bs_bs_banner
Journal of Evaluation in Clinical Practice ISSN 1365-2753
Do violence dynamics matter? David Katerndahl MD MA,1 Sandra Burge PhD,1 Robert Ferrer MD MPH,1 Johanna Becho BA2 and Robert Wood DrPH3 1 Professor, 2Research Associate, 3Biostatistician, Department of Family & Community Medicine, University of Texas Health Science Center at San Antonio, San Antonio, Texas, USA
Keywords battered women, domestic violence, intimate partner violence, non-linear dynamics, outcomes research, systems theory Correspondence Dr David Katerndahl Family & Community Medicine University of Texas Health Science Center at San Antonio 7703 Floyd Curl Drive San Antonio, TX 78229 USA E-mail: [email protected] Accepted for publication: 22 May 2014 doi:10.1111/jep.12216
Abstract Rationale, aims and objectives Intimate partner violence is a complex, non-linear phenomenon. The purpose of this study was to determine whether violence dynamics (pattern, degree of non-linearity, optimal non-linearity) contributed to outcomes in violent relationships. Methods The study was conducted in six primary care clinics, enrolling 200 adult women in violent relationships. In addition to baseline and end-of-study interviews, women completed daily telephone assessments of household environment and partner violence using interactive verbal response. Three non-linearity measures of violence were computed with ‘optimal’ non-linearity estimated using Z-transformations. Assignment of dynamic patterns (periodic, chaotic, random) was made based upon Lyapunov exponent and correlation dimension. Outcomes across dynamic patterns were analysed using analysis of variance. In addition, stepped multiple linear regression explained factor-analysed outcomes, adjusting for demographic, childhood, mental health and marital variables; attitudinal/behavioural outcomes were also adjusted for when explaining clinical outcomes. Results Women experiencing periodic violence recognized the importance of violence and used their active coping to seek mental health care. Those with chaotic dynamics recognized that they were not responsible, experienced fewer psychological symptoms and emotional role limitations, and did not seek help. Those experiencing random violence recognized its unpredictability and uncontrollability. Violence non-linearity predicted negative coping, positive appraisals and hope/support in regression analyses, while optimal non-linearity contributed to readiness for change and symptoms functioning. Of the nine outcomes investigated, violence non-linearity contributed to five outcomes. Conclusion Dynamic pattern of violence, degree of violence non-linearity and optimal non-linearity correlated with several attitudinal/behavioural and clinical outcomes. Knowledge of violence dynamics may have applications when working with violent couples.
Introduction Intimate partner violence (IPV) is a non-linear phenomenon. Quantitatively, the day-to-day level of violence displays non-linear dynamics in most relationships. Although the dynamic pattern may speak to aetiological factors [1], is non-linearity important to patient outcomes? An emerging belief is that non-linearity may be critical to health and well-being [2]. When biological systems are healthy, they behave in a non-linear way, exhibiting adaptability with resistance to external stressors that might disrupt their healthy dynamics [3].
However, when systems transition into periodicity due to illness, their trajectories become regular and potentially amenable to intervention, enabling clinicians to restore them to healthy, adaptable, non-linear dynamics. This transition from healthy non-linear dynamics to unhealthy linear dynamics can be understood by realizing that non-linear systems display a variety of dynamics, depending upon their resources and constraints, interconnectedness, and feedback. In the case of individual people, as the number of chronic medical problems increases, the patient’s resourcefulness, flexibility and adaptability (and, hence, nonlinearity) may decrease, leading to linear dynamics and poor
Journal of Evaluation in Clinical Practice 20 (2014) 719–727 © 2014 John Wiley & Sons, Ltd.
719
Do violence dynamics matter?
overall health. As a continuum from linear to non-linear dynamics, the constraining nature of disease severity, co-morbidity and chronic stress should suppress healthy trends towards nonlinearity and adaptability, thus leading to linear dynamics or rigidity. However, because adaptive systems display order and disorder strategically [4], there may exist an optimal degree of non-linearity between the minimum of periodicity and the maximum of randomness. Just as extreme variability may be detrimental within an illness [5], so too may extreme non-linearity be dysfunctional [6]. In the case of family systems, family health as a function of members’ communication, cohesion and flexibility [7]. On the flexibility continuum, extreme values are considered unhealthy. Inflexible families are rigid with strict rules and expectations for family members; they are linear, predictable and not adaptable to change. Overly flexible families are erratic with no rules, no clear roles, and no leadership; they are random, unpredictable and unstable. Thus, the relationship between system dynamics and outcome may be curvilinear [8]. We examined dynamic patterns of violence in heterosexual couple relationships to determine whether non-linearity of violence contributed to attitudes, behaviours or clinical outcomes in women. Specifically, do measures of non-linearity account for significant variance in outcomes? Is there a mid-range level of violence non-linearity that predicts these outcomes?
Methods Sample In this study, as previously described [9], 200 women with a recent history of husband-to-wife physical abuse were recruited from six primary care clinics in San Antonio, Texas. Adult (18–64 years old), non-pregnant women seen without their husbands present at the time of the visit were asked to complete the 6-item brief Conflict Tactic Scale [10] in the examination room while they waited to see their doctor. If they had experienced violence from their partner in the past 30 days, they next responded to a Danger Assessment Screen. Women in extremely dangerous relationships were excluded.
Procedure Once enrolled, subjects completed a baseline survey. For 12 weeks, they were asked to complete a daily telephone survey using interactive verbal response (IVR) to report their household environment and severity of violence, as measured by the modified Conflict Tactic Scale for men and women. At the end of the 12-week period, subjects completed an end-of-study interview, including reassessment of outcomes.
Measurement The baseline survey addressed demographics and basic information about the relationship (duration of the relationship, duration of the marriage, duration of abuse). In addition, it included the following measures. The 21-item version of the Abusive Behavior Inventory assessed the husband’s use of various control strategies [11]. The 17-item Adverse Childhood Experiences instrument assessed childhood emotional, physical and sexual abuse as well as 720
D. Katerndahl et al.
the level of violence and mental illness in the household when the subject was a child [12]. The 5-item Dyadic Splitting Scale assessed dysfunction in the relationship [13]. A single question from a health behaviours questionnaire assessed the husband’s and wife’s daily illicit drug use. Subjects reported the estimated days that they and their husbands used an illicit drug during the previous 30 days [14]. The 9-item Patient Health Questionnaire assessed the severity of major depression [15]. The 7-item Generalized Anxiety Disorder assessed generalized anxiety disorder [16]. The 15-item Patient Health Questionnaire-Panic Disorder assessed panic disorder [17]. Based upon prior study, IPV could impact subjects’ health, attitudes and/or behaviours [18]. Hence, we assessed attitudes and behaviours (hope, coping strategies, appraisal, support), as well as clinical outcomes (health care utilization, functional status, symptom levels, readiness for change) at baseline and at the end of the study. Hope was assessed using the 12-item Herth Hope Scale [19]. Coping was assessed using the 53-item COPE, which has 12 sub-scales measuring active versus avoidant coping mechanisms. Five scales measure problem-focused coping, four scales measure emotion-focused coping and three scales measure maladaptive coping [20]. Violence appraisal was assessed using the Appraisal Dimension Scales (ADS). This 24-item instrument assesses six dimensions of appraisal (control, salience, novelty, duration, causality and predictability). Appraisals were directed at a primary stressor (violence) [21]. Family support and stress were assessed using the 22-item Duke Social Support & Stress Scale [22]. Health Care Utilization was assessed using a 10-item health care utilization instrument which documented the number of visits to general medical (emergency departments, minor emergency rooms, clinics, medical doctor offices) and mental health (psychiatrists, psychologists, social workers, therapists) settings in the previous 2 months [23]. Functional status was measured using the Medical Outcomes Study Short Form-36 (MOS SF-36). The SF-36 consists of 36 items forming eight subscales and was used to assess functional status and quality of life [24]. Symptomatology was evaluated using the Biopsychosociospiritual Inventory. This 29-item instrument assesses physical, psychological, social and spiritual symptoms [25]. Finally, readiness for change was assessed using questions based on the transtheoretical model of change. This model states that behaviour change moves through five stages: precontemplation, contemplation, preparation, action and maintenance [26]. Subjects were asked to report their plans to (1) seek help for the violence; and (2) leave the relationship using a 5-point scale from ‘I am not interested’ to ‘I am trying to leave (get help) now’ [15].
Analysis Measures of daily violence severity were computed. First, the frequencies of husband- and wife-perpetrated violence were used. Second, the mean levels of episode severity for husbands and wives were calculated across all days in which any violence occurred. To quantitatively assess dynamics, time series data must be complete. Daily reports were made on a mean of 63.2 (± 15.9 SD) days with 50% of subjects reporting on 80% of days or more; missing data was imputed using the nstep procedure in the TISEAN software package [27] to maintain any non-linear
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
characteristics. The nstep approach to imputation has been shown to least distort non-linear characteristics of time series when compared with traditional methods [28,29]. When the initial data points in the time series were insufficient to apply nstep (generally n ≤ 4), the mode was inserted until the time series was long enough to use nstep.
Assessment of non-linearity Three types of non-linearity measurements are available and we used one example of each type [30]. Algorithmic complexity (a measure of the amount of information needed to describe the data) was measured by LZ complexity [31]. Regularity (or the lack of it) was measured by approximate entropy (ApEn) [32]. Finally, sensitivity to initial conditions (speed with which two adjacent points diverge over time) was measured with the largest Lyapunov’s exponent [33]. Using the time series of the daily assessments of severity of husband-to-wife violence, we calculated LZ complexity and Lyapunov exponents using the Chaos Data Analyzer software (Physics Academio Software, Raleigh, NC, USA) for each subject’s time series. We calculated approximate entropy using the ApEn module in OCTAVE. Stable estimates requires as few as 50 data points for approximate entropy [34–36] and 30 data points for LZ complexity [37]; Lyapunov exponents are resistant to the effects of missing data if non-linearly imputed data (≤15%) were used in datasets with underlying periodic or chaotic dynamics [29]. ‘Optimal (mid-range) non-linearity’ were derived for each non-linearity measure using absolute values of Z-transformed measures.
Do violence dynamics matter?
Table 1 Outcome factors Attitudinal/behavioural outcomes Variable (factor loading)
Clinical outcomes Variable (factor loading)
Positive coping (COPE) Planning (0.842) Active coping (0.723) Reinterpretation (0.599) Negative coping (COPE) Denial (0.811) Behavioural disengagement (0.796) Mental disengagement (0.655) Acceptance (0.598) Positive appraisal (ADS) Importance (−0.750) Not temporary (−0.748) Control (0.630) Negative appraisal (ADS) Novelty (0.790) Predictable (−0.537) Not self-caused (0.510) Hope and support Seeks support (COPE) Emotional reasons (0.814) Instrumental reasons (0.538) Total social support (DUSOCS) (0.635) Hope (HHI) (0.319)
Symptoms and functioning Psychological symptoms (BIOPSSI) (0.837) Poor mental health (SF-36) (0.819) Physical symptoms (BIOPSSI) (0.795) Social functioning (SF-36) (−0.757) Emotional role functioning (SF-36) (−0.705) Physical functioning (SF-36) (−0.588) Medical utilization Primary care settings (0.736) Emergency settings (0.688) Readiness for change To leave relationship (0.840) To seek help (0.720) Social symptoms and mental health care Mental health utilization (−0.764) Social symptoms (BIOPSSI) (0.648)
Assessment of dynamic pattern Assignment of dynamic patterns (periodic, chaotic, random) was made based upon whether the Lyapunov exponent was positive or negative (presence or absence of sensitivity to initial conditions), and whether the correlation dimension (presence or absence of a low-dimensional attractor) saturated with increasing embedding dimensions. A periodic pattern would have a negative Lyapunov exponent (not sensitive to initial conditions) and the presence of an attractor. A chaotic pattern would have a positive Lyapunov exponent (sensitive to initial conditions), and the presence of an attractor. A random pattern would have a positive Lyapunov exponent, but no attractor. However, this method was designed for analysing long time series (n > 1000) and thus should be interpreted with caution when smaller datasets are used. Dynamic patterns have been assigned using datasets with 100 data points [38]; studies of corporate innovations have used 50, 74 and 102 data points [39].
Relationship between dynamics and outcomes To assess differences in outcomes across dynamic patterns, we used one-way analysis of variance with Ryan-Einot-GabrielWelsch F post-hoc testing. Groups were defined by their dynamic pattern: periodic, chaotic or random. To reduce the number of outcomes for regression analyses, we conducted principal components factor analysis with varimax rotation for attitudinal/ behavioural and clinical outcomes. Factor loadings ≥0.3 were considered significant. Factor analysis resulted in five attitudinal/
© 2014 John Wiley & Sons, Ltd.
behavioural and four clinical outcome factors; measures of these factors consisted of their factor scores (see Table 1). Staged multiple regression was used to assess the relationship between nonlinearity measures, their optimal estimates and outcomes in a two-step process. First, only childhood, mental health and relationship variables that were significantly (P ≤ 0.25) correlated with each outcome were retained for possible inclusion in each regression analysis. Second, a staged regression analysis was conducted with each outcome factor using stepwise methods to minimize collinearity effects. In the first stage, demographic variables (age, Hispanic ethnicity, socio-economic status score [40]) were entered. In the second stage, significant childhood, mental health, marital and household environmental variables were entered. For clinical outcome regressions, significant attitudinal/behavioural outcomes were entered next. Then, violence characteristics (frequency, mean episode severity, and the three non-linearity measures) were entered. To then determine whether an optimal level of non-linearity would contribute to the remaining variance, the three optimal measures of non-linearity were entered. A 0.05 < P-value ≤ 0.10 was deemed trending towards significance.
Results The sample was predominantly low income and Hispanic. Many (43%) women were in common law marriages, and reported relationship duration of 9.6 years [± 8.9 (SD) years]. Duration of abuse was 5.5 (± 6.5 SD) years with age-of-abuse-onset of 32.6 (± 11.3 SD) years. 721
Do violence dynamics matter?
Outcome Attitudinal/behavioural outcomes Appraisal of violence Control Importance Novelty Not temporary Not self-caused Predictable Hope Total support Coping Active coping Planning Seeks support for instrumental reasons Seeks support for emotional reasons Reinterpretation Acceptance Denial Behavioural disengagement Mental disengagement Disengagement via substances Clinical outcomes Readiness To seek help To leave relationship Symptomatology Physical symptoms Psychological symptoms Social symptoms Health care utilization Primary care settings Mental health settings Emergency settings Functional status Physical functioning Emotional role functioning Social functioning Poor mental health AB
D. Katerndahl et al.
Periodic (n = 16)
Chaotic (n = 40)
Random (n = 79)
F (P)
10.56B 12.94A 7.25 9.47 7.50B 8.31B 36.8 8.38
10.69B 10.45B 6.59 8.36 9.13A 8.11B 36.1 9.26
9.55A 11.36B 5.81 9.12 8.11B 6.88A 35.4 8.91
4.01(.020) 3.84(.024) 1.43(.243 1.05(.354) 2.42(.093) 5.31(.006) 0.64(.531) 0.22(.800)
8.91A 8.29 6.86
6.66B 7.28 5.20
6.81B 7.58 6.00
2.52(.086) 0.44(.642) 1.09(.341)
8.00
6.06
6.36
1.50(.228)
8.29 7.43 5.07 4.15 6.62 0.46
7.57 6.74 3.60 4.31 6.50 0.53
7.13 7.52 4.75 5.59 7.00 0.69
0.81(.449) 0.61(.544) 1.13(.326) 2.10(.127) 0.40(.673) 0.45(.637)
2.63B 0.69
1.44A 0.43
2.35B 0.63
5.45(.005) 0.78(.459)
18.9 11.3B 23.2 1.54AB 1.93A 1.07 12.5 0.94B 4.88 13.3
15.5 8.7A 23.7
17.5 10.7B 24.2
Table 2 Outcomes across dynamic patterns (mean)
0.78(.459) 3.07(.050) 0.09(.915)
2.20B 0.36B 1.18
1.27A 0.79B 0.48
3.29(.041) 3.66(.029) 1.34(.265)
14.3 1.60A 5.40 11.2
13.7 1.01B 4.62 12.6
0.47(.628) 3.12(.047) 1.39(.252) 1.57(.213)
= Post-hoc differences.
Table 2 presents differences in outcomes across dynamic patterns. Based upon post-hoc analysis, those women in periodically violent relationships believed the violence as most important, compared with women in non-periodic relationships, and were most likely to use active coping, while those in chaotic relationships were most likely to believe that they were not the cause of the violence. Women in relationships with random violence were least likely to believe that the violence was predictable or controllable. Among clinical outcomes, several differences were observed. In three of the five outcomes in which significant differences were found, those with chaotic violence differed from the others; these women reported fewer psychological symptoms, better emotional role functioning and less willingness to seek help. Table 3 presents the results of the regression analyses for the attitudinal/behavioural outcomes. For two outcomes (positive 722
coping and negative appraisal), the combination of childhood, mental health, relationship and violence variables failed to account for significant variance. However, in the remaining three outcomes, the degree of violence non-linearity (ApEn and Lyapunov exponent) contributed significantly to the models beyond violence frequency and episode severity. Although all four models of clinical outcomes were significant (see Table 4), in none did violence non-linearity contribute; the only violence variable that was significant was that violence frequency was related to low social symptoms with high mental health utilization. To evaluate whether the non-linearity–outcome relationships were curvilinear with a mid-range optimal level of non-linearity, each regression model discussed above was rerun with the three optimal non-linearity estimates added. Optimal approximate
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
Do violence dynamics matter?
Table 3 Regression analyses of attitudinal/behavioural outcomes [β (P)] Coping Predictors Childhood Emotional abuse Physical abuse Sexual abuse Witnessing violence Family dysfunction Mental health Illegal drug use Women’s use Men’s use Depression severity Anxiety severity Panic disorder Relationship Marital characteristics Number of children Duration of relationship Duration of marriage Abuse duration Forced into sex Household environment Coercion Intimidation Emotional abuse Isolation Minimization Use of children Male privilege Economic Dyadic splitting Husband’s violence Frequency Episode severity Non-linearity Approximate entropy LZ complexity Lyapunov exponent Model F (P-value) Adjusted R2
Positive
Appraisal Negative
Positive
Negative
Hope and support
−0.144 (0.115)
−0.173 (0.047) 0.293 (0.004)
−0.210 (0.046) −0.201 (0.072) −0.148 (0.164) −0.136 (0.146)
−0.160 (0.055)
−0.252 (0.014)
0.274 (0.004)
−0.250 (0.005) −0.205 (0.027) −0.180 (0.086) −0.280 (0.009)
0.232 (0.009)
1.98 (0.140) 0.039
.226 (.014)
0.179 (0.038)
4.49 (0.000) 0.197
6.44 (0.000) 0.259
0.079 (0.971) −0.027
4.23 (0.001) 0.232
Adjusted for age, socio-economic status and race/ethnicity (Hispanic).
entropy was related to both symptoms and functioning (β = 0.092, P = 0.077) and readiness for change (β = 0.167, P = 0.033).
Discussion Women experiencing periodic violence recognized its importance and used their active coping to seek mental health care. Those with chaotic dynamics recognized that they were not responsible, experienced fewer psychological symptoms and emotional role limitations, and did not seek help. Those experiencing random violence recognized its unpredictability and uncontrollability. Measures of non-linearity contributed to five outcome factors.
© 2014 John Wiley & Sons, Ltd.
Attitudinal/behavioural outcomes In terms of coping, regression analysis did little to explain positive coping. While this study found an association with the husband’s use of control [41], other studies have found that problem-focused coping is positively related to the amount and severity of violence [41] and may lessen perceived distress [42]. Understanding negative coping was more complicated. Regression analysis found that Lyapunov non-linearity was an independent predictor. IPV has previously been linked to such negative coping strategies as acceptance, avoidance, denial and detachment [43]. While this study found that negative coping was related to husband’s drug 723
Do violence dynamics matter?
D. Katerndahl et al.
Table 4 Regression analyses of clinical outcomes
Predictors Childhood Emotional abuse Physical abuse Sexual abuse Witnessing violence Family dysfunction Mental health Illegal drug use Women’s use Men’s use Depression severity Anxiety severity Panic disorder Relationship Marital characteristics Number of children Duration of relationship Duration of marriage Abuse duration Age-of-violence onset Forced into sex Household environment Coercion Intimidation Emotional abuse Isolation Minimization Use of children Male privilege Economic Dyadic splitting Attitudinal/behavioural outcomes Appraised importance Seeks support for emotional reasons Hope Total support Husband’s violence Frequency Episode severity Non-linearity Approximate entropy LZ complexity Lyapunov exponent Model F (P-value) Adjusted R2
Symptoms and functioning
Medical utilization
Readiness for change
Social symptoms and mental health utilization
0.247 (0.011) 0.149 (0.028)
−0.176 (0.054) 0.142 (0.160)
0.549 (0.000)
0.037 (0.688)
0.236 (0.001)
−0.339 (0.001)
0.209 (0.080)
0.176 (0.009)
0.118 (0.210)
0.165 (0.061)
0.187 (0.057) −0.387 (0.000) −0.230 (0.011) −0.271 (0.011) −0.151 (0.080)
23.41 (0.000) 0.671
5.62 (0.000) 0.157
4.91 (0.000) 0.240
8.83 (0.000) 0.445
Adjusted for age, socio-economic status and race/ethnicity (Hispanic).
use, wife’s depression and dyadic splitting, others also found a relationship with depression [44]. Although a relationship between childhood abuse and negative coping was found in this study [45], that relationship was lost when other factors were entered. In terms of violence appraisal, negative appraisal could not be explained via regression analysis. Negative expectations have been linked to partner-reported IPV [46]. Although self-blame is common in IPV [47], such negative appraisals have not been well 724
studied. Understanding positive appraisal is more complicated with regression analysis finding that two forms of violence nonlinearity (ApEn and Lyapunov exponent) independently predicted positive coping. Positive expectancies can lead to anger and relational adjustment [46], possibly related to the associations with coercion and dyadic splitting noted in this study. When modelling hope and support, husband’s drug use, wife’s depression and violence severity were inversely related to hope
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
and support, so too was ApEn; yet, optimal ApEn contributed significantly as well. Prior IPV research has shown that not only are women with adequate social support less likely to have a history of IPV, but those with such a history are less likely to be involved in a second violent relationship if they have support [48]. In fact, long-term abuse is decreased when supportive individuals provide information that improves access to resources [49]. Women in more severely violent relationships may use caution when seeking support due to family, cultural and societal sanctions, fear, sense of isolation, sense of shame or a lack of perceived benefit when seeking that support [50].
Clinical outcomes Medical utilization was unrelated to violence in this study of women in violent relationships; it was associated only with a history of child sexual abuse [51]. Fletcher [52] reported that the association between childhood abuse and health care utilization, as noted in our regression analysis, was no longer observed when adjusted for confounders. Previous work has suggested that physical IPV is linked to higher medical utilization [53], which continued even after the IPV stopped [54]. Among women in a battered women’s shelter, social support intervention reduced psychological distress and health care utilization [55]. Social symptoms and lack of mental health utilization depended upon attitudinal/behavioural outcomes; only violence infrequency predicted social symptoms. While social isolation is common in IPV [56], mental health service use is often increased [53]. Although self-imposed isolation and lack of mental health service utilization [57] could be due to concerns about disclosure [58], many women will use those services if offered [59]. Symptoms and dysfunction depended upon a variety of factors from childhood family dysfunction and current depression and panic disorder to use of coercion by the husband and age-of-violence onset. However, optimal ApEn also contributed. Although possibly inflated by confounders [52], associations may exist between IPV and both symptoms [60] and dysfunction [61]. While the relationship to depression found in this study was noted before [60], the links between coping [62], appraisal [63], support [56] and symptoms or dysfunction found previously were not seen. Readiness for change was related to appraised violence importance, short relationship duration, and optimal ApEn. The current study measured factors previously associated with help seeking in IPV – prior trauma [57,64], coping [65], forced sex, witnessing childhood violence [64], appraisal [66], presence of children [67], social support [68] and abuse severity [67]. We also assessed factors associated with leaving the relationship – alcohol intake [69], presence of children [70], witnessing childhood violence [71], financial independence [71], relationship quality [72], social support [73], lack of forgiveness [74], duration of abuse [72] and increasing violence [75]. But these predictors were not found to be related to readiness for change in this study. However, appraised importance was a significant predictor of readiness for change as previously found [75]. In addition, the fact that duration of relationship is inversely related to readiness in this study may reflect willingness for action in women who have a lower investment in the relationship [76].
© 2014 John Wiley & Sons, Ltd.
Do violence dynamics matter?
Implications Although a potential straightforward, positive relationship between non-linearity and positive outcomes may exist, its characteristic unpredictability may explain why the non-linearity of violence can be associated with both positive appraisal (violence as temporary, unimportant and controllable) and negative coping, with positive appraisal and lack of hope or support. Non-linearity as a continuous measure predicts three attitudinal/behavioural outcomes only, while optimal non-linearity predicts two clinical outcomes only, suggesting a curvilinear relationship with clinical outcomes. Thus, for attitudinal/behavioural outcomes, unpredictability may directly impact coping, appraisal and hope, but for clinical outcomes, there exist optimal levels of predictability/ unpredictability. Readiness for change may be sensitive to the interplay between constraints-limiting activity and resourcesenabling activity [76] in a process of decisional balance [77]. The importance of non-linearity with its unpredictability in five such outcomes in this study reinforces the need to individualize management and coping strategies [78].
Limitations This study has several limitations. First, the sample size is small for time series analysis, especially for determining dynamic pattern. Second, it is unclear whether women accurately report the level of violence perpetrated by their partners or themselves. However, Regan et al. [79] found that violence reports by husbands and wives were highly correlated. Finally, the predominance of Hispanics within the sample may limit the generalizability of the findings. Caetano et al. [80] found that, over a 5-year period, Hispanics were 2.5 times as likely as Anglos to initiate IPV, while reporting a violence recurrence rate four times higher. In addition, Hispanic women in violent relationships may differ from other ethnic women in their use of emergency departments [81] and avoiding IPV disclosure to providers [82]. In fact, acculturation is important in help seeking for such women [81].
Conclusion Measures of violence non-linearity contributed to the variance of three attitudinal/behavioural outcomes, and optimal non-linearity contributed to two clinical outcomes. Knowledge of violence dynamics may have applications when working with violent couples.
Acknowledgements This study was funded by the National Science Foundation (#0826812). We wish to thank Stephanie Mitchell, Kelli Giacomini, Robert Mesec and Wilson Pace at the University of Colorado, Department of Family Medicine for their invaluable assistance.
References 1. Katerndahl, D., Burge, S., Ferrer, R., Becho, J. & Wood, R. (2010) Complex dynamics in intimate partner violence. Primary Care Companion to the Journal of Clinical Psychiatry, 12 (4), e1–e12.
725
Do violence dynamics matter?
2. Albrecht, G., Freeman, S. & Higginbotham, N. (1998) Complexity and human health. Culture, Medicine and Psychiatry, 22 (1), 55–92. 3. Wallace, R. & Wallace, D. (2004) Structured psychosocial stress and therapeutic failure. Journal of Biological Systems, 12, 335–369. 4. Waldrop, M. M. (1992) Complexity: The Emerging Science at the Edge of Order and Chaos. New York: Simon & Schuster. 5. Hirsch, I. B. & Brownlee, M. (2005) Should minimal blood glucose variability become the gold standard of glycemic control? Journal of Diabetes and its Complications, 19, 178–181. 6. Harbourne, R. T., Deffeyes, J. E., Kyvelidou, A. & Stergiou, N. (2009) Complexity of postural control in infants. Nonlinear Dynamics, Psychology, and Life Sciences, 13, 123–144. 7. Olson, D. H. & Gorall, D. M. (2003) Circumplex model of marital and family systems. In Normal Family Processes, 3rd edn (ed. F. Walsh), pp. 514–547. New York: Guilford Press. 8. Guastello, S. J. (2012) Self-complexity and the interpretation of entropy and information statistics. Society for Chaos Theory in Psychology and Life Sciences Newsletter, 19 (2), 5–10. 9. Burge, S. K., Becho, J., Ferrer, R. L., Wood, R. C., Talamantes, M. & Katerndahl, D. A. (2014) Safely examining complex dynamics of intimate partner violence. Families, Systems and Health: The Journal of Collaborative Family Healthcare, (in press). 10. Straus, M., Gelles, R. & Steinmetz, S. (1980) Behind Closed Doors. New York: Anchor Books. 11. Shepard, M. F. & Campbell, J. A. (1992) The Abusive Behavior Inventory. Journal of Interpersonal Violence, 7, 291–305. 12. Felitti, V. J., Anda, R. F., Nordenberg, D., et al. (1998) Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: the Adverse Childhood Experiences (ACE) study. American Journal of Preventive Medicine, 14, 245–258. 13. Siegel, J. P. (2006) Dyadic splitting in partner relational disorders. Journal of Family Psychology, 20, 418–422. 14. Glasgow, R. E., Ory, M. G., Klesges, L. M., Cifuentes, M., Fernald, D. H. & Green, L. A. (2005) Practical and relevant self-report measures of patient health behaviors for primary care research. Annals of Family Medicine, 3, 73–81. 15. Nease, D. E. & Malouin, J. M. (2003) Depression screening. The Journal of Family Practice, 52, 118–126. 16. Spitzer, R. L., Kroenke, K., Williams, J. B. W. & Lowe, B. (2006) Brief measure for assessing generalized anxiety disorder: GAD-7. Archives of Internal Medicine, 166, 1092–1097. 17. Spitzer, R. L., Kroenke, K. & Williams, J. B. (1999) Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study: primary care evaluation of mental disorders: patient health questionnaire. Journal of the American Medical Association, 282, 1737–1744. 18. Koss, M. P. & Heslet, L. (1992) Somatic consequences of violence against women. Archives of Family Medicine, 1, 53–59. 19. Herth, K. A. (1992) An abbreviated instrument to measure hope. Journal of Advanced Nursing, 17, 1251–1259. 20. Carver, C. S., Scheier, M. F. & Weintraub, J. K. (1989) Assessing coping strategies: a theoretically based approach. Journal of Personality and Social Psychology, 56, 267–283. 21. Vitaliano, P. P. (1985) Manual for Appraisal Dimension Scale and Revised Ways of Coping Checklist. Seattle: Department of Psychiatry, University of Washington. 22. Duke University (1986) Duke Social Support and Stress Scale. Durham: Duke University. 23. Katerndahl, D. A. & Realini, J. P. (1997) Use of health care services by persons with panic symptoms. Psychiatric Services: A Journal of the American Psychiatric Association, 48, 1027–1032. 24. Ware, J. E. Jr & Sherbourne, C. D. (1992) The MOS 36-item short form health survey (SF-36), I. Conceptual framework and item selection. Medical Care, 30, 473–483.
726
D. Katerndahl et al.
25. Katerndahl, D. & Oyiriaru, D. (2007) Assessing the biopsychosociospiritual model in primary care. International Journal of Psychiatry in Medicine, 37, 393–414. 26. Prochaska, J. O. & DiClemente, C. C. (1986) Toward a comprehensive model of change. In Treating Addictive Behaviors (eds W. R. Miller & N. Heather), pp. 3–27. New York: Plenum Press. 27. Heggler, R., Kantz, H. & Schreiber, T. (1999) Practical implementation of nonlinear time series methods: the TISEAN package. Chaos (Woodbury, N.Y.), 9, 413–435. 28. Kreindler, D. M. & Lumsden, C. J. (2006) Effects of the irregular sample and missing data in time series analysis. Nonlinear Dynamics, Psychology, and Life Sciences, 10, 187–214. 29. Kreindler, D. M. & Lumsden, C. J. (2007) Effects of irregular sampling and missing data on largest Lyapunov exponents. Nonlinear Dynamics, Psychology, and Life Sciences, 11, 401–412. 30. Land, B. & Elias, D. (2005) Measuring the ‘complexity’ of a time series. Available at: http://www.nbb.cornell.edu/neurobio/land/ PROJECTS/Complexity/index.html (last accessed 22 April 2014). 31. Ziv, J. & Lempel, A. (1978) Compression of individual sequences via variable-rate coding. IEEE Transactions on Information Theory, IT-24, 530–536. 32. Pincus, S. M. (2006) Approximate entropy as a measure of irregularity for psychiatric serial metrics. Bipolar Disorders, 8, 430–440. 33. Wolf, A., Swift, J. B., Swinney, H. L. & Vastano, J. A. (1985) Determining Lyapunov exponents from a time series. Physica D. Nonlinear Phenomena, 16, 285–317. 34. Gevers, E., Pincus, S. M., Robinson, I. C. A. F. & Veldhuis, J. D. (1998) Differential orderliness of the GH release process in castrate male and female rats. The American Journal of Physiology, 274, R437–R444. 35. Yeragani, V. K., Pohl, R., Mallavarapu, M. & Balon, R. (2003) Approximate entropy of symptoms of mood. Bipolar Disorders, 5, 279–286. 36. Pincus, S. M., Hartman, M. L., Roelfsema, F., Thorner, M. O. & Veldhuis, J. D. (1999) Hormone pulsatility discrimination via coarse and short-time sampling. The American Journal of Physiology, 277, E948–E957. 37. Zhang, X. S., Zhu, Y. S., Thakor, N. V. & Wang, Z. Z. (1999) Detecting ventricular tachycardia and fibrillation by complexity measure. IEEE Transactions on Biomedical Engineering, 46, 548–555. 38. Rosenstein, M., Collins, J. & DeLuca, C. (1993) Practical method of calculating largest Lyapunov exponents from small data sets. Physica D. Nonlinear Phenomena, 65, 117–134. 39. Cheng, Y. T. & Van de Ven, A. H. (1996) Learning the innovation journey: order out of chaos? Organization Science, 7 (6), 593–614. 40. Hollingshead, A. & Redlich, F. (1958) Social Class and Mental Illness. New York: John Wiley & Sons. 41. Sabina, C. & Tindale, R. S. (2008) Abuse characteristics and coping resources as predictors of problem-focused coping strategies among battered women. Violence Against Women, 14, 437–456. 42. Yoshihama, M. (2002) Battered women’s coping strategies and psychological distress. American Journal of Community Psychology, 30, 429–452. 43. Orzeck, T. L., Rokach, A. & Chin, J. (2010) Effects of traumatic and abusive relationships. Journal of Loss and Trauma, 15, 167– 192. 44. Calvete, E., Corral, S. & Estevez, A. (2007) Cognitive and coping mechanisms in the interplay between intimate partner violence and depression. Anxiety, Stress and Coping: An International Journal, 20, 369–382. 45. Taft, C. T., Resick, P. A., Panuzio, J., Vogt, D. S. & Mechanic, M. B. (2007) Examining the correlates of engagement and disengagement coping among help-seeking battered women. Violence and Victims, 22, 3–17.
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
46. Meis, L. A., Murphy, C. M. & Winters, J. J. (2010) Outcome expectancies of partner abuse. Assessment, 17, 30–43. 47. Carlson, B. E. (1997) Stress and coping approach to intervention with abused women. Family Relations, 46, 291–298. 48. Plazaola-Castano, J., Ruiz-Perez, I. & Montero-Pinar, M. I. (2008) Grupo de studio para la violencia de genero. Gaceta Sanitaria, 22, 527–533. 49. Bosch, K. & Schumm, W. R. (2004) Accessibility to resources. Journal of Sex and Marital Therapy, 30, 357–370. 50. Rose, L. E., Campbell, J. & Kub, J. (2000) Role of social support and family relationships in women’s responses to battering. Health Care for Women International, 21, 27–39. 51. Chou, K. L. (2012) Childhood sexual abuse and psychiatric disorders in middle-aged and older adults. The Journal of Clinical Psychiatry, 73, e1365–e1371. 52. Fletcher, J. (2010) Effects of intimate partner violence on health in young adulthood in the United States. Social Science and Medicine, 70, 130–135. 53. Bonomi, A. E., Anderson, M. L., Rivara, F. P. & Thompson, R. S. (2009) Health care utilization and cost as associated with physical and nonphysical-only intimate partner violence. Health Services Research, 44, 1052–1067. 54. Rivara, F. P., Anderson, M. L., Fishman, P., Bonomi, A. E., Reid, R. J., Carrell, D. & Thompson, R. S. (2007) Healthcare utilization and costs for women with a history of intimate partner violence. American Journal of Preventive Medicine, 32, 89–96. 55. Constantino, R., Kim, Y. & Crane, P. A. (2005) Effects of a social support intervention on health outcomes in residents of a domestic violence shelter. Issues in Mental Health Nursing, 26, 575–590. 56. Latta, R. E. (2009) Struggling to define my role. Dissertation Abstracts International: Section B: The Sciences and Engineering, 69 (9-B), 5783. 57. Johnson, D. M. & Zlotnick, C. (2007) Utilization of mental health treatment and other services by battered women in shelters. Psychiatric Services: A Journal of the American Psychiatric Association, 58, 1595–1597. 58. Trotter, J. L. (2009) Dynamic investigation of domestic violence survivors’ experiences with their informal social networks. Dissertation Abstracts International: Section B: The Sciences and Engineering, 69 (11-B), 7154. 59. Cerulli, C., Gellman, R. A., Nichols, C., Hall, D., Conner, K. R. & Caine, E. D. (2011) Mental and physical health symptoms of family court intimate partner violence petitioners. International Journal of Law and Psychiatry, 34, 94–98. 60. Matud, M. P. (2005) Psychological impact of domestic violence on Spanish women. Journal of Applied Social Psychology, 35, 2310– 2322. 61. Straus, H., Cerulli, C., McNutt, L. A., Rhodes, K. V., Conner, K. R., Kemball, R. S., Kaslow, N. J. & Houry, D. (2009) Intimate partner violence and functional health status. Journal of Women’s Health, 18, 625–631. 62. Calvete, E., Corral, S. & Estevez, A. (2008) Coping as a mediator and moderator between intimate partner violence and symptoms of anxiety and depression. Violence Against Women, 14, 886–904. 63. Martinez-Torteya, C., Bogat, G. A., Von Eye, A., Levendosky, A. A. & Davidson, W. S. (2009) Women’s appraisal of intimate partner violence stressfulness and their relationship to depressive and posttraumatic stress disorder symptoms. Violence and Victims, 24, 707–722.
© 2014 John Wiley & Sons, Ltd.
Do violence dynamics matter?
64. Henning, K. R. & Klesges, L. M. (2002) Utilization of counseling and supportive services by female victims of domestic abuse. Violence and Victims, 17, 623–636. 65. Shannon, L., Logan, T. K., Cole, J. & Medley, K. (2006) Help-seeking and coping strategies for intimate partner violence in rural and urban women. Violence and Victims, 21, 167–181. 66. Petersen, R., Moracco, K. E., Goldstein, K. M. & Clark, K. A. (2004) Moving beyond disclosure. Women’s Health, 40, 63–76. 67. Djikanovic, B., Wong, S. L. F., Jansen, H. A., Koso, S., Simic, S., Otasevic, S. & Lagro-Janssen, A. (2012) Help-seeking behavior of women who experienced intimate partner violence. Family Practice, 29 (2), 189–195. Available at: http://www.ncbi.nlm.nih.gov/pubmed/ 21896506 (last accessed 18 June 2014). 68. Wright, C. V. & Johnson, D. M. (2009) Correlates for legal helpseeking. Violence and Victims, 24, 771–785. 69. Ramisetty-Mikler, S. & Caetano, R. (2005) Alcohol use and intimate partner violence as predictors of separation among U.S. couples. Journal of Studies on Alcohol, 66, 205–212. 70. Rhodes, K. V., Cerulli, C., Dichter, M. E., Kothari, C. L. & Barg, F. K. (2010) ‘I didn’t want to put them through that’. Journal of Family Violence, 25, 485–493. 71. Kim, J. & Gray, K. A. (2008) Leave or stay? Journal of Interpersonal Violence, 23, 1465–1482. 72. Ulrich, Y. C. (1998) What helped most in leaving spouse abuse. In Empowering Survivors of Abuse (ed. J. C. Campbell), pp. 70–78. Thousand Oaks: Sage Publications. 73. Wilcox, P. (2000) ‘Me mother’s bank and me nanan’s, you know, support!’. Women’s Studies International Forum, 23, 35–47. 74. Gordon, K. C., Burton, S. & Porter, L. (2004) Predicting the intentions of women in domestic violence shelters to return to partners. Journal of Family Psychology, 18, 331–338. 75. Chang, J. C., Dado, D., Hawker, L., Cluss, P. A., Buranosky, R., Slagel, L., McNeil, M. & Scholle, S. H. (2010) Understanding turning points in intimate partner violence. Journal of Women’s Health, 19, 251–259. 76. Burton, S. J. (2004) To go or not to go, that is the question. Dissertation Abstracts International: Section B: The Sciences and Engineering, 64 (9-B), 4605. 77. Burke, J. G., Denison, J. A., Gielen, A. C., McDonnell, K. A. & O’Campo, P. (2004) Ending intimate partner violence. American Journal of Health Behavior, 28, 122–133. 78. Hetzel-Riggin, M. D. & Meads, C. L. (2011) Childhood violence and adult partner mistreatment. Journal of Family Violence, 26, 585–593. 79. Regan, K. V., Bartholomew, K., Kwong, M. J., Trinke, S. J. & Henderson, A. J. Z. (2006) Relative severity of acts of physical violence in heterosexual relationships. Personal Relationships, 13, 37–52. 80. Caetano, R., Field, C. A., Ramisetty-Mikler, S. & McGrath, C. (2005) 5-year course of intimate partner violence among white, black, and Hispanic couples in the United States. Journal of Interpersonal Violence, 20, 1039–1057. 81. Lipsky, S., Caetano, R., Field, C. A. & Larkin, G. L. (2006) Role of intimate partner violence, race, and ethnicity in help-seeking behaviors. Ethnicity and Health, 11, 81–100. 82. Kelly, U. A. (2010) Symptoms of PTSD and major depression in Latinas who have experienced intimate partner violence. Issues in Mental Health Nursing, 3, 119–127.
727
Journal of Evaluation in Clinical Practice ISSN 1365-2753
Do violence dynamics matter? David Katerndahl MD MA,1 Sandra Burge PhD,1 Robert Ferrer MD MPH,1 Johanna Becho BA2 and Robert Wood DrPH3 1 Professor, 2Research Associate, 3Biostatistician, Department of Family & Community Medicine, University of Texas Health Science Center at San Antonio, San Antonio, Texas, USA
Keywords battered women, domestic violence, intimate partner violence, non-linear dynamics, outcomes research, systems theory Correspondence Dr David Katerndahl Family & Community Medicine University of Texas Health Science Center at San Antonio 7703 Floyd Curl Drive San Antonio, TX 78229 USA E-mail: [email protected] Accepted for publication: 22 May 2014 doi:10.1111/jep.12216
Abstract Rationale, aims and objectives Intimate partner violence is a complex, non-linear phenomenon. The purpose of this study was to determine whether violence dynamics (pattern, degree of non-linearity, optimal non-linearity) contributed to outcomes in violent relationships. Methods The study was conducted in six primary care clinics, enrolling 200 adult women in violent relationships. In addition to baseline and end-of-study interviews, women completed daily telephone assessments of household environment and partner violence using interactive verbal response. Three non-linearity measures of violence were computed with ‘optimal’ non-linearity estimated using Z-transformations. Assignment of dynamic patterns (periodic, chaotic, random) was made based upon Lyapunov exponent and correlation dimension. Outcomes across dynamic patterns were analysed using analysis of variance. In addition, stepped multiple linear regression explained factor-analysed outcomes, adjusting for demographic, childhood, mental health and marital variables; attitudinal/behavioural outcomes were also adjusted for when explaining clinical outcomes. Results Women experiencing periodic violence recognized the importance of violence and used their active coping to seek mental health care. Those with chaotic dynamics recognized that they were not responsible, experienced fewer psychological symptoms and emotional role limitations, and did not seek help. Those experiencing random violence recognized its unpredictability and uncontrollability. Violence non-linearity predicted negative coping, positive appraisals and hope/support in regression analyses, while optimal non-linearity contributed to readiness for change and symptoms functioning. Of the nine outcomes investigated, violence non-linearity contributed to five outcomes. Conclusion Dynamic pattern of violence, degree of violence non-linearity and optimal non-linearity correlated with several attitudinal/behavioural and clinical outcomes. Knowledge of violence dynamics may have applications when working with violent couples.
Introduction Intimate partner violence (IPV) is a non-linear phenomenon. Quantitatively, the day-to-day level of violence displays non-linear dynamics in most relationships. Although the dynamic pattern may speak to aetiological factors [1], is non-linearity important to patient outcomes? An emerging belief is that non-linearity may be critical to health and well-being [2]. When biological systems are healthy, they behave in a non-linear way, exhibiting adaptability with resistance to external stressors that might disrupt their healthy dynamics [3].
However, when systems transition into periodicity due to illness, their trajectories become regular and potentially amenable to intervention, enabling clinicians to restore them to healthy, adaptable, non-linear dynamics. This transition from healthy non-linear dynamics to unhealthy linear dynamics can be understood by realizing that non-linear systems display a variety of dynamics, depending upon their resources and constraints, interconnectedness, and feedback. In the case of individual people, as the number of chronic medical problems increases, the patient’s resourcefulness, flexibility and adaptability (and, hence, nonlinearity) may decrease, leading to linear dynamics and poor
Journal of Evaluation in Clinical Practice 20 (2014) 719–727 © 2014 John Wiley & Sons, Ltd.
719
Do violence dynamics matter?
overall health. As a continuum from linear to non-linear dynamics, the constraining nature of disease severity, co-morbidity and chronic stress should suppress healthy trends towards nonlinearity and adaptability, thus leading to linear dynamics or rigidity. However, because adaptive systems display order and disorder strategically [4], there may exist an optimal degree of non-linearity between the minimum of periodicity and the maximum of randomness. Just as extreme variability may be detrimental within an illness [5], so too may extreme non-linearity be dysfunctional [6]. In the case of family systems, family health as a function of members’ communication, cohesion and flexibility [7]. On the flexibility continuum, extreme values are considered unhealthy. Inflexible families are rigid with strict rules and expectations for family members; they are linear, predictable and not adaptable to change. Overly flexible families are erratic with no rules, no clear roles, and no leadership; they are random, unpredictable and unstable. Thus, the relationship between system dynamics and outcome may be curvilinear [8]. We examined dynamic patterns of violence in heterosexual couple relationships to determine whether non-linearity of violence contributed to attitudes, behaviours or clinical outcomes in women. Specifically, do measures of non-linearity account for significant variance in outcomes? Is there a mid-range level of violence non-linearity that predicts these outcomes?
Methods Sample In this study, as previously described [9], 200 women with a recent history of husband-to-wife physical abuse were recruited from six primary care clinics in San Antonio, Texas. Adult (18–64 years old), non-pregnant women seen without their husbands present at the time of the visit were asked to complete the 6-item brief Conflict Tactic Scale [10] in the examination room while they waited to see their doctor. If they had experienced violence from their partner in the past 30 days, they next responded to a Danger Assessment Screen. Women in extremely dangerous relationships were excluded.
Procedure Once enrolled, subjects completed a baseline survey. For 12 weeks, they were asked to complete a daily telephone survey using interactive verbal response (IVR) to report their household environment and severity of violence, as measured by the modified Conflict Tactic Scale for men and women. At the end of the 12-week period, subjects completed an end-of-study interview, including reassessment of outcomes.
Measurement The baseline survey addressed demographics and basic information about the relationship (duration of the relationship, duration of the marriage, duration of abuse). In addition, it included the following measures. The 21-item version of the Abusive Behavior Inventory assessed the husband’s use of various control strategies [11]. The 17-item Adverse Childhood Experiences instrument assessed childhood emotional, physical and sexual abuse as well as 720
D. Katerndahl et al.
the level of violence and mental illness in the household when the subject was a child [12]. The 5-item Dyadic Splitting Scale assessed dysfunction in the relationship [13]. A single question from a health behaviours questionnaire assessed the husband’s and wife’s daily illicit drug use. Subjects reported the estimated days that they and their husbands used an illicit drug during the previous 30 days [14]. The 9-item Patient Health Questionnaire assessed the severity of major depression [15]. The 7-item Generalized Anxiety Disorder assessed generalized anxiety disorder [16]. The 15-item Patient Health Questionnaire-Panic Disorder assessed panic disorder [17]. Based upon prior study, IPV could impact subjects’ health, attitudes and/or behaviours [18]. Hence, we assessed attitudes and behaviours (hope, coping strategies, appraisal, support), as well as clinical outcomes (health care utilization, functional status, symptom levels, readiness for change) at baseline and at the end of the study. Hope was assessed using the 12-item Herth Hope Scale [19]. Coping was assessed using the 53-item COPE, which has 12 sub-scales measuring active versus avoidant coping mechanisms. Five scales measure problem-focused coping, four scales measure emotion-focused coping and three scales measure maladaptive coping [20]. Violence appraisal was assessed using the Appraisal Dimension Scales (ADS). This 24-item instrument assesses six dimensions of appraisal (control, salience, novelty, duration, causality and predictability). Appraisals were directed at a primary stressor (violence) [21]. Family support and stress were assessed using the 22-item Duke Social Support & Stress Scale [22]. Health Care Utilization was assessed using a 10-item health care utilization instrument which documented the number of visits to general medical (emergency departments, minor emergency rooms, clinics, medical doctor offices) and mental health (psychiatrists, psychologists, social workers, therapists) settings in the previous 2 months [23]. Functional status was measured using the Medical Outcomes Study Short Form-36 (MOS SF-36). The SF-36 consists of 36 items forming eight subscales and was used to assess functional status and quality of life [24]. Symptomatology was evaluated using the Biopsychosociospiritual Inventory. This 29-item instrument assesses physical, psychological, social and spiritual symptoms [25]. Finally, readiness for change was assessed using questions based on the transtheoretical model of change. This model states that behaviour change moves through five stages: precontemplation, contemplation, preparation, action and maintenance [26]. Subjects were asked to report their plans to (1) seek help for the violence; and (2) leave the relationship using a 5-point scale from ‘I am not interested’ to ‘I am trying to leave (get help) now’ [15].
Analysis Measures of daily violence severity were computed. First, the frequencies of husband- and wife-perpetrated violence were used. Second, the mean levels of episode severity for husbands and wives were calculated across all days in which any violence occurred. To quantitatively assess dynamics, time series data must be complete. Daily reports were made on a mean of 63.2 (± 15.9 SD) days with 50% of subjects reporting on 80% of days or more; missing data was imputed using the nstep procedure in the TISEAN software package [27] to maintain any non-linear
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
characteristics. The nstep approach to imputation has been shown to least distort non-linear characteristics of time series when compared with traditional methods [28,29]. When the initial data points in the time series were insufficient to apply nstep (generally n ≤ 4), the mode was inserted until the time series was long enough to use nstep.
Assessment of non-linearity Three types of non-linearity measurements are available and we used one example of each type [30]. Algorithmic complexity (a measure of the amount of information needed to describe the data) was measured by LZ complexity [31]. Regularity (or the lack of it) was measured by approximate entropy (ApEn) [32]. Finally, sensitivity to initial conditions (speed with which two adjacent points diverge over time) was measured with the largest Lyapunov’s exponent [33]. Using the time series of the daily assessments of severity of husband-to-wife violence, we calculated LZ complexity and Lyapunov exponents using the Chaos Data Analyzer software (Physics Academio Software, Raleigh, NC, USA) for each subject’s time series. We calculated approximate entropy using the ApEn module in OCTAVE. Stable estimates requires as few as 50 data points for approximate entropy [34–36] and 30 data points for LZ complexity [37]; Lyapunov exponents are resistant to the effects of missing data if non-linearly imputed data (≤15%) were used in datasets with underlying periodic or chaotic dynamics [29]. ‘Optimal (mid-range) non-linearity’ were derived for each non-linearity measure using absolute values of Z-transformed measures.
Do violence dynamics matter?
Table 1 Outcome factors Attitudinal/behavioural outcomes Variable (factor loading)
Clinical outcomes Variable (factor loading)
Positive coping (COPE) Planning (0.842) Active coping (0.723) Reinterpretation (0.599) Negative coping (COPE) Denial (0.811) Behavioural disengagement (0.796) Mental disengagement (0.655) Acceptance (0.598) Positive appraisal (ADS) Importance (−0.750) Not temporary (−0.748) Control (0.630) Negative appraisal (ADS) Novelty (0.790) Predictable (−0.537) Not self-caused (0.510) Hope and support Seeks support (COPE) Emotional reasons (0.814) Instrumental reasons (0.538) Total social support (DUSOCS) (0.635) Hope (HHI) (0.319)
Symptoms and functioning Psychological symptoms (BIOPSSI) (0.837) Poor mental health (SF-36) (0.819) Physical symptoms (BIOPSSI) (0.795) Social functioning (SF-36) (−0.757) Emotional role functioning (SF-36) (−0.705) Physical functioning (SF-36) (−0.588) Medical utilization Primary care settings (0.736) Emergency settings (0.688) Readiness for change To leave relationship (0.840) To seek help (0.720) Social symptoms and mental health care Mental health utilization (−0.764) Social symptoms (BIOPSSI) (0.648)
Assessment of dynamic pattern Assignment of dynamic patterns (periodic, chaotic, random) was made based upon whether the Lyapunov exponent was positive or negative (presence or absence of sensitivity to initial conditions), and whether the correlation dimension (presence or absence of a low-dimensional attractor) saturated with increasing embedding dimensions. A periodic pattern would have a negative Lyapunov exponent (not sensitive to initial conditions) and the presence of an attractor. A chaotic pattern would have a positive Lyapunov exponent (sensitive to initial conditions), and the presence of an attractor. A random pattern would have a positive Lyapunov exponent, but no attractor. However, this method was designed for analysing long time series (n > 1000) and thus should be interpreted with caution when smaller datasets are used. Dynamic patterns have been assigned using datasets with 100 data points [38]; studies of corporate innovations have used 50, 74 and 102 data points [39].
Relationship between dynamics and outcomes To assess differences in outcomes across dynamic patterns, we used one-way analysis of variance with Ryan-Einot-GabrielWelsch F post-hoc testing. Groups were defined by their dynamic pattern: periodic, chaotic or random. To reduce the number of outcomes for regression analyses, we conducted principal components factor analysis with varimax rotation for attitudinal/ behavioural and clinical outcomes. Factor loadings ≥0.3 were considered significant. Factor analysis resulted in five attitudinal/
© 2014 John Wiley & Sons, Ltd.
behavioural and four clinical outcome factors; measures of these factors consisted of their factor scores (see Table 1). Staged multiple regression was used to assess the relationship between nonlinearity measures, their optimal estimates and outcomes in a two-step process. First, only childhood, mental health and relationship variables that were significantly (P ≤ 0.25) correlated with each outcome were retained for possible inclusion in each regression analysis. Second, a staged regression analysis was conducted with each outcome factor using stepwise methods to minimize collinearity effects. In the first stage, demographic variables (age, Hispanic ethnicity, socio-economic status score [40]) were entered. In the second stage, significant childhood, mental health, marital and household environmental variables were entered. For clinical outcome regressions, significant attitudinal/behavioural outcomes were entered next. Then, violence characteristics (frequency, mean episode severity, and the three non-linearity measures) were entered. To then determine whether an optimal level of non-linearity would contribute to the remaining variance, the three optimal measures of non-linearity were entered. A 0.05 < P-value ≤ 0.10 was deemed trending towards significance.
Results The sample was predominantly low income and Hispanic. Many (43%) women were in common law marriages, and reported relationship duration of 9.6 years [± 8.9 (SD) years]. Duration of abuse was 5.5 (± 6.5 SD) years with age-of-abuse-onset of 32.6 (± 11.3 SD) years. 721
Do violence dynamics matter?
Outcome Attitudinal/behavioural outcomes Appraisal of violence Control Importance Novelty Not temporary Not self-caused Predictable Hope Total support Coping Active coping Planning Seeks support for instrumental reasons Seeks support for emotional reasons Reinterpretation Acceptance Denial Behavioural disengagement Mental disengagement Disengagement via substances Clinical outcomes Readiness To seek help To leave relationship Symptomatology Physical symptoms Psychological symptoms Social symptoms Health care utilization Primary care settings Mental health settings Emergency settings Functional status Physical functioning Emotional role functioning Social functioning Poor mental health AB
D. Katerndahl et al.
Periodic (n = 16)
Chaotic (n = 40)
Random (n = 79)
F (P)
10.56B 12.94A 7.25 9.47 7.50B 8.31B 36.8 8.38
10.69B 10.45B 6.59 8.36 9.13A 8.11B 36.1 9.26
9.55A 11.36B 5.81 9.12 8.11B 6.88A 35.4 8.91
4.01(.020) 3.84(.024) 1.43(.243 1.05(.354) 2.42(.093) 5.31(.006) 0.64(.531) 0.22(.800)
8.91A 8.29 6.86
6.66B 7.28 5.20
6.81B 7.58 6.00
2.52(.086) 0.44(.642) 1.09(.341)
8.00
6.06
6.36
1.50(.228)
8.29 7.43 5.07 4.15 6.62 0.46
7.57 6.74 3.60 4.31 6.50 0.53
7.13 7.52 4.75 5.59 7.00 0.69
0.81(.449) 0.61(.544) 1.13(.326) 2.10(.127) 0.40(.673) 0.45(.637)
2.63B 0.69
1.44A 0.43
2.35B 0.63
5.45(.005) 0.78(.459)
18.9 11.3B 23.2 1.54AB 1.93A 1.07 12.5 0.94B 4.88 13.3
15.5 8.7A 23.7
17.5 10.7B 24.2
Table 2 Outcomes across dynamic patterns (mean)
0.78(.459) 3.07(.050) 0.09(.915)
2.20B 0.36B 1.18
1.27A 0.79B 0.48
3.29(.041) 3.66(.029) 1.34(.265)
14.3 1.60A 5.40 11.2
13.7 1.01B 4.62 12.6
0.47(.628) 3.12(.047) 1.39(.252) 1.57(.213)
= Post-hoc differences.
Table 2 presents differences in outcomes across dynamic patterns. Based upon post-hoc analysis, those women in periodically violent relationships believed the violence as most important, compared with women in non-periodic relationships, and were most likely to use active coping, while those in chaotic relationships were most likely to believe that they were not the cause of the violence. Women in relationships with random violence were least likely to believe that the violence was predictable or controllable. Among clinical outcomes, several differences were observed. In three of the five outcomes in which significant differences were found, those with chaotic violence differed from the others; these women reported fewer psychological symptoms, better emotional role functioning and less willingness to seek help. Table 3 presents the results of the regression analyses for the attitudinal/behavioural outcomes. For two outcomes (positive 722
coping and negative appraisal), the combination of childhood, mental health, relationship and violence variables failed to account for significant variance. However, in the remaining three outcomes, the degree of violence non-linearity (ApEn and Lyapunov exponent) contributed significantly to the models beyond violence frequency and episode severity. Although all four models of clinical outcomes were significant (see Table 4), in none did violence non-linearity contribute; the only violence variable that was significant was that violence frequency was related to low social symptoms with high mental health utilization. To evaluate whether the non-linearity–outcome relationships were curvilinear with a mid-range optimal level of non-linearity, each regression model discussed above was rerun with the three optimal non-linearity estimates added. Optimal approximate
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
Do violence dynamics matter?
Table 3 Regression analyses of attitudinal/behavioural outcomes [β (P)] Coping Predictors Childhood Emotional abuse Physical abuse Sexual abuse Witnessing violence Family dysfunction Mental health Illegal drug use Women’s use Men’s use Depression severity Anxiety severity Panic disorder Relationship Marital characteristics Number of children Duration of relationship Duration of marriage Abuse duration Forced into sex Household environment Coercion Intimidation Emotional abuse Isolation Minimization Use of children Male privilege Economic Dyadic splitting Husband’s violence Frequency Episode severity Non-linearity Approximate entropy LZ complexity Lyapunov exponent Model F (P-value) Adjusted R2
Positive
Appraisal Negative
Positive
Negative
Hope and support
−0.144 (0.115)
−0.173 (0.047) 0.293 (0.004)
−0.210 (0.046) −0.201 (0.072) −0.148 (0.164) −0.136 (0.146)
−0.160 (0.055)
−0.252 (0.014)
0.274 (0.004)
−0.250 (0.005) −0.205 (0.027) −0.180 (0.086) −0.280 (0.009)
0.232 (0.009)
1.98 (0.140) 0.039
.226 (.014)
0.179 (0.038)
4.49 (0.000) 0.197
6.44 (0.000) 0.259
0.079 (0.971) −0.027
4.23 (0.001) 0.232
Adjusted for age, socio-economic status and race/ethnicity (Hispanic).
entropy was related to both symptoms and functioning (β = 0.092, P = 0.077) and readiness for change (β = 0.167, P = 0.033).
Discussion Women experiencing periodic violence recognized its importance and used their active coping to seek mental health care. Those with chaotic dynamics recognized that they were not responsible, experienced fewer psychological symptoms and emotional role limitations, and did not seek help. Those experiencing random violence recognized its unpredictability and uncontrollability. Measures of non-linearity contributed to five outcome factors.
© 2014 John Wiley & Sons, Ltd.
Attitudinal/behavioural outcomes In terms of coping, regression analysis did little to explain positive coping. While this study found an association with the husband’s use of control [41], other studies have found that problem-focused coping is positively related to the amount and severity of violence [41] and may lessen perceived distress [42]. Understanding negative coping was more complicated. Regression analysis found that Lyapunov non-linearity was an independent predictor. IPV has previously been linked to such negative coping strategies as acceptance, avoidance, denial and detachment [43]. While this study found that negative coping was related to husband’s drug 723
Do violence dynamics matter?
D. Katerndahl et al.
Table 4 Regression analyses of clinical outcomes
Predictors Childhood Emotional abuse Physical abuse Sexual abuse Witnessing violence Family dysfunction Mental health Illegal drug use Women’s use Men’s use Depression severity Anxiety severity Panic disorder Relationship Marital characteristics Number of children Duration of relationship Duration of marriage Abuse duration Age-of-violence onset Forced into sex Household environment Coercion Intimidation Emotional abuse Isolation Minimization Use of children Male privilege Economic Dyadic splitting Attitudinal/behavioural outcomes Appraised importance Seeks support for emotional reasons Hope Total support Husband’s violence Frequency Episode severity Non-linearity Approximate entropy LZ complexity Lyapunov exponent Model F (P-value) Adjusted R2
Symptoms and functioning
Medical utilization
Readiness for change
Social symptoms and mental health utilization
0.247 (0.011) 0.149 (0.028)
−0.176 (0.054) 0.142 (0.160)
0.549 (0.000)
0.037 (0.688)
0.236 (0.001)
−0.339 (0.001)
0.209 (0.080)
0.176 (0.009)
0.118 (0.210)
0.165 (0.061)
0.187 (0.057) −0.387 (0.000) −0.230 (0.011) −0.271 (0.011) −0.151 (0.080)
23.41 (0.000) 0.671
5.62 (0.000) 0.157
4.91 (0.000) 0.240
8.83 (0.000) 0.445
Adjusted for age, socio-economic status and race/ethnicity (Hispanic).
use, wife’s depression and dyadic splitting, others also found a relationship with depression [44]. Although a relationship between childhood abuse and negative coping was found in this study [45], that relationship was lost when other factors were entered. In terms of violence appraisal, negative appraisal could not be explained via regression analysis. Negative expectations have been linked to partner-reported IPV [46]. Although self-blame is common in IPV [47], such negative appraisals have not been well 724
studied. Understanding positive appraisal is more complicated with regression analysis finding that two forms of violence nonlinearity (ApEn and Lyapunov exponent) independently predicted positive coping. Positive expectancies can lead to anger and relational adjustment [46], possibly related to the associations with coercion and dyadic splitting noted in this study. When modelling hope and support, husband’s drug use, wife’s depression and violence severity were inversely related to hope
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
and support, so too was ApEn; yet, optimal ApEn contributed significantly as well. Prior IPV research has shown that not only are women with adequate social support less likely to have a history of IPV, but those with such a history are less likely to be involved in a second violent relationship if they have support [48]. In fact, long-term abuse is decreased when supportive individuals provide information that improves access to resources [49]. Women in more severely violent relationships may use caution when seeking support due to family, cultural and societal sanctions, fear, sense of isolation, sense of shame or a lack of perceived benefit when seeking that support [50].
Clinical outcomes Medical utilization was unrelated to violence in this study of women in violent relationships; it was associated only with a history of child sexual abuse [51]. Fletcher [52] reported that the association between childhood abuse and health care utilization, as noted in our regression analysis, was no longer observed when adjusted for confounders. Previous work has suggested that physical IPV is linked to higher medical utilization [53], which continued even after the IPV stopped [54]. Among women in a battered women’s shelter, social support intervention reduced psychological distress and health care utilization [55]. Social symptoms and lack of mental health utilization depended upon attitudinal/behavioural outcomes; only violence infrequency predicted social symptoms. While social isolation is common in IPV [56], mental health service use is often increased [53]. Although self-imposed isolation and lack of mental health service utilization [57] could be due to concerns about disclosure [58], many women will use those services if offered [59]. Symptoms and dysfunction depended upon a variety of factors from childhood family dysfunction and current depression and panic disorder to use of coercion by the husband and age-of-violence onset. However, optimal ApEn also contributed. Although possibly inflated by confounders [52], associations may exist between IPV and both symptoms [60] and dysfunction [61]. While the relationship to depression found in this study was noted before [60], the links between coping [62], appraisal [63], support [56] and symptoms or dysfunction found previously were not seen. Readiness for change was related to appraised violence importance, short relationship duration, and optimal ApEn. The current study measured factors previously associated with help seeking in IPV – prior trauma [57,64], coping [65], forced sex, witnessing childhood violence [64], appraisal [66], presence of children [67], social support [68] and abuse severity [67]. We also assessed factors associated with leaving the relationship – alcohol intake [69], presence of children [70], witnessing childhood violence [71], financial independence [71], relationship quality [72], social support [73], lack of forgiveness [74], duration of abuse [72] and increasing violence [75]. But these predictors were not found to be related to readiness for change in this study. However, appraised importance was a significant predictor of readiness for change as previously found [75]. In addition, the fact that duration of relationship is inversely related to readiness in this study may reflect willingness for action in women who have a lower investment in the relationship [76].
© 2014 John Wiley & Sons, Ltd.
Do violence dynamics matter?
Implications Although a potential straightforward, positive relationship between non-linearity and positive outcomes may exist, its characteristic unpredictability may explain why the non-linearity of violence can be associated with both positive appraisal (violence as temporary, unimportant and controllable) and negative coping, with positive appraisal and lack of hope or support. Non-linearity as a continuous measure predicts three attitudinal/behavioural outcomes only, while optimal non-linearity predicts two clinical outcomes only, suggesting a curvilinear relationship with clinical outcomes. Thus, for attitudinal/behavioural outcomes, unpredictability may directly impact coping, appraisal and hope, but for clinical outcomes, there exist optimal levels of predictability/ unpredictability. Readiness for change may be sensitive to the interplay between constraints-limiting activity and resourcesenabling activity [76] in a process of decisional balance [77]. The importance of non-linearity with its unpredictability in five such outcomes in this study reinforces the need to individualize management and coping strategies [78].
Limitations This study has several limitations. First, the sample size is small for time series analysis, especially for determining dynamic pattern. Second, it is unclear whether women accurately report the level of violence perpetrated by their partners or themselves. However, Regan et al. [79] found that violence reports by husbands and wives were highly correlated. Finally, the predominance of Hispanics within the sample may limit the generalizability of the findings. Caetano et al. [80] found that, over a 5-year period, Hispanics were 2.5 times as likely as Anglos to initiate IPV, while reporting a violence recurrence rate four times higher. In addition, Hispanic women in violent relationships may differ from other ethnic women in their use of emergency departments [81] and avoiding IPV disclosure to providers [82]. In fact, acculturation is important in help seeking for such women [81].
Conclusion Measures of violence non-linearity contributed to the variance of three attitudinal/behavioural outcomes, and optimal non-linearity contributed to two clinical outcomes. Knowledge of violence dynamics may have applications when working with violent couples.
Acknowledgements This study was funded by the National Science Foundation (#0826812). We wish to thank Stephanie Mitchell, Kelli Giacomini, Robert Mesec and Wilson Pace at the University of Colorado, Department of Family Medicine for their invaluable assistance.
References 1. Katerndahl, D., Burge, S., Ferrer, R., Becho, J. & Wood, R. (2010) Complex dynamics in intimate partner violence. Primary Care Companion to the Journal of Clinical Psychiatry, 12 (4), e1–e12.
725
Do violence dynamics matter?
2. Albrecht, G., Freeman, S. & Higginbotham, N. (1998) Complexity and human health. Culture, Medicine and Psychiatry, 22 (1), 55–92. 3. Wallace, R. & Wallace, D. (2004) Structured psychosocial stress and therapeutic failure. Journal of Biological Systems, 12, 335–369. 4. Waldrop, M. M. (1992) Complexity: The Emerging Science at the Edge of Order and Chaos. New York: Simon & Schuster. 5. Hirsch, I. B. & Brownlee, M. (2005) Should minimal blood glucose variability become the gold standard of glycemic control? Journal of Diabetes and its Complications, 19, 178–181. 6. Harbourne, R. T., Deffeyes, J. E., Kyvelidou, A. & Stergiou, N. (2009) Complexity of postural control in infants. Nonlinear Dynamics, Psychology, and Life Sciences, 13, 123–144. 7. Olson, D. H. & Gorall, D. M. (2003) Circumplex model of marital and family systems. In Normal Family Processes, 3rd edn (ed. F. Walsh), pp. 514–547. New York: Guilford Press. 8. Guastello, S. J. (2012) Self-complexity and the interpretation of entropy and information statistics. Society for Chaos Theory in Psychology and Life Sciences Newsletter, 19 (2), 5–10. 9. Burge, S. K., Becho, J., Ferrer, R. L., Wood, R. C., Talamantes, M. & Katerndahl, D. A. (2014) Safely examining complex dynamics of intimate partner violence. Families, Systems and Health: The Journal of Collaborative Family Healthcare, (in press). 10. Straus, M., Gelles, R. & Steinmetz, S. (1980) Behind Closed Doors. New York: Anchor Books. 11. Shepard, M. F. & Campbell, J. A. (1992) The Abusive Behavior Inventory. Journal of Interpersonal Violence, 7, 291–305. 12. Felitti, V. J., Anda, R. F., Nordenberg, D., et al. (1998) Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: the Adverse Childhood Experiences (ACE) study. American Journal of Preventive Medicine, 14, 245–258. 13. Siegel, J. P. (2006) Dyadic splitting in partner relational disorders. Journal of Family Psychology, 20, 418–422. 14. Glasgow, R. E., Ory, M. G., Klesges, L. M., Cifuentes, M., Fernald, D. H. & Green, L. A. (2005) Practical and relevant self-report measures of patient health behaviors for primary care research. Annals of Family Medicine, 3, 73–81. 15. Nease, D. E. & Malouin, J. M. (2003) Depression screening. The Journal of Family Practice, 52, 118–126. 16. Spitzer, R. L., Kroenke, K., Williams, J. B. W. & Lowe, B. (2006) Brief measure for assessing generalized anxiety disorder: GAD-7. Archives of Internal Medicine, 166, 1092–1097. 17. Spitzer, R. L., Kroenke, K. & Williams, J. B. (1999) Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study: primary care evaluation of mental disorders: patient health questionnaire. Journal of the American Medical Association, 282, 1737–1744. 18. Koss, M. P. & Heslet, L. (1992) Somatic consequences of violence against women. Archives of Family Medicine, 1, 53–59. 19. Herth, K. A. (1992) An abbreviated instrument to measure hope. Journal of Advanced Nursing, 17, 1251–1259. 20. Carver, C. S., Scheier, M. F. & Weintraub, J. K. (1989) Assessing coping strategies: a theoretically based approach. Journal of Personality and Social Psychology, 56, 267–283. 21. Vitaliano, P. P. (1985) Manual for Appraisal Dimension Scale and Revised Ways of Coping Checklist. Seattle: Department of Psychiatry, University of Washington. 22. Duke University (1986) Duke Social Support and Stress Scale. Durham: Duke University. 23. Katerndahl, D. A. & Realini, J. P. (1997) Use of health care services by persons with panic symptoms. Psychiatric Services: A Journal of the American Psychiatric Association, 48, 1027–1032. 24. Ware, J. E. Jr & Sherbourne, C. D. (1992) The MOS 36-item short form health survey (SF-36), I. Conceptual framework and item selection. Medical Care, 30, 473–483.
726
D. Katerndahl et al.
25. Katerndahl, D. & Oyiriaru, D. (2007) Assessing the biopsychosociospiritual model in primary care. International Journal of Psychiatry in Medicine, 37, 393–414. 26. Prochaska, J. O. & DiClemente, C. C. (1986) Toward a comprehensive model of change. In Treating Addictive Behaviors (eds W. R. Miller & N. Heather), pp. 3–27. New York: Plenum Press. 27. Heggler, R., Kantz, H. & Schreiber, T. (1999) Practical implementation of nonlinear time series methods: the TISEAN package. Chaos (Woodbury, N.Y.), 9, 413–435. 28. Kreindler, D. M. & Lumsden, C. J. (2006) Effects of the irregular sample and missing data in time series analysis. Nonlinear Dynamics, Psychology, and Life Sciences, 10, 187–214. 29. Kreindler, D. M. & Lumsden, C. J. (2007) Effects of irregular sampling and missing data on largest Lyapunov exponents. Nonlinear Dynamics, Psychology, and Life Sciences, 11, 401–412. 30. Land, B. & Elias, D. (2005) Measuring the ‘complexity’ of a time series. Available at: http://www.nbb.cornell.edu/neurobio/land/ PROJECTS/Complexity/index.html (last accessed 22 April 2014). 31. Ziv, J. & Lempel, A. (1978) Compression of individual sequences via variable-rate coding. IEEE Transactions on Information Theory, IT-24, 530–536. 32. Pincus, S. M. (2006) Approximate entropy as a measure of irregularity for psychiatric serial metrics. Bipolar Disorders, 8, 430–440. 33. Wolf, A., Swift, J. B., Swinney, H. L. & Vastano, J. A. (1985) Determining Lyapunov exponents from a time series. Physica D. Nonlinear Phenomena, 16, 285–317. 34. Gevers, E., Pincus, S. M., Robinson, I. C. A. F. & Veldhuis, J. D. (1998) Differential orderliness of the GH release process in castrate male and female rats. The American Journal of Physiology, 274, R437–R444. 35. Yeragani, V. K., Pohl, R., Mallavarapu, M. & Balon, R. (2003) Approximate entropy of symptoms of mood. Bipolar Disorders, 5, 279–286. 36. Pincus, S. M., Hartman, M. L., Roelfsema, F., Thorner, M. O. & Veldhuis, J. D. (1999) Hormone pulsatility discrimination via coarse and short-time sampling. The American Journal of Physiology, 277, E948–E957. 37. Zhang, X. S., Zhu, Y. S., Thakor, N. V. & Wang, Z. Z. (1999) Detecting ventricular tachycardia and fibrillation by complexity measure. IEEE Transactions on Biomedical Engineering, 46, 548–555. 38. Rosenstein, M., Collins, J. & DeLuca, C. (1993) Practical method of calculating largest Lyapunov exponents from small data sets. Physica D. Nonlinear Phenomena, 65, 117–134. 39. Cheng, Y. T. & Van de Ven, A. H. (1996) Learning the innovation journey: order out of chaos? Organization Science, 7 (6), 593–614. 40. Hollingshead, A. & Redlich, F. (1958) Social Class and Mental Illness. New York: John Wiley & Sons. 41. Sabina, C. & Tindale, R. S. (2008) Abuse characteristics and coping resources as predictors of problem-focused coping strategies among battered women. Violence Against Women, 14, 437–456. 42. Yoshihama, M. (2002) Battered women’s coping strategies and psychological distress. American Journal of Community Psychology, 30, 429–452. 43. Orzeck, T. L., Rokach, A. & Chin, J. (2010) Effects of traumatic and abusive relationships. Journal of Loss and Trauma, 15, 167– 192. 44. Calvete, E., Corral, S. & Estevez, A. (2007) Cognitive and coping mechanisms in the interplay between intimate partner violence and depression. Anxiety, Stress and Coping: An International Journal, 20, 369–382. 45. Taft, C. T., Resick, P. A., Panuzio, J., Vogt, D. S. & Mechanic, M. B. (2007) Examining the correlates of engagement and disengagement coping among help-seeking battered women. Violence and Victims, 22, 3–17.
© 2014 John Wiley & Sons, Ltd.
D. Katerndahl et al.
46. Meis, L. A., Murphy, C. M. & Winters, J. J. (2010) Outcome expectancies of partner abuse. Assessment, 17, 30–43. 47. Carlson, B. E. (1997) Stress and coping approach to intervention with abused women. Family Relations, 46, 291–298. 48. Plazaola-Castano, J., Ruiz-Perez, I. & Montero-Pinar, M. I. (2008) Grupo de studio para la violencia de genero. Gaceta Sanitaria, 22, 527–533. 49. Bosch, K. & Schumm, W. R. (2004) Accessibility to resources. Journal of Sex and Marital Therapy, 30, 357–370. 50. Rose, L. E., Campbell, J. & Kub, J. (2000) Role of social support and family relationships in women’s responses to battering. Health Care for Women International, 21, 27–39. 51. Chou, K. L. (2012) Childhood sexual abuse and psychiatric disorders in middle-aged and older adults. The Journal of Clinical Psychiatry, 73, e1365–e1371. 52. Fletcher, J. (2010) Effects of intimate partner violence on health in young adulthood in the United States. Social Science and Medicine, 70, 130–135. 53. Bonomi, A. E., Anderson, M. L., Rivara, F. P. & Thompson, R. S. (2009) Health care utilization and cost as associated with physical and nonphysical-only intimate partner violence. Health Services Research, 44, 1052–1067. 54. Rivara, F. P., Anderson, M. L., Fishman, P., Bonomi, A. E., Reid, R. J., Carrell, D. & Thompson, R. S. (2007) Healthcare utilization and costs for women with a history of intimate partner violence. American Journal of Preventive Medicine, 32, 89–96. 55. Constantino, R., Kim, Y. & Crane, P. A. (2005) Effects of a social support intervention on health outcomes in residents of a domestic violence shelter. Issues in Mental Health Nursing, 26, 575–590. 56. Latta, R. E. (2009) Struggling to define my role. Dissertation Abstracts International: Section B: The Sciences and Engineering, 69 (9-B), 5783. 57. Johnson, D. M. & Zlotnick, C. (2007) Utilization of mental health treatment and other services by battered women in shelters. Psychiatric Services: A Journal of the American Psychiatric Association, 58, 1595–1597. 58. Trotter, J. L. (2009) Dynamic investigation of domestic violence survivors’ experiences with their informal social networks. Dissertation Abstracts International: Section B: The Sciences and Engineering, 69 (11-B), 7154. 59. Cerulli, C., Gellman, R. A., Nichols, C., Hall, D., Conner, K. R. & Caine, E. D. (2011) Mental and physical health symptoms of family court intimate partner violence petitioners. International Journal of Law and Psychiatry, 34, 94–98. 60. Matud, M. P. (2005) Psychological impact of domestic violence on Spanish women. Journal of Applied Social Psychology, 35, 2310– 2322. 61. Straus, H., Cerulli, C., McNutt, L. A., Rhodes, K. V., Conner, K. R., Kemball, R. S., Kaslow, N. J. & Houry, D. (2009) Intimate partner violence and functional health status. Journal of Women’s Health, 18, 625–631. 62. Calvete, E., Corral, S. & Estevez, A. (2008) Coping as a mediator and moderator between intimate partner violence and symptoms of anxiety and depression. Violence Against Women, 14, 886–904. 63. Martinez-Torteya, C., Bogat, G. A., Von Eye, A., Levendosky, A. A. & Davidson, W. S. (2009) Women’s appraisal of intimate partner violence stressfulness and their relationship to depressive and posttraumatic stress disorder symptoms. Violence and Victims, 24, 707–722.
© 2014 John Wiley & Sons, Ltd.
Do violence dynamics matter?
64. Henning, K. R. & Klesges, L. M. (2002) Utilization of counseling and supportive services by female victims of domestic abuse. Violence and Victims, 17, 623–636. 65. Shannon, L., Logan, T. K., Cole, J. & Medley, K. (2006) Help-seeking and coping strategies for intimate partner violence in rural and urban women. Violence and Victims, 21, 167–181. 66. Petersen, R., Moracco, K. E., Goldstein, K. M. & Clark, K. A. (2004) Moving beyond disclosure. Women’s Health, 40, 63–76. 67. Djikanovic, B., Wong, S. L. F., Jansen, H. A., Koso, S., Simic, S., Otasevic, S. & Lagro-Janssen, A. (2012) Help-seeking behavior of women who experienced intimate partner violence. Family Practice, 29 (2), 189–195. Available at: http://www.ncbi.nlm.nih.gov/pubmed/ 21896506 (last accessed 18 June 2014). 68. Wright, C. V. & Johnson, D. M. (2009) Correlates for legal helpseeking. Violence and Victims, 24, 771–785. 69. Ramisetty-Mikler, S. & Caetano, R. (2005) Alcohol use and intimate partner violence as predictors of separation among U.S. couples. Journal of Studies on Alcohol, 66, 205–212. 70. Rhodes, K. V., Cerulli, C., Dichter, M. E., Kothari, C. L. & Barg, F. K. (2010) ‘I didn’t want to put them through that’. Journal of Family Violence, 25, 485–493. 71. Kim, J. & Gray, K. A. (2008) Leave or stay? Journal of Interpersonal Violence, 23, 1465–1482. 72. Ulrich, Y. C. (1998) What helped most in leaving spouse abuse. In Empowering Survivors of Abuse (ed. J. C. Campbell), pp. 70–78. Thousand Oaks: Sage Publications. 73. Wilcox, P. (2000) ‘Me mother’s bank and me nanan’s, you know, support!’. Women’s Studies International Forum, 23, 35–47. 74. Gordon, K. C., Burton, S. & Porter, L. (2004) Predicting the intentions of women in domestic violence shelters to return to partners. Journal of Family Psychology, 18, 331–338. 75. Chang, J. C., Dado, D., Hawker, L., Cluss, P. A., Buranosky, R., Slagel, L., McNeil, M. & Scholle, S. H. (2010) Understanding turning points in intimate partner violence. Journal of Women’s Health, 19, 251–259. 76. Burton, S. J. (2004) To go or not to go, that is the question. Dissertation Abstracts International: Section B: The Sciences and Engineering, 64 (9-B), 4605. 77. Burke, J. G., Denison, J. A., Gielen, A. C., McDonnell, K. A. & O’Campo, P. (2004) Ending intimate partner violence. American Journal of Health Behavior, 28, 122–133. 78. Hetzel-Riggin, M. D. & Meads, C. L. (2011) Childhood violence and adult partner mistreatment. Journal of Family Violence, 26, 585–593. 79. Regan, K. V., Bartholomew, K., Kwong, M. J., Trinke, S. J. & Henderson, A. J. Z. (2006) Relative severity of acts of physical violence in heterosexual relationships. Personal Relationships, 13, 37–52. 80. Caetano, R., Field, C. A., Ramisetty-Mikler, S. & McGrath, C. (2005) 5-year course of intimate partner violence among white, black, and Hispanic couples in the United States. Journal of Interpersonal Violence, 20, 1039–1057. 81. Lipsky, S., Caetano, R., Field, C. A. & Larkin, G. L. (2006) Role of intimate partner violence, race, and ethnicity in help-seeking behaviors. Ethnicity and Health, 11, 81–100. 82. Kelly, U. A. (2010) Symptoms of PTSD and major depression in Latinas who have experienced intimate partner violence. Issues in Mental Health Nursing, 3, 119–127.
727
