Childhood exposure to violence is associated with risk for mental disorders and adult’s weight status: a community-based study in Tunisia

Abstract

Background

We sought to investigate the relationship between social violence and adult overweight/obesity and the role of common mental disorders (CMD) in mediating this relationship.

Methods

A cross-sectional study was conducted from January to June 2016 in Tunisia. Participants were selected from randomly selected Primary Health Care Centers. The Arabic version of the Adverse Childhood Experiences-International Questionnaire (ACE-IQ) was used.

Results

A total of 2120 participants were included. Women exposed to social ACEs had higher rates of overweight/obesity than men (13.5 versus 9.5%; P = 0.004). For women, statistically significant partial mediation effects of CMD were observed for exposure to community violence (% mediated = 17.7%). For men, partial mediation was found for the exposure to peer violence (% mediated = 12.5%).

Conclusion

Our results provide evidence of the independent increase of overweight/obesity after exposure to social ACEs. Efforts to uncover and address underlying trauma in health care settings may increase the effectiveness of obesity interventions.

Introduction

Overweight and obesity are defined as abnormal or excessive fat accumulation that may impair health.¹ The global prevalence of overweight has increased substantially in recent decades² and obesity has more than doubled since 1980.¹ The World Health Organization (WHO) estimates that more than 1.9 billion adults worldwide are overweight, of these; at least 600 million are obese.¹ Overweight and obesity are associated with multiple health conditions for adults^3–5 causing 3.4 million deaths, 3.9% of years of life lost and 3.8% of disability adjusted life years (DALYs) globally.⁶ In Tunisia, the prevalence of overweight and obesity remain high at an estimated 25.4% (37% in women and 13.3% in men).⁷ Most of the Tunisian studies done on this subject have only evaluated the role of lifestyle in obesity development.⁸

The etiology of obesity is complex involving interactions among genetic, environmental and behavioral factors.⁹ One such factor is the adverse childhood experiences (ACEs) such as abuse, neglect, or exposure to family dysfunction.¹⁰ This early exposure to adversity increases the risk of obesity, which may not be present in childhood but may appear during early adulthood.¹¹ Shin and Miller¹² demonstrated that adolescents who reported being maltreated had faster BMI growth rates into young adulthood than adolescents who reported no maltreatment. Min et al.¹³ studied poor, urban African American women and found that childhood maltreatment was not associated with adult obesity. Midei and Matthews¹⁴ conducted a meta-analysis examining the role of interpersonal domestic violence in childhood and obesity and found that most studies demonstrated an association between obesity and at least one form of childhood adversity. The sex difference reported by Fuemmeler et al.¹¹ is intriguing and requires further examination. While they propose that their finding of an association for men, but not women, rests on their use of a community sample rather than clinical sample, their focus on young adults and sex differences in social pressure regarding weight, we propose that the nature of the traumas studied may also account for their findings. Most of these studies, however, have examined the association between adult obesity and adversities faced within the family. It has been demonstrated that the traditional ACE factors fail to capture a significant proportion of citizen exposure to community level adversities (witnessing violence, bullying…).¹⁵ Childhood adversities that occur outside of the family (referred to as social ACEs hereafter) can include such adversities as peer victimization, bullying and community violence in the form of crime. While the original work showed a relationship between intra-familial ACEs and obesity¹⁶ there is growing support for an association between social ACEs and obesity in both childhood and later adulthood.¹⁴

Earlier studies, have not considered another form of social ACEs that affects millions worldwide; namely exposure to collective violence or war. Armed conflicts, and attacks against civilian populations are at an all-time high and known to raise mental health problems, morbidity and mortality.¹⁷ A longitudinal study of Kuwaiti children estimated that 6% of the variance in young adult BMI could be explained by the direct association with exposure to war as a preadolescent.¹⁸

Previous studies have demonstrated the role of intra-familial ACEs and common mental disorders (CMD) especially anxiety and depression.¹⁹ Such mental disorders are likely mediators of the adversity and obesity link.¹⁴ According to De Venter et al.²⁰ strong correlations exist between child abuse or neglect and later CMD. There appears to be a bidirectional relationship between CMD and maladaptive eating patterns.^21,22 The source of this CMD and weight association remains poorly explained.²³ Far less is known about the potential role of mental disorders in mediating the relationship between social ACEs and overweight/obesity.

This study sought to contribute to the emergent literature in this area by examining the relationship between social ACEs and adult overweight/obesity. In addition, we sought to contribute to the literature by examining CMD in mediating the relationship between social ACEs and overweight/obesity. We also evaluated the independent effect of social ACEs by adjusting for intra-familial ACEs and other covariates such as age and physical activity. This will be done allowing for a comparison of these effects by gender among a community-based sample of Tunisian adults.

Methods

Setting and sampling

The study was carried out in the region of East Central Tunisia including three cities (Monastir, Mahdia and Kairouan). We performed a cross-sectional study including a random sample of adults from January to June 2016.

Recruitment took place in Primary Health Care Centers (PHC) and eligible participants were adults (ages 18 years or older). PHC were chosen from multiple parts of the study region to ensure representation from the various socio-economic groups. We selected a random sample of 15 PHC in the study region. Each center provided many preventive and curative health services, such as family planning, vaccination, child development assessment, breast and cervical cancer screening, pregnancy follow-up and management of acute and chronic diseases. Thus, people attending these PHC are representative of the general population.

We used the formula of cross-sectional studies to calculate the prevalence of ELA (N = Z²P (1−P)/i²). In this formula N is the sample size, Z is the statistic corresponding to level of confidence, P is the expected prevalence, and i is precision (corresponding to effect size).²⁴

Based on the frequency of childhood adversities among adults of 79.5% according to previous studies,²⁵ the minimal sample size required for the study was 1562

After sampling procedures, eligible participants were invited to participate in the study by responding to a self-administered questionnaire. A total of 2120 adults agreed to participate; each one of them signed a written consent form approved by the University hospital of Monastir, Tunisia Ethics Committee.

Study instruments

Measurement of childhood adversities

Data were collected using the Arabic version of the Adverse Childhood Experiences-International Questionnaire (ACE-IQ) developed by the WHO²⁶ then validated in Saudi Arabia⁵ and adapted to the Tunisian context. The use of the ACE-IQ has demonstrated a strong reliability in other Arabic countries (Cronbach’s alpha > 0.80²⁷) and previous utility has been demonstrated in the study of adversity and chronic disease, risky health behavior, and addiction.^28,29

The ACE-IQ includes 13 questions that are scored into eight categories of childhood adversities (see Appendix 1), divided in two sections:

• – Intra-familial ACEs (ACEs experienced in the home): including conflictual relationship to parents/caregivers; neglect; household dysfunction; physical abuse and sexual abuse.

• – Social ACEs (ACEs experienced in the society): including peer violence, witnessing community violence and exposure to collective violence.

All 13 questions about ACEs were related to the respondents’ first 18 years of life and respondents indicate whether or not the adversity applies to them. This produces binary scores for the 13 items that are then used to compute the five intra-familial ACEs and three Social ACEs (see Appendix 1). The WHO provides further information, including scoring and additional validity studies, on the ACE-IQ (Organization, 2000). This version also includes the Health Appraisal Questions (HAQ), which collects self-report of chronic health conditions (diabetes, hypertension, cancer, etc.), depression, anxiety and other mental illnesses. Part VI, Sections A and B of the HAQ were used to gather complaints of low mood, anxiety and sleep disruption. This instrument has been used reliably in other studies of Arabic countries.²⁷

Measurement of body mass index

The measurement of height and weight was performed by the health professionals. Body mass index (BMI) was calculated as weight/height² (kg/m²). National Heart, Lung and Blood Institutes recommendations for BMI were used to determine categories of underweight (BMI < 18.5 kg/m²), normal weight (BMI 18.5–24.9 kg/m²), overweight (BMI 25.0–29.9 kg/m²) and obesity (BMI > 30.0 kg/m²).³⁰

Statistical analysis

Data entry and statistical analysis were performed using the SPSS version 21.0. The descriptive analysis involved an examination of the demographic characteristics and the rate of each category of ACEs for women and men.

A logistic regression analysis was used to estimate the relationship between overweight and obesity (estimated by the BMI, then divided in a binary variable) and social ACEs adjusted to age, intra-familial ACEs and physical activity. Analysis were stratified by gender, taking into account the results of previous studies suggesting gender differences in ACEs and their relationship with obesity,^9,10 the data were analyzed and displayed for each social ACE factor. Spearman correlations were used to assess the zero-order relationships among social ACEs, CMD and weight status. Mediation modeling was performed to determine the presence of a significant mediation (or indirect effect) of CMD in the relationship between social ACEs and overweight/obesity. CMD was considered as a potential mediating variable when the inclusion of CMD into the model resulted in a partial or total diminution of the relationship between social ACEs as the independent variable and overweight/obesity was the dependent variable.³¹

Mediation analyses were conducted using SPSS version 21.0 and the PROCESS macro developed by Andrew F. Hayes.³² Results were also adjusted to age, intra-familial ACEs and physical activity. Verification of the indirect effect was assessed using the Sobel equation.³³ Confidence intervals for the indirect effects are reported from bootstrap resampling. To apply the mediation model, the relationship between the independent variable (social ACEs) and the dependent variable (weight status as a continuous variable) must be significant (pathway a); second, the variable of mediation (CMD) must be significantly associated with the weight status and with the social ACEs (pathway b); and finally, the relationship between the social ACEs and CMD must be significant (pathway c). Significant mediation occurs when pathway c is reduced significantly (partial mediation) or no longer significant (full mediation) by the inclusion of the mediator into the assessment of pathway c (pathway c′).³² In order to mitigate against Type I error, we chose to use the more conservative P-value of 0.01.

Results

Characteristics of the study sample

Table 1 summarizes the baseline characteristics of the study sample. A total of 2120 participants were included in the study with 1120 were females (52.9%). The mean age of the study population was 36.2 ± 12.5 years. The prevalence of obesity was 11.6% with significant differences between women and men (13.5 and 9.5%, respectively, <0.001).

Table 1.

Baseline characteristics of the study sample by gender (N = 2120)

Characteristics	Male (n = 1000)	Female (n = 1120)	P
Age, mean (SD) (missing = 5)	38.2 ± 13.2	34.5 ± 11.5	<0.001
Level of education, n (%) (missing = 10)			0.043
Unschooled/Primary	454 (45.4)	495 (44.2)
Secondary education	462 (46.2)	501 (44.7)
Post-secondary education	84 (8.4)	124 (11.1)
Working status, n (%) (missing = 13)			0.045
Employed, n (%)	914 (99.3)	1028 (98.4)
Unemployed, n (%)	6 (0.7)	17 (1.6)
Obesity status, n (%)	93 (9.5)	149 (13.5)	0.004
Smoking status, n (%)	721 (74.5)	105 (9.7)	<0.0001
Alcohol consumption, n (%) (missing = 6)	412 (43.3)	38 (3.5)	<0.0001
Illicit drugs use, n (%) (missing = 22)	91 (9.6)	19 (1.8)	<0.0001
Regular physical activity, n (%)	183 (18.9)	113 (10.5)	<0.0001
Depression/anxiety, n (%) (missing = 15)	69 (7.2)	128 (11.8)	<0.0001
Other mental disorders, n (%) (missing = 34)	106 (11)	176 (16.2)	<0.001

Among men and women, 75 and 9.7% were smokers, respectively (P < 0.0001). Significant crosstab differences were found between women and men for mental disorders. In fact, the prevalence of anxiety and depression was 11.2% in women and 7.2% in men (P < 0.0001).

ACEs reporting

As shown in Table 2, nearly 99% of participants experienced at least one ACE. Women and men reported similar prevalence of household dysfunction (68.5 versus 65.6%; P = 0.18) and sexual abuse (5.3 versus 7.7%; P = 0.10). Men were more likely to have experienced all categories of social ACEs as well as physical and emotional abuse than females. For example, 69.5% of men and 36.4% of women experienced peer violence during childhood (P < 0.0001).

Table 2.

Distribution of ACEs categories by gender (N = 2120)

Categories of ACEs	Male (n = 1000)	Female (n = 1120)	P
Intra-familial ACEs (missing = 19)
Neglect	132 (7.2)	152 (4.7)	0.02
Household dysfunction	616 (68.5)	659 (65.6)	0.18
Emotional abuse	639 (67.1)	602 (56.2)	<0.0001
Sexual abuse	53 (5.3)	80 (7.7)	0.10
Physical abuse	459 (49)	363 (34.5)	<0.0001
Social ACEs (missing = 15)
Peer violence	657 (69.5)	375 (36.4)	<0.0001
Witnessing community violence	519 (54.9)	376 (36.5)	<0.0001
Exposure to war/collective violence	158 (16.7)	61 (5.9)	<0.0001

As shown in Fig. 1, compared to the group with no ACEs, participants exposed to social ACEs reported significantly higher rates of overweight/obesity which differed by sex. Women exposed to social ACEs have higher rates of overweight/obesity than men. To illustrate, 53.8% of women and 18.1% of men who experienced three social ACEs qualified as overweight or obese (P < 0.001).

Fig. 1.

Prevalence of overweight and obesity by social ACEs status and gender.

Regression analysis of social ACEs and overweight/obesity

The analysis of each gender uncovered a significant sex difference in the association between adult overweight/obesity and the three categories of social ACEs (Table 3). For women, the distribution of social ACEs in the unadjusted model was significantly associated to the weight status during adulthood. For example, exposure to peer violence increased significantly the coefficient of overweight/obesity (β = 1.50, CI: 1.17–1.84), and community violence increased the probability of overweight/obesity by 1.68 (95% CI: 1.35–2.01). After adjustment for age, physical activity and intra-familial ACEs, the risk of women overweight/obesity remained significantly associated to the occurrence of all social ACEs (Table 3). To illustrate, adjustment in the exposure to peer violence reduced the coefficient of overweight/obesity from 1.50 (95% CI: 1.17–1.84) to 1.18 (95% CI 0.84–1.52, P < 0.001), adjustment in the model for witnessing collective violence reduced the coefficient of overweight/obesity from 3.59 (95% CI: 3.02–4.16) to 2.97 (95% CI: 2.40–3.55, P < 0.001). In men, all categories of social ACEs also increased the likelihood of adult overweight/obesity. In fact, exposure to peer violence led to an increase of the overweight/obesity by 0.46 (95% CI = 0.20–0.73, P<0.001). Also, witnessing collective violence increased slightly the same coefficient by 0.29 (95% CI: 0.04–0.50, P=0.021). After adjustment for age, physical activity and intra-familial ACEs, the risk of adult overweight/obesity among men remained significantly associated to peer violence (β = 0.36, 95% CI: 0.08–6.32), witnessing collective violence (β = 0.67, 95% CI: −0.050.46). No significant relationship was found between community violence and overweight/obesity among men after adjustment (Table 3).

Table 3.

Unadjusted and adjusted relationships between social ACEs and overweight/obesity, by gender (N = 2120)

ACE	Women		Men
	Unadjusted	Adjusteda	Unadjusted	Adjusteda
	β (95% CI)	β (95% CI)	β (95% CI)	β (95% CI)
Peer violence
Yes	1.50 (1.17–1.84)***	1.18 (0.84–1.52)***	0.46 (0.20–0.73)***	0.36 (0.08–6.32)*
No	Referent	Referent	Referent	Referent
Community violence
Yes	1.68 (1.35–2.01)***	1.37 (1.05–1.71)***	0.29 (0.04–0.50)*	0.20 (−0.05–0.46)
No	Referent	Referent	Referent	Referent
Collective violence
Yes	3.59 (3.02–4.16)***	2.97 (2.40–3.55)***	0.75 (0.43–1.07)**	0.67 (0.33–1.01)*
No	Referent	Referent	Referent	Referent

^aAdjusted to age, physical activity and intra-familial ACEs.

*P < 0.05, **P < 0.01, ***P < 0.001.

Mediation analyses

Zero-order Spearman correlations between the three study variables showed that in women and men, social ACEs and CMD were independently associated with weight status, and ACEs were significantly associated with CMD.

The results of the mediation model for both women and men were displayed in Table 4. For women, after adjustment for age, physical activity and intra-familial ACEs and after accounting for the indirect effect of CMD, statistically significant partial mediation effects were observed for exposure to community violence (P = 0.016; % mediated = 17.7%). Although no statistically significant indirect effects were found for peer violence (P = 0.06; % mediated = 3.5%) and witnessing collective violence (P = 0.10; % mediated = 2.8%). The overall mediation model using the cumulative number of social ACEs was not significant (P = 0.13; % mediated = 4.4%).

Table 4.

Adjusted mediation model of the relationship of social ACEs on weight status and common mental disorders by gender (N = 2120)

	Coefficientsa				Sobel test		% Mediatedb	Bootstraping BCa 95% CIc
Type of social ACE	a	b	c	c’	SE	P		Lower	Upper
Women
Peer violence	0.22 (3.10)	0.18 (2.31)	1.15 (6.73)	1.11 (6.47)	0.04 (1.80)	0.06	3.5	0.002	0.12
Community violence	0.12 (1.77)	0.21 (2.56)	1.58 (8.03)	1.30 (7.88)	0.02 (1.38)	0.016	17.7	−0.0006	0.10
Collective violence	0.57 (4.60)	0.13 (1.60)	2.88 (9.76)	2.80 (9.41)	0.07 (1.48)	0.10	2.8	−0.011	0.25
Cumulative social ACEs	0.17 (4.34)	0.11 (1.40)	1.15 (12.27)	1.10 (11.94)	0.03 (2.21)	0.13	4.4	−0.009	0.07
Men
Peer violence	−0.13 (−1.67)	0.42 (6.90)	0.40 (2.96)	0.35 (2.48)	−0.05 (−1.61)	0.09	12.5	−0.14	0.005
Community violence	0.28 (2.20)	0.41 (6.78)	0.21 (1.84)	0.19 (1.51)	−0.04 (−1.20)	0.07	9.5	−0.10	0.10
Collective violence	0.13 (1.33)	0.40 (6.56)	0.66 (3.89)	0.61 (3.68)	0.05 (1.29)	0.09	7.6	−0.014	0.13
Cumulative social ACEs	−0.03 (−0.94)	0.42 (6.88)	0.28 (4.02)	0.25 (3.69)	−0.01 (−0.92)	0.10	10.7	−0.98	0.32

Note: Bold values are the values of the Sobel test used to assess the significance of the mediation analysis.

^aModel adjusted to the effects of age, physical activity and intra-familial ACEs.

^b% Mediated = c – c′/c.

^cBCa 95% CI, bias corrected and accelerated; 5000 bootstrap samples 95% confidence Interval.

After accounting for the indirect effect of CMD and adjustment for age, physical activity and intra-familial ACEs, the overall mediation analysis was not significant for men (P=0.10; % mediated = 10.7%). For men, the analysis of each social ACE showed that no statistically significant mediation effects were found for peer violence (P < 0.09; % mediated = 12.5%), exposure to community violence (P = 0.07; % mediated = 9.5) and witnessing collective violence (P = 0.09; % mediated = 7.6).

Discussion

Main findings of the study

The present study is among the few studies to investigate simultaneously the relationship between exposure to social violence during childhood and adult overweight/obesity by gender controlling for exposure to intra-familial adversities. We found that the prevalence of overweight/obesity was significantly higher in females than males. Women with a history of exposure to social ACEs had a strong increase in risk of obesity ranging from nearly triple to quadruple the risk, while for men witnessing collective violence predicted double the risk for overweight/obesity in the adjusted model. The mediation analysis suggested that the increased risk of overweight/obesity in women was due to community violence is partially mediated by poor mental health, but there was no such mediation for peer violence, collective violence or the cumulative effect of all social ACEs. For men, peer violence and overweight/obese was an indirect effect of poor mental health, but no other association was indirect. These results are partially consistent with a study of Gulf War exposure in preadolescent Kuwaiti children³⁴ where they showed a direct association between war exposure and BMI in young adulthood. However, Llabre and Hadi¹⁸ did not show mediation of the war exposure and BMI direct effect via psychological distress.

What is already known on this topic

Many previous studies showed that ACEs were associated with significant morbidity and mortality decades later.^30,35 In fact, ACEs have been linked to multiple health outcomes in adulthood, such as ischemic heart disease, diabetes, liver disease, autoimmune disease, chronic obstructive pulmonary disease (COPD), depression and obesity.^36,37 In our study, the relationship between ACEs and overweight/obesity were stronger for women than for men both in coefficients and in the patterns of association (all three social ACEs for women but only two for men). These results were consistent with results of previous studies^38–40 showing that, in females, some intra-familial ACEs categories such as sexual abuse and physical abuse were significantly associated with adult obesity after adjusting for other forms of maltreatment and covariates.^41,42

What this study adds

Several studies found that all types of childhood intra-familial abuse^10,43 and household dysfunction⁴⁴ were significantly associated with adult obesity in both males and females. Nevertheless, the absence of relationship between some social ACEs and obesity (after adjustment) in men in this study contrasted with other studies finding specifically that men who reported a history of sexual abuse,⁹ or that emotional abuse⁴⁵ during childhood were more likely to be obese in adulthood. The difference in this case could be explained by the inclusion of intra-familial ACEs categories studied in literature as opposed to the inclusion of both intra-familial and social ACEs in our study. In addition, it could be explained that men were more resistant than females to chronic or severe stress which is shown to be the common risk factor for obesity.⁴⁶ In terms of abuse severity, we found that more than three ACEs were significantly associated with greater risk for obesity compared with no ACEs. The results of this study align with previous studies exploring dose–response effects finding that high frequent abuse was significantly associated with greater risk for adult obesity compared with light and moderate abuse.^7,8

A strong aspect of this study is the use of a community sampling to explore the relationship between ACEs and overweight/obesity in Tunisia for the first time. The large sample size allows for subgroup comparisons and the control for intra-familial ACEs in the examination of mediation of social ACEs.

Limitations of this study

The data were obtained from a retrospective, self-report measure in an adult population, which carries the risk of potential biases and underestimate the true association. Further, well-known risk factors for obesity include behavior factors (e.g. diet) and socio-economic status, which were not assessed in the current study. Finally, childhood weight is a predictor of adulthood weight, which was not available here. Future studies should examine the associations between ACEs and overweight/obesity while controlling for the potential effects of behavior factors, childhood weight and socio-economic status.

Conclusion

This study provides novel results that also align with recent research that established the association between ACEs and overweight/obesity. The impact of ACEs on obesity needs to be evaluated in clinical settings by asking sensitive questions about ACEs. Among people with eating disorders, treatment programs might be more effective when tailored to individuals with a history of childhood exposure to violence. Such programs might be incorporated into primary care with the collaboration of family physicians.

Appendix 1

List of questions related to intra-familial and social ELAs in the Adverse Childhood Experiences-International Questionnaire

ELA Type	No. of questions	Example of questions	Criteria for positive case
Parent/caregiver relationship	2	Parents understand children’s problems/worries; know what children do with their free time	Responses of ‘many times’ or ‘sometimes’
Neglect	4	Insufficient food even when available; parents taking drinks or drugs; refusing to send child to available school; insufficient use of available medical care	Responses of ‘many times’ or ‘sometimes’
Household dysfunction	1 (2 part)	Part 1—living with substance abuser, mentally ill or imprisoned household members; parental separation or death Part 2—witnessing household violence	Response was ‘yes’ to any of Part 1 or ‘many times’ and ‘sometimes’ for the Part 2
Physical abuse	4	Parents push, grab, slap or throw something at children or hit their children so hard resulting in mark or injury	Responses of ‘many times’ or ‘sometimes’
Sexual abuse	4	During the first 18 years of life an adult, relative, family friend or stranger: – ever touched or fondled your body in a sexual way – had you touch their body in a sexual way – attempted to have any type of sexual intercourse with you (oral, anal or vaginal) – effectively had any type of sexual intercourse with you (oral, anal or vaginal)	Responses of ‘many times’ or ‘sometimes’
Peer violence	2	Bullied or physical fight	Responses of ‘many times’ or ‘sometimes’
Community violence	3	Beaten up; stabbed/shot or threatened with a knife/gun in real life	Responses of ‘many times’ or ‘sometimes’
Collective violence	4	Wars, terrorism, political or ethnic conflicts, genocide, repression, disappearances, torture and organized violent crime such as banditry and gang warfare	Responses of ‘many times’ or ‘sometimes’