Psychological Distress Among Lebanese Women During the COVID-19 Connement: Any Mediating Effect of Violence?

Background: Violence among women is associated with poor psychological health and increased internalizing symptoms, such as depression, stress, and anxiety. The primary objective was to evaluate the association of marital conict, pregnancy status, socioeconomic status, and stressful life events with violence, depression, anxiety, and stress. The secondary objective was to evaluate the mediating effect of violence on psychological distress. Methods: A cross-sectional study was conducted among 369 women between June 8 and August 1, 2020. The questionnaire was developed on Google Forms and distributed online, using the snowball technique. Results: Higher marital conict (Beta=0.159), nancial diculties (Beta=2.572), a history of child abuse (Beta=2.546), and higher violence scores (Beta=0.121) were signicantly associated with higher depression scores. Higher marital conict (Beta=0.358) and having nancial diculties (Beta=3.859) were signicantly associated with higher anxiety scores. Higher marital conict (Beta=0.070) and nancial diculties (Beta=1.483) were signicantly associated with higher perceived stress scores. Composite abuse scale partially mediated the association between pregnancy status (25.28%), nancial diculties (7.71%), history of child abuse (21.92%), abuse by a family member (50.50%), and anxiety scale. Also, the composite abuse scale partially mediated the association between marital conict (21.42%), nancial diculties (14.17%), history of child abuse (20.05%), abuse by a family member (33.57%), and depression scale. Conclusion: Our main ndings suggest that violence mediated the association between pregnancy status, marital conicts, nancial diculties, and depression and anxiety but did not mediate between these factors and stress. Further studies are needed to understand better the factors that mediate the relations between domestic violence and mental well-being among Lebanese women.


Introduction
Stress, another mental health disorder, is a highly prevalent condition in women, particularly during pregnancy (24). Pregnant women are highly prone to stress due to the feeling of imbalance when they cannot cope with demands and worries (25). Other in uential factors of increased risk of stress among women include low socioeconomic status, exposure to violence, stressful life events, lack of social support, and con icts (26)(27)(28).
Previous literature supports the fact that violence among women is related to several factors, including low socioeconomic level, previous violence in the family, pregnancy status, and social isolation (29). Furthermore, women exposed to violence had poor psychological health and increased internalizing symptoms, such as depression, stress, and anxiety (30,31). Previous studies reported various mediating factors for violence. However, only a few examined the mediating effect of violence on psychological distress in women (32,33). Our research uses a conceptual model based on a study done by Ceballo et al. (Fig. 1) (29).
While these problems are reported in the literature and have different health effects, no studies have yet evaluated the magnitude of depression, anxiety, and stress among Lebanese women. Therefore, our primary objective was to assess the association of marital con ict, pregnancy status, socioeconomic status, and stressful life events with violence, depression, anxiety, and stress. Our secondary objective was to evaluate the mediating effect of violence on psychological distress.

Study design and sampling
A cross-sectional online study conducted between June 8 and August 1, 2020, enrolled a total of 369 women. The questionnaire used was developed on Google Forms and distributed on social media and WhatsApp groups, using the snowball technique; it required 40 minutes to complete. Eligibility criteria were the following: married women aged between 18 and 51, currently living with their partner, and with internet access. Single, widowed, or divorced women, and those with a fertility problem, were excluded. The inclusion criteria were stated in the consent form at the beginning of the survey. Participation in this study was anonymous and voluntary, and participants received no compensation in return.

Sample size calculation
The Epi info software (Centers for Disease Control and Prevention, Epi Info™) calculated a minimum sample of 233 participants, taking into account a Lebanese female population of 2,294,260 (34), a prevalence of 10.9% women with depression, according to a previous study (35), a con dence level of 95%, and adding a 4% margin of error. A sample of 500 women was targeted to allow for missing values. The nal sample size consisted of 369 participants.
Translation procedure A forward and backward translation was conducted for all the items of the questionnaire. One translator did the translation from English into Arabic, and a second one performed the back translation. Discrepancies between the two English versions were resolved by consensus.

Questionnaire
The online questionnaire consisted of two sections, including closed-ended questions available in English and Arabic. The questionnaire was developed for this study and it is provided as Additional File 1. The rst one assessed the sociodemographic details of participants (age, educational level, the region of residence, religion, working status, monthly income, smoking and alcohol status, and physical activity). The monthly income was divided into four levels: no income, low < 1,000 USD, intermediate 1,000-2,000 USD, and high income > 2,000 USD.
The second section consisted of the following measures:

Household crowding index
The household crowding index was calculated by dividing the number of persons living in the household by the number of rooms, excluding bathrooms and kitchen, to assess socioeconomic status (36).

Stressful life events
Negative life events were assessed using three dichotomous questions (Yes/No) about nancial di culties, history of child abuse, and history of family member abuse.

Marital con ict
The Relationship Dynamics Scale was utilized to measure the frequency of con icts in relationships (37). The scale consists of eight items scored on a 5-point Likert scale, (1 = Never or almost never to 5 = Always) (37). A higher score indicates a higher con ict between partners.

Pregnancy status
The current pregnancy status was assessed using binary questions (Yes/No). The current pregnancy status re ects the ability of women to choose to reproduce or be coerced into an unwanted pregnancy.

Partner violence
Violence was assessed using the Composite Abuse Scale (Revised) -Short Form (CASR-SF). This 15-item scale evaluates the existence, extent, and severity of physical, sexual, or psychological abuse (38). The total score is calculated by summing the 15 responses. Items are graded on a Likert scale from 1 to 6, where a higher score indicates a higher intensity/occurrence of abuse. The author of the questionnaire, Professor Marilyn Ford-Gilboe, granted permission to use the scale.

Psychological distress
Stress, anxiety, and depression were measured using three scales, re ecting psychological distress among women. The Perceived Stress Scale-4 (PSS-4) evaluated the occurrence and intensity of stress. It consists of four questions graded on a Likert scale from 1 to 5. The total score was calculated by summing the four items, ranging from 0 to 16, with higher scores indicating a higher intensity/occurrence of perceived stress (39). The Lebanese anxiety scale (LAS; developed and validated in Lebanon) is a 10-item tool graded on a Likert scale from 1 to 5. It assessed the occurrence and intensity of anxiety. The total score was obtained by summing all the responses, with higher scores indicating higher anxiety (40). The PHQ-9, validated in Lebanon (41), assesses the level of depression. It consists of 9 items graded on a Likert scale from 1 to 4. The PHQ-9 score can range from 0 to 27, with higher scores indicating a more severe depression (42).
All the scales used in this study do not require any license for their use except the CASR-SF where a permission from the author was granted.

Statistical analysis
Data were analyzed on SPSS software version 25 (SPSS Inc., Chicago, IL, USA). A descriptive analysis was performed, using the absolute frequency and percentages for categorical variables and mean and standard deviation for quantitative measures. Student t-test and ANOVA F tests were used to assess the association of continuous variables with two or more levels, and Pearson correlation was used for linear correlation between continuous variables. In the multivariable analysis, four linear regressions were performed, taking stress, anxiety, depression, and violence as the dependent variables, meeting all the assumptions for normality and homoscedasticity. Regressions took into account the variables with a value of P < 0.2 in the bivariate analysis to minimize residual confounding.
The PROCESS SPSS Macro version 3.4 model four was used to evaluate the mediation of violence using the CASR-SF through three steps.
Step 1 determined the regression coe cient for the effect of marital con ict, household crowding index, pregnancy status, nancial di culties, history of child abuse, and abuse by a family member.
Step 2 examined the association between violence and psychological distress (stress, anxiety, and depression), and Step 3 estimated the direct effect of each of the marital con ict, household crowding index, pregnancy status, nancial di culties, history of child abuse, and abuse by a family member on psychological distress. Each independent variable was entered in a separate model since the variables were not highly correlated. The macro generated bias-corrected bootstrapped 95% con dence intervals (CI) to test the signi cance of the indirect effect (43,44). Mediation was considered signi cant when the CI around the indirect effect did not include zero (44). The mediation effect was calculated by dividing the indirect effect of X by the direct effect of Y. The covariates included in the mediation model were those that showed signi cant associations with psychological distress in the bivariate analysis.
Statistical signi cance was set at a value of P < 0.05.

Sample description
The mean age of women was 32.5 ± 6.4 years. The majority (83.9%) had a university level of education, 59.9% were employed, 23.3% had no income, and 42.5% practiced physical activities. Only 31.2% of them were smokers, and 10.8% consumed alcohol. The mean duration of con nement was 71.0 ± 42.8 days, and the mean fear of poverty was 5.8 ± 3.2.
Bivariate analysis: correlates of psychological scales A signi cantly higher mean violence score was found in non-pregnant as compared to pregnant women (M non−pregnant women =2. .08, P = 0.003). Moreover, higher marital con ict (r = 0.338), higher stress (r = 0.503), depression (r = 0.785), and violence (r = 0.141) were signi cantly associated with higher anxiety.

Multivariable analysis
The rst linear regression taking the violence score as the dependent variable showed that higher marital con icts (Beta = 0.485), abuse by a family member (Beta = 2.988), and a higher number of pregnancies (Beta = 1.470) were signi cantly associated with a higher violence score. Whereas, higher household crowding index (Beta=-1.551) was signi cantly associated with a lower violence score ( Table 2, Model 1).
A second linear regression taking the depression score as the dependent variable showed that higher marital con icts (Beta = 0.159), nancial di culties (Beta = 2.572), a history of child abuse (Beta = 2.546), and higher violence scores (Beta = 0.121) were signi cantly associated with a higher depression score ( Table 2, Model 2).
A third linear regression taking the anxiety score as the dependent variable showed that higher marital con icts (Beta = 0.358) and nancial di culties (Beta = 3.859) were signi cantly associated with higher anxiety scores ( A fourth linear regression taking the stress score as the dependent variable showed that higher marital con icts (Beta = 0.070) and nancial di culties (Beta = 1.483) were signi cantly associated with higher perceived stress scores (   Table 3 presents the mediation analysis conducted on perceived stress, anxiety, and depression. The CASR-SF partially mediated the association between pregnancy status (25.28%), nancial di culties (7.71%), history of child abuse (21.92%), abuse by a family member (50.50%), and anxiety. Also, the CASR-SF partially mediated the association between marital con icts (21.42%), nancial di culties (14.17%), history of child abuse (20.05%), abuse by a family member (33.57%), and depression. No mediation effect of the CASR-SF was found between the independent variables and perceived stress (Table 3). Step 1: Taking the anxiety scale as the dependent variable and the composite abuse scale as a mediation factor Step Step 1: Taking the anxiety scale as the dependent variable and the composite abuse scale as a mediation factor Beta t p Beta t P Beta t p Step

Discussion
To our knowledge, this study is the rst to assess the factors related to psychological distress among Lebanese women during the COVID-19 con nement. The results showed that marital con icts and nancial di culties were associated with high stress, anxiety, and depression, while higher violence was related to higher depression among women. Also, violence was a mediating factor between pregnancy status and anxiety and between marital con ict and depression.
In this study, marital con icts were related to higher depression, anxiety, and stress among women. Consistently, previous ndings have also associated high levels of psychological distress with marital dissatisfaction (45)(46)(47)(48). A recent study conducted in China during the COVID-19 pandemic showed that emotional well-being in married people was lower than in unmarried people (49). Evidence suggests that stressful or life-threatening events, such as natural disasters, can lead to a decline in the quality of relationships due to stress and con ict or, instead, improve the quality of relationships and intimacy (50)(51)(52). Also, during the COVID-19 con nement, several factors can lead to increased tension between romantic partners con ned together, such as disturbed everyday activities, the fear of the pandemic, loss of physical activity, loss of outside social networks, lack of access to non-essential health services, and reduced physical interaction in overcrowded households (53). Previous studies have shown that the persistence of tense verbal or non-verbal exchanges between partners can lead to physiological and psychological changes, eventually progressing to disease states (54,55). Once health conditions develop or intensify, they can be experienced as a traumatic life occurrence or persistent stress, resulting in adverse mental health consequences, such as depression and anxiety (56). Our results also showed that marital con icts were associated with higher violence, consistent with those of previous studies (57)(58)(59). The severity of marital con icts ranges from minor disagreements to disputes that partners cannot overcome, which harms their relationship and could lead to violence (60).
This study showed that violence was associated with higher depression among women, in agreement with previous studies showing that women subject to behavioral, physical, and sexual violence by an intimate partner were more likely to exhibit signs of depression (61)(62)(63). The core mechanisms that explain why violence can cause subsequent depression in women are traumatic and psychological stress reactions (64). Sustained psychological stress due to social threats or rejection can affect women's cognition, mood, and behavior, leading to symptoms of depression (65). Also, domestic environments have become a suitable place for intimate partner abuse because of lockdown restrictions (66). Abusive individuals may use the con nement to exert power over the vulnerable ones, especially women (66). Lockdown restrictions can contribute to increased stress and further lead to exacerbation of anxiety, depressive symptoms, and violence (67). Thus, the perpetrator disturbed psychological health could adversely affect that of abused individuals (67). The persistent exposure of vulnerable persons to abusive individuals constrains the victim to cope with the situation, which further aggravates depressive symptoms (68).
Our ndings revealed that nancial di culties were associated with higher depression, stress, and anxiety, consistent with those of a recent study showing a correlation between nancial hardship and emotional impact due to the pandemic (69). Other studies have also reported that nancial di culties due to the lockdown might impair mental health and increase the incidence of psychological stress (70)(71)(72). During times of crisis and con nement, women face plenty of problems, including high risks of economic poverty. Also, they are over-represented in health services, continue to do most unpaid household care jobs, and face increased risks of violence, assault, or harassment (73). Additionally, their employment, businesses incomes, and living conditions could be more vulnerable than men to the predicted systemic economic effects of the pandemic, placing them at greater risk of nancial instability. All these issues have a considerable impact on the health and psychological well-being of women (73). Psychological di culties expected by nancial pressures include higher psychological distress and intensi ed issues and relations with family members, friends, and in the workplace (74).
Our results showed that a higher number of pregnancies and abuse by a family member were related to higher violence against women. Similarly, other studies found that a large family is a risk factor for frequent domestic violence, empowered by lower socioeconomic status, deprivation of maternal care, neglect of children, lack of family planning, and parenteral con ict (75)(76)(77)(78).
Furthermore, a history of child abuse was related to higher depression among women, consistently with other ndings (79)(80)(81)(82). Hence, the loss of attachment and the absence of a secure life were the main features for women to experience depression (83,84). Moreover, women victims of childhood abuse are more likely to endure violence later in their life (85)(86)(87). Identifying women with a history of childhood abuse and assaults would provide e cient maternal care and increase awareness against domestic violence.
Our results showed that violence mediated the association between pregnancy status, marital con icts, nancial di culties, and depression and anxiety but did not mediate between these factors and stress. Contrary to our ndings, a recent study conducted in Lebanon during the COVID-19 lockdown among 502 adults found that female gender and physical violence at home were associated with higher stress during the pandemic (88). It also showed that a previous higher socioeconomic status was associated with higher stress, while current nancial wellness was correlated with lower stress (88). However, a previous study among 215 mothers found that stressful life events were directly linked to psychological distress, particularly depression through domestic violence, while marital and pregnancy statuses were not related to the presence of domestic violence (29). Several other studies found that dissatis ed married women would present a high risk of mental distress (89)(90)(91). The sudden disruption of routine due to the COVID-19 outbreak has affected the populations globally, leading to exacerbation of stress in the majority of people. However, the Lebanese people exhibited psychological resilience during the con nement, as they had faced and continue to endure political instabilities, a decline of essential services, in addition to several types of violence (92); they adapt to abuse and work through it more than they oppose it (92). Regarding gender equality in Lebanese society, women continue to face discrimination at numerous levels, such as decision-making structures, socio-cultural values, and security issues. Men are raised in a manner that gives them a sense of authority over women and makes them expect women to recognize their superiority. Thus, unhappy marriages, marital con icts, and economic insecurity make women more vulnerable to mental problems (48,60,80). Future studies are necessary to determine the psychological distress faced by Lebanese married women, particularly during con nement.

Limitations
This study has several limitations. Its ndings could not be generalized to the entire Lebanese female population due to the small sample size. Also, its cross-sectional design makes it di cult to nd causal relations. The study relies on self-report data, which may be biased by memory recall and social desirability effects. A selection bias is also likely because the sample was not randomly chosen but rather obtained using the snowball sampling method. Residual confounding bias is also possible since there could be variables linked to psychological distress that were not measured. The violence scale was not validated in Arabic among women prior to our investigation.

Conclusion
Our main ndings suggest that violence mediated the association between pregnancy status, marital con icts, nancial di culties, and depression and anxiety but did not mediate between these factors and stress. During the COVID-19 pandemic, reducing the effect of stress in relationships and attempting to solve existing con icts are essential steps to assist partners and families to adapt to new and challenging conditions. Further studies are needed to understand better the factors that mediate the relations between domestic violence and mental well-being among Lebanese women.

Declarations
Ethics approval and consent to participate The Psychiatric Hospital of the Cross ethics committee approved the study protocol (HPC-018-2020). Online Informed consent was obtained from each participant involved in this study on the rst page of the questionnaire.

Consent for publication
Not applicable.

Availability of data and materials
Data can be made available under reasonable request form the corresponding author.