Longitudinal Changes in Posttraumatic Stress Disorder After Resettlement Among Yazidi Female Refugees Exposed to Violence

Key Points Question How does posttraumatic stress disorder (PTSD) change during a 1-year period in female refugees who survived mass atrocities, and what factors are associated with PTSD course? Findings This cohort study with 116 female refugee survivors of captivity and genocide found high PTSD severity 2 years after resettlement in Germany with no significant change 1 year later. Factors associated with severe PTSD were earlier symptoms of intrusions and longer time spent in traumatic situations, whereas strengthening in faith and social relationships were associated with symptom relief over time. Meaning This study suggests that female refugee survivors of genocide and captivity are at high risk for severe and chronic PTSD beyond the initial years of resettlement.


Introduction
Mental illnesses, such as posttraumatic stress disorder (PTSD), are prevalent after mass atrocities and displacement. Early studies 1-7 investigating 1 of the most recently affected populations, displaced Yazidis from northern Iraq after the 2014 genocide, indicate an alarming prevalence of PTSD (42.9%-100%), with women having higher prevalence rates than men. Previous research [8][9][10][11][12] with genocide survivors (eg, in Rwanda or Bosnia) found that severe mental health effects can last years, even decades, after genocide.
The greater prevalence of PTSD among women is also seen in genocide-affected and refugee populations. 10,13,14 A potential explanation might be that sexualized violence is disproportionately committed against women in armed conflicts. [15][16][17] This explanation aligns with the recently published finding that exposure to gender-based violence, including sexual slavery while held in captivity by the nonstate armed organization known as the Islamic State (IS), was associated with PTSD in Yazidi women after the 2014 genocide. 18 Although several factors associated with PTSD onset have been identified, studies investigating factors associated with the longitudinal PTSD course, meaning the trajectory of posttraumatic symptoms over several years, remain rare. 19 A systematic review 19 of naturalistic prospective cohort studies with trauma survivors found social relationships and support to be preventive of a severe, chronic PTSD symptom trajectory. The review 19 also found that female sex, older age, minority status, trauma severity, and PTSD symptoms (particularly higher hyperarousal) at baseline are associated with an unfavorable PTSD course. Other studies [20][21][22][23][24] have yielded inconsistent findings regarding whether individual PTSD symptoms are associated with subsequent PTSD. Because only a few studies in the systematic review 19 involved genocide survivors or refugees and other crosssectional studies 25,26 indicate the existence of distinct PTSD symptomatologic patterns in refugees, a deeper understanding of the course of PTSD and potential risk and resilience factors is essential for this field of research and clinical practice.
Previous cross-sectional studies 27,28 have highlighted the effective use of coping strategies (eg, sense of purpose in life, use of social support, or religious coping) as preventive for PTSD after trauma and as associated with lower PTSD symptoms and even PTSD recovery. Coping refers to cognitive and behavioral efforts to help an individual master, reduce, or tolerate specific external and/or internal demands that are perceived as overwhelming. 29,30 Two early qualitative studies 31,32 with small samples suggest collective and religious coping strategies to be salient among Yazidi genocide survivors.
Current understanding of the longitudinal course of PTSD and the effectiveness of coping strategies is incomplete, particularly with regard to the distinctive needs of female refugees after surviving mass atrocities. Longitudinal studies in this area are essential for research and clinical practice. Because such studies are currently lacking, the current study combines 2 main aims. The first aim, following a pathogenic approach, is to identify the course of PTSD in resettled female survivors of the 2014 genocide and potential factors associated with PTSD severity and course over time. The second aim, following a salutogenic approach, is to identify preferred coping strategies and protective factors during the long-term course of PTSD in this high-risk group. On the basis of the previous research discussed above, 19,27,28,31,32 we hypothesized that earlier PTSD symptoms and higher severity of trauma exposure (for Yazidi women, the amount of time spent in IS captivity) are associated with an unfavorable PTSD course. Positive posttraumatic changes in religious and social factors are hypothesized to be significantly associated with a favorable PTSD course.

Study Design
The prospective cohort study included a baseline (September 1, 2017, to January 12, 2018) and a 1-year follow-up (August 29, 2018, to January 15, 2019) assessment. The study proceeded via interpreter-aided interviews. Results of the baseline investigation regarding psychosomatic symptoms and perspectives on justice are presented in previous publications. 33

Participants and Setting
In August 2014, IS attacked and overran the Sinjar Mountains of the Nineveh governorate in northern Iraq. The mountain region has historically been home to the Yazidis (Êzidî), who were explicitly and brutally targeted by IS. 36  through January 2016, a total of 1000 women and children were flown out of Iraq to Germany. 40 The beneficiaries received secure housing in 19 cities in Baden-Württemberg and financial support.
Moreover, medical, psychotherapeutic, and psychiatric support was available to the women. 39,[43][44][45] In our sample, approximately 73% had seen a psychologist in the first 2 years of resettlement. 39 All adult HAP beneficiaries were eligible to participate in the current study because they were already preselected as severely affected by trauma through the program. Social workers of each HAP accommodation center were invited to inform their clients about participation starting from July 2017. Voluntary recruitment within this special sample aimed to enable participation without creating pressure. When women confirmed interest, the research team provided detailed study information in Kurdish-Kurmanji, Arabic, or German. All 116 study participants had experienced the 2014 IS attacks in northern Iraq.

Impact of Event Scale-Revised
To measure PTSD severity, the Impact of Event Scale-Revised (IES-R) 46,47 was used. This self-report measure consists of 22 items that assess the degree of distress caused by PTSD symptoms during the past 7 days on a 5-point Likert scale (0 to 4), with 0 indicating not at all and 4 indicating extremely.
The IES-R raw sum scores (ranging from 0 to 88) as well as 3 subscale scores (intrusions, avoidance, and hyperarousal) can be derived, with higher scores indicating higher levels of distress. All 3 subscale scores range from 0 to 4, with the intrusion subscale score consisting of the mean of 8 item scores (eg, regarding intrusive thoughts or nightmares), the avoidance subscale score referring to the mean of another 8 item scores (eg, regarding the avoidance of feelings or reminders), and the hyperarousal subscale score defined as the mean of 6 item scores (eg, regarding irritability). 47 The IES-R shows high internal consistency (α = .96). 48 Even though the IES-R is not a diagnostic tool, there is evidence that it might discriminate between individuals with and without probable PTSD, and cutoff scores have been cited in previous literature. 49 Best diagnostic accuracy was found with a cutoff score of 33. 48 Because this cutoff has been used in previous literature within different samples, including refugees, 48,50,51 we report the percentage of IES-R raw sum scores above 33 to enable comparison with other studies.

Context-Specific Questionnaire Items and Coping
To assess context-specific details, the study team, consisting of epidemiologists, psychologists, psychotherapists, and physicians from Harvard University and the University Hospital of Tübingen experienced in the research field of genocide, developed questionnaire items that contained sociodemographic characteristics (eg, ethnicity), HAP specifics (eg, time spent in captivity), and potential outcomes associated with trauma (eg, changes in faith through trauma and feelings of exclusion from the community) (eAppendix in the Supplement). Moreover, the perceived helpfulness of different emotion-focused coping strategies according to Folkman and Lazarus 29,30 in the aftermath of trauma was assessed ("How much do the following strategies help you cope with the effects of IS violence? 1. Belief in collective strength, e.g. strength of the Yazidi community or your family, 2. Belief in personal strength, e.g., belief in yourself and your own strength, 3. Praying, 4. Social retreat, e.g., spending time alone, 5. Exchange trauma contents with others, 6. Seeking professional help, e.g., doctors, psychotherapist, 7. Seeking help within the Yazidi community"). During follow-up, the same questionnaire was used with some additions (eg, 1 item assessing changes in social relationships during the past year).
The questionnaire was developed in German and English and then translated into Kurdish-Kurmanji, the language spoken by participants. The Kurdish translations were discussed, revised, and agreed on by a multidisciplinary expert team that included Kurdish and Yazidi members. The final version was piloted and discussed with 2 Kurdish/Yazidi women to ensure comprehensibility and cultural appropriateness.

Study Implementation
Interviewers were female mental health professionals accompanied by female interpreters. Before data collection, interviewers and interpreters received several days of training. Interviews took place in private rooms of the HAP accommodation in 14 German cities. To facilitate the selection of responses to quantitative questions and to ensure accuracy, participants could show their answer on a graphic representation of the Likert scale. At baseline, interviews were audio-recorded, and the Kurdish segments of the recordings, instead of the interpreters' spontaneous translations, were translated again and transcribed to validate the data entry process and to allow the research team to analyze the interviews qualitatively. 39

Statistical Analysis
For sample description, means, valid percentages, and distributions are reported. Repeated-measure analyses of variance were calculated with Greenhouse-Geisser adjustments for lack of sphericity and Bonferroni-adjusted post hoc analyses to counteract the problem of multiple comparisons.
Furthermore, Pearson correlations, independent t tests, Mann-Whitney tests, and paired-samples t tests were used. A first multiple linear regression analysis was performed to test the hypothesis that the severity of distinct PTSD symptom clusters at baseline is associated with overall PTSD severity 1 year later. Because previous studies [20][21][22][23][24] suggest that past PTSD severity is associated with subsequent PTSD severity but findings can be inconsistent regarding individual PTSD symptom clusters as factors associated with subsequent PTSD, we included all 3 baseline IES-R subscale scores as independent variables, controlling for age 52,53 and number of days spent in captivity. 42 Given that the authors of the IES-R recommend the use of means of the different item scores rather than raw sum scores 46 and that we aimed to avoid an overrepresentation of the 8-item subscales over the 6-item subscale in the sum score for PTSD severity for this analysis, we used the sum of all 3 subscales (IES-R subscale sum score range, 0-12) 54,55 at follow-up instead of the IES-R raw sum score as the dependent variable. A second multiple linear regression was performed to test the hypothesis that posttraumatic changes in faith and in social relationships are associated with a favorable PTSD course, 19 even when age 52,53 and the number of days in captivity are controlled for. 42 The PTSD    Table 2.

Follow-up Assessment Education and Employment
At of the baseline sample).

Posttraumatic Stress Disorder
At follow-up, the mean (SD) IES-R raw sum score was 59 Table 1.

Associations With PTSD Severity at Follow-up
The data met the assumptions for regression analysis (Durbin-Watson statistic = 2.099). The analysis found that the model of symptom severity of intrusions, avoidance, and hyperarousal at baseline, age, and the number of days spent in captivity was statistically significantly associated with PTSD severity at follow-up (R 2 = 0.312, adjusted R 2 = 0.270, P < .001). As given in Table 4, standardized β values are highest for baseline intrusions (β = 0.389, P = .007) and number of days spent in captivity (β = 0.218, P = .02).

Coping
Over time, 2 statistically significant changes occurred in participants' ratings of coping strategies (

Associations With PTSD Course
A multiple linear regression analysis was performed to assess the extent to which the number of days spent in captivity as well as posttraumatic changes in faith and social relationships could explain the variance of PTSD symptom course. To control for relevant sociodemographic influences, age was included in the model. 53 The data met the assumptions for regression analysis (Durbin-Watson statistic = 2.190). A statistically significant regression equation was found (R 2 = 0.177, adjusted R 2 = 0.136, P = .003).
Results are given in Table 5 and show an association of strengthening in faith (β = −0.206, P = .05) and social relationships (β = −0.221, P = .03) with PTSD symptom relief and an association between more days spent in captivity and PTSD aggravation (β = 0.227, P = .04).

Discussion
This cohort study investigated PTSD severity and coping in female genocide survivors 2 and 3 years after resettlement. Findings suggest that survivors experience severe psychological symptoms for years, even when participating in a HAP that aims to reduce postmigration stressors and provides mental health care. Longer captivity and severe intrusions are associated with unfavorable PTSD course, whereas a strengthening in faith through the traumatic event and positive changes in social relationships were identified as protective factors.
Even though approximately 73% of participants accepted psychotherapeutic help within the first 2 years of the program, 39 most continued to experience high distress from PTSD symptoms up to 3 years after resettlement. This chronicity is consistent with research investigating different samples of refugees 14,56 and genocide survivors. [8][9][10] Moreover, the severity of PTSD in the current sample is comparable to Rwandan genocide survivors who experienced traumatic crisis during commemoration activities 16 years after the genocide (IES-R raw sum scores, 59-62). 57 Distress resulting from intrusions, hyperarousal, and avoidance behavior each remained persistently high over time. However, the findings of this study suggest a special focus on intrusions when identifying high-risk groups in refugees because intrusions were rated as the most distressing of the PTSD symptom clusters. Previous research with other samples found acute intrusions to be associated with other PTSD symptoms shortly after a traumatic event 58 and associated with PTSD 6 months later. 24 The current study suggests that this association can still be found in intrusions of highly traumatized women years after the traumatic event. Avoidance and hyperarousal were not The coping strategies perceived as most helpful in the current sample underline the importance of religion, community, and self-efficacy in the aftermath of trauma, which supports earlier qualitative findings. 31,32 With only 2 significant changes over time, these preferences in coping can be interpreted as robust findings. Interestingly, the helpfulness ratings of prayer decreased, whereas self-efficacy, or belief in personal strength, increased during 1 year. Perceived personal strengthening might be a sign of adaptation to the new life in Germany and aligns with the current study's findings that more women were literate, employed, and seeking employment than at the initial assessment.
Even though the perceived helpfulness of prayer decreased slightly within the study's assessment period, praying was still considered one of the most helpful ways to cope with trauma 3 years after resettlement. Moreover, most women reported an increase in faith through the experienced trauma, which could be identified as a protective factor regarding PTSD chronicity.
Drawing strength from the community was found to be another highly preferred coping strategy, whereas social retreat was considered least helpful at both assessment times. Moreover, an improvement in social relationships was associated with PTSD symptom relief. This result supports a previous systematic review 61 that found that social support is well established as an important factor for trauma recovery. Because survivors of war-time rape are often rejected by their community and family, 2,62 the current finding can be seen as an encouragement to focus on community-based interventions 63 that strengthen social relationships in traumatized refugees.

Limitations
This study has several limitations. Because of the preselected sample of HAP beneficiaries as particularly affected by severe and enduring trauma by IS fighters and the voluntary recruitment method, generalization to other populations might be limited. 34 Moreover, because the IES-R is not a diagnostic tool, a valid PTSD prevalence rate could not be assessed in the current study. Future studies could benefit from a clinically assessed PTSD diagnosis. Limitations based on the use of translated versions of questionnaires and self-developed questionnaire items in an interpreter-aided interview setting should be considered when interpreting these results. However, because participants' answers were given verbally in Kurdish and were translated afterward, errors in data collection attributable to interpreters' spontaneous translation in the interview should be limited.
Nevertheless, a bias in answers attributable to social desirability in the interview setting cannot be fully ruled out.

Conclusions
The current study is unique because it depicts the longitudinal course of PTSD in a homogeneous sample in a setting in which postmigration stressors are limited and mental health services are available. Findings suggest that female refugee survivors of captivity are at high risk for severe and chronic PTSD beyond the initial resettlement period. The results also suggest that specifically assessing and targeting symptoms of intrusion while simultaneously fostering self-efficacy, faith, and social support may be promising strategies for similar samples in psychotherapy, as communitybased interventions, and/or as scalable interventions for a global use. These strategies should be further investigated in future longitudinal studies.