The mechanism of HP occurrence has been examined based on different theoretical perspectives focused on intergenerational transmission. Studies based on a social learning perspective suggest intergenerational continuities in angry and aggressive behaviors [11] [12]. The social information processing perspective implies a connection between acceptance and the use of physical disciplinary strategies, the propensity for disciplinary escalation, and the risk for abuse perpetration [13]. The association between positive evaluations of harsh discipline and parents’ aggressive behavior toward their children has been shown to be universal [14]. These studies hypothesize that the intergenerational transmission of HP occurs through direct learning of parenting behaviors and the mediation of positive evaluations of HP. On the other hand, as for the mechanism by which partner aggression and IPV increase HP, in social learning, learned violence becomes a family norm [15], and in stress theory, partner aggression and IPV spill overstress to the mother [16] [17].

Thus, both learning and stress theories explain how childhood HP experiences and current IPV contribute to HP. Interestingly, studies based on social developmental models using prospective data have showed a continuum of HP across generations, simultaneously pointing out the existence of moderators that interfere with the continuum [18]. In this study, we focused on the sense of coherence (SOC) based on the salutogenic theory proposed by Antonovsky [19] and examined the moderating effect of SOC on the propagation of domestic violence.

SOC refers to a person’s ability to use existing and potential resources to combat stress and promote health. SOC is based on one’s perception of manageability, meaning, and comprehensibility [20]. Regarding the association between SOC and violence, the combined experience of violence/abuse during adulthood and low SOC increase the odds ratio (OR) of common symptoms [21], whereas high SOC appears to act as a mediator of the experience of violence and stress reactions at workplace [22]. Studies focusing on the relationship between parenting and SOC have reported that mothers with higher SOC experience lower stress, even when their children have severe autism symptoms [23]. Furthermore, lower SOC is associated with higher use of corporal punishment among mothers of children with intellectual disabilities [24]. These studies indicate that SOC may modulate mothers’ HP through their perceptions and coping styles toward stressful violent experiences. Thus, we believe it is reasonable to assume that SOC is an indicator of resilience in mothers with violent experiences and a moderator of the cycle of violence.

The purpose of this study was to investigate the association between a mother’s experiences of violence and HP of children aged between 1 and 2 years (when HP has been reported to increase) [16] and to examine the effect of SOC in modifying this association.

The participants included mothers who attended a 4-month infant checkup at public health centers in Sapporo City, Hokkaido Prefecture, Japan between May 2001 and June 2001. The researcher personally handed out the baseline questionnaire to 1500 mothers (10% of the annual births) at the health checkup site. Written and verbal information regarding the purpose and methods of the study was provided to the participants; they were informed that participation in the survey was voluntary and answering the questionnaire was considered as their consent to participate. This study was approved by the Sapporo Medical University Ethics Committee (Approval no. 21-2-4).

This research was part of a prospective cohort study. Data were collected in three phases by self-administrated questionnaires. A baseline questionnaire was given to participants at the 4-month infant checkup that included questions about demographic data, unwished-for birth, and child health status; in total, 487 (32.5%) participants responded to the questionnaire. One year after childbirth, the second questionnaire was mailed to each participant that included questions about SOC, IPV experiences during the 1 year after childbirth, childhood experiences of severe corporal punishment by parents, and childhood experiences of witnessing physical violence between their parents. A third follow-up questionnaire was mailed to mothers 18 months after childbirth that focused on HP behaviors, life events, and satisfaction with the husbands’ support. For each phase, reminders were sent to the participants twice.

Data were analyzed using multivariate logistic regression analysis. ORs were used to estimate the bivariate associations between childhood and adult experiences of violence and HP behaviors. To assess the potential confounding effects of the relationship between the experience of violence and HP, the variables for which statistical significance was suggested in chi-square analysis, were entered into three separate multivariate logistic regression models based on violent experience type. To clarify the effect modification of SOC on the relationship between the experience of violence and HP, the following groups were used: mothers with childhood experience of corporal punishment but with a strong or moderate SOC, mothers with no childhood experience of corporal punishment but with a weak SOC, and mothers with childhood experience of corporal punishment and a weak SOC. ORs were used to estimate relative risks. In the same manner, effect modification of witnessing physical IPV during childhood and the experience of IPV during 1 year after childbirth was examined. The Statistical Package for the Social Sciences (IBM, Armonk, NY, USA) was used for all statistical analyses. Statistical significance was determined at the 95% confidence interval (CI) level.

Demographic characteristics of the effectual respondents were compared with those of the non-effectual respondents using the chi-square test. The percentage of mothers aged ≥ 26 was 87% in the effectual respondents group and 77.3% in the non-effectual respondents group (p = 0.010). Of the effectual and non-effectual respondents, 36.9% and 48.1% mothers, respectively, did not graduate from high school (p = 0.015); 89.2% and 75.4%, respectively, were housewives (p < 0.001); and 10.5% and 16.9%, respectively, were economically insecure (p = 0.042). All the above mentioned parameters were statistically significant.

According to the definition described above, the HP group included 69 mothers (24.9%) (shaded text in Table 1). In total, 69 mothers (24.9%) experienced IPV, 93 mothers (33.6%) experienced severe corporal punishment by their parents during childhood and 92 (33.2%) mothers witnessed physical IPV between their parents during childhood (Table 2). The total SOC scores ranged from 17 to 89, and the mean score was 63.2 ± 12.6. The weak SOC group included 86 mothers (31.4%).

*: P < 0.05, **: P < 0.01, ***: P < 0.001. OR: Odds ratio; CI: Confidence interval.

The associations between demographic and psychosocial variables and HP are presented as crude ORs with 95% confidence intervals in Table 2. Severe corporal punishment in childhood (OR = 4.97; 95% CI 1.71 - 7.32, p < 0.001), witnessed physical IPV between parents in childhood (OR = 2.13; 95% CI 1.22 - 3.74, p = 0.012), and IPV during 1 year after childbirth (OR = 2.78; 95% CI 1.54 - 5.02, p = 0.001) were associated with HP. Strong or moderate SOC had a significantly lower OR for HP than a weak SOC (OR = 0.29; 95% CI 0.16 - 0.51, p < 0.001). Young age, dissatisfaction with the husband’s support, and life events were also significantly associated with HP.

A multiple logistic regression analysis was performed to examine the relationship between the experience of violence and HP, while controlling for age, satisfaction with husband’s support, life events, and SOC (Table 3). In the three models of the different types of violence, the OR for the experience of severe corporal punishment (OR = 3.85; 95% CI 2.09 - 7.09, p < 0.001), age (OR = 3.19;

*: P < 0.05, **: P < 0.01, ***: P < 0.001. OR: Odds ratio; CI: Confidence interval; SOC: Sense of coherence; IPV: Intimate partner violence.

95% CI 1.40 - 7.31, p = 0.006) and for SOC (OR = 0.45; 95% CI 0.24 - 0.84, p = 0.013) remained statistically significant. The OR for childhood experience of witnessing physical violence (OR = 1.92; 95% CI 1.04 - 3.52, p = 0.036), for age (OR = 3.60; 95%, CI 1.63 - 7.92, p = 0.001), husband’s support (OR = 0.37; 95% CI0.14 - 0.97, p = 0.043), and for SOC (OR = 0.41; 95% CI 0.22 - 0.77, p = 0.005) remained significant. The OR for childhood experience of IPV during 1 year after childbirth (OR = 2.05; 95% CI 1.08 - 3.88, p = 0.027), for age (OR = 3.05; 95% CI 1.38 - 6.73, p = 0.006), and for SOC (OR = 0.41; 95% CI 0.22 - 0.75, p = 0.004) remained statistically significant. In all multiple logistic models by type of violence, the statistical significance of each violence experience and SOC was maintained.

The effects of concurrent weak SOC on the association between experience of violence and HP were investigated (Table 4). Being exposed to corporal punishment during childhood together with a weak SOC increased the OR for HP (adjusted for age, life events, and the husband’s support, OR = 8.67; 95% CI 3.69 - 20.37, p < 0.001). In the same manner, having witnessed physical IPV in childhood and having a weak SOC increased the OR for HP (OR = 4.43; 95% CI 1.94 - 10.12, p < 0.001). For being exposed to IPV during 1 year after birth and

*P < 0.05; **P < 0.01; ***P < 0.001. OR: Odds ratio; CI: Confidence interval; SOC: Sense of coherence; IPV: Intimate partner violence.

having a weak SOC, the OR increased (OR = 4.93; 95% CI 2.07 - 11.73, p < 0.001). For mothers with a strong or moderate SOC, childhood witnessing of physical IPV and IPV during 1 year after birth lost significance. However, childhood experience of punishment remained statistically significant (OR = 3.78; 95% CI 1.70 - 8.39, p = 0.001).

Participants who completed follow-up had fewer high-risk groups for child abuse, such as poor, young, and poorly educated. In addition, the mean SOC of this study was 63, which is higher than the reported mean of 54 for the Japanese population aged 25 to 34 [26]. The inclusion of participants in the study may have been biased toward healthier mothers.

Each of the three types of violent experiences increased the OR for HP, and parental HP was associated with childhood experiences of HP [11] [12] [13] [14] [18] [27] [28] and experiences of domestic violence [15] [16] [17], consistent with the results of previous studies. Furthermore, mothers witnessing physical violence between parents during childhood were associated with HP, whereas mothers having a good relationship with a current partner were found to mitigate HP. Wood and Sommers [29] reported that girls who witnessed IPV in childhood may become aggressive in self-defense, or they may become aggressive to control a difficult situation. Support from husbands seemed to alleviate mothers’ instinct for self-protection and parenting difficulties and reduce their aggression toward their children, but this needs to be further verified in future investigations.

The results of the present study showed that maternal SOC independently moderates the association between experiences of violence (i.e., witnessing corporal punishment and IPV as a child, and current IPV) and HP. These findings are similar to those of Kimura and Yamazaki [24], who found that SOC influences HP practices.

Furthermore, the way SOC mitigates the association between experiences of violence and HP depended on the type of violence experienced, with severe corporal punishment in childhood being significantly associated with HP, even if the SOC was “strong” or “moderate.” We believe that these data indicate that mothers perceive corporal punishment in childhood as a correct method of discipline. SOC reflects a person’s outlook on life, their ability to cope with stressful situations, their level of trust in identifying and benefiting from people and resources, and personal attitudes and behaviors [30]. Moreover, severe corporal punishment is possibly perceived as a useful resource for mothers. Russa and Rodriguez [13] reported that parents who resort to physical discipline strategies to address their children’s misbehaviors and noncompliance may escalate to abusive levels in emotional and confrontational interactions. The highest OR of HP in childhood among the three violent experiences means positive perceptions of physical punishment and emotional escalation. Furthermore, SOC may not be able to moderate positive perceptions of physical punishment. The mechanism of the association between IPV and HP has been described as stress spillover [15] [16] [17]. Moreover, witnessing IPV is experienced by women as stress that can cause depression and aggression [29]. SOC, which is believed to take effect under high stress, may confer mitigating effects under these conditions.

This study showed that mothers who maintained high SOC despite various experiences of violence were able to use internal and external resources, control their own disadvantage and stress spillover, and choose healthy parenting with nonviolent strategies. Furthermore, the findings indicate that SOC can indicate resilience in mothers.

SOC is a useful measure of resilience and predictor of risky parenting behaviors and recovery. SOC indicates the ability to utilize resources and trust people, which may affect the acceptance and effect of support. In particular, mothers with low SOC may have difficulties in accepting health education and support programs. First, a trauma-informed approach that supports the mother’s ability to trust others may be necessary, and screening for SOC may help in the selection of intervention strategies.

To break the cycle of violence, victims need to relearn safe parenting and accept support from the people around them. Opportunities are needed not only for mothers but also for their partners, families, schools, and community to learn about the negative health effects of HP and alternative positive parenting practices. A society that can effectively curb violence will enhance the moderating function of SOC, reduce HP among women who have been victims of violence, and contribute greatly to improving people’s mental health.

Although the data for this survey was collected in 2001, the results are not affected by time because the study purpose is not to reveal the distribution of abuse, but to reveal the mechanism of the abuse chain. However, the data were collected from a single city in Japan, and the collection rate was low. Therefore, the data are not representative of the general population, but rather represent the results of a relatively healthy population. In contrast, the number of 277 effectual responses is 10 times more than the number of explanatory variables (14 in this study), which is considered to be the standard sample size for logistic regression analysis [31]; hence, we believe that it is a sufficient sample size to obtain valid analysis results. Furthermore, the HP measurement tool is a prototype, and its reliability needs to be further verified.