|Home | About | Journals | Submit | Contact Us | Français|
Among US women, 18 years or older, approximately 5.3 million intimate partner violence (IPV) victimizations occur each year, with 2 million injuries, 550 000 of which require medical attention.1 The IPV can be physical and/or sexual violence, threats of physical/sexual violence, and/or psychological harm by a current or former partner or spouse. The IPV can occur on a continuum and vary in frequency and severity.2 Nearly 1800 women are killed per year by men. Femicide, murder of women, is the seventh leading cause of premature death for women in the United States and the second leading cause of death for young African American and American Indian/Alaska Native women, aged 15 to 35 years. The cost is loss of precious lives as well as nearly $1 billion in lifetime earnings from victims of IPV homicide.1,3 The largest perpetrator category (30%–55%) of these femicides are intimate partners (IPs): husbands; boyfriends; partners; or ex-partners.4 Physical IPV was reported to have preceded the homicide in 65% to 80% of the IP femicide cases.4,5 Physical, emotional, and near-fatal violence is highly traumatic for victims and results in well-documented physical and mental health sequela,6,7 including posttraumatic stress disorder (PTSD) symptoms.8,9 However, our knowledge about near-lethal and lethal violence and the pathways that lead to PTSD is still incomplete. Understanding these interactions will help to form a more realistic and holistic approach for addressing the needs of IPV victims who experience severe violence.9
The purpose of this study was to examine the factors associated with PTSD symptomatology among a sample of women who experienced severe and near-lethal violence, called police, and had an officer come to the scene of an IPV incident to take a report, that is, police-involved IPV (PI-IPV). Using Dutton’s empowerment model as a framework,10 we hypothesized that, controlling for demographic characteristics, the likelihood of experiencing PTSD symptomatology will increase with (1) increased risk of future violence, (2) increased severity of violence, (3) the presence of psychological abuse and controlling behavior, (4) poor health and injury due to violence, and (5) increased commitment to the abusive relationship (marriage and children in common).
Dutton’s empowerment model10 provides a basis for understanding the social, political, cultural, and economic context of battered women’s responses. According to this model, IPV is viewed not only as a discrete episode but also as involving repeated occurrences of violent, abusive, and controlling behaviors by an abusing partner over months and years with traumatic effects. Goals for interventions, therefore, are for safety, shifting the power/control back to the victim, and helping victims heal from the complex series of physical and psychological traumas that have occurred. The model also explicitly posits the psychological effects of abuse, including traumatic effects such as PTSD symptoms and physiological reactivity, with identified mediators of the relationship of IPV and PTSD, namely, institutional response, personal strengths and inner resources, tangible resources and social support, medical factors, current stressors, and positive aspects of the relationship. Although the model was advanced early in the development of science around women’s responses to IPV, it is considered a classic and has a holistic approach congruent with nursing science, considering both psychological and physical factors. Current research on PTSD is primarily congruent with the model as originally conceptualized.
To investigate and understand the traumatic sequelae of battering, Dutton10 suggests that investigators examine 4 dimensions that distinguish battering from other traumas: (1) threat of ongoing traumatization; (2) repeated exposures to trauma; (3) history of similar trauma exposures; and (4) the nature of the relationship with the perpetrator. Of the relationships theorized in Dutton’s10 model, we were able to examine the current IPV relationship, the threat of ongoing victimization (victim’s assessment of risk of future violence and indicators of potential lethality), and the relationship with the perpetrator. As part of Dutton’s social, cultural, and economic context,10 we hypothesized that some individual characteristics (such as race/ethnicity and education) would impact PTSD symptomatology. We have updated our understanding and adapted the model using more recent research as summarized later (Figure 1).
Adverse physical and emotional health consequences of IPV are well documented, both nationally and internationally, but less understood are the consequences of severe near-lethal and lethal violence and its interrelationship with physical/emotional health and PTSD symptom outcome.
Posttraumatic stress disorder, depression, and anxiety are the most prevalent mental health sequela of IP trauma documented in Western literature.6,11 Prevalence of PTSD has been found to be higher in samples of battered women as compared with women who have not experienced battering. As posited in Dutton’s model,10 severity of abuse, previous trauma, and partner control were found to be important precursors to PTSD development.6 A meta-analysis of US studies found that 31% to 84.4% of women who experienced IPV met PTSD criteria and that the risk for developing IPV-related PTSD and depression was higher than that resulting from childhood sexual assault,11 thus supporting Dutton’s model because of the threat of ongoing traumatization in a continuing relationship with a partner. The California Women’s Health Survey, a population-based study, showed positive associations among physical and sexual assault, PTSD symptoms, and binge drinking.12 The PTSD was also associated with higher rates of psychiatric symptomatology in a sample of sheltered battered women.13
There is also much research evidence that women exposed to IPV have multiple physical health problems that are violence related,6,14 and strongly associated with fear15,16 and PTSD.8,9 Abused women with PTSD living in shelters showed health-related problems in the following 4 areas: neuromuscular; stress; sleep; and gynecological problems. The more severe the PTSD, the more health problems these women experienced.8 Women recruited from crisis shelters and community agencies with concurrent IPV, health symptoms, and PTSD showed poorer global sleep quality and nighttime disruptive behaviors.17 In a population-based sample of California women, those with chlamydia were 5 times more likely to have had a history of multiple abuses.18 Another population-based sample of more than 6000 California women showed an association among IPV, PTSD, and unemployment, but the association was found with psychological and not physical abuse.19 Analysis of the Chicago Women’s Health Survey showed that psychological abuse, physical violence, sexual coercion, and PTSD were highly correlated with miscarriages.12
Recent inquires about IPV and PTSD have focused on possible mediators that affect the strength of the relationship of IPV exposure and the development, severity of PTSD symptomatology, as well as the relationship among these variables.20 This line of study has been productive to begin to identify pathways, models, and predictive factors associated with IPV exposure and PTSD that will eventually lead to preventive interventions. Violence severity and chronicity,21 victim emotional coping,22 immune system integrity, attachment anxiety,23 depression,21 fear16, and childhood maltreatment24 are the variables found in IPV populations that have been implicated in the development of PTSD symptom outcomes. These are congruent with Dutton’s model, although she considers some of these as characteristics of the trauma.
Studies with different racial, ethnic, and cultural groups have shown a variety of results when exploring the relationship of IPV exposure and PTSD. In the Chicago Women’s Health Risk Survey, pregnant black abused women had the highest rate of IPV as compared with all other women yet had the lowest probability for developing PTSD as compared with nonpregnant black and other ethnic-background women. Hispanic abused women, on the contrary, had decreased probability of both IPV and PTSD as compared with nonpregnant Hispanic women. Study of 234 American Indian/Alaska Native primary care patients revealed that severe IPV was associated with PTSD, anxiety, and major depressive disorder but not with substance abuse.25 Internationally, 56% of the 105 women presenting to National Institute of Mental Health and Neurosciences Adult Psychiatric Clinic in South India reported severe violence and sexual coercion positively associated with PTSD symptoms.26 Using the Diagnostic and Statistical Manual for Mental Disorders (DSM) criteria, risk for PTSD in a sample of South African men and women was similar for men and women, but women experienced more IPV and childhood sexual assault than men.27 A review of PTSD studies of Chinese abused women revealed only 4 studies, cautioning comparisons with Western PTSD DSM-IV T-R diagnostic criteria not yet validated for cultural specificity.28 Thus, knowledge about cultural group variations in the development of IPV-related PTSD and physical/mental health disorders is still in its infancy.
Recent research29 has focused on particular PTSD symptoms or clusters of symptoms that have differential health outcomes, thus moving the research to analyzing the interplay of the relationships of IPV, PTSD, and physical/emotional health. Our analysis sought to add to the body of knowledge about the predictors of PTSD symptomatology from among the factors chosen from the Dutton’s empowerment framework.10
Data for this cross-sectional survey research study were collected through structured telephone interviews with victims of PI-IPV. Interviews lasted between 45 minutes and 1 hour, and participants were provided with a $15 gift card to thank them for their time. This study was approved by the institutional review boards of the University of Oklahoma Health Sciences Center, Arizona State University, Johns Hopkins University, the Oklahoma State Department of Health, and the Cherokee Nation.
Police officers recruited eligible participants at the scene of domestic violence incidents across 7 jurisdictions in a single Southwestern state.30 Women were eligible for recruitment if they had experienced IPV and the officer had reason to believe that either (1) this was a repeat instance of PI-IPV or (2) the victim was in danger from the abuser. When an IPV victim met either one of these criteria, the officer read an advisement statement asking the victim if she would be willing to be contacted by researchers. If the victim was willing to be contacted, the officer gathered 1 to 2 safe telephone numbers and a safe time for researchers to call. This information was forwarded to researchers by the police department.30 On average, these potential participants were called within 24 hours of researcher receipt of the referral. In total, police departments provided names and contact information for 1137 victims of IPV over approximately 18 months. Of these referrals, 486 (42.74%) were not able to be contacted because of unanswered, disconnected, or wrong numbers and 47 (4.1%) were not eligible (eg, not an IPV victim or were underage). Of the 604 eligible referrals able to be contacted, 164 (27.2%) declined to participate and 440 (72.8%) provided informed consent and participated in the research study.
Of the 440 participants, 37 (8.4%) were missing data on pertinent variables. Twenty cases had limited missing data, and conditional mean imputation was used to insert missing values based on predicted probabilities developed by using regression models. These models predicted known cases, with an average of 81.14% accuracy (range, 73.66%–87.50%). While this technique is neither perfect nor entirely free from bias,31 it is an improvement on list-wise deletion and unconditional mean imputation as strategies for handling missing values.32 The 17 participants deleted from the analysis due to missing data were not different than participants included in terms of demographic information, relationship characteristics, or the independent and dependent variables used in this analysis.
The final model utilized data from 423 (96.1%) of the 440 participants interviewed. For logistic regression models, a sample size of 423 had 80% power to detect an odds ratio of 0.5 or 1.5 for binary or categorical predictor variables and 1.3 or 0.7 for linear independent variables.
After making contact with the referred victim, the interviewer explained the purpose of the study and what participation entailed. If the victim was willing to be part of the research study, the interviewer obtained informed consent verbally. While participants could request that a copy of the consent document be sent to a safe address, having a document tying the participant to a research study on violence in their intimate relationship might be dangerous, and, therefore, verbal consent was deemed a safer option. The telephone interview was conducted at that time or at another time that was safe and convenient for the participant. The telephone survey was confidential (not anonymous). Participants’ identifying information was kept in a secure location; databases were stripped of identifying information as soon as the research was completed, and interviewers were instructed not to share identifying information with others. A small gift certificate for time spent in participating in the study was sent to a safe address in a nondescript envelope, with a letter thanking the participant for participating in a women’s health survey. Finally, to ensure that the confidentiality of participants was protected, a privacy certificate was obtained from the National Institute of Justice.
Interviewers were trained by using telephone safety precautions, first developed for the Canadian DV survey and since used by this research team in the Risk Assessment Validation Study (RAVE).33,34 The interviewer’s first priority during all contacts with the participant was to maintain participant safety. To achieve this, only female interviewers were used to interview female participants and a cover story was developed (participation in a women’s health survey). At the beginning of every conversation, participants were asked whether it was a safe time to talk, and participants were instructed to say “it is not a good time for me to talk now” if, at any time during the conversation, she felt unsafe. Interviewers were also trained to listen for any interruption, disturbance, altercation, or eavesdropping. If, during the phone conversation, the participant indicated that it was not a safe time to talk or the interviewer suspected that it was not a safe time to talk, the interviewer reverted to a list of yes/no questions. Depending on the situation, these questions were about the participant’s health (if, eg, eavesdropping was suspected) or about whether the participant was safe and would like the interviewer to contact police (if, eg, the interviewer heard an altercation). Throughout the study period, interviewers met to debrief and share difficult and educational experiences.
The dependent variable, PTSD symptomatology, was measured by using the Primary Care Posttraumatic Stress Disorder (PC-PTSD) Screen.35 This instrument is a recommended screen36 and has been used in population-based research examining the relationship of PTSD to childhood and adult victimization,37 as well as to screen for PTSD in research among war veterans.38 The PC-PTSD consists of 4 items for examining the symptoms of PTSD: (1) reexperiencing the event (you have had nightmares or thought about it when you did not want to); (2) numbing (you felt numb or detached from others, activities, or your surroundings); (3) avoidance (you tried hard not to think about it or went out of your way to avoid situations that reminded you of it); and (4) hyperarousal (you were constantly on guard, watchful, or easily startled). Participants were asked whether they had ever experienced these symptoms as a result of their partner’s abuse. For each question, a yes response was coded as 1 and a no response was coded as 0. A cutoff score of 3 was used in this analysis and has been found to be optimal with good sensitivity (0.70), specificity (0.84), and efficiency (0.81) among women. Utilizing the clinician-administered PTSD scale as a gold standard, the PC-PTSD was found to correctly classify 78% of cases.39 Nevertheless, it is important to note that the PC-PTSD is not a diagnostic tool, and, thus, this research examined the outcome of PTSD symptomatology and not a PTSD diagnosis.
Following our adaptation of Dutton’s10 empowerment model (Figure 1), it is hypothesized that, controlling for demographic characteristics, the following independent variables will be associated with an increase in the likelihood of PTSD symptomatology: (1) increased commitment to the abusive relationship (marriage and children in common); (2) risk of future violence; (3) poor health and injury due to violence; (4) increased severity of violence; and (5) the presence of psychological abuse and controlling behavior. The measurement of these independent variables is explicated later.
Respondents reported their age in years. Respondents self-reported their race/ethnicity, and this was collapsed into the categories: white, African American, Native American, Latina, and other. Participants self-reported their current employment status as employed full/part time or not employed full/part time. Finally, respondents were asked to report their highest level of education: less than high school; high school graduate; some college; college graduate; or any graduate school; this was collapsed into high school degree or less (=0) and some college or higher (=1).
Participants were asked to report their current legal marital status (single, married, or separated/divorced). Being married to their partner is conceptualized as increased commitment over being single, and reporting their legal marital status as separated/divorced is conceptualized as decreased commitment over being single. Participants were asked the number of children in common with their partner. This was dichotomized as not having children with their partner (=0) and having children with their partner (=1); having children with their abusive partner was conceptualized as increased commitment to the relationship.
The danger assessment (DA), a clinical and research instrument that assists battered women in assessing the risk of being murdered by their IP, was administered to participants. The DA psychometrics have been evaluated in 6 major studies,40 supporting predictive validity for actual and attempted femicide (receiver operating characteristic score = 0.90) and reoffending (receiver operating characteristic score = 0.67)33; thus, this instrument is ideal for assessing the threat of ongoing traumatization from severe violence. The DA consists of 20 dichotomous items (yes = 1/no = 0), which are weighted and summed to produce an overall score between 0 and 37 (www.dangerassessment.org). The DA score was treated as a linear variable in this analysis, but scores can also be placed into the following categories: variable danger (0–7); increased danger (8–13); severe danger (14–17); and extreme danger (18 or higher). In addition, to assess self-perceived risk of future injurious victimization, participants were asked to rate the likelihood (on a scale of 0–10, with 0 being no chance and 10 being sure to happen) that their partner would seriously hurt them in the next year.
A single item from the Health Status Questionnaire was utilized to measure physical health status; this is one of the most common measures of self-reported health, is strongly associated with other indicators of health, and is generally considered to be both reliable and valid.41 Participants were asked to respond to the following question taken from the Health Status Questionnaire42: “In general, would you say your health is: excellent, very good, good, fair, or poor.” This variable was subsequently dichotomized for the purpose of this analysis into excellent, very good, and good (=1) and fair and poor (=0). One dichotomous (yes = 1/no = 0) item regarding experience of physical injury due to IPV, taken from the revised Conflict Tactics Scale (CTS-2),43 was significant in the final model, “Have you ever experienced physical pain that still hurt the next day because of a fight with your partner?”
The physical assault subscale of the CTS-243 was used to assess acts of physical violence that the participant experienced in her relationship. Because of the high levels of violence in this sample (97.4% of participants experienced 1 or more types of physical violence), severe, near-lethal and sexual violence were examined. Severe violence included the following: your partner used a knife or gun on you, punched you, hit you with something that could hurt, choked you (strangulation), beat you up, burned or scalded you on purpose, and kicked you. Near-lethal violence was assessed with 2 dichotomous questions: “Has your partner tried to kill you?” and “Has your partner ever done anything that might have killed you or nearly killed you, whether or not he intended to?” Finally, 3 questions from the sexual coercion subscale of the CTS-2 were used to assess sexual violence: “Has your partner used force to make you have sex?” “Has your partner made you have sex without a condom?” and “Has your partner insisted on sex when you did not want to?” Responses were dichotomized into “this has happened” (=1) and “this has never happened” (=0).
Five of the 10 items from the Women’s Experience of Battering (WEB) scale44 were included in the analysis. These were used as single items, rather than as a scale, to assess specific aspects of the psychological effects of violence on participants. These were (1) “My partner makes me feel unsafe even in my own home,” (2) “I feel ashamed of the things my partner does to me,” (3) “my partner makes me feel like I have no control over my life, no power, no protection,” (4) “I hide the truth from others because I am afraid,” and (5) “my partner can scare me without ever laying a hand on me.” These items were dichotomized such that 1 indicated that the respondent agreed with the statement and 0 indicated that the respondent disagreed with the statement.
Univariate analyses were used to describe the sample, including individual demographic characteristics and relationship with the perpetrator, risk of future violence/lethality/injury, health status and physical pain due to IPV, violence exposure, and the psychological effects of violence. Bivariate analyses with the outcome of PTSD symptomatology were examined, and those variables significant at the P < .10 level were examined for inclusion in the final multivariate model. To test the hypotheses, multivariate logistic regression was used to examine the effect of (1) increased commitment to the abusive relationship (marriage and children in common); (2) poor health and injury due to violence; (3) increased severity of violence; (4) the presence of psychological abuse and controlling behavior; and (5) risk of future violence on the probability of experiencing PTSD symptomatology controlling for demographic characteristics. Logistic regression was utilized for this analysis, as it is able to take into account the effect of multiple linear and binary independent variables on a single binary dependent variable (PTSD symptomatology). Although regression models predict the likelihood of an outcome based on the independent variables, it is important to note that the cross-sectional nature of these data makes it impossible to know whether the significant variables in the model are causally related to the outcome. The relationships found between independent and dependent variables are associations.
As described in Table 1, of the 423 women included in the analysis, mean respondent age was 32.64 years (SD = 9.47). Approximately 45% of the survey participants described themselves as white, 33% as African American, 13% as Native American, 7% as Latina, and 7% were placed in the “other” category. The majority of the participants were not currently married either because they had never been married (58.16%) or because they were divorced/separated (18.20%). Slightly less than half (47.04%) of the participants had a child with the perpetrator. In regard to education, approximately half (49.65%) of the participants had some college education or higher. Less than half (41.64%) of the participants were employed full or part time.
As shown in Table 1, this is a sample that has experienced high levels of IPV; 88.65% had experienced severe physical abuse. This includes 18.91% (n = 80) of participants who reported having a knife or gun used on them; 63.03% (n = 266) who reported being hit with something that could hurt; 71.63% (n = 303) who had been strangled; 67.22% (n = 283) who reported being beat up; 7.58% (n = 32) who reported being burned or scalded on purpose; and 38.77% (n = 164) who reported being punched or kicked. Near-lethal violence was reported by nearly half of the participants (45.86%). Sexual abuse was also common with 43.03% of the participants reporting sexual abuse, including 18.05% (n = 76) who were forced to have sex with their partner. Finally, psychological effects of violence were prevalent as well, with between 27% and 57% of the participants responding affirmatively to at least 1 item on the WEB scale.
The dependent variable, PTSD symptomatology, is a binary variable (0/1), where participants with 3 or 4 PTSD symptoms were considered to have PTSD symptomatology and participants who reported 2 or fewer symptoms were not considered to have PTSD symptomatology. Of the 423 participants, 53.19% were classified as experiencing PTSD symptomatology by the PC-PTSD. Eighty-five (20.09%) participants experienced no PTSD symptoms; 44 (10.40%) participants exhibited 1 symptom; 69 (16.31%) participants exhibited 2 symptoms; 88 (20.80%) participants exhibited 3 symptoms; and 137 (32.39%) participants exhibited all 4 symptoms.
Table 2 contains the results of the multivariate logistic regression. The majority of participants’ demographic and relationship characteristics were not associated with PTSD symptomatology in this multivariate analysis but were retained as control variables. Race/ethnicity is an exception, with African American participants experiencing 60% less PTSD symptomatology than white participants. Consistent with the hypothesis, separated/divorced participants are significantly less likely (54%) to report PTSD symptomatology. The direction of the relationships of PTSD symptomatology with being married and having children in common with the perpetrator, while not significant, were contrary to what was expected.
Risk of future violence/lethality, as measured by the DA, was significantly associated with increased likelihood of PTSD symptomatology. A 1-point increase in the DA score was associated with an adjusted odds ratio of 1.07. Holding all other variables at the mean, for each 4-point increase in the DA score, the likelihood of PTSD symptomatology increases between 4.62% and 6.86%. Similarly, holding all other variables at the mean, a participant with the mean variable danger score (rounded = 4) has a 31.8% chance of experiencing PTSD symptomatology, and a participant with the mean extreme danger score (rounded = 23) has nearly double this chance (63.31%). An increase in the participant’s estimation of the likelihood that her partner will seriously hurt her in the next year is also associated with a significant increase in the likelihood of PTSD symptomatology (adjusted odds ratio = 1.15 for each 1-point increase). Holding all other variables at the mean, for each 1-point increase in participant’s estimation of the likelihood that her partner will seriously hurt her in the next year, the likelihood of PTSD symptomatology increases between 2.76% and 3.5%. A participant who believes that there is no chance that her partner will seriously hurt her in the next year has a 41.77% chance of PTSD symptomatology, holding all other variables constant. This increases to a 62.42% chance if her estimation is slightly higher than neutral and a 74.46% chance of PTSD symptomatology if she has no doubt that her partner will hurt her in the next year.
Consistent with hypotheses, physical health was significantly and negatively related to PTSD symptomatology in the multivariate analysis; participants reporting good, very good, or excellent health were 49% less likely to experience PTSD symptomatology. Experiencing physical pain due to abuse that hurt the following day was significantly associated with PTSD symptomatology, with more than an 8-fold increase associated with a positive response on this item. While severe violence, near-lethal violence, and sexual assault were associated with PTSD symptomatology in bivariate analyses, contrary to hypotheses, these variables did not remain significant in multivariate analyses. Post hoc analyses found significant relationships between DA score and severe violence (t = −6.90, P < .001), near-lethal violence (t = −10.80, P < .001), and sexual violence (t = −9.64, P <.001), as well as between participant estimation of the likelihood of their partner causing serious injury in the next year and severe violence (t = −2.54, P < .01), near-lethal violence (t = −5.28, P < .001), and sexual violence (t = −3.68, P < .001).
Two items from the WEB scale were associated with an increase in PTSD symptomatology in the multivariate analysis, though all items were associated with PTSD in bivariate analyses. This is consistent with the hypotheses. If the participant reported that she is ashamed of the abuse that her partner inflicts on her, there is an associated 3-fold increase in PTSD symptomatology. Similarly, there is a significant increase in PTSD symptomatology if the participant reports feeling unsafe due to the abuse, with an adjusted odds ratio of 2.70.
Of the 423 participants, 53% were classified as experiencing PTSD symptomatology by the PC-PTSD. This is within, but at the high end of, the range found in other studies. The women in this sample have experienced high levels of severe violence and were in contact with the police due to these experiences of violent victimization, and this is congruent with research findings that show an association of PTSD with severity of battering. White women had significantly higher PTSD scores than the African American women. The lack of association between PTSD and other ethnicities in our study may be due to relatively low numbers of participants in the Native American, Latina, and other groups. Since other studies do not show similar results, further investigation of the effects of race/ethnicity on PTSD symptomatology in women with severe IPV is warranted.
Our findings support prior research, indicating that fear is a powerful predictor of PTSD.16,45,46 In fact, in this sample, while experiences of severe, near-lethal, and sexual violence were significant in bivariate analyses, these were not significant in the multivariate analysis. Rather, risk of future violence/lethality, self-perceived risk of injury due to abuse, and feeling unsafe were significant predictors of PTSD; these factors reasonably invoke or indicate serious fear. Thus, in this sample, fear appears to be a more powerful predictor of PTSD symptomatology than experiences of violent victimization. In addition, similar to previous research demonstrating that victims are accurate in assessing their own risk,40,46 we found a linear relationship such that the more a woman believes that her partner is likely to hurt her in the next year, the higher her levels of PTSD symptomatology. This finding is consistent with the theoretical model proposed by Dutton10 and demonstrates that fear should be taken seriously by nurse practitioners in clinical assessments of IPV, as it is a predictor of PTSD.
Similarly, the overall DA score, measuring risk of homicide, was clearly associated with PTSD symptomatology, as one would expect, given the level of danger in the situation and the women’s fairly accurate appraisal of their risk. The risk factors measured in the DA include variables such as abuse during pregnancy, forced sex, and threats of homicide. The association of the DA score with PTSD symptomatology speaks to the traumagenic properties of these forms of IPV for women. These have been supported in prior research, but the implications for nursing theory, practice, and research have not been well considered. The need for assessment of IPV among pregnant women is fairly well accepted in nursing but is focused more on the physical health and safety of the mother and unborn child than on mental health implications. Similarly, sexual assault is a frequent part of IPV for women, and although considered in terms of gynecological and reproductive health by practitioners, it is less considered in psychological and psychophysiological interactions.
It is important to note that this research was limited because of the cross-sectional nature of the data. The lack of ability to be able to determine time ordering in cross-sectional analyses, such as ours, is a frequent feature of trauma and health outcomes research and will be solved with only more longitudinal research. Another limitation of this research was that we screened for PTSD among this sample and did not diagnose PTSD and thus measured PTSD symptomatology rather than PTSD. While the PC-PTSD is a valid and reliable screen for PTSD, it is not a diagnostic tool, and future research should examine the variables associated with PTSD diagnosis or longer PTSD symptom measures. We did not ask women in this sample about childhood experiences of abuse or other types of violent victimization. Because there is a developing literature on PTSD as a mediator between childhood and adult victimization, future research should take into account multiple life experiences of violent victimization.
These limitations, however, are balanced by the strengths of this study. Women were recruited after being in contact with police due to IPV. This is a unique sample that has experienced relatively severe forms of violence and is at high risk for homicide. Risk of homicide and future violence was measured with a reliable and valid lethality risk-assessment measure as well as using women’s perceptions of their own risk. Future research should seek to replicate the relationship between risk for future violence and PTSD.
Our findings support Dutton’s model10 that distinguishes battering from other traumas and suggests adding a fifth dimension to consider: the risk of near-lethal and/or lethal physical harm. Not all traumas (eg, a drive-by shooting) have the risk for future near-lethal and/or lethal violence, but as found in this study, that risk was high for women who experienced severe IPV, and it is significantly associated with psychological outcomes—specifically, the increased likelihood of PTSD symptomatology. Near-lethal and/or lethal violence risk are a distinct potential for a large proportion of women who experience IPV and can be studied as a differentiating factor with IPV traumas for further development within the Dutton model.
Dutton’s mediators of battered women’s responses to abuse should be tested to see the interrelationships of how each affects and changes specific psychological and physical responses. For example, based on this sample and holding all other variables at the mean, a woman who rates her health as fair/poor, with a DA score in the extreme range (eg, 23) and has a belief, however slight, that her partner will hurt her within the next 6 months has approximately an 80% chances of screening positive for PTSD symptomatology. However, a woman who rates her health as good/very good/excellent, has a DA score in the variable range (eg, 4), and has no belief that her partner will hurt her within the next 6 months has approximately a 20% chance of screening positive for PTSD symptomatology. Therefore, changes in the DA score, health, and self-perceived risk can have a major impact on the likelihood of PTSD symptomatology. Dutton classifies these as mediators, and while this suggests a mediating relationship, future research should examine specifically these relationships.
Mediators can also inform development of IPV preventive practice strategies such as use of lethality assessment protocols (an example of an institutional support) within hospitals, clinics, and community settings that can be investigated for the changes and effects they cause on battered women’s psychological and physical well-being.
Since this study supported that PTSD symptomatology was best interpreted as a sequelae of violence rather than a precursor, future research can examine this relationship in both bivariate and multivariate analyses, with the possibility that PTSD mediates the relationship of severe violence and fear with physical health outcomes. Our finding that African American women in IPV scored lower on PTSD symptomatology suggests that more investigation is needed with culturally diverse populations to test associations of IPV, PTSD, and physical/mental health, so we can determine evidence-based and culturally specific practices.47
Our finding that physical health and experiences of pain in women with severe IPV are associated with PTSD symptomatology supports and emphasizes the need for nurses in clinical practice to use IPV danger/lethality assessments and primary care screens for PTSD. However, the co-occurrences among PTSD, IPV, fear, and women’s emotional/physical health form complex interrelationships that may impede a woman’s ability to ask for help and access resources for IPV care and therein affect her safety in an already-dangerous situation. Help for severely abused victims such as the women in this sample is even more crucial because of their high risk for lethality.48 Consistent findings of the co-occurrence of IPV with multiple mental and physical health problems (eg, depression and anxiety) and, in this study, with PTSD, should alert health care practitioners and advocates to screen all women (not just women who self-report) for IPV and PTSD in hospitals, clinics, shelters, and community settings.13 Unfortunately, PTSD symptomatology is often not recognized as IPV sequela and is frequently treated with psychotropic medications without recognizing or intervening with the IPV and/or PTSD directly, thus potentially putting victims at an even-greater risk for lethal IPV.49
The intersection of multiple abuses seen in this study supports an expanded clinical conceptualization of PTSD that encompasses the experiences of victims of IP violence, not only with severe, prolonged, and repeated abuses but also with abuses that are diverse and intersect with one another. Herman50 suggests inclusion of a new classification of PTSDs called complicated or complex PTSD with type I and type II categorizations to reflect severe, prolonged, and diverse abuses. This type of classification would enhance evidence-based diagnostic categories that more accurately reflect the experiences of abuse survivors.
Advanced practice nurses in primary care settings, emergency rooms, medical surgical units, obstetrics/gynecology, and mental health and clinic settings have a pivotal role in health care screening, assessment, and referral for the physical and emotional sequelae of IPV-related PTSD. However, to do this, nurses must know not only about IPV but also about the relationship of IPV to PTSD and how they influence the assessment and intervention process, from the perspective of both victim and nurse. Presenting psychological symptoms in women should be assessed in the context of IPV and PTSD and not so readily diagnosed as psychiatric disorders without first assessing for trauma. Also, more trauma-focused treatment resources, rather than psychiatric services, should be available and more research is needed to gather evidence of the efficacy of such treatment modalities for IPV victims.
This study was funded by a grant from the National Institute of Justice (2008-WG-BX-0002).