Participants of the screening interview were selected from a sample of 666 previously identified and electronically registered ex-combatants on the government payroll of the Republic of Somaliland (North-Western Somalia1) and selected to be participants of the Pilot Demobilization and Reintegration Program (DRP). Out of this group, 195 people were randomly selected and asked to participate in an interview, i.e. the screening for PTSD (from now on referred to as 'screening interview'). Despite the fact that they were on the government payroll, 47 could not be tracked due to their nomadic life style and lack of permanent residence. In addition, one person had died, four had moved to another town or country, and two were imprisoned. The remaining 141 ex-combatants were contacted by project staff, informed about the intention and procedure of the assessment, and invited to take part in the screening interview. Four individuals refused to participate. Of the remaining 137 participants, 2 did not complete the screening interview (response rate 135 of 141, i.e. 95.7%). The remaining 135 subjects were screened for PTSD symptoms by trained local staff (all non-experts). Additional topics of the screening interview consisted of demographic and clinical data such as the consumption of the stimulant drug khat. At the Horn of Africa and the neighboring regions, khat is a traditionally consumed substance with amphetamine-like properties [20, 21], which is not illegal.
The first 64 interviewees of the screening interview were asked to participate in a second assessment conducted 2 to 14 days later. 62 of them completed this interview (response rate 96.9%). This assessment included a structured clinical interview conducted by a team of international researchers and clinicians who specialize in trauma. Trained interpreters assisted with these interviews. This second interview will be referred to below as the 'validation interview'. All interviewers and interpreters of the validation interview were blind with respect to the outcome of the screening interview.
Of the 135 participants of the screening interview, 133 were men and two were women. Their ages ranged from 19 to 70 years. Participants were involved in three different sections of the Somaliland armed forces: army, police, and custodian corps (prison wards)2. All participants were former members of the 'Somali National Movement' (SNM) and were receiving a monthly salary from the armed forces at the time of the study.
From this sample, 62 men and the 2 women were selected for the validation interview. This sample did not differ from the 71 ex-combatants who only participated in the screening interview with respect to age (M = 34.0, SD = 9.5 years vs. M = 34.3, SD = 10.2; t = 0.190, df = 132, p = .849), body mass index (M = 19.2, SD = 2.6 vs. M = 19.2, SD = 3.0; t = -0,072, df = 129, p = .943), military branch (army: 51.6% vs. 43.7%, police: 25.0% vs. 32.4%, prison wards: 23.4% vs. 23.9%; χ2 = 1.084, df = 2, p = .582), or on the average amount of money spent per day on khat in the week preceding the screening interview (M = 1.04, SD = 1.66 US$ vs. M = 0.66, SD = 0.99 US$; t = -1.507, df = 108, p = .135). Importantly, the sum score of the screening instrument (Somali-PDS) did not differ either (M = 11.0, SD = 9.9 vs. M = 10.3, SD = 9.2, t = -.456, df = 133, p = .649).
The validation interview revealed that the average age of the ex-combatants when they started to actively fight in the war was 18.6 years (range 11 to 32 years; SD = 5.3; n = 54). At the time of their first military operation 69% were 18 or younger, 43% were 16 years or younger and 30% 15 or younger. Thus, a large fraction of the sample comprised former child soldiers. At the time of the validation interview, ex-combatants had an average of 5.2 years of formal education (SD = 4.2); 53% of them were married and their household included on average of 8.7 persons (SD = 5.1).
The screening interview assessed for symptoms of PTSD using a modified version of the Posttraumatic Stress Diagnostic Scale (PDS; ). The scale had been adapted to the Somali language, culture, and Islamic religion (Somali-PDS) according to recommendations for cultural adaptation . The PDS is a widely used self-report instrument for the assessment of PTSD according to the DSM-IV criteria with good psychometric properties and validity [23–25]. According to Foa , the instrument achieved a Cronbach's Alpha of .92, test-retest reliability of .83, and a kappa of .74 (compared to the SCID-PTSD module) in a sample of 248 treatment-seeking individuals. These results are similar to those derived from two samples of general psychiatric outpatients  and of battered women .
In the first part of the instrument, a list of potentially traumatic events is presented and the respondent is asked to mark those event types that he or she experienced during his or her life. The participant is then asked to briefly describe the worst of these events and to indicate whether or not he or she felt extreme anxiety or helplessness during the event. In the second part of the screening interview, the 17 DSM symptoms of PTSD are assessed in reference to the worst event. Participants are asked to rate the frequency of each symptom for the past four weeks on a 4-point scale (0 'not at all/only one time' to 3 'five or more times a week/almost always'; a symptom is counted if a score of 1 or higher is selected), as well as to indicate how long they have been experiencing these symptoms and how soon the symptoms began following the event. The next segment assesses difficulties in everyday functioning related to these symptoms. The scoring method established by Edna Foa is based on DSM-IV criteria for PTSD and was applied in this study: A positive screening case must fulfill all seven criteria indicated in the DSM-IV. The screening interview also included the assessment of demographic information (name, age, gender, military branch) and khat consumption (average money spent daily on khat during the last week).
Because many of the participants were illiterate, the self-report scoring of the PDS was adapted to an 'assisted self-report'. All items and answer categories were read to respondents by the interviewer. The interviewers marked the answers on the form without further probing. If a respondent indicated that he or she did not understand the meaning of the item the interviewer repeated the exact wording. If the respondent was not able to provide an answer the interviewer assisted by offering an alternative wording of the item without actively inquiring or probing. Interviewers received extensive training for this procedure.
The validation interview used the Composite International Diagnostic Interview for the DSM-IV (CIDI; WHO, 1997). This included the PTSD module (section K), and 13 items of the schizophrenia module (section G; G1, G2, G4, G6, G10, G14, G17, G18, G19, G20, G21) – the latter because psychotic symptoms are frequently co-morbid in veterans with PTSD  and their development was related to excessive khat chewing [1, 27]. The CIDI has already been used in cross-cultural studies  and its excellent psychometric properties have been reported [29, 30]. The former DSM-III-R PTSD module has been criticized for being less sensitive in detecting disorders  and has been extensively modified to meet the DSM-IV criteria . Other studies criticized the strict skipping rules . Based on these criticisms, clinicians in our study were instructed to ask and probe all items of the PTSD module and all selected items of the schizophrenia module. Psychotic symptoms were only included if the symptom was not related to dissociative phenomena or flashbacks.
In addition to PTSD and psychotic phenomena, symptoms of anxiety and depression were measured using the Self-Report Questionnaire-20 (SRQ-20; [33, 34]). Items were read to the participant and the interviewer recorded the answers. The validity of answers to SRQ items were examined by probing questions .
In order to assess for exposure to traumatic events, a standard list, which asked for 15 situations with high face validity for the Somali military context, was used ('yes-no' format): fighting in combat (reported by 82%), witnessing combat (76%), killing or wounding enemies in combat (48%), being confronted with dead bodies in combat (88%), experiencing a life threatening accident or explosion (35%), witnessing serious accident or explosion (53%), suffering an injury by weapon (56%), witnessing injury by weapon (83%), witnessing violent death of relative or friend (52%), witnessing murder not in combat (35%), experiencing severe beatings or torture (26%), witnessing beatings or torture (21%), experiencing violent confiscation of property by officials (32%), experiencing harassment by armed personnel (36%), experiencing imprisonment (62%). The internal consistency of this list was satisfactory (Cronbach's α = .76).
Socio-demographic information and minor physical symptoms in the preceding month (cough, diarrhea, fever, hyperventilation, constipation, other; Cronbach's α = .67) were also assessed in the validation interview. Because it is well documented that ex-combatants with PTSD abuse psychotropic drugs more frequently compared to the ones without PTSD  khat consumption was quantified by items that already proved to be valid in field studies . We assessed the money spent on khat in the week prior to the interview and the average time spent chewing khat per day. We also assessed the average number of cigarettes consumed per day as khat consumers usually smoke when chewing . The average number of hours of sleep per day in the previous week was also assessed as patients with PTSD often have sleep difficulties [38, 39].
Cultural adaptation and translation of the PDS
The translation of the PDS to the Somali language as well as its cultural adaptation was carried out by groups of local bilingual and international experts, all of whom had received education in trauma-related concepts. In addition, group discussions and consultations with external specialists were dedicated to culturally specific meanings of items and typical experiences in Somalia. For example, we found no adequate Somali terms for the concepts 'stress' and 'trauma' and needed to circumscribe the meaning or find similar words. For example we translated 'stressful event' as 'difficult event', or 'traumatic event' as 'fearful incident' or 'reliving the traumatic event' as 'behaving as if you are once again in the situation that has caused you fear'. As a result, items such as those concerning rape and sexual experiences were modified to meet cultural and religious requirements. We could not ask directly about sexual abuse and violent sexual experiences, but had to design a hierarchical set of consecutive questions. Due to cultural and religious restrictions, we assessed only for rape and did not inquire into sexual contacts and molestation during childhood. Each subsequent question was asked only if the answer to the question before was positive. First, we asked whether a respondent had ever heard about a rape, then whether he had witnessed a rape. The next question would have been whether he knew the victim, and lastly whether he himself was the victim. The process of translation included a back translation, which was performed by independent professional translators. Items that were judged to be problematic were subjected to extensive discussions and retranslations, and were independently discussed with a second group of local staff. The process of back translation occurred as many times as necessary until all items had a clear and correct meaning.
Training of local interviewers
Six local interviewers underwent a 10-day training by international researchers and clinicians. The training included theoretical education and practical exercises. Additionally, during the first two interviews they were directly supervised. Subsequently, team supervision was continuously provided throughout the four weeks of the screening exercise. The interpreters participated in the same theoretical education as interviewers, and translated and discussed all items of the CIDI and the structured clinical interview with expert team members as part of their training.
The screening interview lasted 20–40 minutes. All interviews took place in the Somalia Demobilization and Reintegration Program center in February and March 2002. Prior to the screening interview, local interviewers read a standardized explanation of the procedure to the participants, answered remaining questions, and asked them to sign an informed consent form. An additional informed consent was obtained for the validation interview, which was accomplished with the help of trained local interpreters and lasted approximately two hours.
The design, procedures and psycho-diagnostic instruments of the study were ethically approved by the Somaliland National Demobilization Commission (NDC) and the Ministry of Resettlement and Rehabilitation, Government of Somaliland, as well as by the German Technical Cooperation (GTZ).
Reliability of the Somali-PDS was evaluated using Cronbach's α (internal consistency). Convergent validity of the screening outcome was evaluated using kappa and coefficients of sensitivity and specificity. Receiver-operator curve (ROC) analysis was conducted to examine the diagnostic utility of the screening instrument compared to the CIDI . Group differences were confirmed by student's t-Test (or Wilcoxon's test when not applicable) and Chi2 test (or when appropriate Fisher's test). The data was analyzed using SPSS, version 11 for Macintosh.