We conducted a case-control study in which cases met the threshold for having significant personality pathology using the International Personality Disorder Examination (IPDE) screening questionnaire and controls did not. The study was conducted in south Croatia because it is an area that has been affected by repeated military conflicts, most recently during the 1991–1995 war. Followed by the collapse of the former Yugoslavia and subsequent economic crisis, the war could be described as the bloodiest armed conflict in Europe for the last 50 years. It affected several republics of the former Yugoslavia and resulted in several hundred thousand people being killed, over 3 millions being uprooted from their homes with considerable amount of infrastructure being destroyed [
20].
Participants
Both cases and controls were recruited from inpatient and outpatient mental health and general hospital services in Split (southern Croatia). This included inpatient and outpatient mental health services, medical outpatient clinics and several wards at the department of internal medicine. In mental health services, we asked clinicians to refer people who were primarily being treated for personality-related problems, in general hospitals we simply asked to be referred patients who were currently well enough to complete the study interview. The reason for recruiting controls from medical settings was to provide an estimate of the levels of exposure in the same population we recruited cases from: people in contact with healthcare services. To take part in the study potential participants had to have lived in Croatia during the 1991–1995 war and only those who provided written informed consent were recruited. We excluded people suffering from an acute psychotic episode, chronic psychotic illness or from personality change due to organic brain damage, disease and dysfunction. All assessments took place between November 2010 and October 2011. Ethical approval for the study was obtained from the University Hospital Split Ethics Committee and the School of Medicine Ethics Committee, University of Split prior to the start of data collection.
Measures
The primary outcome measure was the presence of personality disorder (PD) assessed with the 77-item International Personality Disorder Examination (IPDE) screening questionnaire. The IPDE was derived from the original version of Personality Disorder Examination (PDE) and found to have good inter-rater reliability (0.71–0.91) and intertemporal reliability (0.55–0.84) [
21,
22]. The IPDE-77 Screening Questionnaire used in this study is a self-report screen containing 77 items written at a 9 years of age reading level that measure personality pathology according to the DSM-IV. The screen has 10 PD subcategories, each containing 7–8 items, except for borderline and narcissistic subgroups which have 9 items. The IPDE requires dichotomous ‘true/false’ responses and the questions are interspersed between different PD subcategories with some items reversed. In this way the likelihood of participants guessing and choosing desirable answers is reduced. We used a more conservative approach for scoring participants answers, so a score of three and below meant ‘negative’ for a PD category and a score of four and above meant ‘positive’ for that personality subgroup. The IPDE screen has been shown to be reliable in clinical and non-clinical populations [
23‐
25].
Traumatic war-related experience and symptoms of post-traumatic stress were assessed by using the Harvard Trauma Questionnaire (HTQ) [
26,
27], a self-report measure that has been widely translated and used in traumatised refugees and civilian population, war veterans and victims of torture throughout the world including the communities of the former Yugoslavia [
13,
28‐
30]. The HTQ items related to symptoms of post-traumatic stress are consistent with DSM IV PTSD criteria based on three sub-domains: re-experiencing traumatic events, avoidance and numbing, and increased arousal. A cut off score of ≥2.5, which was initially derived from Indochinese population, was generally considered to be “checklist positive” for PTSD. Although a study conducted in the former Yugoslavia [
31] indicated that the cut-off point of ≥2.5 was too high and recommended a cut-off score of ≥2.0 for PTSD ‘positive’ cases, we used the former and a more conservative cut off point of ≥2.5 to reduce the likelihood of making a false positive PTSD diagnosis and to make our findings comparable to wider international communities.
As we were interested in studying the exposure to severe trauma, and in the absence of an established definition of this concept, we have decided to define severe (catastrophic) trauma based on the ICD-10 description of catastrophic stress [
32] and the findings from a survey of trauma experts [
33]. Our decision to use the ICD-10 classification in this instance (rather than DSM IV) was based on this concept being described in more detail in the ICD 10 which also recognised that personality change may occur following the experience of catastrophic stress [
32]. Based on these it was assumed that severe (catastrophic) trauma would involve prolonged exposure to life-threatening circumstances with imminent possibility of being killed (for example exposure to war trauma, concentration camp experience, being tortured, hostage situations and sexual assault). Two authors (JM and MC) independently assessed 47-items of HTQ trauma events (Part I) and selected those items in the HTQ that they thought would meet the criteria for severe trauma. Any disagreements were resolved by further discussions. Out of the 47 war-related traumatic events listed in the Harvard Trauma Questionnaire, 17 items (36%) were considered to be of the severity that could be described as ‘severe’ war-related trauma (according to the above description of catastrophic trauma) and are presented in Table
2.
Symptoms of depression and anxiety were assessed using the Hopkins Symptom Checklist-25 (HSCL-25) [
27]. The HSCL-25 is a widely used self-report inventory which has been translated and culturally adapted to different populations across the world including communities of the former Yugoslavia [
29]. The HSCL −25 consists of 25 self-report items which are divided into a 10-point anxiety scale and 15-item scale of depressive symptoms that have been experienced in the week prior to the assessment. The ten anxiety symptoms included in the HSCL-25 are consistent with the DSM diagnosis of generalized anxiety disorder, whilst the 15 depression items are applicable to the DSM diagnosis of major depression. We used the recommended cut-off point of ≥1.75 for the HSCL-25 diagnosis of depression and anxiety [
27]. The HSCL-25 has been extensively validated in numerous studies on refugees, has high test-retest reliability and good validity in predicting depression and anxiety [
29,
30,
34].
The Social Functioning Questionnaire (SFQ) was used to assess participants’ levels of social dysfunction. This is an eight-item self-report measure that was developed from the Social Functioning Schedule [
35]. It was found to have good and robust psychometric properties and has been used in a variety of studies and was found to have good test-retest and inter-rater reliability as well as construct validity [
36,
37]. The SFQ score of 10 or more indicates poor social functioning and has been found to be positively associated with a diagnosis of personality disorder [
36].
Current alcohol and drug misuse were screened by two questions asking participants whether they were using alcohol and drugs (‘yes ‘or ‘no’ answers) at the time of data collection. If the answer was positive to either question, participants were also asked about the type of alcohol/drugs used and weekly amount they consumed.
Data analysis
Characteristics of the study sample were examined using univariate descriptive statistics. The relationship between categorical explanatory and outcome variables was examined using contingency tables. Differences in proportions were calculated with 95% confidence intervals (CI 95%). The statistical significance of differences was calculated using Chi square (X2) tests. Fisher exact test was used if any one cell had an expected frequency of <5.
Standard binary logistic regression was used to examine the relationship between exposure to severe war trauma among cases and controls, controlled for potential confounding effects of other variables (demographic factors). Odds ratios with accompanying 95% confidence intervals were calculated. Caseness (IPDE positive vs IPDE negative) was used as the dependent variable and predictor variables (after being checked for multicollinearity) were entered into a logistic regression model using standard (enter) method.