Violence Against Women and Perceived Health: An Observational Survey of Consultants in Family Planning Centers in the Paris, France, Region.

Background. Suffering violence has both short- and long-term negative effects on women’s health. We set out to study the relationship between socio-demographic characteristics of women attending three Family Planning Centers (FPCs) in France and the violence they experienced and to the association between women’s perception of their health status and their exposure to violence. Methods. We conducted an observational survey of clinical practice from December 2018 to February 2019 at three FPCs in Seine-Saint-Denis (Ile-de-France). All women patients aged 18 years or older were eligible. We solicited data on social characteristics, precariousness, and suffering violence. We measured health status on a 10-point scales for six different symptoms. Results. Of the 274 women who participated, 28% had experienced violence. This percentage was higher among women living in socially precarious situations compared to the not or slightly precarious (39.3% vs 15.8%, respectively). Experiencing violence was signicantly associated with being between 25 and 44 years old, being in a temporary or irregular administrative situation, being unemployed or seeking a job, and being in a precarious to very precarious situation. The proportion of women reporting domestic violence was more than twice as high in those consulting in FPC-1 (42%) than in FPC-2 (20%) or FPC-3 (16.2%). Regardless of age, history of violence was signicantly associated with poor sleep, poor diet, low morale, and poor health in general for all FPCs and all levels of precariousness. Conclusion. Our study reveals that reports of violence against women increased with migratory status and precariousness, unemployment or job-seeking, and a poorer state perception of one’s own health in a population drawn from three Family Planning Centers in Seine-Saint-Denis. It also suggests a structure specically dedicated to the management of women victim of violence that FPC-1, may be perceived more attractive by women exposed to domestic violence.

physical and/or sexual partner or ex-partner violence each year. These estimates likely undercount domestic violence survivors.
Violence against women has serious short-and long-term consequences for the mental, physical and sexual health of victims. , The WHO estimates that surviving violence costs women one to four years of healthy life. Reproductive health is particularly threatened since a violent partner may force sexual intercourse, not use a condom during intercourse, and restrict women's access to health care facilities. Women victims of domestic or sexual violence are more likely to seek abortions, , have spontaneous miscarriages, contract sexually transmitted infections, suffer sexual dysfunction, and develop pelvic in ammatory disease than women who have not suffered violence. , In addition, women who suffered violence have more anxiety disorders, sleep disorders, depression, post-traumatic stress disorder (PTSD), , and a higher risk of suicide. They also have more sexual, gynecological, and obstetrical disorders. They have more encounters with health professionals and consume more medication (anxiolytics, analgesics) than non-violent women who have not experienced domestic violence.
The proportion of women who have survived violence among those who visit a family planning Centre (FPC) in France has not been well characterized yet, but is likely to be high especially FPCs located in poorer districts. The department with the highest poverty rate in mainland France is Seine-Saint-Denis, located next to Paris. At one of its largest hospitals, the Delafontaine Hospital Centre in Saint-Denis city, 58% of women who visit are in a precarious or very precarious situation. Seine-Saint-Denis was thus an ideal site for conducting the rst French study to examined the relationship between social characteristics and overall health status of abused women who attend FPCs.
We set out to explore the relationship between socio-demographic characteristics, women's perception of their own health, and violence they had suffered. We included women who consulted at three FPCs in the Seine-Saint-Denis department and sought to determine their perception of their own health according to their previous experience of violence.

Data sources and study population
We carried out an observational survey of clinical practice from December 2018 to February 2019 in three FPCs in the French department of Seine-Saint-Denis: the "Maison des femmes," adjacent to and a liated with the Delafontaine hospital (FPC-1); "Les Moulins," a municipal health center in Saint-Denis city (FPC-2); and, a municipal health center in Aubervilliers, a neighboring town (FPC-3). The Maison des Femmes (https://www.lamaisondesfemmes.fr/) is a house, adjacent to the hospital, that was created speci cally to accommodate all women who are vulnerable or victims of violence. Women attending FPC1 can meet doctors, midwives, psychologists, social workers, but also police o cers, lawyers, and sports and cultural leaders. This multidisciplinary team aims to accompany women in their journey by providing full medical and social support. All women aged 18 years or older were eligible. Participation was voluntary. Adult women were offered a paper questionnaire (see supplementary material); those who had di culty understanding could ask for assistance from volunteer staff dedicated to this survey. The standardized questionnaire was identical in all three units and had four parts: 1) social and demographic characteristics, including a measure of social precariousness; 2) perceived health status; 3) women's knowledge about sexuality, contraception and use of voluntary termination of pregnancy; and, 4) a set of questions about violence they may have suffered.
The Committee for the Protection of Persons (CPP) of Ile de France 6 (located at Pitié-Salpêtrière Hospital, Paris) authorized this study.
Oral informed consent was obtained from all the women.
Questionnaires were self-administered and sent back to us by the women after they have left the FPC so that they did not feel pressured to answer questions with which they would feel uncomfortable.

Indicators
We included the following socio-demographic variables: place of residence; age; country of birth; administrative situation; living as a couple or not; having children; level of education; employment status; health insured; housing; and, assessment of living situation base on an individual indicator of precariousness (the Evaluation de la Précarité et des Inégalités de Santé dans les Centres d'Examens de Santé/ Evaluation of Health Precariousness and Inequalities in Health Examination Centers or EPICES score). , EPICES score is made up of 11 yes or no questions that sum up 90% of a subject's precarity. Each answer is assigned a coe cient and the 11 answers are summed to provide the EPICES score. The score is continuous, ranging from 0 (absence of precariousness) to 100 (maximum precariousness). We categorized the EPICES score into four classes corresponding to the quartiles of the overall sample.
Perceived health status was measured with the question, "How would you rate your health?". Participants responded on a Likert scale from 0 (very poor) to 10 (very good) for each of these symptoms: sleep; diet; mood; concentration; memory; and, general health. To simplify, we constructed an indicator for each system in three categories to sort responses into bad (0-4), average (5-7), or good (8-10) states. To summarize health data, we constructed an indicator by summing the scores attributed to the 6 symptoms (the global index of perceived health). The global index is graduated from 0 to 60, and also broken into three classes that correspond to the tertiles of the global sample.
At the end of the questionnaire participants were asked about violence they may have suffered: "Currently or in the course of your life, have you suffered violence?" (yes/no). Women who had suffered violence were asked to answer these questions: "Do you think that this violence can have, or may have had, an impact on your health? (yes/no/don't know)" and "Have you ever been able to talk about it before? (yes/no/ don't know).

Analysis strategy
No formal sample size calculation was made as this was a pilot of the ongoing AVEC-L study. we initially planned to include 100 women in each of the FPCs.
We described the characteristics of the sample in terms of distribution frequency: socio-demographic characteristics; perceived health status; violence suffered by respondents; and, the effect violence had on their health. We used bivariate and then multivariate analyses (adjustment on age, precarious situation and center) to study the associations between socio-demographic characteristics and having suffered violence; we calculated raw odds ratios (OR) and adjusted ORs (aOR) expressed with their 95% con dence intervals (CI) in logistic regression models. A p value of < 0.05 was considered signi cant. We expressed the relationship between having suffered violence and state of perceived health as raw and adjusted ORs, based on the 6 health status domains and the global index of perceived health.
The Pearson Chi-2 test was used to compare the percentages; statistical signi cance of the ORs was derived from Wald's Chi-2 test. All analyses were performed with SAS® version 9.4 software.

Results
Socio-demographic characteristics of the sample A total of 274 women completed the questionnaire at the three FPCs and were included in the study (100 at FPC-1; 100 at FCP-2; 74 at FPC-3). Their characteristics are summarized in Table 1. Of these, 234 (85%), lived in Seine-Saint-Denis, 82 (30,4%) were less than 25 years old; 143 (53.4) were born outside France, most (n = 62, 23.1%) in sub-Saharan Africa; 41 (15.5%) were in a temporary or irregular administrative situation. Only 113 (42,3%) lived as a couple; 160 (62%) had at least one child; 35 (13%) had not reached the college level and 96 (37%) had a level of education higher than a bachelor's degree. Of the women we included, 85 (33%) were homemakers or not in the labor force and 45 (18%) were seeking a job or unemployed; 86 (32%) of the women lived in sheltered housing; 177 (73.1%) were in a precarious situation. in the other two centers. Not living as a couple was associated with a non statiscally1.6 (OR1.64; 0.94-2.87, p = 0.082) increased risk of having been exposed to violence. Education degree below college was associated with the same a non statiscally1.6 increased risk. (See Table 2 for a summary of these results.) After adjustment, women with a very precarious status (determined by EPICES score) had an almost ve-fold greater risk of having experienced violence (aOR = 4.99; 1.86-13.34) and women who visited FPC-1 had a three-fold higher risk of having suffered violence (aOR = 3.37; 1.60-7.11).  Table 3. The sum of the six symptoms (summary of health data) gives a mean of 40 out of 60 with a standard deviation of 11.77, corresponding to "average" perceived health status.   Table 2 for a detailed summary.

Discussion
Violence was associated with reduced perceived state of health in the women we surveyed.
More than one fourth of women (27%) who visited the three FPCs claimed to have suffered violence at least once in their lives.
Despite being close to each other and all located in the same deprived area, the proportion of women reporting domestic violence in FPC1 (42.0%) was more than twice as high as in FPC2(20.0%) or in FPC3 (16.2%). Women aged 25-34 years and 35-44 years were around 3 times more likely than women below 25 years or above 45 years of having experienced violence in their lifetime.
Women from Sub-Saharan Africa had a 1.71 higher risk of violence (unadjusted OR = 1.71; 0.98-3.30); this rate was 2.3 higher among those who arrived in France less than 3 years ago than those who arrived in France more than 15 years ago (unadjusted OR = 2.30; 0.85-6.20).
Our nding that 27% of women had suffered violence, a proportion slightly higher than the 23% rate reported in a study conducted in France in 2009, 16 and close to the 30% reported by the WHO. We also reported higher prevalence of violence perpetrated by a partner or ex-partner than was found by studies in other countries, ranging from 4.0% in Australia 12 to 11.4% in India. But these studies only surveyed violence women suffered in the past year, while we considered lifetime history of violence. A European prevalence study found a lifetime prevalence of violence by a partner of 6.1%, which is much lower than in our result.
In our study, women aged 25-44 years were most likely to have suffered violence in their lifetime. International studies published in the past decade report that youth is an individual risk factor for suffering violence in women. ,, Women under the age of 25 years are the most likely to experience violence, including sexual violence. 26 Our questionnaire asked women of all ages about their lifetime history of violence so, even if the pattern of abuse of younger women holds true, older women were still more likely to have experienced violence in their lifetimes than younger women. Younger women may also more likely to minimize domestic violence, which intensi es and changes modalities over time. Older women may thus be more ready and able to identify domestic violence.
We may hypothesize that young women who attend FPCs may be more likely to visit to obtain contraception or end undesired pregnancies, and may turn elsewhere if they are victims of violence. A study by Sanz-Barbero et al., 26 using the same de nition of exposure to domestic violence as ours found that the prevalence of domestic violence was greater among women aged between 18 and 24 years. These discrepancies in both prevalence and risk factors illustrate the very partial knowledge we have on the subject and argue in favor of continuing observational studies in different contexts.
The association between low socio-economic level and higher risk of violence against women is often mentioned in the literature.
Our results support the argument that women in vulnerable social situations are at increased risk of violence, especially related to precariousness, unemployment or job search, and immigrant status with irregular administrative status. In the major reference survey in France on violence against women ENVEFF (Enquête Nationale sur les Violences Faites aux Femmes en France), the women most exposed to violence were unemployed, retired, or working part-time. In violent relationships, the aggressor is often economically dominant and may make it di cult his partner to seek and hold a steady job. Financial independence is an obstacle to eeing an abusive relationship. Aggressors may perceive a partner's search for a job as a manifestation of autonomy, which could trigger domestic violence.
Immigrant status is a risk factor for violence, whether in the country of origin, during migration route, or in the country of arrival. Our study aligns with other studies of immigration and violence that have demonstrated this. Immigrant women are often dependent upon their partner (con scation of papers, lack of economic independence) and may nd it hard to seek help (language barrier, social isolation, reluctance to call the police in irregular situations). Gender inequalities, a source of reproductive violence, may also be more marked in an immigrant's country of origin. Cultural factors affect feelings of shame, fear of stigmatization, and perceptions of violence and knowledge of the possibilities of external aid. − Family Planning Center 1, La Maison des Femmes, is a house speci cally dedicated to the management of vulnerable women or women exposed to violence. It offers care combining the health, social and judicial aspects in a single place. As FPC-1 is more than twice as likely than the 2 others, despite all being in the same deprived Paris' conurbation, our study suggests that the "Maison des Femmes" has achieved its rst objective, i.e. to be well identi ed by the some women of the department and neighboring departments as a preferred structure for dealing with violence.
Our nding that the women in our study reported poorer health status if they had a history of violence aligns with the results of many international studies that found abused women suffered more mental and physical health problems than women who were not victims of domestic violence. In particular, they have more anxiety disorders, sleep disorders, depression, post-traumatic stress disorder (PTSD) , and a higher risk of suicide. They also have more sexual, gynecological, and obstetrical disorders.

Study limitations
Our study was limited by the small number of participants (274) so we could not make a detailed comparison of the characteristics of women from each of the three centers. The on-going AVEC-L study will increase the number of included women in order to strengthened robustness of our background results. Women may minimize or fail to identify violence in self-reports, which may lead to classi cation bias. Our survey did not ask women to share a chronology of violence or specify the types of violence, severity or repetition, and we did not ask them to classify their aggressors. This may explain the discrepancy as for example Sanz-Barbero et al. suggested that if younger women are more frequently exposed to intimate partner violence, these are of less severity. 26 Selection bias may also be present; it is possible that women who refused to answer may have suffered more violence than by women who were willing to complete the survey. But this is unlikely because the rate of violence reported here is at least of the same magnitude than in the international literature.
This cross-sectional study does not allow us to conclude on causal links between antecedents of violence, socio-demographic characteristics and the health status perceived by the women respondents.
A cross-sectional study is not designed to identify causal links between antecedents of violence, socio-demographic characteristics, and perceived health status. A causal link between level of precariousness and surviving violence is impossible to draw from this data, and the causal connection may run in either direction or both. The amplitude of association may be the sum of an effect in both directions. Causal relationship between poor health and a history of violence is equally hard to determine. The relationship between health status and surviving violence against may be cumulative (the more traumatic episodes a woman experiences, the greater the impact on her health), adaptive (a woman's environment and experience may positively or negatively modulate the effects of violence on her health over the short and long term).
Data obtained in this study and in the on-going AVEC-L study will provide robust background evidence to assess the medical and social utility of structures speci cally dedicated to the management of domestic violence like "La Maison-des-Femmes" (FPC1)

Conclusion
Our study reveals that reports of violence against women increased with migratory status and precariousness, unemployment or job-seeking, and a poorer state perception of one's own health in a population drawn from three Family Planning Centers in Seine-Saint-Denis. Availability of data and materials: "The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request." Competing interests:

None
The authors declare they have no competing interests. Funding: