Mental and physical health outcomes among intimate partner violence survivors in Taiwan: A nationwide registry-based case control study

Abstract

BACKGROUND

Intimate partner violence (IPV) is a major public health concern linked to psychiatric and physical morbidity. However, evidence from East Asia, particularly from registry-based and sex-inclusive analyses, remains limited. We hypothesized that IPV survivors would have elevated risks of psychiatric disorders and stress-related physical conditions.

AIM

To assess psychiatric and physical health risks following IPV exposure in Taiwan.

METHODS

We conducted a nationwide, registry-based case control study using data from Taiwan’s Health and Welfare Data Science Center. Adults aged 18-64 years with a first IPV report in 2019 (n = 43393) were matched 1:1 by sex and age to controls. Incident diagnoses within 1 year were identified from claims data. Conditional logistic regression was used to estimate adjusted odds ratios (AORs).

RESULTS

Compared to controls, IPV survivors had higher risks of depressive disorders [AOR = 4.18, 95% confidence interval (CI): 3.78-4.60, P < 0.001], bipolar disorder (AOR = 4.81, 95%CI: 3.83-6.10, P < 0.001), schizophrenia (AOR = 1.75, 95%CI: 1.46-2.10, P < 0.001), and alcohol/substance use disorders (AOR = 5.98, 95%CI: 2.21-8.50, P < 0.001). The risk of asthma was modestly elevated (AOR = 1.31, 95%CI: 1.08-1.60, P = 0.006). No significant association was observed for irritable bowel syndrome (P = 0.94).

CONCLUSION

IPV survivors in Taiwan face substantially increased psychiatric risk and a modestly elevated risk of asthma, warranting early screening and integrated mental and physical health care.

Key Words: Intimate partner violence; Depression; Bipolar disorder; Substance use disorder; Asthma; Nationwide registry

Core Tip: This nationwide, registry-based case control study is the first in Taiwan to comprehensively assess both psychiatric and physical health risks among adults with documented intimate partner violence (IPV) exposure. The findings demonstrate that IPV survivors, regardless of sex, face substantially increased risks of major psychiatric disorders and asthma within 1 year, highlighting the urgent need to integrate mental health screening and multidisciplinary care into routine services for IPV-affected individuals.

INTRODUCTION

Intimate partner violence (IPV) is a major global public health concern that substantially affects individuals’ lifelong health, developmental opportunities, and overall well-being[1]. IPV is closely associated with a spectrum of chronic physical illnesses involving the cardiovascular, musculoskeletal, digestive, reproductive, and nervous systems[2-5]. Physically, it can cause both fatal and non-fatal injuries, such as contusions, lacerations, sprains, visceral damage, and fractures, most commonly to the head, neck, and face. Moreover, prolonged IPV exposure leads to psychological stress and neurobiological dysregulation, which may contribute to chronic pain, gastrointestinal (GI) distress, nausea, vomiting, and stress-related asthma[4,6-11].

While most prior research has focused predominantly on female survivors[7,12-14], emerging evidence indicates that men may also experience comparable health consequences from IPV[4,8,15-17]. In Taiwan, the proportion of male IPV reports has increased in recent years[18], highlighting the need for a more inclusive and sex/gender-sensitive research approach. This study addresses this gap by examining both male and female adult survivors concurrently, a strategy still uncommon in the East Asian context.

Sex/gendered socialization, division of labor, and sexlinked biology jointly shape IPV exposure, disclosure, and downstream morbidity, supporting the need for a sexinclusive analytic framework. Sociocultural expectations often place women in caregiving and economically dependent roles, increasing coercive control, chronic stress, and barriers to helpseeking, whereas masculine norms of stoicism and selfreliance can suppress disclosure among men and promote externalizing coping behaviors (e.g., hazardous alcohol/substance use) after violence[4,14,17,19,20]. Corresponding cognitive-emotional response patterns after trauma, greater internalizing symptoms in women and externalizing behaviors in men, map onto distinct neuroendocrine, inflammatory, and healthbehavior pathways, with implications for psychiatric and stresssensitive somatic disease[6,8]. Occupational stratification further modifies vulnerability, with women being over-represented in precarious care work and men being more likely to be involved in physically hazardous or shiftbased jobs, shaping exposure contexts, access to care, and cardiometabolic or circadian stress[2-4,17]. Disease susceptibility also differs by sex, with women having higher risks of depressive and anxiety disorders and men carrying greater burdens of alcohol or substance use. Moreover, asthma and functional GI symptoms show sexspecific prevalence and exacerbation patterns that may interact with traumarelated neuroimmune activation[4,8,14,17,19,20].

In Taiwan, rising male IPV victimization underscores the need to examine these gendered pathways alongside persistent urban-rural and socioeconomic disparities in access and stigma, thereby motivating concurrent analyses of men and women and the focus on psychiatric and stresslinked physical outcomes within a 12month window[18,21]. Empirical evidence from East Asian populations on the full spectrum of psychiatric and physical health outcomes following IPV exposure remains limited. Existing studies are often constrained by small study populations, reliance on self-reported data, or a focus on specific disease outcomes[7,12,13,16]. Large-scale, registry-based studies involving both sexes and comprehensive health endpoints are still scarce.

To address these knowledge gaps, we conducted a nationwide, registry-based case control study in Taiwan, linking government-reported IPV with National Health Insurance (NHI) records to explore the association between documented IPV and subsequent risks of major psychiatric and physical disorders in adults aged 18-64 years, regardless of sex. By examining the integrated health impact among both men and women, our study offers novel, population-level evidence to guide future clinical screening, intervention strategies, and policymaking for IPV survivors in diverse settings.

MATERIALS AND METHODS

Study design and setting

This study employed a nationwide registry-based case control design to examine the association between documented IPV exposure and subsequent mental and physical health outcomes among adults in Taiwan. All procedures were conducted in accordance with the STROBE guidelines for case control studies. The study protocol was approved by the Institutional Review Board of National Yang Ming Chiao Tung University, No. YM111102E.

Data sources

Data were obtained and linked within the secure environment of the Health and Welfare Data Science Center (HWDC), Ministry of Health and Welfare, Taiwan, using encrypted identifiers across four nationwide datasets: (1) The Family Violence Report Registry (2018-2019), to identify IPV notifications based on the “relationship between parties” field; (2) The NHI Enrollment File (2019), to capture demographic and insurance premium information; (3) The NHI Outpatient and Inpatient Claims Databases (2019-2020), to ascertain diagnoses, visits, and admissions using International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) codes; and (4) The National Mortality Registry (2019-2020), to determine vital status for exclusions. Data linkage was performed internally by HWDC, and only analysis-ready de-identified data were made available for on-site processing.

Study population and case definition

The study population comprised all individuals enrolled in Taiwan’s NHI program in 2019. Cases were defined as adults aged 18-64 years who were officially reported and confirmed as victims of IPV in the Family Violence Report Registry between January 1, 2019 and December 31, 2019. A total of 43393 individuals met these criteria and were included in the case group.

Control selection

Controls were selected from the pool of NHI beneficiaries aged 18-64 years who remained alive throughout the study period and had no record of a family violence report in 2019. To minimize potential confounding and survival bias, individuals with family violence reports in either 2018 or 2019 or who died between 2019 and 2020 were excluded from both the case and control pools. After applying these exclusion criteria, the final eligible population comprised 16418438 individuals.

Matching and analytic approach

To control for strong confounding by age and sex while preserving interpretability, we performed individual 1:1 matching on sex and narrow 3-year age bands, followed by conditional logistic regression to estimate odds ratios (ORs) within matched strata, thereby aligning the design and analysis stages. Although propensity score-based matching and related matching methods are widely used in large administrative and insurance datasets to improve covariate balance when randomization is infeasible, such methods cannot remove unmeasured confounding and may introduce selection bias when trimming or caliper restrictions exclude non-overlapping observations[22-24]. In this study, we therefore: (1) Used exact matching on sex and age bands to minimize model dependence for primary demographic confounders; (2) Retained all matchable pairs without caliper trimming to limit selection distortion; (3) Evaluated post-match balance and adjusted for residual socioeconomic imbalances (insurance premium, low-income status, urbanization) in conditional models; and (4) Interpreted estimates in light of potential unmeasured confounding inherent to observational designs. This integrated approach reflects best practices for balancing comparability and generalizability in population-scale registry analyses[22-24].

Exposure and covariates

The primary exposure of interest was documented IPV, ascertained using the 2019 Family Violence Report Registry. Individuals were classified as IPV cases if the “relationship between parties” field indicated a current spouse, ex-spouse, current cohabiting partner, or ex-cohabiting partner[25]. To ensure incident IPV, only those with a first report in 2019 were included, while those with any family violence report in 2018 were excluded. For individuals with multiple reports during 2019, the date of the first notification was designated as the index date for subsequent analyses.

A set of covariates was selected a priori based on prior literature and their potential to confound the association between IPV and health outcomes. These included sex, age [calculated as of July 1, 2019, and grouped into five categories: 18-24 (reference), 25-29, 30-39, 40-49, and 50-64 years], monthly insurance premium as a proxy for household income [low (≤ 28860 New Taiwan Dollar; reference), medium (28861-43642 New Taiwan Dollar), and high (> 43642 New Taiwan Dollar), as defined by the government’s family income survey], low-income household status, as determined by specific insurance codes, and urbanization level of the insured residential area [recategorized from the original seven levels into high (highly and moderately urbanized), medium (emerging and general towns), and low (aging, agricultural, and remote areas)][21]. These covariates were included in all models to control for demographic and socioeconomic differences between groups.

Outcome measures

The primary outcome variables were incident health conditions, classified into two main categories: Psychiatric disorders and physical diseases. Health outcomes were identified from both primary and secondary diagnosis codes from the NHI outpatient and inpatient claims databases within 1 year following the index date of IPV reporting. A condition was considered incident if the corresponding ICD-10-CM diagnosis appeared at least twice in outpatient claims or at least once in inpatient claims during the follow-up period[26]. Psychiatric disorders included depressive disorders (F32-F39)[27], alcohol and substance use disorders (F10-F16, F18, F19)[28], schizophrenia (F20-F29)[27], and bipolar disorder (F30, F31)[27]. Physical diseases included irritable bowel syndrome (K58)[29] and asthma (J45)[30,31]. Outcome selection followed prespecified criteria prioritizing measurability and coding validity in claims data with established algorithms or registry precedent, clinical and public health relevance as plausible short- to medium-term sequelae of IPV, detectability within a 12month window, and minimization of ascertainment bias by excluding endpoints tightly coupled to IPV reporting or acute injury care. Based on these principles, we included major psychiatric disorders and two sentinel physical conditions, asthma and irritable bowel syndrome, reflecting stress-immune and neurogastrointestinal pathways supported by prior evidence and reliably captured through claims coding. In contrast, we excluded outcomes with limited short-term low-validity entities (e.g., complex pain syndromes without validated claims algorithms), long-latency cardiometabolic diseases unlikely to present within 1 year, and acute trauma endpoints that may artificially inflate detection independent of underlying risk[6,8,10,11,14,20].

All diagnoses were classified according to the ICD-10-CM. This classification and ascertainment approach ensured the robust and systematic identification of incident psychiatric and physical health outcomes following IPV exposure. Recent validation studies in Taiwan’s NHI Research Database have demonstrated good performance of ICD-10-CM codes for psychotic and affective disorders, supporting the reliability of this approach[32].

Statistical analysis

Statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, United States). Baseline characteristics before and after matching were summarized descriptively, and group differences were assessed using χ² tests. To preserve sample size and representativeness and avoid overmatching, 1:1 individual matching was prespecified on sex and narrow 3-year age-bands only; socioeconomic indicators were intentionally excluded from the matching algorithm and addressed analytically. Primary associations were estimated using conditional logistic regression, which accounts for the matched design, and results are reported as ORs and adjusted ORs (AORs) with 95% confidence intervals (CIs). Multivariable models adjusted for monthly insurance premium (proxy for household income), low-income household status, and urbanization level. Given the nationwide registry design, all eligible 2019 IPV notifications were included, and 1:1 matched controls were selected; therefore, no a priori power calculation was performed. Missingness was minimal; the small “missing” category in urbanization was retained as a separate level in all models to avoid case-wise deletion.

RESULTS

Baseline characteristics

Before propensity score matching, clear demographic and socioeconomic disparities were observed between individuals with and without IPV exposure. The IPV group (n = 43393) was predominantly female (82.4%), while the non-IPV group had only 49.8% female participants. The age distribution also differed, as follows: 33.2% of IPV survivors were aged 30-39 years, compared to 21.4% among controls, while the youngest age group (18-24 years) was less represented in the IPV cohort (7.0% vs 12.4%). Socioeconomic differences were evident, with 72.6% of IPV survivors in the lowest-income premium bracket (≤ 28860 New Taiwan Dollar) compared to 56.6% of controls, and membership in low-income households was more common (3.6% vs 1.1%). Furthermore, a greater proportion of IPV survivors resided in areas with low urbanization (8.9% vs 6.1%). All differences were statistically significant (P < 0.0001; Table 1).

Table 1 Baseline demographic characteristics of the study population before matching, according to intimate partner violence exposure status, n (%).

Variables	No IPV, n = 16418438	IPV, n = 43393	P value1
Sex			< 0.0001
Male	8244135 (50.22)	7637 (17.6)
Female	8174303 (49.78)	35756 (82.4)
Age in years			< 0.0001
18-24	2029362 (12.36)	3046 (7.02)
25-29	1566086 (9.53)	4162 (9.59)
30-39	3515627 (21.41)	14426 (33.24)
40-49	3777280 (23)	12835 (29.58)
50-64	5530083 (33.68)	8924 (20.57)
Monthly insurance premium			< 0.0001
Low (≤ 28860 New Taiwan dollar)	9290890 (56.58)	31485 (72.56)
Medium (28861-43642 New Taiwan dollar)	3010419 (18.33)	5770 (13.3)
High (> 43642 New Taiwan dollar)	4117129 (25.07)	6138 (14.15)
Low-income household			< 0.0001
No	16245178 (98.94)	41825 (96.39)
Yes	173260 (1.06)	1568 (3.61)
Urbanization level			< 0.0001
High	9841750 (59.94)	24471 (59)
Medium	4748681 (28.92)	13315 (32.1)
Low	1008816 (6.14)	3690 (8.9)
Missing	819191 (4.98)	1917 (4.42)

¹Group differences were examined using the χ² test.

IPV: Intimate partner violence.

After matching by sex and age (3-year intervals), the IPV and control groups each included 43393 individuals with identical distributions for these core demographics, with 82.4% female patients in both groups and similar age profiles. Nonetheless, imbalances persisted in socioeconomic status and urbanization; the IPV group had higher proportions in the lowest-income premium bracket (72.6% vs 56.7%), low-income household status (3.6% vs 1.0%), and residence in low-urbanization areas (8.9% vs 5.8%). These differences remained statistically significant after matching (P < 0.0001; Table 2). Although socioeconomic and urbanization differences persisted after sex- and age-matching, all AORs for primary outcomes were estimated using conditional logistic models that included monthly insurance premium, low-income household status, and urbanization level as covariates.

Table 2 Demographic characteristics after propensity score matching, n (%).

Variables	No IPV, n = 43393	IPV, n = 43393	P value1
Sex			1.000
Male	7637 (17.6)	7637 (17.6)
Female	35756 (82.4)	35756 (82.4)
Age in years			0.947
18-24	3070 (7.07)	3046 (7.02)
25-29	4138 (9.54)	4162 (9.59)
30-39	14334 (33.03)	14426 (33.24)
40-49	12859 (29.63)	12835 (29.58)
50-64	8992 (20.72)	8924 (20.57)
Monthly insurance premium			< 0.0001
Low (≤ 28860 New Taiwan dollar)	24582 (56.65)	31485 (72.56)
Medium (28861-43642 New Taiwan dollar)	8541 (19.68)	5770 (13.3)
High (> 43642 New Taiwan dollar)	10270 (23.67)	6138 (14.15)
Low-income household			< 0.0001
No	42964 (99.01)	41825 (96.39)
Yes	429 (0.99)	1568 (3.61)
Urbanization level			< 0.0001
High	26596 (64.61)	24471 (59)
Medium	12188 (29.61)	13315 (32.1)
Low	2379 (5.78)	3690 (8.9)
Missing2	2230 (5.14)	1917 (4.42)

¹Group differences were examined using the χ² test.

²Urbanization included a small “missing” category (this level was retained in adjusted models).

IPV: Intimate partner violence.

Association between IPV and health outcomes

The relationship between IPV exposure and subsequent health outcomes was analyzed using conditional logistic regression. For each mental and physical health outcome, both crude ORs and AORs were calculated, adjusting for sex, age, monthly insurance premium, low-income household status, and urbanization. In unadjusted models, IPV survivors showed markedly higher risks for several outcomes, namely depressive disorders (OR = 4.58, 95%CI: 4.15-5.06), alcohol and substance use disorders (OR = 7.34, 95%CI: 5.18-10.4), schizophrenia (OR = 2.12, 95%CI: 1.78-2.53), bipolar disorder (OR = 2.12, 95%CI: 1.78-2.53), and asthma (OR = 1.38, 95%CI: 1.15-1.65; Table 3).

Table 3 Association between intimate partner violence exposure and mental or physical health outcomes.

Outcome1	Prevalence		Crude OR (95%CI)	P value	Adjusted OR (95%CI)	P value
	No IPV	IPV
Psychiatric disorders
Depressive disorders	1.14	5.01	4.58 (4.15-5.06)	< 0.0001	4.18 (3.78-4.60)	< 0.0001
Alcohol and substance use disorders	0.08	0.61	7.34 (5.18-10.4)	< 0.0001	5.98 (2.21-8.50)	< 0.0001
Schizophrenia	0.42	0.89	2.12 (1.78-2.53)	< 0.0001	1.75 (1.46-2.10)	< 0.0001
Bipolar disorder	0.22	1.13	2.12 (1.78-2.53)	< 0.0001	4.81 (3.83-6.10)	< 0.0001
Physical diseases
Irritable bowel syndrome	0.17	0.09	1.01 (0.74-1.40)	0.935	1.01 (0.72-1.40)	0.943
Asthma	0.46	0.63	1.38 (1.15-1.65)	0.0006	1.31 (1.08-1.60)	0.0056

¹For all outcomes, the reference group (odds ratio = 1.00) consisted of controls without documented intimate partner violence exposure. Adjusted odds ratios were obtained from conditional logistic regression models adjusted for sex, age group, monthly insurance premium (low, medium, high), low-income household status, and urbanization level.

CI: Confidence interval; IPV: Intimate partner violence; OR: Odds ratio.

Upon adjustment for covariates, these associations largely persisted, though the effect sizes were somewhat attenuated for most outcomes, namely depressive disorders (AOR = 4.18, 95%CI: 3.78-4.60), alcohol and substance use disorders (AOR = 5.98, 95%CI: 2.21-8.50), schizophrenia (AOR = 1.75, 95%CI: 1.46-2.10), bipolar disorder (AOR = 4.81, 95%CI: 3.83-6.10), and asthma (AOR = 1.31, 95%CI: 1.08-1.60). In contrast, no significant association was observed for irritable bowel syndrome (AOR = 1.01, 95%CI: 0.72-1.40). The differences between crude ORs and AORs indicate that certain demographic and socioeconomic factors confound the relationship between IPV and health, but the elevated risks remain prominent even after adjustment (Table 3). Given that sex was matched, these findings apply comparably to both male and female survivors, underscoring the broad relevance of IPV-related health risks. A study flow diagram is provided in Supplementary Figure 1, summarizing participant eligibility, exclusions (e.g., prior family violence reports in 2018-2019, deaths in 2019-2020), the matching process, and the final analytic sample.

DISCUSSION

Key findings and interpretation

This nationally representative case control study provides robust evidence that IPV is strongly associated with increased risks of multiple adverse health outcomes, including depressive disorders, alcohol and substance use disorders, schizophrenia, bipolar disorder, and asthma, among adults in Taiwan. The associations remained significant after adjustment for key demographic and socioeconomic factors, consistent with previous international research linking IPV exposure to widespread psychological and physical morbidity[1-4,6-9,17].

Mechanisms and interpretation

Beyond statistical associations, several convergent pathways plausibly link IPV to the observed psychiatric and physical outcomes. Chronic exposure to threat and coercive control activates stress circuitry, including the hypothalamic-pituitary-adrenal axis and the sympathetic-adrenomedullary system, and alters immune signaling, promoting neuroinflammation and synaptic plasticity changes implicated in major depression, bipolar disorder, and psychosis-spectrum presentations. Concomitant sleep disruption, pain, and hazardous coping strategies (e.g., alcohol or drug use) further entrench risk and impede recovery[6,8,33]. Interpersonal trauma is also associated with heightened allostatic load and autonomic dysregulation, which can exacerbate respiratory inflammation and contribute to stressreactive phenotypes of asthma, especially under recurrent or multimodal exposures[8,20]. In Taiwan, universal insurance coverage coexists with persistent socioeconomic and urban-rural disparities in access, stigma, and service availability. These factors may modulate both care-seeking behavior and diagnostic capture after IPV, shaping the clinical phenotypes that enter routine claims systems.

Disease-specific findings and comparison with previous research

Our findings reveal that IPV survivors are at particularly high risk for a range of psychiatric disorders, with the AORs for depressive disorders (AOR = 4.18) exceeding effect sizes reported in some previous studies (e.g., 2.30-3.02 for women and 2.63-4.37 for men)[13-15]. This increased risk suggests that individuals identified in official IPV registries, who are more likely to experience more severe or ongoing violence, may bear even greater psychological burdens. Individual factors, such as high neuroticism and low extraversion, are known to increase vulnerability to depression under major stress, further amplifying the mental health impact of IPV exposure[12]. Moreover, the predominance of formally reported cases in our cohort, involving intervention by medical, police, or social services, may explain the higher risk observed compared to community- or self-report samples.

The risks of alcohol and substance use disorders among IPV survivors were also markedly elevated (AOR = 5.98), surpassing the 1.6-3.2-fold increases reported in Western cohorts[4,14,17,19,34]. Interviews with Taiwanese female survivors have described alcohol use as a coping mechanism to relieve distress and helplessness in persistent abusive environments[35]. However, while such use may provide temporary relief, it substantially increases the long-term risk of substance dependence and associated deterioration in mental and physical health, reinforcing a self-perpetuating negative feedback loop.

For severe psychiatric disorders, particularly schizophrenia and bipolar disorder, our study found that IPV is independently associated with sharply increased risks. Notably, the AOR for bipolar disorder (AOR = 4.81) was even higher than that in unadjusted models, suggesting a suppressor effect of demographic and socioeconomic risk factors that, when controlled for, clarify the direct impact of IPV. Kroon et al[36] reported a bimodal age distribution for bipolar disorder, with incidence peaks at 15-24 years and 45-54 years, and higher incidence in disadvantaged areas. Similarly, Rowland and Marwaha[37] demonstrated that social determinants such as financial hardship, instability, and inadequate support are closely linked to bipolar risk. Our adjustments for income and low-income status partially accounted for these underlying socioeconomic gradients. Importantly, most prior research has focused on the risk of violence victimization among individuals with existing severe mental illness[4,17,33,38], whereas our study, by analyzing the effect of IPV exposure on new psychiatric diagnoses, uniquely expands the current understanding of these causal pathways between violence and severe mental illness.

Physical health effects were also evident in our study. The moderate but significant association between IPV and asthma (AOR = 1.31) is consistent with prior estimates reported elsewhere and aligns with models implicating chronic stress, trauma, and neuro-immune activation in airway inflammation and disease[8,20]. Although modest in magnitude, this effect is clinically pertinent: Small relative increases can translate into substantial absolute burdens at the population level, given the high prevalence of both IPV and asthma. Even modest elevations may flag a subgroup with exacerbation-prone disease, poorer asthma control, and higher acute-care utilization, particularly when co-occurring with depression, anxiety, or substance use[8,10,11,20]. Notably, the risk of asthma increases with the number and types of violence experienced, e.g., men exposed to four distinct violence types may have a 3-fold elevated risk, and our estimate likely reflects conservative ascertainment within a 12-month window, as violence chronicity, multiplicity of abuse types, and environmental co-exposures (e.g., air pollution, smoking) may amplify risk beyond what administrative codes capture in a short follow-up[8,10,11].

In contrast, no significant association was observed between IPV and irritable bowel syndrome in our cohort (AOR = 1.01), a finding that warrants cautious interpretation. Claims-based ascertainment of irritable bowel syndrome is challenged by heterogeneity in symptom criteria, diagnostic substitution, and the need for exclusionary work-ups. Our stringent definition (inpatient or repeated outpatient diagnoses within 1 year) prioritized specificity but likely under-captured functional GI syndromes, compared with questionnaire-based or broader GI outcome studies that report links between stress and GI dysfunction[8,9,14,39]. Methodological factors, including a narrower ICD-10-CM code set and the short 12-month follow-up, alongside unmeasured factors such as psychological distress trajectories, diet, and healthcare-seeking patterns, may attenuate detectable associations. Taken together, the null finding is compatible with either a true weak short-term association or outcome misclassification. Future research should harmonize outcome definitions, incorporate validated symptom scales and GI-specific clinical data, and extend follow-up to examine dose-response relationships and latency.

Comparison with Western evidence and East-West context

Relative to Western cohorts, our estimates for depressive disorders and substance use outcomes are at the upper end of reported ranges, while findings for schizophrenia and asthma align with prior work[8,10,11,14,17,19,20]. This pattern may reflect the greater severity and persistence of cases captured through official registries, cultural differences in help-seeking and coding practices, and Taiwan’s near-universal coverage, increasing detection once patients engage with care. Conversely, the null association for irritable bowel syndrome differs from some questionnairebased reports in Western samples, consistent with methodological differences in outcome definitions and the conservative nature of claims-based phenotyping. Overall, the Taiwanese results are broadly in line with Western evidence for psychiatric sequelae and stress-reactive respiratory morbidity, while highlighting how administrative case definitions and follow-up horizons can influence observable effect sizes.

Strengths and methodological considerations

This study capitalizes on large-scale, linked national administrative and clinical databases to deliver an objective, population-level assessment of IPV-related health effects, using physician-assigned diagnoses within a system of comprehensive health coverage to minimize self-report bias and enhance generalizability. The inclusion of both men and women addressed a key evidence gap and generated findings directly relevant for sex/gender-inclusive policy and practice. Employing a matched case control design with rigorous adjustment for demographic and socioeconomic confounders, we demonstrated robust associations between IPV and multiple psychiatric disorders (depressive disorders, bipolar disorder, schizophrenia, and alcohol/substance use) and asthma, supporting the external validity of estimates across diverse segments of the Taiwanese adult population and aligning with international evidence.

To minimize overmatching and preserve representativeness in this registry-based design, socioeconomic status was not included in the matching process; instead, post-match imbalances were addressed via multivariable adjustment. This strategy is common in population-scale studies, although residual confounding from unmeasured factors related to socioeconomic status (e.g., education, occupation, household crowding) may persist and should be considered when interpreting the results.

Policy and practice

Building on these findings, translating evidence into practice requires coordinated responses across clinical, public health, and social care systems. Policy and practice should translate multisector collaboration into concrete, measurable steps: First, implement routine IPV informed screening in primary care, emergency, and obstetric/gynecologic settings using validated tools [e.g., HITS (Hurt, Insult, Threaten, Scream scale)/HARK (Humiliation, Afraid, Rape, Kick scale) for IPV; Patient Health Questionnaire-9, Generalized Anxiety Disorder-7, Alcohol Use Disorders Identification Test-Consumption/Drug Abuse Screening Test-10 for common comorbidities], and ensure that positive screens receive tiered, same day warm handoffs to behavioral health and social services. Screening protocols should explicitly include male survivors and provide low threshold, same day referral options. Second, establish standardized, closed loop interagency referral pathways linking healthcare, social welfare, and justice sectors. A single electronic referral embedded in clinical workflows should document safety planning; connect patients to shelter, legal advocacy, and financial supports; and schedule follow up within 7-14 days with confirmation of receipt. Third, embed brief, evidence based interventions within IPV care pathways to address high risk comorbidities, such as motivational interviewing for hazardous drinking and contingency management or facilitated referral to medication assisted treatment for substance use. Care pathways should be sex informed, prioritizing substance use comorbidity in men and mood disorders in women. Fourth, incorporate trauma informed elements into asthma management, including screening for recent IPV, stress, and depression; optimizing controller therapy; and addressing tobacco exposure, given the observed stress reactive respiratory morbidity. Finally, monitor implementation with pragmatic indicators, such as screening coverage, referral completion, and 30-90 days follow up engagement, to evaluate fidelity and clinical outcomes and to guide continuous improvement.

These sex inclusive, registry based findings are directly actionable within a universal coverage setting, supporting standardized IPV informed screening and closed loop referral pathways that explicitly include male survivors. Future research in Taiwan should prioritize depth over breadth by extending follow ups to delineate long term trajectories, enriching exposure assessment (type, severity, and chronicity), and evaluating the effectiveness and scalability of system level strategies within the NHI framework. Such implementation focused work is well positioned to reduce the short term psychiatric and respiratory burdens identified in this study while building a stronger evidence base for sustainable, sex/gender inclusive care.

Study limitations and future directions

Several issues constrain inference. First, reliance on officially registered IPV enhances procedural validity but introduces selection and measurement limitations; registry capture depends on disclosure and administrative processing, which may overrepresent acute or situational episodes, and co-occurs with social and clinical adversities (e.g., poverty, prior trauma). Thus, observed associations may partly reflect correlated risk environments rather than the isolated effect of IPV, and residual confounding is, therefore, likely despite matching and adjustment[6,33]. Second, the database lacked granularity on violence severity, type (physical, sexual, psychological, economic), frequency or chronicity, coercive control, victim appraisal, and perpetrator intent, dimensions known to shape dose-response and heterogeneity of health sequelae, potentially biasing estimates toward the null via nondifferential misclassification or obscuring highrisk subgroups[3,8]. Third, although sex matching enabled sex-inclusive interpretation, the male subgroup was small, yielding imprecise estimates. Consequently, sex-stratified analyses were not performed, which is a key study limitation. Future work should increase male representation (e.g., targeted recruitment or linkage) and prespecify sex-stratified models with small-sample-robust methods to enable reliable sex-specific inference[19,20]. Fourth, the 12-month follow-up emphasized short-term, stress-responsive outcomes and may have missed delayed or chronic sequelae (e.g., recurrent mood disorders, cardiometabolic effects of allostatic load). Extended longitudinal follow-up with repeated exposure assessment and time-varying confounder control is warranted[6,14]. Finally, claims-based ascertainment is contingent on care access and coding practices, and lacks clinical detail on baseline severity and treatment adherence, risking under-ascertainment and residual confounding. Linking claims to clinical records, standardized symptom measures, and victim-services data, alongside richer social determinant variables, would strengthen causal interpretation and external validity[3,8,33]. Future research should expand on our findings by utilizing longer follow-up periods to assess chronic and delayed-onset health outcomes, adopting more nuanced measures of violence type and severity, and incorporating baseline health status to clarify causal pathways. To minimize multiplicity in this initial report, we did not perform sensitivity analyses nor formal interaction tests; future studies should evaluate robustness to alternative specifications and conduct prespecified sex-stratified and interaction analyses.

CONCLUSION

In this nationwide, registry-based case control study, documented IPV was associated with markedly increased 12-month risks of major psychiatric disorders (depressive disorders, bipolar disorder, schizophrenia, and alcohol/substance use disorders) and a modest but significant increase in asthma among Taiwanese adults, independent of demographic and socioeconomic factors. These population-scale, clinician-assigned findings extend predominantly Western evidence to an East-Asian context and underscore that the mental and physical sequelae of IPV transcend sex/gender boundaries[3,4,6,8,17]. Importantly, this study delineates nearterm, clinically actionable targets, depression, bipolar disorder, psychosis, substance use, and asthma, where health systems can intervene promptly within the first year after exposure.

By leveraging linked registry data and physician-assigned diagnoses within a universal health coverage system, this study provides large-scale empirical evidence from East Asia and confirms that IPV is a significant driver of both psychiatric and stress-related physical morbidity. The findings highlight the importance of improving IPV identification and timely health service engagement to mitigate downstream disease burden. Future research should extend follow-up periods to examine chronic and delayed outcomes, incorporate more precise exposure measures such as violence type, severity, and chronicity, and integrate data on pre-existing vulnerabilities to clarify causal mechanisms. Additionally, evaluating the impact of IPV-informed care pathways and coordinated response models within the NHI system will be essential to strengthen the evidence base for sustainable, gender-inclusive intervention strategies in Taiwan and comparable settings.

ACKNOWLEDGEMENTS

The authors thank the Health and Welfare Data Science Center, Ministry of Health and Welfare, Taiwan, for providing access to the data.