Growing recognition of experiences of domestic violence (EDV) in childhood and the associated trauma has led to the explicit acknowledgement of children as victim-survivors of domestic violence (including intimate partner violence between caregivers) in their own right ^{Reference Sawyer and McFarlane1} and is now well-recognised as a distinct type of child maltreatment. ^{Reference Mathews, Pacella, Scott, Finkelhor, Meinck and Higgins2} Despite increasing awareness and understanding of the victimisation of children through EDV, little attention has been given to the associated health outcomes later in life. Among studies that have examined the adverse health outcomes associated with childhood EDV, many have relied on measures that only capture incidents of physical domestic violence. ^{Reference Agrawal, Lei, Shah, Bui, Halfon and Schickedanz3–Reference McLaughlin, Basu, Walsh, Slopen, Sumner and Koenen6} The focus on physical violence does not give sufficient consideration to non-physical forms of violence, even though they are common and harmful. These non-physical forms of domestic violence are often referred to as coercive control (a course of conduct aimed at dominating and controlling another through an imbalance of power and fear), a core underpinning of EDV that does not require physical violence to be present. ⁷ New legislation has been introduced in multiple Australian states and territories that prohibit coercive control, which includes diverse types of physical, sexual and psychological violence, economic coercion, surveillance and other forms of abuse within a domestic relationship. ^8,9

Qualitative interviews with children show that they are capable of recognising and articulating experiences of coercive control, ^{Reference Lapierre, Sazgar and Cardeal10} including the persistence of coercive behaviours such as manipulation, control of time, movements and activities, and undermining of the relationship with their other parent beyond the point of parental separation. ^{Reference Thiara and Humphreys11} Recent systematic review evidence has shown that coercive control in childhood is associated with numerous adverse child wellbeing outcomes including emotional and behavioural problems, social and relational difficulties and poorer physical health, independently of other forms of EDV. ^{Reference Xyrakis, Aquilina, McNiece, Tran, Waddell and Suomi12} Given that many studies that examine the health outcomes associated with childhood EDV fail to capture experiences of coercive control, which encompasses diverse forms of violence including financial, sexual, relational or financial for example, the true extent of the attributable harm is likely underestimated.

In a landmark study of mother-child victim-survivors, Katz ^{Reference Katz13} provided empirical evidence of how children are directly harmed by non-physical forms of domestic violence and the way it can profoundly impact them. The study by Katz demonstrated that children can experience harm in the absence of physical forms of domestic violence. Importantly, unlike physical violence, coercive control by a parent – through acts such as surveillance, isolation and economic abuse – can persist long after a relationship has ended and is often perpetrated through the weaponisation of children, particularly through the family courts. ^{Reference Wilde, Sheeran and Douglas14} Perpetrators’ tactics can also include manipulating children into sharing information about the protective parent’s whereabouts, undermining the child’s attachment bond with their protective parent ^{Reference Lapierre, Sazgar and Cardeal10,Reference Thiara and Humphreys11} and withholding financial support such as child support payments. ^{Reference Cook, Byrt, Edwards and Burgin15} Domestically violent parents are also more likely to engage in emotionally and physically abusive or neglectful behaviours towards their children. ^{Reference Heward-Belle16} These experiences can lead to chronic stress, hypervigilance, poor sleep, lower educational attainment and altered neuro-development. ^{Reference Doroudchi, Zarenezhad, Hosseininezhad, Malekpour, Ehsaei and Kaboodkhani17,Reference Tomoda, Nishitani, Takiguchi, Fujisawa, Sugiyama and Teicher18} Compounding children’s EDV are common feelings of being powerless, having no agency over their own lives and not being taken seriously or being involved in decision-making post-separation. ^{Reference Noble-Carr, Moore and McArthur19}

Chronic stress and trauma during childhood significantly increases the risk of mental disorders (e.g. post-traumatic stress disorder (PTSD) and depressive and anxiety disorders) and health risk behaviours (e.g. suicide, substance use and obesity) across the lifespan. ^{Reference De Bellis and Zisk20} In the context of non-physical childhood EDV, few studies examine these health outcomes and even fewer examine them in adulthood. In a US population-based cohort study of adolescents, maternally-reported childhood EDV was associated with higher odds of substance use in adolescence. ^{Reference James, Donnelly, Brooks-Gunn and McLanahan21} In an Australian data-linkage study, early childhood police-recorded EDV was associated with a three-fold increase in the odds of a mental-health related hospitalisation (including for alcohol use) prior to the age of 18. ^{Reference Orr, Fisher, Preen, Glauert and O’Donnell22} In terms of adult outcomes, a Swedish population-based study found that self-reported childhood EDV was associated with a significant increase in PTSD, anxiety, depression and suicide in early adulthood. ^{Reference Cater, Miller, Howell and Graham-Bermann23} Although these studies incorporate non-physical forms of EDV, albeit to varying degrees, none examined the independent influence of each form of EDV on health outcomes. Generating evidence of the harms associated with coercive control and other forms of non-physical forms of EDV in childhood is vital for ensuring that children are adequately protected and supported both during the parental relationship and beyond separation. Screening measures of post-separation EDV have been criticised for failing to detect non-physical EDV such as coercive and controlling behaviours, which can inadvertently place children at continued risk of physical and/or psychological harm. ^{Reference Shorey and Baladram24}

The nuances, subtleties and context-specific nature of coercive control can make it difficult to detect by external observers compared to more overt forms of EDV, and hinder the provision of protection, support and therapeutic intervention for these children. In Australia, a 2021 survey conducted of a nationally representative sample found the national prevalence of childhood EDV among people aged 16 and over is 39.6%. ^{Reference Mathews, Pacella, Scott, Finkelhor, Meinck and Higgins2} Additional analyses of these data from the Australian Child Maltreatment Study (ACMS) have shown intimidation and control (reflecting coercive and controlling behaviours) is experienced more frequently, over a longer period of time, and at older ages, than other forms of domestic violence like physical violence, threats of serious physical harm and damage to property (details available from the author on request). In addition, this subsequent analysis found that intimidation and control presented the strongest risk for experiencing other types of child maltreatment, particularly for emotional abuse, independently of physical domestic violence. There is a need to understand the associations between the different forms of childhood EDV and long-term mental and physical health outcomes to identify areas and opportunities to support recovery and disrupt poor health trajectories for children who experience domestic violence. This paper utilised data from the ACMS, to quantify the outcomes associated with childhood EDV. Specifically, we aim to examine associations between four different forms of childhood EDV (physical violence, serious threats of harm, damage to property or harm to pets and intimidation or control) and four mental health disorders (PTSD, generalised anxiety disorder (GAD), major depressive disorder (MDD) and alcohol use disorder (AUD)), and six physical health risk behaviours (suicide, self-harm, binge drinking, obesity, smoking and cannabis dependence) in adulthood.

Results

Mental disorders

Figure 1 and Supplementary Table 2 present the reported count and prevalence (as at 2021) of each mental disorder among those with and without childhood EDV (and for each form), for the total sample and separately for men and women. All mental disorders were more common among those with any childhood EDV compared to those without. MDD was the most prevalent (any EDV: 27.0%; no EDV: 12.7%), followed by GAD (any EDV: 18.5%; no EDV: 7.2%). The prevalence of PTSD was almost four times greater in those with childhood EDV (9.6%) compared to those without (2.6%).

Fig. 1 Estimated prevalence* of mental disorders in Australians 16 years and over by childhood experience of domestic violence. *Proportions (with 95% CIs) are weighted by age group, gender, Indigenous status, country of birth (Australia or overseas), highest educational level and residential socioeconomic status (Relative Socio-economic Advantage and Disadvantage quintiles). PTSD, post-traumatic stress disorder; GAD, generalised anxiety disorder; MDD, major depressive disorder; AUD, alcohol use disorder; mod, moderate; sev, severe.

Women with any childhood EDV had the highest prevalenceFootnote ^b of PTSD, GAD, MDD, compared to women without EDV and men with and without EDV, whereas men with childhood EDV had the highest prevalence of mild, moderate and severe AUD. A notable gender difference was observed for PTSD prevalence among those with childhood EDV (women: 11.0%; men: 7.5%), but not among those without childhood EDV (women: 2.5%; men: 2.6%).

Figure 2 and Supplementary Table 3 present the odds of each mental disorder for each form of childhood EDV compared to those without. In the fully adjusted model, intimidation or control was associated with increased odds of PTSD (aOR (adjusted odds ratio) 2.30, 95% CI 1.77–2.98), GAD (aOR 1.65, 95% CI 1.36–1.99), and MDD (aOR 1.22, 95% CI 1.04–1.43). Damage to property or pets was associated with PTSD (aOR 1.49, 95% CI 1.15–1.94), MDD (aOR 1.29, 95% CI 1.11–1.50) and severe AUD (aOR 1.76, 95% CI 1.36–2.27). Neither physical violence nor threats of serious harm significantly predicted a mental disorder in adulthood. Adjusting for current and childhood financial hardship, and socioeconomic disadvantage in the fully adjusted models did not significantly attenuate associations.

Fig. 2 Odds ratios and 95% CIs for each form of childhood experience of domestic violence (EDV) and mental disorders. Odds ratios are adjusted for other forms of EDV, age group, the other four types of child maltreatment, current financial strain, socioeconomic disadvantage and childhood economic hardship. PTSD, post-traumatic stress disorder; GAD, generalised anxiety disorder; MDD, major depressive disorder; AUD, alcohol use disorder.

The results of the gender-stratified analysis can be seen in Fig. 2. Serious threats of harm, damage to property or pets, and intimidation or control were associated with a 1.54 to 2.03 increased odds of PTSD for women. Serious threats of harm were also associated with higher odds of mild AUD (aOR 1.54, 95% CI 1.06–2.24) but lower odds of severe AUD (aOR 0.57, 95% CI 0.36–0.89). For men, intimidation or control was the only form of EDV associated with PTSD, GAD and MDD (aOR 1.33–2.55). Damage to property significantly increased the odds of severe AUD (aOR 1.82, 95% CI 1.30–2.53), whereas physical violence was associated with reduced odds (aOR 0.57, 95% CI 0.36–0.89).

Health risk behaviours

Figure 3 and Supplementary Table 4 present the reported count and prevalence of each health risk behaviour among those with and without EDV (and for each form), for the total sample and separately for men and women. Health risk behaviours were generally more common among those with childhood EDV, though 95% CIs overlapped for those with and without EDV for both binge drinking and obesity. Obesity was the most prevalent (any EDV: 27.3%; no EDV: 24.6%), followed by smoking (any EDV: 21.9%; no EDV: 14.3%). The prevalence for suicide attempt was almost four times greater in those with any EDV (1.9%) compared to those with no EDV (0.5%). Women with childhood EDV had a higher prevalence of suicide attempt and self-harm, while men with childhood EDV had a higher prevalence of current smoking, cannabis dependence and binge drinking.

Fig. 3 Estimated prevalence* of health behaviours in Australians 16 years and over by childhood experience of domestic violence. *Proportions (with 95% CIs) are weighted by age group, gender, Indigenous status, country of birth (Australia or overseas), highest educational level and residential socioeconomic status (Relative Socio-economic Advantage and Disadvantage quintile).

Figure 4 and Supplementary Table 5 present the odds of each health risk behaviour by each form of EDV. In the fully adjusted model, intimidation or control was associated with higher odds of self-harm (aOR 1.49, 95% CI 1.07–2.07) and cannabis dependence (aOR 1.64, 95% CI 1.07–2.51), but lower odds of current smoking. Damage to property or pets was associated with an increased odds of binge drinking (aOR = 1.24, 95% CI 1.03–1.50) and smoking (aOR = 1.36, 95% CI 1.17–1.59), but lower odds of obesity. Neither physical violence nor threats of serious harm were significantly associated with any health risk behaviour. No form of EDV type was significantly associated with suicide attempt. Adjusting for current and childhood financial hardship, and socioeconomic disadvantage in the fully adjusted model did not significantly attenuate associations.

Fig. 4 Odds ratios and 95% CIs for each form of childhood experience of domestic violence (EDV) and health risk behaviours. Odds ratios are adjusted for other forms of EDV, age group, the other four types of child maltreatment, current financial strain, socioeconomic disadvantage and childhood economic hardship.

The results of the gender-stratified analysis can be seen in Fig. 4. Significant associations for women included physical violence and binge drinking (aOR 1.57, 95% CI 1.08–2.28), intimidation or control and self-harm (aOR 1.98, 95% CI 1.29–3.03) and obesity (aOR 1.31, 95% CI 1.03–1.65), and serious threats of harm and obesity (aOR 1.38, 95% CI 1.04–1.83). For men, damage to property or pets was associated with increased odds of self-harm, binge drinking and smoking (aOR 1.38–2.46) and reduced odds of obesity. Threats of serious harm increased the odds of cannabis dependence (aOR 2.47, 95% CI 1.33–4.60).

A sensitivity analysis was conducted to examine associations with lifetime self-harm and suicide attempt. Intimidation or control was associated with lifetime suicide attempt (aOR 1.39, 95% CI 1.14–1.70, see Supplementary Table 6). This association remained significant for men only in the gender-stratified analysis (aOR 2.06, 95% CI 1.50–2.84).