In recent decades, intimate partner violence against women (IPV) has been clearly shown to be associated with health problems,1,2 both physical3–5 and mental.3,4,6 Early studies centred on analysis of the consequences of physical or sexual IPV;7 it soon became apparent, however, that psychological violence also caused health problems, and nowadays studies that analyse the associations between IPV and health include psychological IPV and stalking.3,6

In addition to the type of violence, other aspects, such as the duration and severity of violence, have been analysed: such research has shown more serious mental health problems in cases of prolonged abuse8 and an increase in physical and psychological symptoms when the abuse is more severe1 or different types of violence are combined.5,6

In spite of the abundance of literature on the health-related effects of different types of severity of IPV, there are no quantitative studies in Spain that examine the association between IPV and health according to whether or not the abuse is perceived as such by the victim herself.

Most abused women pass through a stage in which they do not perceive the abuse as such.9,10 In the 2006 Macrosurvey on gender violence in Spain, about three quarters of the abused women did not consider themselves as abused.11 Qualitative research reveals that women at this stage often complain of depression, anxiety and other physical and psychological symptoms.10,12

Despite its limitations, the transtheoretical model of behaviour change (TTM), initially proposed by Prochaska et al13 to describe the process of smoking cessation, is considered useful to illustrate the different stages through which women go before freeing themselves of abuse.14 In the first or so-called “pre-contemplation” stage —one that may last for years—, the woman fails to recognise the problem. In the second stage, denominated the “contemplation” stage, the woman has become aware of the problem, but does not feel ready to act. In the following stages, she begins to seek help, then implements plans to leave the relationship and finally maintains the changes already attained. The progression through these stages is not always linear: there may be jumps to later stages in response to triggers, as a sense of danger for life, or, conversely, relapses due to inappropriate responses from the environment.15

Our main research question sought to establish whether, even at the stage when women do not consider themselves as abused, IPV might have effects on their health; hence, this study aimed to assess the association between unperceived IPV (uIPV) and various health problems, medication use and health-service utilisation. The second research question sought to study the differences between uIPV and perceived IPV (pIPV) in terms of the magnitude of association with the various health outcomes.

Of the 4862 women initially contacted, 567 did not fulfill the inclusion criteria. Of the remaining 4295, 1460 (34%) refused to participate (1231 initial refusal and 229 withdrawal from the interview). The percentage of refusals to participate was significantly higher among women aged 55 to 70 years than among those aged under 55 years (42.8% vs. 31.1%, p<0.001). Of the 2835 completed interviews, 14 were excluded due to missing data in the violence-related questions, giving a final sample of 2821 valid interviews.

A total of 247 cases of uIPV (prevalence 8.8%) and 96 cases of pIPV (prevalence 3.4%) were detected, with uIPV thus being 2.6 times more frequent than pIPV.

Among women in an IPV situation, whether or not perceived, there was a higher percentage of foreigners and women with socio-economic deprivation, terminated domestic partner relationships, and experience of violence before the age of 15, compared to IPV-free women (Table 1).

Most abused women who did not perceive the abuse as such were subjected to a single type of IPV, which was psychological in 90.7% of cases, physical in 5.4% and sexual in 3.9% (data not shown in the table). In contrast, women who did perceive the abuse, tended to be faced with two to three types of IPV (Table 1).

Table 2 lists all the outcomes that were associated at a value of p<0.1 with uIPV and/or pIPV, taking the violence-free population as a reference. In most of these outcomes, prevalence was higher for pIPV than for uIPV.

For each outcome shown in Table 2, a logistic regression model was constructed adjusting for the covariates: the association with uIPV and/or pIPV at a value of p<0.05 was confirmed for 23 of the 26 problems. The highest OR was observed for PHQ9-assessed depression, which was linked to uIPV with an OR of 3.4 (95%CI: 2.4-4.8) and to pIPV with an OR of 4.1 (95%CI: 2.5-6.8) (Table 3).

In 18 of the 26 problems analysed, the OR of association with respect to absence of IPV proved higher for pIPV than for uIPV (Table 3). Nevertheless, when we focused on the group of abused women and compared pIPV with uIPV with regard to health outcomes, only diagnosis of depressive disorder and visits to the social worker proved significantly associated with perception of the abuse; self-perceived suboptimal health was associated with a p value at the limit of statistical significance. All these associations became non-significant, however, when the variable “number of types of violence” was introduced into the models (Table 4).

Use of weight-loss drugs, in contrast, showed a stronger association with uIPV, and the association remained in evidence even after the variable “number of types of violence” had been introduced into the model.

The results indicate that the situation of uIPV is 2.6 times more frequent than that of pIPV, in line with the 2006 Macrosurvey.11 Edwards et al25 reported a very similar ratio of 2.3 between women in the pre-contemplation stage and those at later stages.

Among our interviewees, uIPV was characterised, in the majority of cases, by a single type of violence, which was almost invariably psychological. According to the qualitative literature, abuse often begins as isolated psychological violence that is not perceived as such,26 with initial episodes being regarded as “normal” domestic rows27 or obsessive control interpreted as love.10

The study's principal results show that uIPV is far from being harmless. The greatest magnitude of association is seen in items relating to mental health, as is widely confirmed by the literature6: when women with uIPV were compared to abuse-free women, they displayed a higher prevalence of PHQ-rated depression, diagnosis of depressive disorder and anxiety disorder, use of anxiolytics and antidepressants, counselling support/psychotherapy, and visits to the psychiatrist.

The deterioration of these women's health, however, affects areas that go beyond the purely mental: perception of suboptimal health, presence of limitation on activity due to health problems, presence of asthma, stomach ulcer, accidents in the preceding 12 months and smoking habit were all more prevalent in women with uIPV than among abuse-free women. The intake of commonly used medications, cold and influenza remedies, antibiotics and drugs for the gastro-intestinal system was likewise more frequent.

The second research question sought to ascertain whether there were differences between uIPV and pIPV in terms of the magnitude of association with the various health outcomes. Diagnosis of depression proved significantly more frequent in pIPV than in uIPV, in line with the findings of Edwards et al,25 who observed more severe mental health symptoms in the later stages. Self-perceived suboptimal health was also associated with perception of abuse, though without attaining statistical significance.

Qualitative research has shown that, when other types of violence -and physical violence in particular- are added to psychological violence, women become aware of the abuse and give the correct name to their partner's abusive behavior.10,23 When the variable “number of types of IPV” was taken into account in the analysis, the differences between uIPV and pIPV disappeared, indicating that the greater frequency of depressive disorder and self-perceived suboptimal health in pIPV is due to the accumulation of psychological, physical and sexual violence. These results agree with those published in the literature.1,5,6

With regard to the greater intake of weight-loss drugs among women who do not feel abused, our data are insufficient to advance any interpretations. While the known association between eating disorders and post-traumatic stress disorder related to sexual trauma28 may help one understand the relationship between the desire to lose weight and IPV, it does not explain why this should only be found in uIPV. The starting point for any subsequent research into this aspect should perhaps be the qualitative literature on the topic, which describes the effort made by women to maintain an ideal body image in attempt to please their abusive partners.10

Our study has the limitations specific to cross-sectional designs, with the ensuing impossibility of establishing temporal and causal relationships. Longitudinal studies have shown that IPV influences depression,29 just as depression can precede and facilitate IPV.30 This latter relationship is plausible in some cases of mental disease, since a depressed woman may well be more vulnerable and more exposed to IPV. However, it is inapplicable to other aspects of health, with it being very difficult to explain why asthma or smoking would increase IPV. The high number and the types of items positively associated with IPV leads us to surmise that, in general, IPV precedes and has a causal role in health problems, a conclusion supported by the literature, which documents the deterioration in overall health when IPV persists over time31 and the improvement in mental health when violence ceases.32

Another limitation is that it is not possible to say whether a woman who states that she does not feel abused perceives in reality the relationship as abusive, but is either unwilling or unable to reveal this to others. We should therefore consider that the uIPV category includes women who do not want to disclose the abuse.

Although population-based surveys are considered the best way to obtain accurate data on the prevalence of IPV33 and Caetano et al34 found that violence related variables are not the main factors determining participation in the survey, some degree of bias due to non-response rate of 34% could exist, especially because the participation rate was not homogeneous through the age groups.

The TTM used as a framework has a number of limitations we have to take into account. First of all, an abusive relationship is not comparable to substance addiction or other dysfunctional behaviours to which the TTM has been applied. As Chang et al15 observe, it is the perpetrator, and not the woman, the one who has a problem behaviour. Even if there is for the woman some room for action, her behaviour changes occur in the context of a relationship, where the counter-reaction of the abuser determines the subsequent evolution. The TTM probably is insufficient to capture the complexity of IPV; nonetheless it is a useful tool to guide the professional in supporting the woman according with her timeline.35

The documentation in this study attesting to the high prevalence of uIPV and its association with the same number of health problems as pIPV highlights the need for the efforts of health professionals and public health to be targeted at unperceived abuse, so as to help women recognise their situation and take the necessary decisions to preserve their health. Health professionals, aware of the fact that certain health problems may be linked to unperceived abuse, could contribute to awareness of IPV by helping women establish a link between their physical and psychological problems and their partner's abusive behavior.14

As can be seen from the results, asking a woman directly in a survey whether she felt abused would result in the majority of cases going undetected. In all likelihood, this direct question would prove equally inappropriate in clinical practice, where it is advisable to use suitable interview techniques adjusted to the stage in which the woman finds herself, as suggested by the guidelines drawn up to address IPV in a health-care context.35,36

M. Sonego and A. Gandarillas contributed to conception and design, analysis and interpretation of data, drafting the manuscript and final approval of the version published. B. Zorrilla, L. Lasheras, and M. Pires contributed to conception and design, acquisition of data, interpretation of data, revising the article critically for important intellectual content and final approval of the version published. A. Anes and M. Ordobás contributed to analysis and interpretation of data, revising the article critically for important intellectual content and final approval of the version published.

None.

None.