ACE exposure (97.8% vs 89.9%). The median total ACE score was also higher in the self-harm group (median 5, Q1-Q3 3–7). A greater proportion of participants with self-harm reported exposure to four or more ACEs (74.0% vs 40.4%). Most individual ACE items were significantly more common among adolescents with self-harm, particularly violence against household members, emotional abuse, bullying, physical abuse, sexual abuse and witnessing community violence. In contrast, coping-related protective factors were consistently less frequently reported among adolescents with self-harm, including ability to cope with stress (41.8% vs 74.2%), social support (70.4% vs 89.3%), realistic life goals or future plans (59.1% vs 74.5%) and positive therapeutic relationships (71.5% vs 82.8%).

Table 2. Adverse childhood experiences and coping-related protective factors by self-harm status among adolescents referred to psychiatry and mental health services in Malaysia (n=1164).

Domain and factor	Self-harm		P value*
	No self-harm n=337	Self-harm n=827
Any ACE exposure			<0.001
No ACE	34 (10%)	18 (2.2%)
At least one ACE	303 (90%)	809 (98%)
Total ACE score†	3 (2, 5)	5 (3, 7)	<0.001
ACE burden category			<0.001
0 ACE	34 (10%)	18 (2.2%)
1 ACE	49 (15%)	31 (3.7%)
2–3 ACEs	118 (35%)	166 (20%)
≥4 ACEs	136 (40%)	612 (74%)
Individual ACE items
Emotional neglect	46 (14%)	280 (34%)	<0.001
Physical neglect	58 (17%)	216 (26%)	0.002
Household substance misuse	29 (8.6%)	93 (11%)	0.176
Household mental illness or suicidality	33 (9.8%)	130 (16%)	0.011
Incarcerated household member	29 (8.6%)	92 (11%)	0.213
Parental separation, divorce or loss	88 (26%)	287 (35%)	0.006
Violence against household members	195 (58%)	671 (81%)	<0.001
Emotional abuse	166 (49%)	650 (79%)	<0.001
Physical abuse	109 (32%)	456 (55%)	<0.001
Sexual abuse	50 (15%)	270 (33%)	<0.001
Bullying	179 (53%)	605 (73%)	<0.001
Witnessing community violence	102 (30%)	395 (48%)	<0.001
Exposure to collective violence	12 (3.6%)	53 (6.4%)	0.075
Coping protective factors
Ability to cope with stress	250 (74%)	346 (42%)	<0.001
Strongly held religious/cultural beliefs	261 (77%)	533 (64%)	<0.001
Realistic life goals or future plans	251 (74%)	489 (59%)	<0.001
Responsibility to children/loved ones/pets	219 (65%)	463 (56%)	0.006
Social support	301 (89%)	582 (70%)	<0.001
Positive therapeutic relationship	279 (83%)	591 (71%)	<0.001

^*Pearson’s χ2 test was used to compare proportions between self-harm groups. Values are presented as n (%) unless otherwise stated.

^†Total ACE score is presented as median, Q1, first quartile; Q3, third quartile and the Wilcoxon rank-sum test was used to compare total ACE scores.

ACEs, adverse childhood experiences.

Table 3 presents the primary multivariable logistic regression model examining the associations of ACE burden and coping-related protective factors with self-harm. After adjustment for age group, gender, ethnicity, education level, household income and coping-related protective factors, higher ACE burden remained associated with self-harm. Compared with participants with no ACE exposure, those with two to three ACEs and those with four or more ACEs had higher odds of self-harm aOR 2.54, 95% CI 1.27 to 5.20 and aOR 6.14, 95% CI 3.13 to 12.40, respectively. Exposure to one ACE was not significantly associated with self-harm.

Table 3. Multivariable analysis of adverse childhood experiences and coping factors associated with self-harm among adolescents in Malaysian mental health services (n=1164).

Predictor	Crude OR	95% CI	P value	aOR	95% CI	P value
ACE burden category
0 ACE	Reference
1 ACE	1.20	(0.58 to 2.50)	0.631	1.18	(0.52 to 2.73)	0.698
2–3 ACEs	2.66	(1.45 to 5.02)	0.002	2.54	(1.27 to 5.20)	0.009
≥4 ACEs	8.50	(4.72 to 15.80)	<0.001	6.14	(3.13 to 12.4)	<0.001
Ability to cope with stress
No	Reference
Yes	0.25	(0.19 to 0.33)	<0.001	0.32	(0.23 to 0.45)	<0.001
Strongly held religious/cultural beliefs
No	Reference
Yes	0.53	(0.39 to 0.70)	<0.001	1.02	(0.71, 1.48)	0.899
Realistic life goals or future plans
No	Reference
Yes	0.50	(0.37 to 0.65)	<0.001	0.88	(0.62 to 1.26)	0.492
Responsibility to children/loved ones/pets
No	Reference
Yes	0.69	(0.53 to 0.89)	0.005	0.71	(0.51 to 0.98)	0.039
Social support
No	Reference
Yes	0.28	(0.19 to 0.41)	<0.001	0.63	(0.40 to 0.97)	0.038
Positive therapeutic relationship
No	Reference
Yes	0.52	(0.38 to 0.71)	<0.001	0.85	(0.58 to 1.26)	0.432

Crude ORs were estimated using univariable logistic regression. Adjusted ORs were estimated using multivariable logistic regression with lifetime self-harm as the outcome. ACE burden category, all six coping-related protective factors, age group, gender, ethnicity group, education level and household income were entered simultaneously in the adjusted model. For binary predictors, ‘No’ was the reference category unless otherwise stated.

ACE, adverse childhood experience; aOR, adjusted odds ratio.

Several coping-related protective factors were associated with lower odds of self-harm. Participants who reported the ability to cope with stress had lower odds of self-harm (aOR 0.32, 95% CI 0.23 to 0.45). Responsibility to children, loved ones or pets (aOR 0.71, 95% CI 0.51 to 0.98) and social support (aOR 0.63, 95% CI 0.40 to 0.97) were also associated with lower odds of self-harm. While strongly held religious or cultural beliefs, realistic life goals or future plans and positive therapeutic relationships were not significantly associated with self-harm after adjustment.

Among the covariates retained in the model, older age and female gender were associated with higher odds of self-harm. Secondary education, compared with primary education, was also associated with higher odds of self-harm, while tertiary education, ethnicity and household income were not significantly associated after adjustment. Model diagnostics suggested acceptable model performance. The Hosmer–Lemeshow goodness-of-fit test showed acceptable fit (p=0.761), and the model showed good discrimination, with an area under the curve of 0.815 (95% CI 0.788 to 0.842).

In the sensitivity analysis, additional adjustment for the presence of psychiatric disorder did not materially change the main findings (see online supplemental table S3). The presence of a recorded psychiatric disorder was not significantly associated with self-harm after adjustment (aOR 1.00, 95% CI 0.69 to 1.43). In the exploratory secondary analysis, the ACE burden category was replaced by individual ACE items while retaining the coping-related protective factors and sociodemographic covariates in the model. Violence against household members, emotional abuse, sexual abuse, bullying and witnessing community violence were independently associated with higher odds of lifetime self-harm (online supplemental table S4). Among the coping-related protective factors, only the ability to cope with stress remained significantly associated with lower odds of lifetime self-harm. Other individual ACE items and coping-related protective factors were not statistically significant after adjustment.

Discussion

In this Malaysian clinical referral sample seen in the psychiatry and mental health services, self-harm prevalence is high. While not uncommon, it is clinically important because referred young patients often present with higher symptom burden and multiple psychosocial difficulties by the time they reach specialist care. We included participants aged 10–24 years because contemporary developmental frameworks¹⁵ recognise that adolescence and emerging adulthood often extend beyond the age of 19 and are shaped by later educational transitions, delayed entry into workforce and prolonged dependence on family structures.^{15 24}

The main finding of this study was the strong association between cumulative ACE burden and self-harm. Compared with participants with no ACE exposure, those with two to three ACEs had more than twice the odds of self-harm, while those with four or more ACEs had substantially higher odds. This supports a cumulative-risk framework, where repeated or multiple forms of adversity may progressively increase vulnerability through chronic stress, impaired emotion regulation, reduced perceived safety and disruption of supportive relationships.^{25 26} Similar patterns have been reported in adolescent studies. In Asia, a South Korean school-counselling-linked sample found higher odds of self-harm among children and adolescents with two or more ACEs,¹¹ while a large Chinese adolescent study linked exposure to four or more ACEs with non-suicidal self-injury and suicidality.²¹ Evidence from Western settings also supports a cumulative relationship, with each additional ACE increasing the likelihood of self-harm by 88% in a Northern Ireland youth sample²⁷ and four or more ACEs showing the strongest associations with suicide-related outcomes in US adolescent surveillance data.²⁸

In our study, coping-related protective factors also remained relevant after adjustment. The ability to cope with stress was associated with lower odds of self-harm. This aligns with the view that self-harm may occur when distress exceeds available coping resources and that skill-based therapies can reduce self-harm by strengthening emotion regulation and problem-solving.²⁹ Social support may be protective by shaping whether young people disclose distress, whether caregivers notice change early and whether professional help is accessed in time.³⁰ Social support acts both as a buffer against distress and as a pathway to timely care.²⁴ In our study, responsibility towards loved ones or pets is a significant protective factor which is consistent with the ‘reasons for living’ framework where connection and obligation can anchor adolescents during periods of distress.³¹ While evidence on pets is mixed with inconclusive findings for suicide attempts in a recent review,³² a systematic review reported that reasons for living, particularly responsibility to family, are protective against suicidal ideation and attempts.³³

Although ACE burden was the primary exposure in this study, the exploratory secondary analysis using individual ACE items provided additional insight into which experiences may have contributed to this cumulative pattern. Emotional abuse and sexual abuse showed the strongest associations with self-harm, consistent with existing studies.34,36 In Malaysia, sexual abuse against children remains a major child protection concern, yet substantial under-detection and delayed disclosure are common and are often shaped by shame, fear and dependence on adults.^{37 38} These dynamics may intensify distress and disrupt emotion regulation, which is a key functional pathway linking adversity to self-harm.³⁹

Bullying also emerged as an important adversity domain. This is relevant because bullying is not confined to the home environment and may reflect wider peer, school, neighbourhood and digital risk contexts. Evidence from Asian populations suggests that that bullying victimisation remains a significant adolescent health issue, and prevention needs to start well before tertiary care.⁴⁰ The regional literature is also moving beyond ‘traditional’ bullying, with systematic reviews reporting consistent associations between cyberbullying and self-harm in children and young people.^{41 42} This supports prevention and early identification in the education system and primary care, while also integrating online safety and digital well-being into youth mental health strategies rather than treating cyberbullying separately.

Stigma may help explain why some young people reach specialist services only after distress has become severe. In many Asian settings, emotional distress is easily interpreted as personal weakness, attention-seeking behaviour or disciplinary problems, which may delay help-seeking and contribute to social withdrawal.⁴³ When these perceptions are compounded by shame-based family dynamics and concerns about bringing embarrassment to the family, disclosure may be perceived as particularly risky from an adolescent’s perspective.⁴⁴ Local evidence further indicates that adolescents encounter both practical and attitudinal barriers to accessing mental healthcare, including those related to stigma, acceptability and concerns about confidentiality.²⁹

These stigma-related dynamics often operate within families. Parental beliefs about caregiving, emotional expressions and discipline influence how adolescent emotional distress is perceived and managed. Many Asian parents were raised in authoritarian and performance-focused environments that discourage open emotional expression and routine discussion of feelings.⁴⁵ Concurrently, public discourse around gentle parenting, emotional literacy and mental health awareness has begun to change expectations and practices.^{46 47} Consequently, a number of families find themselves caught between contrasting pressures from extended relatives, school systems and social media, as well as the changing children’s developmental needs, which may intensify household tensions, undermine validation of adolescents’ emotions and delay or deter timely disclosure of distress.^{46 48}

These dynamics underscore the importance of family-sensitive approaches that strengthen emotion regulation and communication skills not only among adolescents but also among parents, alongside improved mental health literacy and supportive monitoring.49,51 In the Malaysian context, emotion regulation warrants particular emphasis; however, engagement may be limited when distress is interpreted through prevailing social expectations and religious or cultural narratives that prioritise endurance, discretion or resolution within the family rather than seeking formal support.⁵² Integrating emotion regulation and mental health skills early through education systems and established parenting touchpoints, such as school-based engagements, routine maternal and child health services, community and faith-based programmes such as pre-marital courses and workplace well-being initiatives, may improve acceptability, accessibility and overall reach.

At the policy level, Malaysia’s decriminalisation of attempted suicide, enforced from September 2025, marks a shift towards a health-oriented response and may reduce fear-based barriers over time.⁵³ However, legal reform alone is unlikely to produce earlier care without practical changes in the settings for adolescents, including schools, online platforms, primary care and community services. Hence, a multifaceted prevention approach consistent with WHO frameworks, combining population-level actions with targeted detection and clear referral pathways, is very much supported.⁵⁴ Priority includes universal antibullying and cyber-safety efforts, targeted screening for adversity and coping difficulties in high-risk groups and trauma-informed service responses that prioritise safe disclosure pathways, emotion regulation skills and strengthening supportive relationships starting from early education.⁴¹ Finally, screening should not be limited to the presence of psychiatric diagnosis alone. The sensitivity analysis showed that additional adjustment for psychiatric disorder did not materially change the main associations, suggesting that the relationship between ACE burden, coping-related protective factors and self-harm was not fully explained by the diagnosis alone. This supports trauma-informed assessment and coping-focused approaches into routine components of youth mental healthcare.

Strengths and limitations

Several limitations should be considered. The cross-sectional design limits causal inference and cannot establish a temporal sequence between ACE exposure, coping factors and self-harm. Tools were measured by retrospective self-report and may be affected by recall bias. The clinical referral sample also limits generalisability to community populations, as participants seen in specialist services are likely to have higher symptom burden and more complex psychosocial needs. Therefore, the findings should not be interpreted as prevalence estimates for Malaysian adolescents and young adults in the general population. Rather, they are most applicable to young people referred to government psychiatry and mental health services, particularly those with similar clinical and psychosocial profiles. Finally, binary measures of complex constructs such as coping, support and therapeutic relationship may reduce sensitivity and obscure dose–response effects. Future work could strengthen inference using prospective designs. More detailed measurement of adversity (timing, chronicity, severity) and protective factors (source and quality of support, family functioning) may identify more actionable intervention targets. Mixed-methods research in Malaysian settings could also clarify cultural meanings of distress, disclosure and help-seeking and support the design of youth-friendly, trauma-informed care pathways.

Conclusions

This study found a high prevalence of lifetime self-harm among adolescents and young adults referred to psychiatry and mental health services in Malaysia. Higher cumulative ACE burden was associated with greater odds of self-harm, while ability to cope with stress, perceived social support and responsibility towards loved ones or pets were associated with lower odds. Secondary analyses suggested that violence against household members, emotional abuse, sexual abuse, bullying and witnessing community violence may be relevant adversity domains. These findings support trauma-informed and coping-focused approaches that strengthen safe disclosure, family engagement, social support and early identification across youth-serving settings.

Data availability statement

Data are available upon reasonable request.