The cross-sectional survey was conducted in Y City, Henan Province, China, in July 2023, where there is frequent labor migration to urban areas for work (Figure S1 in Multimedia Appendix 1). For participant recruitment, a multistage stratified random sampling method was used. First, 7 towns were chosen: 3 low-, 2 medium-, and 2 high-economic towns. Second, 102 survey sites were selected within these towns. With support from local Women’s Federation staff or social workers of these selected survey sites, participants were invited to complete a face-to-face questionnaire administered by trained field investigators in meeting rooms of village committee service centers. Data on women’s sociodemographic and lifestyle information, violence exposure, and mental health outcomes were self-reported and collected following the World Health Organization recommendations to maximize participant safety and disclosure [26]. To ensure data quality, daily debriefing sessions were held with all field investigators to address potential questions arising from interviews based on their submitted questionnaires. The statistics team performed daily checks on antifraud questions, minimum completion time thresholds, and logical consistency checks (eg, implausible age values).

Eligible participants were rural married women aged 18 years or older with normal communication and understanding ability, whose husbands had been absent for at least 6 months within the past year. A total of 1525 questionnaires were finally retained for the survey, of which 1519 were valid. Participants with missing or ambiguous data on violence (n=3) or mental health outcomes (n=13) were excluded, resulting in a final sample of 1503 women (Figure S1 in Multimedia Appendix 1).

The study protocol was approved by the Ethics Review Committee of Zhejiang University School of Public Health (approval number: ZGL202306-5). All participants provided written informed consent for participation and publication. Study data were anonymized prior to analysis. Each participant received a health gift package valued at approximately ¥20 (approximately US $2.80). No identifiable images of participants are included in the manuscript or Multimedia Appendix files.

The questionnaire used to assess various forms of VAW in this study was developed based on the World Health Organization’s VAW Instrument [27]. The adapted questionnaire was validated through a 2-round Delphi survey with 18 experts, ensuring its credibility. The lifetime VAW was divided into 2 primary forms according to the sources: IPV (violence from partners) and NPV (violence from other family members or outside the family) (Figure 1). Given that technology-driven, geographically distant abuse is unique for rural left-behind women, remote IPV was specifically measured. IPV was measured using 25 items across 6 types (social IPV, economic IPV, emotional IPV, physical IPV, sexual IPV, and technology-facilitated IPV). Remote IPV was assessed using 16 items across 4 types (social remote IPV, economic remote IPV, emotional remote IPV, technology-facilitated remote IPV). NPV was measured through 19 items across 5 types (social NPV, economic NPV, emotional NPV, physical NPV, and sexual NPV) (Table S1 in Multimedia Appendix 1).

Depressive symptoms were measured using the 10-item Center for Epidemiological Studies Depression scale [28], which demonstrated good internal consistency (Cronbach α=0.80) [29]. Participants were asked how frequently they experienced a series of depressive symptoms over the past week, with response options ranging from 0 (rarely or none of the time) to 3 (most or all of the time). The total 10-item Center for Epidemiological Studies Depression Scale score ranged from 0 to 30, with a cutoff score of 10 indicating the presence of depressive symptoms [30].

Anxiety symptoms were assessed using the generalized anxiety disorder-7 scale [31,32], which also showed excellent internal consistency (Cronbach α=0.92) [33]. The Generalized Anxiety Disorder-7 includes 7 items rated on a 4-point scale: 0 (not at all) to 3 (nearly every day), with a cutoff score of 5 indicating anxiety symptoms [34].

CDA was defined as the simultaneous presence of both depressive and anxiety symptoms. In this study, CDA, depressive symptoms, and anxiety symptoms were the primary mental health outcomes of interest.

Resilience was measured using the 10-item Connor-Davidson Resilience Scale [24,35,36], with a Cronbach α of 0.98. Items were rated on a 0‐4 scale, with higher scores indicating greater resilience (range 0‐40). Social support was measured using the Social Support Rating Scale, including objective support (3 items), subjective support (4 items), and utilization of support (3 items), with a total score of 66 points. The Social Support Rating Scale has demonstrated acceptable internal consistency and reliability, with a Cronbach α of 0.70 in this study [37].

Covariates include women’s sociodemographic, lifestyle, and left-behind related variables. The age of women was assessed as a continuous variable. Education was classified into primary school and below, middle school, and high school and above. Work was categorized into no job, farmer, formal job, casual job, or others. Annual household income in CNY was divided into below ¥30,000, ¥30,001‐50,000, and ¥50,001 and above (all currency values in Chinese yuan [¥] are based on the conversion rate of ¥7.1 = US $1, the approximate exchange rate at the time of the study). An unhealthy lifestyle consists of smoking, drinking, and poor sleep quality. BMI was calculated as weight divided by the square of height and categorized into 3 levels (normal weight or under: <23.9 kg/m², overweight: 23.9‐27.8 kg/m², and obesity: ≥27.9 kg/m²) [38]. Chronic disease history was self-reported physician diagnosis of the common noncommunicable diseases, including hypertension, dyslipidemia, diabetes or hyperglycemia, gastrointestinal disorders, liver diseases, heart diseases, stroke, urological disorders, respiratory system diseases, arthritis, or rheumatism. Length of being left behind was divided into below 3 years, 3‐6 years, 6 years and above.

Descriptive statistics summarized rural left-behind women’s characteristics by CDA, depressive, and anxiety symptoms. Skewed continuous variables were reported as medians (IQRs) and compared via the Wilcoxon rank-sum test. Categorical variables were presented as frequencies (%) and analyzed using Pearson chi-square or Fisher exact tests.

Multivariable logistic regressions were used to analyze associations of any VAW and its forms (IPV [including remote IPV] and NPV) as well as specific types with CDA, depressive, and anxiety symptoms. Models were adjusted for age, education, occupation, household income, healthy lifestyle, BMI, chronic disease history, and length of being left behind. Restricted cubic splines were used to investigate the dose-response associations between the number of VAW items and mental health outcomes.

A 4-way decomposition analysis was performed to quantify how much of the association between VAW and mental health outcomes could be attributed to mediation (the sum of the mediated interaction and the pure indirect effect) or interaction (the sum of the reference interaction and the mediated interaction) of low resilience and social support [39]. In addition, separate mediation and additive and multiplicative interaction analyses were also performed. Detailed methods are described in Multimedia Appendix 1.

Population attributable fractions (PAFs) were calculated to estimate the proportion of CDA, depressive symptoms, and anxiety symptoms cases that could potentially be attributable to VAW based on the following equation[40]: PAF=Pe(RR-1)1+PeRR-1 . Our study replaced risk ratios (RRs) in the formula with odds ratios (ORs) obtained in the above logistic regression. The fraction of participants being exposed to the risk factor (Pe) and the OR of the outcome associated with the corresponding factors were derived from the study. The 95% CI of PAF was calculated using bootstrap [41]. Furthermore, we explored pathway-specific PAFs of potential mediators between VAW and outcomes, which indicated the relative reduction in outcome prevalence if a certain mediating pathway was disabled. The “graphPAF” R package was used to obtain pathway-specific PAFs [42,43].

All statistical analyses were performed using R software (version 4.2.1; R Foundation for Statistical Computing). Statistical significance was considered to be <.05 with a 2-sided, or 95% CIs of OR that did not cross 1.00.

The characteristics of rural left-behind women stratified by CDA are shown in Table 1. Of the 1503 rural left-behind women, 12.8% (n=192) had CDA. Women with CDA were more likely to experience VAW (43.8% vs 29.4%), IPV (40.1% vs 26.6%), remote IPV (12.0% vs 4.4%), and NPV (14.6% vs 5.9%), compared with those without CDA (all P<.05). For VAW types, similar comparisons are also evident.

In the multivariable logistic models presented in Figure 2 and Table S2 in Multimedia Appendix 1, compared with no VAW, VAW exposure was associated with an increased CDA risk (adjusted odds ratio [aOR] 1.84, 95% CI 1.32‐2.54), and this risk rose with each additional VAW item (aOR 1.21, 95% CI 1.13‐1.29). When VAW was categorized by item count, women with 3 or more VAW items showed a markedly higher risk of CDA (aOR 3.41, 95% CI 2.25‐5.15). Similar associations were also observed for depressive and anxiety symptoms in relation to VAW (Figure 2 and Table S2 in Multimedia Appendix 1). Restricted cubic spline curves exhibited dose-response associations between increased VAW numbers and elevated risks of mental health outcomes (Figure S2 in Multimedia Appendix 1).

For the 3 forms of VAW, lifetime IPV (aOR 1.84, 95% CI 1.32‐2.56), remote IPV (aOR 2.79, 95% CI 1.60‐4.74), and NPV (aOR 2.63, 95% CI 1.58‐4.26) were all associated with higher odds of CDA. Among the types of IPV, only economic IPV did not show a significant association with mental health outcomes. Social remote IPV (aOR 3.39, 95% CI 1.71‐6.48), technology-facilitated remote IPV (aOR 5.95, 95% CI 1.07‐30.12), economic NPV (aOR 6.10, 95% CI 1.19‐26.48), and emotional NPV (aOR 3.15, 95% CI 1.86‐5.24) were all associated with higher odds of CDA. Similar findings were observed for depressive symptoms and anxiety symptoms, respectively (Figure 2 and Table S2 in Multimedia Appendix 1).

In a 4-way decomposition analysis, all mediation effects were statistically significant (P<.05), whereas none of the interactions reached significance (P>.05). Overall, 69.4% of the association between VAW and CDA could be explained by low resilience, with 37.9% of this effect attributable to total mediation. Similarly, 45.2% of the association between VAW and CDA could be explained by low social support, with 25.1% attributable to total mediation (Figure 3 and Table S3 in Multimedia Appendix 1).

Separate mediation and interaction analyses validated similar conclusions. Mediation analysis also showed that low resilience and low social support partially mediated the associations between VAW and mental health outcomes (Table S4 in Multimedia Appendix 1). While the additive and multiplicative interactions of “VAW*resilience” and “VAW*social support” on mental health outcomes were not significant, as the 95% CI of relative excess risk due to interaction crossed 0 and the 95% CI of multiplicative scales crossed 1 (Table S5 in Multimedia Appendix 1).

A total of 20.8% of CDA cases, 15.1% of depressive symptoms cases, and 22.7% of anxiety symptoms cases were attributable to VAW (Figure 4). Among the 3 forms of VAW, the PAF of IPV on depressive symptoms (13.9%) is lower than on anxiety symptoms (23.6%), while the PAF of NPV on depressive symptoms (7.7%) is higher than on anxiety symptoms (5.9%) (Table S6 in Multimedia Appendix 1). The pathway-specific PAFs of CDA, depressive symptoms, and anxiety symptoms attributable to VAW through resilience or social support are presented (Figure 4 and Table S6 in Multimedia Appendix 1). As a mediator, resilience accounted for 7.2%, 6.3%, and 3.4% of CDA, depressive symptoms, and anxiety symptoms with VAW, respectively. The pathway-specific PAFs of social support between VAW and CDA, depressive symptoms, and anxiety symptoms were 4.7%, 4.2%, and 1.9%, respectively.

Chinese rural left-behind women’s exposure to lifetime VAW, including IPV (and remote IPV) and NPV, is strongly associated with adverse mental health outcomes, including CDA, depressive, and anxiety symptoms. These associations were partially mediated by low resilience and social support, though interactions were nonsignificant. VAW accounted for a substantial proportion of CDA, depressive, and anxiety symptoms, with differing attributable fractions for IPV and NPV, and resilience and social support partially mediated these attributable proportions.

This study aligns with previous research, reinforcing the association between lifetime VAW and adverse mental health outcomes, including its cumulative effects [44,45]. The left-behind situation among women in rural areas has lasted several decades in China due to disproportionate male labor migration to advanced economic regions. In such a context, women are particularly vulnerable without support from their husbands. Stereotypical gender norms and prevailing beliefs that family matters are private in rural areas constrain women’s autonomy and perpetuate violence [46,47]. Moreover, this vulnerability can be further exacerbated by women’s economic dependence and limited mental health services resources, further deepening their mental health issues [48]. These associations between VAW and mental health issues can also be supported by biological evidence. Repeated violence exposure may lead to dysregulation of the hypothalamic-pituitary-adrenal axis, a critical pathway in the body’s stress response system, contributing to long-term adverse health effects [49]. Moreover, violence exposure can alter neural circuits involved in emotional learning, resulting in impaired memory, heightened threat vigilance, and exaggerated emotional responses to environmental stimuli [50].

The increased risk of adverse mental health outcomes for women exposed to lifetime IPV and NPV is well established and extended by our findings [20,51,52]. Remote IPV has also been found to be an important factor affecting left-behind women’s mental health, calling attention to technology-facilitated remote IPV [16,53]. All IPV types, except economic IPV, were significantly associated with mental health outcomes. The lack of a significant association for economic IPV might be due to the socioeconomic context of rural left-behind women, many of whom are in lower economic strata and depend primarily on their husbands for financial support. Additionally, social and technology-facilitated remote IPV, along with economic and emotional NPV, was associated with higher CDA odds. Our findings expand previous research by comprehensively assessing a broader range of VAW forms, types, and their associations with mental health outcomes.

The associations between VAW and mental health outcomes were partially mediated by low resilience and social support, aligning with previous research. Resilience is thought to buffer the psychological effects of violence, given its ability to access various resources to adapt positively in the face of adversity [54,55]. A study found that empowering women to develop resilience-based coping strategies significantly mediated the impact of IPV on mental health outcomes [56]. Social support is also a vital source of empowerment for women provided by a network of relationships, which helps them cope with stress and maintain well-being. Consistent with our findings, research shows that higher social support buffers the negative impacts of adverse events, helping individuals maintain physical and mental well-being [57,58]. This highlights the importance of strengthening social support networks and fostering resilience as potential interventions to reduce the psychological burden of VAW among rural left-behind women.

Our study estimated that 20.8% of CDA cases, 15.1% of depressive symptoms cases, and 22.7% of anxiety symptoms cases were attributable to VAW. Depressive symptoms were predominantly attributed to IPV, while anxiety symptoms were mainly attributed to NPV. These findings underscore the significant mental health burden associated with VAW, emphasizing the need for clinicians, psychotherapists, and social workers to recognize it as a critical risk factor in rural left-behind women with depression or anxiety. Additionally, pathway-specific PAF revealed that psychological resilience accounted for a greater fraction of mental health outcomes attributable to VAW than social support. This suggests that enhancing resilience may be more effective in reducing the mental health impact of VAW than solely focusing on social support. These findings underscore the importance of incorporating both resilience and social support into public health interventions to address the mental health consequences of VAW.

This study highlights the urgent need for targeted strategies to reduce VAW, foster resilience, and strengthen social support networks through collaborative efforts from communities, health services, and local governments. Such strategies can help to improve mental health outcomes for rural left-behind women in China. For example, implementing educational campaigns to raise prevention awareness of VAW, providing psychological skills training to build resilience, and establishing women’s support groups to strengthen social networks might be practical and effective community-based strategies. Besides, we call for early detection of various forms of VAW and routine mental health screening in both community and health care settings, which can facilitate timely identification and support, such as psychological counseling. Moreover, consistently implementing poverty alleviation and economic empowerment for women living in rural areas in China can further enhance women’s resilience.

Future research should explore prevention and intervention strategies for addressing VAW among left-behind women in rural China. Additionally, intervention studies evaluating the effectiveness of such strategies and policies would provide actionable guidance for improving the mental health of women experiencing violence.

This study also has several limitations. First, the focus on rural left-behind women may limit the generalizability. Second, the use of self-reported data for VAW exposure and covariates introduces potential recall bias, which may underestimate the prevalence of VAW. Third, as a cross-sectional study, this research cannot establish causation, though consistent and cumulative associations suggest potential causal pathways. Finally, the available covariates could not fully capture all relevant factors, such as variables reflecting body fat composition and physical activity, which were not included.

Our study demonstrates that rural left-behind women who experienced lifetime VAW, including IPV (and remote IPV) and NPV, had significantly higher risks of CDA, depressive and anxiety symptoms, with low resilience and inadequate social support partially mediating these associations. Targeted strategies, including reducing VAW, fostering resilience, and strengthening social support networks, are urgently needed to improve mental health outcomes among rural left-behind women. Future research on the development of VAW prevention and intervention strategies, along with mental health support services, could aid in efforts to mitigate the prevalence of VAW and mental health issues for women in rural areas.