The association between youth violence and mental health outcomes in Colombia: A cross sectional analysis

Abstract

Background:

Violence against children and youth poses public health risks regarding mental health symptoms and substance use. Less studied is the relationship between violence and mental health/substance abuse in the Latin American context. This study explored sex-stratified relationships between violence and mental health/substance use among Colombian youth.

Methods:

We analyzed the 2018 Colombian Violence Against Children and Youth Survey, which collected cross-sectional data from Colombian youth (13–24 years) (n = 2706). Exposure variables were (i) binary sexual, emotional, and physical victimization and (ii) poly-victimization. The outcomes were binary suicidal thoughts, self-harm, past-month psychological distress, binge drinking, smoking, and drug use. Sex-stratified, logistic regressions were adjusted for age, primary school, parental presence, relationship status, and witnessing community violence.

Results:

For females, (i) emotional violence (compared to being unexposed) was associated with greater odds of suicidal thoughts, self-harm, and psychological distress and (ii) sexual violence was associated with suicidal thoughts and self-harm. For males, (i) emotional violence (compared to being unexposed) was associated with greater odds of suicidal thoughts and psychological distress, but not self-harm and (ii) sexual violence exposure was associated with suicidal thoughts and self-harm. Physical violence was generally not associated with internalized mental health outcomes for females/males, when emotional and sexual violence were held constant. Poly-victimization was consistently and positively associated with internalized mental health symptoms among females, and to a lesser degree for males. Substance use outcomes for males or females were not associated with violence.

Conclusions:

Findings highlight the internalized mental health burden of emotional and sexual violence.

1. Introduction

A rich body of scholarship has investigated the relationship between adverse childhood experiences (ACE) and health adversity in later adulthood. This literature has reported a greater risk of chronic diseases (type 2 diabetes, cardiovascular disease, chronic obstructive pulmonary disease) (Felitti, 2002; Felitti et al., 1998; Hughes, Bellis, Hardcastle, Sethi, et al., 2017b), poor diet quality (Aquilina et al., 2021, mental health outcomes (depression and attempted suicide) (Chapman et al., 2004; Choi et al., 2017; Felitti et al., 1998; Mersky et al., 2013), sexually transmitted infections (Felitti et al., 1998; Hughes, Bellis, Hardcastle, Sethi, et al., 2017a), and substance abuse (alcohol, nicotine, marijuana, and intravenous drugs) (Merrick et al., n.d.; Dube et al., 2002; Felitti et al., 1998; Hughes, Bellis, Hardcastle, Sethi, et al., 2017b; Mersky et al., 2013) among survivors of ACE compared to their unexposed counterparts. However, investigators have mainly relied on constructing a summed cumulative score of the total number of ACE an individual was exposed to during their childhood, thereby demonstrating associations between the ordinal or numerical summed ACE score and varied health outcomes (Lacey & Minnis, 2020). However, the summed score approach may obscure a more nuanced understanding of how distinct forms of violence perpetrated against children impact health outcomes in differential ways (Lacey & Minnis, 2020). This paper specifically considers sex-stratified relationships between the major forms of violence experienced during childhood (physical, emotional, and sexual violence) and internalized as well as externalized mental health outcomes among Colombian youth.

1.1. Global epidemiology of violence against children and youth

Globally, over half of all children experience some form of violence between the ages of two and seventeen (Hillis et al., 2016). Based on current estimates, interpersonal violence is the fifth leading cause of death among youth between the ages of 10 and 24 and accounts for 3.5 % of disability adjusted life years globally (Vos et al., 2020). Youth face distinct forms of interpersonal violence such as bullying, physical assault, sexual assault, emotional abuse from parents or caregivers, and gang violence. Globally, an estimated 200,000 homicides occur each year among youth 10–29 years of age (WHO, 2023) and 30 % of adolescents (12–17 years) report experiencing bullying in the past 30 days (Biswas et al., 2020). Exposure to non-lethal violence, in all its forms, has consequences that manifest across the life course (Hillis et al., 2016). There is an emerging understanding that childhood and youth victimization is often recurring and occurs in tandem with environmental hardships such as poverty, witnessing violence, or family dysfunction.

1.2. Violence against children/youth and mental health disorders

In particular, exposure to childhood violence is a risk factor for the development and progression of mental health disorders, many of which have first onset during youth (Stansfeld et al., 2017). The body of literature examining the impact of childhood violence (defined as exposure to violence during childhood) on mental health outcomes has traditionally focused on physical and sexual violence. Across a variety of settings, exposure to childhood physical and/or sexual victimization has been linked to internalized mental health outcomes such as psychological distress, suicidal thoughts, depression, and self-harm (Seff et al., 2022; Seff & Stark, 2019; Thoresen et al., 2015). Other research has illustrated that exposure to childhood physical and/or sexual abuse is higher among those with substance misuse histories or who were admitted to inpatient facilities for substance abuse (Westermeyer et al., 2001). Moreover, childhood exposure to sexual and/or physical violence increased the odds of drug use among youth (Chigiji et al., 2018; Kappel et al., 2021).

Recent studies have shown that childhood experiences of emotional violence may be equally or even more harmful for mental health outcomes, compared to physical or sexual violence. For example, Meinck et al. (2017) investigated the consequences of girls’ emotional victimization and found that the odds of both suicidal ideation and feeling depressed were approximately two times higher among girls who experienced emotional abuse during childhood, compared to those who did not. Further, a small number of studies conducted in low- and middle-income countries (LMICs) have shown that childhood emotional violence had a stronger impact on adolescent suicidal behaviors than physical or sexual abuse. For instance, a study conducted in Zimbabwe showed that childhood emotional violence resulted in greater odds of attempting suicide, suicidal ideation, and smoking, than physical violence (Chigiji et al., 2018). Additionally, Seff and Stark (2019) reported that exposure to emotional violence perpetrated by caregivers was consistently and significantly associated with suicidal ideation in Haiti, Kenya, and Tanzania, though physical violence perpetrated by caregivers was not. There is also evidence that in addition to physical and sexual victimization, childhood emotional abuse and overall child maltreatment are also related to higher levels of substance use during adolescence (Moran et al., 2004).

1.3. Traumatic stress theory

To better understand the mental health consequences of exposure to multiple violence typologies, attention must also be given to the harmful effects of poly-victimization, defined as exposure to multiple different victimization types (Greenfield & Marks, 2010; McLaughlin & Sheridan, 2016). Traumatic stress theory posits that exposure to multiple forms of violence during childhood may heighten the risk of poor mental health outcomes (Palermo et al., 2019). Emerging research supports this theory: a study in the United States showed that children who experienced poly-victimization were more likely to have negative mental health outcomes compared to children who only experienced one form of violence victimization or none (Adams et al., 2016). While poly-victimization may be more prevalent and yield a greater mental health burden in socially disadvantaged settings such as LMICs (Kamndaya et al., 2017), most research concerning the differential mental health outcomes as a function of poly-victimization is derived from high-income settings. Analyses specific to LMICs are more limited.

1.4. Gender differences in victimization and mental health disorders

Additionally, although both girls and boys experience childhood victimization, their socialization experiences can lead to health sequela that present differently (Seff et al., 2022). The cultural reinforcement of gender roles has led to a pattern in which males are more likely to manifest poor mental health in terms of substance use whereas girls are more likely to exhibit internalizing behaviors (e.g., anxiety, depression, suicidal ideation) (Dulmus & Hilarski, 2006). At least part of the explanation of why we see these differences by sex may be related to the ways in which boys and girls are socialized. In many patriarchal societies, anger is deemed to be an acceptable emotional response for boys, while girls are expected to please others or cater to others’ needs (Kret & De Gelder, 2012). There is emerging evidence supporting this theory of learned behavior, including a study analyzing children between the ages of six and twelve, where gendered differences in externalizing versus internalizing behaviors became significant after children reached the age of nine (Dulmus & Hilarski, 2006). However, in LMIC, research pertaining to gender differences on the mental health outcomes of children and youth exposed to violence is more limited.

1.5. Colombian context

Colombia is a South American country bordering Venezuela, Brazil, Peru, and Ecuador. The country is home to 48 million people, of which 26 % are between the ages of 15 and 29 (IOM, 2019). Social inequality is high and rising in Colombia: the country’s Gini coefficient is the 12th highest in the world and has risen from 50.8 to 51.7 between 2017 and 2018 (IOM, 2019; World Bank, 2020). In 2018, 27 % of Colombians were living below the national poverty line (Cuesta & Pico, 2020; Katz & Fallon, 2022; World Bank, 2020). In addition to socio-economic inequality and poverty, violence against children is also prevalent in Colombia. For example, in 2018 14 % of females and 12 % of males between the age of 18 to 24 experienced physical violence in the last 12 months (IOM, 2019). The Colombian socio-economic context and violence against children intersect with wider insecurity and armed conflict.

For the past five decades, Colombia has experienced protracted internal armed conflict and civil war. Armed groups, fueled by drug trafficking income, perpetrate human rights violations (i.e., kidnapping, forced recruitment of youth in violence, and gender-based violence) which trigger internal and external forced migration (Chaskel et al., 2015). The legacy of Colombia’s armed conflict affects millions of Colombians, including children and youth, who are impacted by the conflict-related normalization of violence as well as mental health and substance use disorders (Bell et al., 2012; Cuartas Ricaurte et al., 2019; León-Giraldo et al., 2021). A sex-stratified examination of the association between violence and mental health outcomes is particularly relevant for Colombia, where recent estimates suggest that among 18–24-year-olds, approximately 41 % of females and 42 % of males have experienced at least one form of violence before the age of 18 (IOM, 2019). Further, among this same age group, 10 % of females and males experienced two forms of childhood violence and 6 % of females experienced sexual, physical, and emotional violence (IOM, 2019).

Scholarship has begun to explore the relationship between childhood victimization and mental health using nationally representative data from Colombia (Moe et al., 2021). The present study builds upon this emergent scholarship using the 2018 Colombia Violence Against Children and Youth Survey (VACS) to examine the sex-stratified associations between exposure to childhood violence victimization, including poly-victimization, and mental health as well as substance abuse outcomes.

1.6. Current study

This research explored sex-stratified relationships between childhood/youth violence exposure and mental health and substance use outcomes. First, we assessed sex-stratified associations between exposure to sexual, emotional, and physical violence victimization and internalized mental health sequela (suicidal thoughts, self-harm, and psychological distress) as well as substance use (binge drinking, smoking, and drug use). Additionally, we examined sex-stratified associations between poly-victimization and internalized mental health sequela as well as various forms of substance use. We hypothesized that exposure to any violence would increase the odds of internalized mental health outcomes and substance use, with statistically higher odds for emotional violence and poly-victimization compared to other types of violence. We also hypothesized more consistent and significant associations between violence and higher odds of internalized mental health outcomes among females, compared to males. This work adds to the existing literature base concerning the relationship between childhood victimization and mental health/substance use outcomes in Latin America, particularly in the Colombian context where youth experience reduced socio-economic opportunities and a pervasive level of community violence. Our sex-stratified analysis also highlights how gender socialization might give rise to differing patterns of mental health and substance use following victimization.

2. Methods

2.1. Study design

We conducted a secondary data analysis using the 2018 national dataset of the Colombia VACS. The VACS have been implemented in a variety of countries and are a collaborative effort between the United States Centers for Disease Control and Prevention, national governments, as well as other multinational and bilateral organizations. The purpose of the 2018 Colombia VACS was to generate nationally representative prevalence estimates of childhood physical, emotional, and sexual violence. The questionnaire includes approximately 300 questions across a variety of topics, including violence experiences, mental health, substance use, family dynamics, witnessing violence, sexual history, and general demographics. Specific questions adapted for the Colombian context were also included and pertained to internal displacement, migration, carrying weapons, and witnessing community violence. Details pertaining to VACS sampling design and methodology are reported elsewhere (IOM, 2019).

To be eligible for inclusion, participants had to reside within the households that were sampled, be fluent in Spanish, and be between 13 and 24 years of age at the time of survey implementation. The VACS employed a three-stage stratified and clustered sample design. In the first sampling stage, the dataset consisted of randomly selected enumeration areas (EAs) (n = 619), which were further subdivided into male (n = 296) and female (n = 323) respondent EAs using a split sample approach. The split sample approach safeguards participant confidentiality and reduces the possibility that perpetrators and victims from the same community will be surveyed. In the second stage, using equal probability systematic sampling, 24 households were selected within the EAs. In the third and final stage, one participant (male or female depending on the EA) was selected for participation from a household list of eligible participants. In total, 1406 females and 1299 males completed the survey. Youth living with cognitive impairments or inability to communicate verbally, lack of understanding of the language, or individuals who were living in institutions or on the street were excluded from the study.

Informed consent was first obtained from the head of household. When the household head was also a parent or guardian of children under the age of 18, they provided written permission for minors to participate. When a private location was secured, enumerators read the contents of the informed consent and/or assent forms to the participants who then verbally expressed their willingness to participate. Enumerators also documented the consent/assent in writing. Enumerators offered referrals to medical, psychosocial, and legal services when appropriate. Ethical approval for study protocols pertaining to the Colombia VACS were approved by The Ethics and Research Methods Committee of the National Institute of Health of Colombia and the CDC’s Institutional Review Board.

Supplementary materials, Table 1 and Table 1 in the main text present a detailed description of (i) the VACS questions used to form the exposure, outcome, and covariate variables and (ii) the variable operationalization, respectively.

Table 1.

Variable operationalization using selected VACS Colombia question.

Exposures	Outcomes	Covariates
Objective 1  Binary Exposure to Childhood Sexual Violence Victimization • Exposed to sexual violence victimization of any type by any perpetrator type • Unexposed to sexual violence victimization (ref) Binary Exposure to Childhood Emotional Violence Victimization • Exposure to emotional violence victimization of any type by caregivers (including parents) • Unexposed to emotional violence victimization (ref) Binary Exposure to Childhood Physical Violence Victimization • Exposed to physical violence victimization of any type by any perpetrator type • Unexposed to physical violence victimization (ref) Objective 2  Ordinal Poly-victimization Experiences: • Exposure to at least 2 forms of violence (sexual, emotional, or physical) • Exposure to 1 form of violence • Unexposed to violence (ref)	Internalized Metal Health Sequela: Binary Lifetime Suicidal Thoughts • Presence of suicidal thoughts • Absence of suicidal thoughts Binary Lifetime Self Harm • Presence of self-harm • Absence of self-harm Binary Past 30 Days Psychological Distress • Presence of psychological distress (mild/moderate and severe psychological distress) • Absence of psychological distress (no psychological distress) Substance Use: Binary Past 30 Days Binge Drinking • Presence of binge drinking • Absence of binge drinking Binary Current Smoking • Presence of smoking • Absence of smoking Binary Past 30-Day Drug Use • Presence of drug use • Absence of drug use	Age • Years (13–24) Binary Completion of Primary School • Yes • No (ref) Nominal Parental Presence • Mother and father were absent during childhood or deceased • Mother was absent during childhood or deceased • Father was absent during childhood or deceased • Both the mother or father were present during childhood and currently alive (ref) Binary Partner Status • Ever had a romantic partner • Never had a romantic partner (ref) Binary Witnessing Internal Conflict • Witnessed internal conflict • Did not witness internal conflict (ref)

Notes. ref. = referent category; for all outcomes, we modelled the presence of each mental health or substance use outcome.

2.2. Independent variables

Each type of violence victimization—sexual, emotional, and physical—was operationalized as binary: (i) exposed to violence if at least one experience of a specific violence form was reported and (ii) unexposed (referent) if no experiences of a specific violence form were reported.

2.2.1. Sexual violence

Sexual violence was defined to include lifetime exposure (at any point in time), to sexual touching, forced sex attempts, forced sex, and coerced/pressured sexual intercourse from any perpetrator.

2.2.2. Emotional violence

Emotional violence was operationalized as lifetime exposure to insults or verbal threats from parents/caregivers (i.e., parent/caregiver told you that you were not loved, wished you had never been born, ridiculed you or put you down, or threatened to get rid of you, etc.).

2.2.3. Physical violence

Lastly, physical violence was defined to include lifetime exposure to physical intimate partner violence and physical violence perpetrated by peers, parents/caregivers/relatives, and/or adults in the community (i.e., being sapped, punched, shoved, whipped, chocked, etc.).

2.2.4. Poly-victimization

In addition, a three-level ordinal measure of poly-victimization was constructed using exposure at any time, to sexual, emotional, and physical violence, wherein levels denoted: (i) exposure to two or more forms of violence, (ii) exposure to only one form of violence, and (iii) no exposure to violence (referent). This operationalization is in line with Finkelhor et al., (2011, p. 4) who defined poly-victimization as “having experienced multiple victimizations of different kinds”.

2.3. Dependent variables

Three internalized mental health outcomes and three substance use related outcomes were dichotomized (Present vs Absent): lifetime suicidal thoughts, lifetime self-harm, past 30-day psychological distress, past 30-day binge drinking, current smoking, and past 30-day drug use. We excluded suicidal attempts because of low numbers.

2.3.1. Suicidal thoughts

Lifetime suicidal thoughts was assessed using the question: “Have you ever thought about killing yourself?”

2.3.2. Self harm

Lifetime self-harm was measured using the question: “Have you ever intentionally hurt yourself in any way?”

2.3.3. Psychological distress

Psychological distress was assessed using the Kessler 6 Scale. The psychometric properties of the Kessler 6 scale have been previously reported (Kessler et al., 2010). Total scores range from 0 (indicating the absence of distress) to 24 (indicating presence of severe distress). Scores of 0 to 4 were coded as the absence of psychological distress and scores of 5–24 were coded as the presence of psychological distress (mild/moderate and severe). The binary operationalization of psychological distress was found to be 75 % sensitive and 75 % specific in categorizing persons in need of clinical mental health care (Prochaska et al., 2012).

2.3.4. Binge drinking

Binge drinking was assessed using the questions: “How old were you when you drank alcohol for the first time that was more than a few sips” and “In the past 30 days, on how many days did you have 4 or more drinks of alcohol in a row”. If a respondent indicated they never drank alcohol they were coded as 0. If a respondent indicated at least one day of binge drinking in the past 30 days, they were coded as 1.

2.3.5. Smoking

Smoking was assessed using the question: “Do you currently smoke tobacco on a daily basis, less than daily, or not at all?”. If a respondent indicated “not at all” they were coded as 0, otherwise they were coded as 1.

2.3.6. Drug use

Drug use was assessed using the question: “In the past 30 days, have you used drugs such as marijuana, pills, ecstasy, or sniffed any chemical such as petrol or glue?”. If a respondent indicated “no” they were coded as 0, otherwise they were coded as 1.

2.4. Covariates

To address the effect of potential confounding variables, the analysis included covariates based on the conceptualization of a confounder (a factor that differs by exposure levels, is a risk factor for the outcome, and is not on the causal pathway between the exposure and outcome) (Lorriane et al., 2008) previous research, and available VACS variables (Borges et al., 2008; Moe et al., 2021; Seff & Stark, 2019; Wadsworth et al., 2004).

2.4.1. Age

Age was operationalized using participants’ self-reported age in years (continuous).

2.4.2. Education

Education was operationalized as binary using participants’ self-reported completion of primary school (yes, no = referent).

2.4.3. Parental presence

An ordinal variable for parental presence was constructed using participants’ disclosures regarding whether their parents were alive and/or present during their childhood (both mother and father were not present, mother was not present, father was not present, both parents were present = referent),

2.4.4. Partner status

A binary variable for partner status was constructed using participants’ self-reported relationship status(ever had a romantic partner, never had a romantic partner = referent),

2.4.5. Witnessing violence

Witnessing violence was operationalized as binary using participants’ self-reports of whether they had witnessed violence within their community (witnessed internal conflict, did not witness internal conflict = referent).

2.5. Statistical methods

Statistical analyses were conducted in SAS (version 9.4). First, descriptive statistics were computed for all included variables. Two sample independent t-test were computed for age (continuous) by sex and the Rao-Scott design adjusted chi squared tests were computed for all other categorical variables, using weights, and adjusting for the complex sampling design. Following descriptive analysis, we used the Proc Survey Logistic procedure to compute logit models for the six binary mental health outcomes as a function of (i) exposure to sexual, emotional, and physical violence (objective 1), and (ii) exposure to poly violence victimization (objective 2), while holding all covariates constant. The analytic strategy for the multivariate regression used the stratification, cluster (primary sampling unit), and sample weight variables.

Sub-group analysis by sex was conducted to obtain separate odds ratio estimates between violence exposure and mental health-related outcomes for males and females. We could not directly assess gender differences given that we used biological sex as the stratifying variable, but we interpret biological sex as a proxy for gender. We present both adjusted and unadjusted models. Statistical significance was assessed at a type I error (alpha) of 5 %. To determine nationally representative estimates among Colombian youth between the ages of 13 and 24 years, all analyses were weighted to account for the abovementioned sampling strategy. We do not present crude or adjusted odds ratio (OR) estimates for drug use among females due to low cell counts: only 3 % (n = 40) of females reported drug use.

The amount of missing data varied by variable and missing observations were excluded from the analysis. Variables wherein observations were missing for >5 % of the data were: binge drinking (n_missing = 370; n_male = 144, n_female = 226) and parental presence (n_missing = 142; m_male = 75, n_female = 67). Accordingly, we conducted sensitivity analysis on imputed data to ensure key findings were not impacted by potential bias in missingness regarding the binge drinking and parental presence variables. A multiple imputation approach was employed for these predictor variables with missing values using “Proc mi” and “Proc mianalyze” in SAS, where 25 imputations were created for each observation with a missing value (Berglund, 2010). The sample weights were handled after creating the multiple imputed datasets. Reported estimates of imputed values represent the average weighted estimation of the 25 imputations on a logarithmic scale (refer to the Supplemetary Materials). Multicollinearity was not deemed to be a concern because variance inflation factors for all models did not exceed a value of 1.3.

3. Results

Descriptive statistics adjusted for the complex sampling strategy and representative of 13–24-year-old Colombian youth are presented in Table 2. While demographic characteristics were similar for males and females, prevalence of victimization differed between the two groups. Compared to males, female Colombian youth reported greater exposure to sexual violence (21.20 % vs 12.70 %; p < 0.05) and emotional violence (22.80 % vs 15.90 %; p < 0.05). Physical violence was the only form of violence that was higher among Colombian males compared to females (46.70 % vs 35.90 %; p < 0.05). Female Colombian youth also reported greater occurrence of suicidal thoughts (19.70 % vs 4.70 %; p < 0.05) and psychological distress (51.90 % vs 35.70 %; p < 0.05) than males. However, substance use was greater among male Colombian youth: binge drinking (44.10 % vs 28.40 %; p < 0.05), smoking (15.80 % vs 5.80 %; p < 0.05), and drug use (9.60 % vs 1.50 %; p < 0.05).

Table 2.

Descriptive statistics.

Variables	Males (N = 1299)		Females (N = 1406)
	n or μ	SD or %	n or μ	SD or %
Demographics
Age (years)	18.45	0.16	18.51	0.17
Completed primary school	1240	96.80	1331	95.80
Parental Presence
Presence of both parents	703	55.8	759	54.80
Absence of mother only	77	4.00	82	3.90
Absence of father only	299	27.40	328	25.10
Absence of both parents	145	12.90	170	16.20
Ever had a partner	1011	79.30	1076	76.60
Ever witnessed internal conflict	192	16.70	168	16.00
Violence victimization
Lifetime sexual violence victimization	156	12.70*	285	21.20*
Lifetime emotional violence victimization from caregivers	206	15.90*	307	22.80*
Lifetime physical violence victimization	602	46.70*	504	35.90*
Lifetime Poly-victimization	237	17.70	327	21.60
Internalized Mental Health Sequela
Lifetime Suicidal Thoughts	92	4.70*	272	19.70*
Lifetime Self-Harm	95	9.20	205	14.00
Past 30-Day Psychological Distress (mild, moderate, or severe)	466	35.70*	729	51.90*
Substance Use
Past 30 Days Binge Drinking	498	44.10*	346	28.40*
Current Smoking Use	209	15.80*	87	5.80*
Past 30 Days Drug Use	124	9.60*	40	1.50*

Notes. n (%) or mean (μ) [SD: standard deviation]. Percentages and means are weighted to be representative of Colombian youth 13–24 years. Two sample independent t-test computed for age and Rao-Scott design adjusted chi squared tests computed for all other categorical variables, using weights, and adjusting for the complex sampling design.

^*p < 0.05 when comparing males and females.

3.1. Violence and internalized mental health

Findings presented in Table 3 detail how the internalized mental health outcomes among females were associated with various forms of violence, most notably emotional violence. After adjusting for covariates, female Colombian youth who were exposed to emotional violence had statistically higher odds of suicidal thoughts (aOR: 4.23;95 % CI: 2.09, 8.57), self-harm (aOR: 2.52;95 % CI: 1.20, 5.33), and psychological distress (aOR: 2.51;95 % CI: 1.25, 5.04), compared to unexposed females. Exposure to sexual violence was similarly associated with suicidal thoughts (aOR: 3.93;95 % CI: 1.87, 8.24) and self-harm (aOR: 4.00;95 % CI: 1.62, 9.71) among females, but not psychological distress (aOR: 1.65;95 % CI: 0.85, 3.21). Unlike emotional and sexual violence, there was little evidence supporting the association between physical violence and internalized mental health outcomes among females.

Table 3.

Estimating internalized mental health outcomes and substance use from childhood exposure to sexual, emotional, and physical violence victimization among females, 2018 Colombia Violence against Children and Youth Survey.

	Female Youth (13–24) n = 1406
	Internalized Mental Health Outcomes						Substance Use
	Suicidal Thoughts		Self-Harm		Psychological Distress		Binge Drinking		Smoking
	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted
	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value
Lifetime Sexual Violence (Ever vs Never)	3.91* [1.93, 7.94] 0.00	3.93* [1.87, 8.24] 0.00	3.14* [1.21, 8.15] 0.019	3.96* [1.62, 9.71] 0.00	1.93* [1.03, 3.61] 0.04	1.65 [0.85, 3.21] 0.14	0.88 [0.43, 1.77] 0.71	0.81 [0.41, 1.60] 0.54	1.59 [0.57, 4.44] 0.38	0.94 [0.37, 2.42] 0.90
Lifetime Emotional Violence from Caregivers (Ever vs Never)	4.73* [2.27, 9.85] <0.0001	4.23* [2.09, 8.57] <0.0001	2.77* [1.29, 5.95] 0.010	2.52* [1.20, 5.33] 0.02	2.73* [1.35, 5.50] 0.01	2.51* [1.25, 5.04] 0.01	0.87 [0.49, 1.56] 0.64	0.82 [0.44, 1.53] 0.52	1.42 [0.50, 3.98] 0.51	1.59 [0.54, 4.66] 0.40
Lifetime Physical Violence (Ever vs Never)	1.94* [1.04, 3.64] 0.04	1.81 [0.91, 3.59] 0.09	1.73 [0.98, 3.05] 0.06	1.76 [0.94, 3.32] 0.08	1.50 [0.83, 2.70] 0.18	1.35 [0.69, 2.65] 0.38	1.84 [0.99, 3.44] 0.06	1.85 [1.00, 3.42] 0.05	2.72* [1.10, 6.71] 0.03	1.97 [0.86, 4.51] 0.11
n	1398	1303	1402	1306	1404	1307	1179	1105	1403	1306
Max Rescaled Pseudo R2	0.30	0.32	0.18	0.27	0.12	0.12	0.02	0.15	0.07	0.16
c-statistic	0.79	0.79	0.74	0.78	0.64	0.66	0.57	0.64	0.66	0.63

Notes. Models adjusted for the individual characteristics of age, completed primary school, parental presence, and presence of partner, and witnessing community violence; Under 3 % of the female sample reported drug use (n = 40), thus the OR estimates are not presented; OR = Odds Ratio; aOR = Adjusted Odds Ratio.

^*p-value <0.05.

Similar to females, the associations between physical violence and internalized mental health outcomes were not significant among the male sample (See Table 4). However, fewer significant findings concerning emotional and sexual violence among the male sample were detected, compared to the female sample. Exposure to emotional violence was significantly associated with suicidal thoughts (aOR: 4.59;95 % CI: 1.32, 15.93) and psychological distress (aOR: 3.51;95 % CI: 1.74, 7.10), but not self-harm (aOR: 1.91;95 % CI: 0.70, 5.24). Moreover, exposure to sexual violence was associated with suicidal thoughts (aOR: 4.81;95 % CI: 1.84, 12.58) in the adjusted model. The significant association noted between sexual violence and self-harm in the crude model (OR = 2.46, CI = 1.10, 5.46) was marginally insignificant (p-value = 0.05) in the adjusted model (aOR = 2.34, CI = 0.99, 5.52), indicating that sexual violence was significantly associated with self-harm among males.

Table 4.

Estimating internalized mental health outcomes and substance use from childhood exposure to sexual, emotional, and physical violence victimization among males, 2018 Colombia Violence against Children and Youth Survey.

	Male Youth (13–24) n = 1299
	Internalized Mental Health Outcomes						Substance Use
	Suicidal Thoughts		Self-Harm		Psychological Distress		Binge Drinking		Smoking		Drug Use
	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted
	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value	OR; [95 % CI]; p-value	aOR; [95 % CI]; p-value
Lifetime Sexual Violence (Ever vs Never)	4.92* [1.99, 12.21] 0.00	4.81* [1.84, 12.58] 0.00	2.46* [1.10, 5.46] 0.03	2.34 [0.99, 5.52] 0.05	1.62 [0.96, 2.71] 0.07	1.14 [0.56, 2.31] 0.71	0.93 [0.52, 1.68] 0.82	0.68 [0.36, 1.28] 0.23	1.16 [0.51, 2.62] 0.73	0.96 [0.35, 2.61] 0.93	0.95 [0.34, 2.65] 0.93	0.93 [0.33, 2.65] 0.90
Lifetime Emotional Violence from Caregivers (Ever vs Never)	4.44* [1.54, 12.82] 0.01	4.59* [1.32, 15.93] 0.02	2.13 [0.84, 5.37] 0.11	1.91 [0.70, 5.24] 0.21	3.49* [2.06, 5.90] <0.0001	3.51* [1.74, 7.10] 0.00	1.19 [0.52, 2.72] 0.67	1.80 [0.84, 3.86] 0.13	1.11 [0.54, 2.27] 0.77	1.67 [0.76, 3.66] 0.21	1.64 [0.69, 3.91] 0.26	2.15 [0.91, 5.08] 0.08
Lifetime Physical Violence (Ever vs Never)	1.67 [0.49, 5.62] 0.41	1.66 [0.52, 5.33] 0.40	1.29 [0.43, 3.88] 0.65	1.33 [0.41, 4.31] 0.64	1.32 [0.91, 1.92] 0.15	1.48 [0.88, 2.49] 0.14	1.51 [0.86, 2.65] 0.15	1.18 [0.67, 2.07] 0.57	1.43 [0.63, 3.27] 0.40	1.56 [0.65, 3.75] 0.32	1.88 [0.72, 4.90] 0.19	2.06 [0.67, 6.27] 0.21
n	1299	1216	1299	1216	1299	1216	1155	1088	1298	1215	1296	1214
Max Rescaled Pseudo R2	0.19	0.22	0.06	0.13	0.11	0.17	0.02	0.19	0.01	0.14	0.03	0.13
c-statistic	0.76	0.75	0.70	0.67	0.64	0.63	0.56	0.70	0.62	0.71	0.65	0.70

Notes. Models adjusted for the individual characteristics of age, completed primary school, parental presence, and presence of partner, and witnessing community violence; OR = Odds Ratio; aOR = Adjusted Odds Ratio.

^*p-value <0.05.

3.2. Violence and substance use

As Table 4 illustrates, there was no evidence supporting the association between sexual, emotional, and physical violence and substance use outcomes for males. Regarding females, while a significant association between physical violence and smoking was noted (OR: 2.72;95 % CI: 1.10, 6.71), this was a crude model estimate (Table 3). Among females, the association between physical violence and binge bringing was marginally significant (aOR: 1.85; 95%CI: 1.00, 3.42).

3.3. Poly-victimization and internalized mental health

Exposure to poly-victimization was also consistently associated with internalized mental health outcomes among females (Table 5). For example, both the crude and adjusted models displayed significant associations between exposure to poly-victimization and suicidal thoughts (aOR: 11.98; 95 % CI: 5.05, 28.44), self-harm (aOR: 9.49; 95 % CI: 3.67, 24.58), and psychological distress (aOR: 3.15; 95 % CI: 1.66, 6.00).

Table 5.

Estimating internalized mental health outcomes and substance use from childhood exposure poly-victimization among females, 2018 Colombia Violence against Children and Youth Survey.

	Female Youth (13–24) n = 1406
	Internalized Mental Health Outcomes						Substance Use
	Suicidal Thoughts		Self-Harm		Psychological Distress		Binge Drinking		Smoking
	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted
	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value
Unexposed to violence (ref)	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Exposed to 1 Form of Violence	2.94* [1.08, 8.06] 0.04	2.59 [0.93, 7.24] 0.07	1.72 [0.68, 4.32] 0.25	1.59 [0.55, 4.64] 0.392	1.72 [0.91, 3.24] 0.10	1.46 [0.72, 2.94] 0.29	1.00 [0.53, 1.89] 0.99	0.94 [0.48, 1.83] 0.84	1.42 [0.46, 4.41] 0.54	1.08 [0.37, 3.14] 0.89
Exposed to Poly-victimization	16.11* [7.38, 35.18] <0.0001	11.98* [5.05, 28.44] <0.0001	9.78* [4.12, 23.23] <0.0001	9.49* [3.67, 24.58] <0.0001	4.23* [2.14, 8.36] <0.0001	3.15* [1.66, 6.00] 0.00	1.40 [0.80, 2.42] 0.24	1.27 [0.74, 2.20] 0.39	3.99* [1.19, 13.39] 0.03	2.42 [0.76, 7.73] 0.14
n	1399	1303	1403	1306	1406	1308	1180	1105	1404	1306
Max Rescaled Pseudo R2	0.28	0.29	0.20	0.27	0.09	0.10	0.01	0.13	0.06	0.16
c-statistic	0.77	0.79	0.73	0.79	0.63	0.64	0.56	0.64	0.65	0.63

Notes. Models adjusted for the individual characteristics of age, completed primary school, parental presence, and presence of partner, and witnessing community violence. Under 3 % of the female sample reported drug use (n = 40), thus the OR estimates are not presented; OR = Odds Ratio; aOR = Adjusted Odds Ratio.

^*p-value <0.05.

Among males a different response pattern was noted, wherein poly-victimization did not predict self-harm for males in the adjusted model (aOR: 3.3695 % CI: 0.83, 13.61) (Table 6). After adjustment, exposure to poly-victimization was positively associated with suicidal thoughts (aOR: 93.48,95 % CI: 29.63, 294.87) and psychological distress (aOR: 4.64,95 % CI: 2.65, 8.15) among males. However, the adjusted odds ratio between poly-victimization and suicidal thoughts was large in magnitude and corresponded with a wide 95 % confidence interval that precluded meaningful inference and precision. Interestingly, exposure to one form of violence was also positively associated with suicidal thoughts in the adjusted mode (aOR: 16.91,95 % CI: 5.05, 56.66).

Table 6.

Estimating internalized mental health outcomes and substance use from childhood exposure poly-victimization among males, 2018 Colombia Violence Against Children and Youth Survey.

	Male Youth (13–24) n = 1299
	Internalized Mental Health Outcomes						Substance Use
	Suicidal Thoughts		Self-Harm		Psychological Distress		Binge Drinking		Smoking		Drug Use
	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted	Crude	Adjusted
	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value	OR; [95 % CI]; p-value
Unexposed to violence (ref)	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00	1.00
Exposed to 1 Form of Violence	14.67* [4.85, 44.37] <0.0001	16.91* [5.05, 56.66] <0.0001	1.50 [0.45, 5.04] 0.51	1.50 [0.39, 5.72] 0.55	1.48 [0.80, 2.75] 0.21	1.33 [0.76, 2.35] 0.32	1.08 [0.56, 1.97] 0.79	0.82 [0.43, 1.55] 0.54	1.34 [0.56, 3.17] 0.51	1.49 [0.63, 3.55] 0.37	1.35 [0.47, 3.88] 0.58	1.58 [0.41, 6.02] 0.51
Exposed to Poly-victimization	73.54* [25.90, 208.76] <0.0001	93.48* [29.63, 294.87] <0.0001	3.43* [1.06, 11.07] 0.04	3.36 [0.83, 13.61] 0.09	4.68* [2.78, 7.90] <0.0001	4.64* [2.65, 8.15] <0.0001	1.54 [0.81, 2.94] 0.19	1.29 [0.61, 2.72] 0.51	1.56 [0.69, 3.56] 0.29	1.98 [0.80, 4.86] 0.14	2.70 [0.82, 8.94] 0.10	3.39* [1.14, 10.10] 0.03
n	1299	1216	1299	1216	1299	1216	1155	1088	1298	1215	1296	1214
Max Rescaled Pseudo R2	0.22	0.26	0.04	0.12	0.10	0.18	0.01	0.19	0.01	0.14	0.18	0.12
c-statistic	0.75	0.77	0.68	0.66	0.64	0.62	0.56	0.69	0.61	0.72	0.66	0.71

Notes. Models adjusted for the individual characteristics of age, completed primary school, parental presence, and presence of partner, and witnessing community violence; OR = Odds Ratio; aOR = Adjusted Odds Ratio.

^*p-value <0.05.

3.4. Poly-victimization and substance use

As depicted in Table 6, among males, exposure to poly-victimization was only associated with higher odds of drug use in the adjusted model (aOR: 3.39,95 % CI: 1.14, 10.10). Among females, there was little evidence to support the relationship between poly-victimization and substance use Although exposure to poly-victimization was positively associated with smoking for females, this was only true in the crude model (OR: 3.99,95 % CI: 1.19, 13.39) and not the adjusted model (aOR: 2.42,95 % CI: 0.76, 7.73) (Table 5).

All significant findings were robust to sensitivity analyses using multiple imputation except for the association between poly-victimization and drug use among males, which was not found to be statistically significant using multiple imputation (OR:2.39, 95%CI:0.89, 6.44, p = 0.0837) (See Supplementary materials, Table 2).

4. Discussion

The current analysis examines the relationship between exposure to violence (emotional, sexual, physical, and poly-victimization) and adverse mental health outcomes among Colombian youth. Overall, the present analysis revealed that emotional victimization and poly victimization were associated with adverse internalized mental health outcomes among Colombian male and female youth. Exposure to sexual violence was only marginally insignificantly associated with self-harm among males. There was less evidence of associations between substance use and victimization among males and females. By exploring these associations in Colombia, the current paper builds on the emerging evidence base examining mental health and violence burden among children and youth in LMIC, and more specifically Latin America.

Four key findings can be drawn from our analysis. First, there were fewer significant associations between victimization and mental health outcomes among the male youth sample. Accordingly, further research is needed to better understand the mental health consequences of male sexual and emotional victimization and whether childhood/youth violence is associated with self-harm among males. Second, our analysis did not yield strong evidence supporting the association between physical victimization and adverse mental health outcomes (when all three forms of violence were included in regression modelling); instead, emotional victimization was more prominently associated with internalized mental health outcomes for Colombian males and females. Third, after adjustment, our analysis did not reveal consistent significant associations between childhood/youth victimization and substance use outcomes (binge drinking, smoking, and drug use). Lastly, exposure to poly-victimization (two or more forms of violence) was positively associated with internalized mental health outcomes among males and females. The key findings are widely generalizable to Colombian youth (13–24 years of age).

In line with previous analyses conducted in Zimbabwe (Chigiji et al., 2018) and in Haiti, Kenya, and Tanzania (Seff & Stark, 2019), our analysis indicated that emotional violence had a robust impact on suicidal ideation among youth. For example, Chigiji et al. (2018) reported that experiencing emotional violence during childhood increased the odds of feeling depressed in the past month, suicidal ideation, attempted suicide, and alcohol use among male and female youth in Zimbabwe. In contrast, there was no clear pattern of consistent associations between male and female Zimbabwean youth regarding the association between physical violence during childhood and various mental health and substance use outcomes (depression, suicidal ideation, suicide, alcohol, drug use, smoking) (Chigiji et al., 2018). Among females, childhood physical violence was related to feeling depressed and suicidal ideation and among males, childhood physical violence was related to depression, alcohol use, and smoking (Chigiji et al., 2018). The analyses conducted by Chigiji et al. (2018) differ from the present research in that models computed by Chigiji et al. (2018) only held constant age and socio-economic status. Our models considered all three forms of violence (emotional, sexual, physical) and held constant participants’ age, completed primary school, parental presence, and presence of partner, and witnessing community violence. Thus, our estimates of emotional violence account for exposure to physical violence and our estimates of physical violence hold constant exposure to emotional violence.

A similar analytical approach was adopted by Seff and Stark (2019), who modelled suicidal ideation as a function of emotional violence, sexual, and physical violence (while holding age constant). Seff and Stark (2019) reported similar patterns of associations as in the present analysis: childhood emotional violence perpetrated by caregivers was consistently associated with suicidal ideation among male and female youth in Haiti, Kenya, and Tanzania. There were no consistent associations for physical violence and suicidal ideation across Haiti, Kenya, and Tanzania.

In our analysis, female youth who experienced emotional violence from parents/caregivers had 4.23 (95 % CI: 2.09, 8.57) times the odds of ever reporting suicidal thoughts, compared to female youth who never experienced emotional violence; male youth who experienced emotional violence from parents/caregivers had 4.59 (95 % CI: 1.32, 15.93) times the odds of ever reporting suicidal thoughts, compared to male youth who never experienced emotional violence. In contrast, physical violence did not appear to be robustly associated with adverse mental health outcomes. Sexual victimization was significantly and positively associated with suicidal thoughts and self-harm among both male and female Colombian youth. However, it is important to note that the association between sexual victimization and self-harm among males was at the threshold of significance (p = 0.05). A differential effect of sexual victimization on suicidal ideation between males and females was examined by Seff and Stark (2019) who noted that in Haiti exposure to sexual violence was associated with suicidal ideation in females but not males. While exposure to emotional violence was positively associated with suicidal thoughts and psychological distress among both males and females, self-harm was marginally insignificant with respect to emotional victimization among males. In Colombia, the risk factors for self-harm may differ between male and female youth and for males may involve exposures beyond direct emotional, physical, and sexual victimization.

Further, few significant associations between substance use outcomes (binge drinking, smoking, and drug use) were noted in the adjusted models. Despite the high prevalence of substance use disorders among Colombian youth (IOM, 2019), our data did not identify a clear link between emotional, sexual, or physical violence victimization and substance use among Colombian youth. This could be related to the small sample size of those reporting these behaviors. Although the relationship between poly-victimization and drug use among men was significant in the adjusted model, this result was not robust to multiple imputation sensitivity analysis for missing data. Thus, a wider ecology of risk factors for substance use may include exposure to armed conflict, participation in paramilitary groups, exposure to violence within the home between parents/caregivers, and gender norms (Bell et al., 2012; Browne et al., 2021; Cuartas Ricaurte et al., 2019; León-Giraldo et al., 2021). For instance, regarding Colombian males, armed conflict and militarization within Colombia have shaped masculine identities and patriarchal structures (Browne et al., 2021). The degree of conformity to hegemonic masculinity may also affect engagement with substance use among male Colombian youth. Masculinity and femininity norms affect the way boys and girls process and cope with negative exposures, such as childhood violence, leading to differential mental health pathologies and the future perpetration of intimate partner violence (Browne et al., 2021; Dulmus & Hilarski, 2006; Kret & De Gelder, 2012; Shannon et al., 2021). Future research is needed to understand how masculinity and femininity norms within the Colombian context impact mental health pathways, particularly in the Colombian context where machismo may inform gender roles related to help seeking or recognizing mental health symptoms (Shannon et al., 2021).

Lastly, our investigation of poly-victimization revealed significant associations with internalized mental health outcomes but not substance use. This was similar to the associations between emotional violence and internalized mental health outcomes. For example, female youth exposed to poly-victimization had 3.15 (95 % CI: 1.66, 6.00) times the odds of reporting past 30-day psychological distress, compared to females without exposure to poly-victimization; male youth exposed to poly-victimization had 4.64 (95 % CI: 2.65, 8.15) times the odds of reporting past 30-day psychological distress, compared to males without exposure to poly-victimization. While the poly-victimization OR point estimates are larger, the imprecise confidence intervals precluded meaningful inferences regarding differences or population effects.

Similarly, using VACS data, Juan et al. (2019) also noted that poly-victimization plays an important role in shaping mental health outcomes of adolescent girls in Cambodia and Haiti. For example, positive associations between the presence of two or more forms of violence and severe mental distress in the prior 30 days were noted among adolescent girls in Cambodia and Haiti and for suicidal thoughts among adolescent girls in Haiti (Juan et al., 2019). Further, also using VACS data, de Oliveira and Jeong (2021) reported a dose-response relationship between poly-victimization index scores and mental distress, suicidal thoughts, and alcohol use in El Salvador. In Kenya, both male and female youth (19–24 years of age) reported greater odds of anxiety, depression, and suicidal thoughts following exposure to three different types of violence (Seff & Stark, 2019). Among girls, adolescence is a developmental time period where in violence against women and violence against children can intersect, given that adolescent girls face a greater risk of intimate partner violence and a risk for violence perpetrated by family members. Thus, the intersections between violence against women and violence against children among adolescent girls is an important consideration with respect to the relationship between poly-victimization and mental health adversity among female youth (Guedes et al., 2016). Thus, the present analysis contributes to an emerging body of literature examining the negative effects of poly-victimization on adverse mental health among youth in LMIC.

4.1. Implications

Since 2015, Colombia has planned and implemented the VACS, with findings yielding a series of recommendations based on the WHO INSPIRE framework (WHO, 2016). Namely, Colombia joined the End Violence Against Children Partnership and developed National Action Plan for Preventing Violence Against Children, which outlined 184 interventions that are being implemented by the government, civil society, and international organizations (End Violence against Children, 2019). National goals include reducing violence against children by >14 % between 2018 and 2024 (End Violence against Children, 2019). Collectively, the interventions address multiple areas of prevention including, norms and value changes, parental support, income and economic strengthening, the promotion of education and life skills, and bolstering data systems.

The present findings have implications for the existing violence prevention and response strategies being planned and implemented in Colombia, namely work on reducing and treating mental distress among victims of violence and reducing emotional violence by parents/caregivers (IOM, 2019). Our analysis also draws attention to the mental health consequences of emotional violence experienced during childhood and youth. While the present analysis cannot make causal claims, we noted consistent associations between childhood emotional victimization perpetrated by parents/caregiver and adverse internalized mental health outcomes among males and females. Institutional strengthening strategies for screening and responding to mental health problems in a variety of settings may be beneficial for children. These strategies may also encourage healthy parenting practices within the wider context of insecurity and violence. A more detailed understanding of how traditional and harmful gender norms impact differential mental health outcomes following violence is also needed to transform inequitable social norms.

4.2. Limitations

The present analysis must be interpreted alongside certain limitations. First, the cross-sectional nature of the data and use of lifetime violence exposures negatively affect causal inference. Specifically, the correct temporal sequence between victimization and mental health outcomes cannot be established particularly for models wherein mental health outcomes were also lifetime measures (i.e., suicidal thoughts and self-harm). Second, while probability sampling was employed, institutionalized youth, persons living with cognitive impairments and physical disabilities that limited their ability to understand and respond to the questionnaire were not eligible for participation. Given that youth living with disabilities may be more likely to experience violence and co-morbidity with adverse mental health, excluding this group of youth may have resulted in underestimation and limits generalizability to such marginalized groups. Third, given that all data were self-reported, due to social desirability bias respondents may have underestimated experiences of violence victimization, substance use, and adverse mental health. Conversely, asking many behaviorally specific questions likely helped to guard against underreporting.

Fourth, our binary and ordinal operationalization of childhood victimization exposures prevented us from assessing the relationship between frequency and severity of childhood violence on mental health and substance use outcomes among youth. Building on our work, future research may consider more complex operationalizations of childhood violence that consider frequency and severity. Lastly, given that sex differences in health-reporting behaviors have been documented, male participants may have been more likely to underestimate the presence of self-reported mental health outcomes, compared to female participants (Caroli & Weber-Baghdiguian, 2016).

5. Conclusion

Using the Colombian VACS, the present analysis examined sex-disaggregated associations between violence victimization (emotional, sexual, physical, and poly-victimization) and various mental health consequences, including substance use. Overall, findings indicate that emotional victimization and poly victimization are associated with adverse mental health outcomes among Colombian male and female youth. Further, only exposure to sexual violence was marginally associated with self-harm among males and substance use related outcomes were not related to victimization among males or females. To expand on the current findings, further research should consider other violence-related risk factors at the individual and community level, with respect to substance use and self-harm outcomes among males. More nuanced analyses investigating relevant mediators (such as, use of mental health services or social support) could advance the understanding of how childhood/youth violence may lead to adverse mental health among Colombian youth.

Data availability

Data are available upon request from https://www.togetherforgirls.org/request-access-vacs/.