Intimate partner violence, depression, and chronic low‐grade inflammation among middle‐aged women in Cebu, Philippines

Recent discussions in human biology have highlighted how local ecological contexts shape the relationship between social stressors and health across populations. Chronic low‐grade inflammation has been proposed as a pathway linking social stressors to health, with evidence concentrated in high‐income Western contexts. However, it remains unclear whether this is an important pathway in populations where prevalence is lower due to lower adiposity and greater infectious exposures. To investigate this further, we tested associations between multiple types of intimate partner violence (IPV), a highly prevalent stressor and health crisis globally, and C‐reactive protein (CRP), a commonly used measure of chronic low‐grade inflammation, in Cebu, Philippines. For reference, we compared results for CRP to depression, a well‐established and consistently observed health outcome of IPV.


| INTRODUCTION
Recent discussions in human biology have highlighted the importance of considering how local ecological contexts shape the effects of chronic stressors on mental and physical health, leading to findings that are population and context specific (DeCaro & Helfrecht, 2022).Chronic low-grade inflammation has been proposed as one potential pathway underlying the health effects of chronic stressors, as it has been implicated in various risks such as cardiovascular disease (CVD) (Chiang et al., 2022;Liu et al., 2017).However, prior studies have come primarily from US samples, where chronic low-grade inflammation is more prevalent due to a combination of factors including greater adiposity and reduced microbial exposures in early life (McDade et al., 2009;McDade et al., 2010;McDade et al., 2021).It is therefore unclear how generalizable these findings might be to other populations with lower prevalence.For example, in the Cebu Longitudinal Health and Nutrition Survey (CLHNS) birth cohort in the Philippines, indicators of greater microbial exposures in infancy have predicted lower C-reactive protein (CRP), a commonly used measure of chronic low-grade inflammation, and attenuated associations between CRP and parental absence in childhood and current perceived stress in young adulthood (McDade et al., 2010;McDade, Hoke, et al., 2013).Further, a recent meta-analysis of early adversity and chronic low-grade inflammation found that the association was attenuated in samples with lower average adiposity, reflected in the body mass index (BMI) (Chiang et al., 2022), suggesting that contextual factors influencing adiposity in a population might also influence the relationship between social stressors and low-grade inflammation.
Intimate partner violence (IPV) is a highly prevalent stressor globally and has been associated with numerous measures of health risk, including substance use, risk of contracting sexually transmitted infections (El-Bassel et al., 2022;Gibbs et al., 2022), pre-term birth and spontaneous abortion (Desta et al., 2021;Fanslow et al., 2008;Sanchez et al., 2013;Stöckl et al., 2012), depression and post-traumatic stress disorder (Goldberg, 2022), and metabolic risk profiles (Goldberg et al., 2021;Pheiffer et al., 2020;Stubbs & Szoeke, 2022).US-based studies have reported associations between IPV and measures of inflammation including gene expression and plasma CRP (Fernandez-Botran et al., 2011;Li et al., 2022;Newton et al., 2011); however, it is unclear how generalizable this might be across populations given prior work showing links between social stressors and low-grade inflammation are dependent on contextual factors.The closest comparable study from a nonUS sample is a Kenyanbased study of gender-based violence among female sex workers, which was associated with higher hair cortisol, indicating a physiological stress response, but unassociated with CRP or its upstream regulator interleukin (IL)-6 (Heller et al., 2018).
The objectives of this study are to further clarify (1) the health burden of IPV in Cebu and (2) the relationship between social stressors and chronic low-grade inflammation more generally by testing associations between multiple types of IPV (physical, emotional, and controlling behavior) and plasma CRP in a population with infectious disease and energetic contexts distinct from many populations studied in similar past work (Adair, Popkin, et al., 2011).IPV is a highly prevalent social stressor in this population, as the estimated prevalence is 44% for the Central Visayas region where Cebu is located, compared to a recent global estimate of 27% (National Statistics Office & ICF Macro, 2009;Sardinha et al., 2022).As a result, it has been the focus of several studies, including risk factors for exposure (Ansara & Hindin, 2009;Hindin & Adair, 2002;Tsai, 2017) and child consequences, including cognitive development, intergenerational transmission of IPV, and depressive symptoms (Hindin & Gultiano, 2006;Julio et al., 2023;Kim et al., 2020;Mandal & Hindin, 2015).However, prior studies have shown chronic low-grade inflammation in this population is lower compared with US data, due to lower adiposity and greater infectious exposures in early life, suggesting it might not be associated with social stressors like IPV (Chiang et al., 2022;McDade et al., 2008;McDade et al., 2009;McDade et al., 2010;McDade, Hoke, et al., 2013).Given that we hypothesized null results, which could either be due to no relationship or limitations in the IPV measures, we compared results for CRP to depression, a well-established and consistently observed health outcome (White et al., 2024).

| Participants
Data came from the CLHNS, which is an ongoing birth cohort that began in 1983 with the recruitment of 3327 pregnant women across randomly selected rural and urban neighborhoods throughout Metropolitan Cebu.The current study focuses on a follow up survey in 2005, which recruited 2018 of the original 3327 women at ages 35-69 years.Compared with the rest of the sample at baseline, the analysis sample was lower in socioeconomic status (SES) and came from more rural areas (McDade et al., 2008).From this survey, 1601 were currently partnered and had complete data for all variables in the analysis.All survey measures and blood samples were collected concurrently and under conditions of informed consent with institutional review board approval from the University of North Carolina, Chapel Hill.Survey data are publicly available at https://dataverse.unc.edu/dataverse/cebu.

| Intimate partner violence
Following the WHO and prior research on IPV in Cebu, we considered three categories of IPV: physical, emotional, and controlling behavior (Julio et al., 2023;WHO, 2013).Physical IPV was measured using a modified version of the Conflict Tactics Scale (Straus, 1990) and was indicated by whether participants answered yes to any of the following questions referring to their current partner: ever had something thrown at them, was ever pushed, grabbed, or shoved, ever hit (not with anything), ever hit with something hard, or ever harmed enough to need medical attention.The emotional IPV question asked if they were ever yelled at or insulted by their current partner (yes/no).The controlling behavior IPV question asked whether their current partner tells them who they can spend time with (yes/no).

| C-reactive protein
CRP (in milligram per liters) measurement in this sample has been described previously (McDade et al., 2008).Briefly, venipuncture blood samples were collected using EDTA-coated vacutainer tubes in the participants' homes in the morning after an overnight fast, stored in coolers on ice packs for no more than 2 h during transport, and centrifuged to separate plasma at the University of San Carlos.Plasma samples were express shipped to Northwestern University and stored at À80 C until measurement using a high sensitivity immunoturbidimetric method (Synchron LX20, lower detection limit: 0.1 mg/L).

| Depression score
A modified depressive symptoms scale was previously developed and used in the CLHNS based on the Center for Epidemiologic Studies-Depression Scale (McDade, Borja, et al., 2013).The scale included 15 questions asking participants how often during the previous 4 weeks they had difficulty sleeping or eating, felt lonely, worried, people were unfriendly or disliked them, had feelings of worthlessness or suicidal thoughts, felt happy, hopeful about the future, or enjoyed daily life.Possible responses included "none of the time", "occasionally", and "most of the time", which were coded as 1, 2, and 3, respectively.

| Analysis
We used multiple ordinary least squares regression models to test associations between IPV and CRP and the depression score.CRP was positive offset (+0.1) and logtransformed due to positive skew.We considered CRP as both a continuous measure (Model set 1) and a dichotomous high-risk indicator of chronic low-grade inflammation (>3 mg/L and < = 10 mg/L) (Model set 2), whereas we also considered models that excluded participants who reported symptoms of an active infection (e.g., runny nose, cough, fever, diarrhea, sore throat) or with CRP > 10 mg/mL, another possible indicator of an acute response (Model set 3).We adjusted for several potential confounders based on prior studies (Ansara & Hindin, 2009;Hindin & Adair, 2002;McDade et al., 2008), including waist circumference (in centimeters), age (in years), current smoker, alcohol consumption, household size, symptoms of acute infection a previously constructed measure of urbanicity that includes population size and density, telephone and internet availability, infrastructure for transportation, education, health services, and markets for consumer goods (Dahly & Adair, 2007), SES, committing physical or emotional IPV (yes/no), and recalled witnessing parental IPV growing up (yes/no).Witnessing IPV was asked during an earlier 2002 survey.
SES was measured as a combination of annual household income (in pesos), education (highest grade completed), and a scale reflecting owned assets.Based on previous studies measuring asset-based wealth in the CLHNS (Desantis et al., 2015;Varghese et al., 2021), the asset measure was the sum of owned items relevant to social class, including electricity, televisions, refrigerators, air conditioners, tape recorders, electric fans, jeepneys, cars, and type of home construction (light, mixed, permanent structure).Possible values ranged 0-11.Household income was log-transformed due to positive skew, and all three SES measures were standardized (mean = 0, SD = 1) and summed into a single score.Cronbach's α for the SES measure was 0.70.Analyses were performed in R (version 4.2.0), whereas the "jtools" (Long, 2022) package was used to plot results.

| RESULTS
Descriptive statistics are shown in Table 1.Among the three types, emotional IPV was most prevalent (47%), followed by controlling behavior (26%), and physical IPV (20%).Committing physical or emotional IPV was also highly prevalent (66%), whereas 32% reported witnessing parental IPV growing up.In addition, 17% reported experiencing infectious symptoms, whereas 16% were current smokers and 39% reported consuming alcohol.The mean depression score was 22.2, with a standard deviation of 3.8.
Descriptive statistics across IPV exposure groups are shown in Tables S1-S3.Those who reported physical IPV exposure were more likely to also report emotional IPV, controlling behavior, committing IPV themselves, and witnessing parental IPV growing up.They also had lower waist circumference, were more likely to smoke and consume alcohol, and lived in more urban and lower SES households.Those reporting emotional IPV were also more likely to report committing IPV themselves, whereas they also had lower waist circumference, were more likely to consume alcohol, and live in a lower SES household.Finally, those reporting controlling behavior were also more likely to be from lower households.Regression results predicting the depression score are shown in Table 2.All three IPV measures were associated with a higher depression score before and after adding other covariates.Further, committing IPV was also associated with a higher depression score.Regression results predicting continuous and dichotomous CRP are shown in Tables 3 and 4, respectively, which showed no associations with any of the IPV measures.IPV was similarly unassociated with CRP in models excluding those reporting symptoms of an active infection or those with CRP > 10 mg/L (Table 5).Results for the depression score and continuous CRP (Model set 1) are shown together in Figure 1 for comparison.As additional sensitivity analyses, we considered the effect of experiencing multiple types of IPV (reporting any two or all three exposure types), which was also unassociated with CRP (Supplementary Table S4).We also broke up the physical IPV exposure measure into its five individual questions that have different degrees of severity, which were all unassociated with CRP (Supplementary Table S5).There were also survey questions asking whether physical or emotional IPV occurred within the last year (as opposed to anytime) and how frequently (never, a few times, once a month, more than once a month), which were all similarly unassociated with CRP (Supplementary Table S6-S8).Finally, we considered the potential interaction between IPV and waist circumference, which indicated a positive association between emotional IPV and CRP only among those with above-mean waist circumference (Supplementary Table S9).This result was also observed using the BMI (Supplementary Table S10).

| DISCUSSION
IPV is a highly prevalent stressor across populations, and understanding its effects on biological variation and health across ecological and cultural contexts is   F I G U R E 1 Comparing standardized effect sizes between the depression score and continuous C-reactive protein (CRP) across intimate partner violence types.Models are covariate adjusted as specified in Tables 2 and 3.
important to both human biology and global health.
Based on prior studies, including of inflammation in this population, we hypothesized that while IPV would be associated with a higher depression score, associations with CRP would be more complex and potentially null, which was supported by the results.All three types of IPV (physical, emotional, and controlling behavior) were associated with a higher depression score, consistent with a recent meta-analysis (White et al., 2024) and suggesting it is a salient stressor that impacts mental health in this population.However, we find that none of the IPV measures were associated with CRP in the full sample.
Although directly comparable studies are limited, these results are consistent with a Kenyan-based study of gender-based violence against female sex workers, which was associated with greater hair cortisol, indicating a physiological stress response, but unassociated with CRP or IL-6 (Heller et al., 2018).In addition, the positive association between emotional IPV and CRP among those with waist circumferences or BMI above the mean is also consistent with a recent meta-analysis indicating adiposity is a key moderator of the association between social stressors and chronic low-grade inflammation (Chiang et al., 2022).Finally, as CRP is a commonly used biomarker of CVD risk, our results are also consistent with two recent reviews finding mixed evidence for associations between IPV and CVD (Liu et al., 2020;Stubbs & Szoeke, 2022).
Although chronic low-grade inflammation has received considerable attention as a potential pathway linking stress to health, our results suggest this might be a less relevant pathway in populations with lower prevalence.Our results extend on prior work in the CLHNS questioning whether hypothesized relationships among chronic stressors, low-grade inflammation, and depression replicate across populations, as depressive symptoms are also unassociated with CRP in this sample or with IL-6 and CRP in the CLHNS birth cohort members (McDade, Borja, et al., 2013).However, this population is also undergoing rapid lifestyle changes, including adoption of obesogenic diets and sedentary occupations, which is contributing to greater adiposity and might eventually lead to the emergence of associations between chronic stressors and low-grade inflammation (Adair et al., 2018;Adair, Gultiano, & Suchindran, 2011).The interaction tests for waist circumference and BMI provided an early test and support for this possibility.Relatedly, experimental studies among nonhuman primates found that a western dietary intervention exacerbated the physiological stress response as well as primed monocytes toward a pro-inflammatory state (Johnson et al., 2021;Shively et al., 2020).These findings collectively highlight the importance of an evolutionary medicine perspective to the study of social stressors and health across populations, as others have highlighted, in which the relationship is exacerbated by novel western lifestyles and environments (Jaeggi et al., 2021;Sapolsky, 2004).
Our results are also relevant to other measures that rely on CRP in their construction.For example, CRP is commonly used in the construction of allostatic load indices (DeCaro & Helfrecht, 2022).It has also been used in the construction of multiple epigenetic aging clocks, including PhenoAge and DunedinPACE (Belsky et al., 2020;Belsky et al., 2022;Levine et al., 2018).Thus, future studies utilizing these measures to understand the effects of social stressors on health should consider the moderating influence of a population's lifestyle and environmental exposures, which might explain potentially null or complex findings.
Although we considered CRP, further studies are needed for other components of immune function, such as adaptive immune function and aging ("immunosenescence"). Multiple studies among HIV-positive women in South Africa have reported associations between IPV and reduced viral suppression (Gibbs et al., 2022;Hatcher et al., 2021).In addition, IPV has been associated with lower CD4T cell counts among HIV-positive women in a US-based sample (Anderson et al., 2018) as well as greater longitudinal declines in CD4T cell counts among HIV-positive women in South Africa (Jewkes et al., 2015).IPV has also been found to be associated with accelerated immune aging in a US-based study, reflected in lower naïve and higher effector memory CD4T cells (Kalokhe et al., 2016).However, more studies are needed across populations and in samples that are not already immunocompromised to further understand how IPV might affect adaptive immune function and aging.
Although our study has multiple strengths, including the large sample size, multiple types of IPV measured, and CRP measurement with a high-sensitivity method, it is not without limitations.We did not have measures of other commonly used indicators of chronic low-grade inflammation, including IL-6, IL-1β, and tumor necrosis factor (TNF)-α.However, these measures tend to be correlated, because CRP is produced in response to IL-6 and IL-1β, whereas reciprocally IL-6, IL-1β, and TNF-α are produced in response to CRP (Black et al., 2004;Du Clos & Mold, 2004).There is also a potential concern regarding acute versus chronic stress, as the physical and emotional IPV measures referred to whether these occurrences ever happened in the past (single event), whereas the controlling behavior IPV measure referred to current ongoing controlling behavior.To address this concern, we assessed measures of physical or emotional IPV occurring within the last year, which were also included in the survey.Among those reporting any physical IPV exposure, a majority (58%) reported exposure within the last year, whereas 89% of those reporting any emotional IPV in the past also reported exposure within the last year, suggesting these tended to be consistent exposures.In addition, results were similarly null between physical or emotional IPV exposures in the last year and CRP (Supplementary Tables S6-S8).Finally, while the Cronbach's α indicated our depression score measure had acceptable reliability; further studies are needed to measure validity in this cultural context.

| CONCLUSION
This study contributes to a broader understanding of both the global health toll of IPV as well as how ecological contexts shape the relationship between social stressors and biological variation across populations.We find that multiple types of IPV are similarly associated with a higher depression score.However, IPV was not associated with chronic low-grade inflammation, reflected in CRP, in the full sample.Instead, an association between emotional IPV and CRP emerged only among those with above-mean waist circumference and BMI, suggesting this relationship is more complex and depends on lifestyle and environmental factors.Our results highlight the need for follow-up studies as this population and others undergo increasing lifestyle changes and reduced infectious exposures to test whether associations between social stressors and chronic low-grade inflammation begin to emerge as seen in US-based studies.
Model also adjusted for age, SES, urbanicity, waist circumference, smoking, alcohol consumption, household size, and current infection symptoms.Model set 3: Predicting CRP continuous a (excluding those with current infection symptoms or CRP > 10 mg/L, n = 1282).