disorders in long-settled war refugees
3.2.3. Recruitment method: random sampling and linkage sampling
In classic statistical theory, if a random sample is drawn from a population whose underlying distribution is known, it may be assumed that the properties of the sample mirror those of the population i.e. a ‘representative’ sample is obtained. In most cases this means that researcher must have a sampling frame from which the sample is recruited. The challenge arises when a sampling frame is not available to a researcher as the subpopulation of interest is ‘hard-to-reach’ and ‘hidden’ in the general population.
Such a methodological challenge is frequently reported in epidemiological research among refugees (Bloch, 2007; Jacobsen & Landau, 2003; Spring, et al., 2003).
In the present study, a combination of random and non-random sampling approaches was adopted to recruit war refugees from former Yugoslavia in Germany, Italy and the United Kingdom. The challenges in recruitment of research participants and specific recruitment methods adopted in each country are described in detail below.
Recruitment in the UK
In the present study no census or immigration figures were available for refugees coming from the former Yugoslavia and residing in the UK. Neither official government organization (such as Home Office) nor non-governmental organizations (NGOs) have reliable statistical overview of the refugees from the former Yugoslavia and their demographic statistics. The most recent census prior to the study taking place was in 2001 and it contained no information on asylum seekers and refugees. It did report on country of birth but the data on Yugoslavian-born residents was reported under the scope of 'other Eastern Europe' (Office for National Statistics, 2004). An additional challenge was the UK's confidentiality laws (Data Protection Act, 1998), prohibiting release of refugee names and addresses.
In the absence of resident registers in the UK, potential interviewees were contacted through community organizations and snowball sampling. In the first step, all existing community organizations that were involved with populations originating from former Yugoslavia were identified. In total 31 community organizations were identified and
147 contacted, of which 16 accepted to collaborate in the study. Of the 16 collaborating community organizations, 10 were involved with the Albanian population coming from Kosovo, five with the Bosnian population, and one community organization was involved with population originating anywhere from the former Yugoslavia. In order to comply with the UK’s confidentiality law that prohibits release of individual’s name and address (Data Protection Act, 1998), community organizations mailed invitation letters and one reminder to all their members without revealing individuals’ names to the researchers. The invitation letters were provided in both mother tongue and English.
In total, 1493 potential participants were invited to participate.
To widen the sample beyond those in contact with community organizations (e.g.
socially isolated individuals and those who had left the refugee community to live in and affiliate with the majority society), snowball or linkage method was used, in which primary respondents were asked to provide the names and contact details of other refugees coming from the former Yugoslavia. Additionally, eligible refugees were recruited through interviewers’ friends.
Several strategies that have previously been successful with refugee communities were used. These included attending various social and cultural community gatherings, spreading information about the study within the community at community meetings, and receiving referrals from previous participants. Such strategies are believed to be helpful in building trust and relationship between researchers and participants who are in many cases suspicious of outsiders (Bloch, 2007; Spring, et al. 2003). This may be particularly important in the case of refugees suffering from posttraumatic stress.
Research has shown that people with posttraumatic stress who have a high level of avoidance are more likely to refuse to participate because they are avoiding any
reminders of the trauma (Schwarz & Kowalski, 1992). Recruiting refugees via community organizations and snowball sampling may to some extent overcome this problem as people who have a high level of avoidance may take part if approached via friends or a “friend of a friend” (a researcher). The importance of the relationship of the interviewer and participants is further emphasized by Lee and Lu (1989) who found that refugees were more likely to disclose trauma histories and symptoms to familiar, trusted and bilingual interviewers. It has been suggested that this sort of community approach increases response rate (Bhui, et al. 2003). This is central to recruiting as broad and representative a sample as possible and to attaining a high participation rate, as high refusal rates may bias the results (Spring, et al. 2003).
Participants were recruited throughout the UK (including London, Birmingham, Coventry, Newcastle, Derby, Doncaster, Dewsbury, and Leeds).
Recruitment in Italy
In Italy, data registers of persons coming from the former Yugoslavia were provided by the Registry Offices of the Town and City Councils in two Italian regions Reggio Emilia (Modena, Bologna, Forlì-Cesena, Ravenna, Rimini, Ferrara, and Piacenza and Parma) and Friuli Venezia Giulia Region (Trieste). Of the 354 Town Councils in Emilia Romagna Region, 333 were contacted and 169 provided data on potential participants. A single Town Council in Friuli Venezia Giulia Region (Trieste) was contacted and accepted to provide data on potential participants. Potential participants were identified from data registers based on available data on their surnames, country of origin and year of arrival to the host country. In total, 4626 potentially eligible
149 participants were identified. All potentially eligible participants were sent an invitation letter in Italian and inferred mother language. In case of no response, individuals were contacted up to three times at a three weeks interval. The random sampling method was supplemented with a small number of participants recruited through snowball sampling method.
Recruitment in Germany
In Germany recruitment was conducted across the Berlin and Saxony regions (Berlin, Dresden, Leipzig and Chemnitz). Registry offices in these cities provided data for approximately 22,772 potentially eligible participants, from which 5783 participants were randomly selected. As in Italy, potential participants were identified from data registers based on available data on their surnames, country of origin and year of arrival to the host country. Selected individuals were sent a letter with an invitation to participate in the study and up to two reminders. The invitation letter was provided in both mother tongue and German. The random sampling method was supplemented with a small number of participants recruited through snowball sampling method.
3.2.4. Measures
Semi-structured interviews utilizing several cross-culturally validated measures were conducted with all participants.
A short description of the measures used, as well as the rationale for their use, follows.
All measures used are also presented in appendix B.
Socio-demographic data about the participants were obtained on a brief structured questionnaire developed by the study team. Within the questionnaire the following areas were assessed: general socio-demographic data (gender, age, marital status, education level, and employment status) and issues related to exile (country of origin, duration of exile, current legal status and time taken to achieve it).
History of potentially traumatic experiences
War-related Stressors Screening Checklist was used to assess whether the interviewee had experienced a potentially traumatic event related to the war, and if so, was she or he at least 16 years old at the time. The list consisted of 20 potentially traumatic war events. The experience of any of the events on the list was sufficient for inclusion.
The Life Stressor Checklist-Revised (LSCL-R)
The history of potentially traumatic experiences before, during, and after the war was assessed using an amended version of the 24-item Life Stressor Checklist-Revised (Mollica, et al., 1992). This instrument assesses whether participants had ever had one or more of 24 traumatic experiences (e.g. injury, shelling, rape or sexual assault, combat, loss of a family member, torture, natural catastrophe) that would be sufficient to qualify as a traumatic event in DSM-IV. For each event experienced the following data were recorded: how many times; year when it occurred (year of the first time, the last time and the most traumatic time in case of a multiple trauma); duration of event when trauma was prolonged and lasted longer than one day (measured in days); whether it was related to war; and the level of personal distress caused by the event at the time of the trauma (measured on a 5-point Likert scale, ranging from 0 = not at all to 4 =
151 extremely, I could hardly stand it). If the event lasted for a longer time, or had been repeated continually, many times over a longer period of time, the period was registered in which the event took place (e.g. 1991-1995). Cumulative scores were calculated for pre-war, war and post-war experiences. A similar version of the LSCL-R has been previously used with sample from former Yugoslavia (Jankovic Gavrilovic, et al., 2005;
Priebe, et al., 2009).
Post-migration stressors
This is a short checklist constructed to capture different aspects of the participants’ life situation since migration. Questions covered the following areas: a) separation from family, b) employment and financial difficulties, c) inadequate accommodation, and d) access to medical care. A cumulative score of the number of stressors experienced was obtained (range 0-6).
Additional single questions on acculturation were assessed: a) level of perceived acceptance by the host culture, b) host language proficiency, (both measured on Likert scale, ranging from 1 = not at all to 5 = entirely), and c) number of social contacts with countrymen and with friends from the host-country. Previous studies have reported high correlations between single-item self-reported measures of acculturation and objective measures of language ability as well as with other important explanatory variables (Beiser & Hou 2000; Westermeyer & Her, 1996), demonstrating their utility and validity.
Past and current psychopathology
The Mini International Neuropsychiatric Interview (MINI)
The MINI (Sheehan, et al., 1998) was used to assess current and lifetime psychiatric diagnoses. It is a brief structured diagnostic interview eliciting all the symptoms listed in the symptom criteria for DSM-IV and ICD-10 for 15 major Axis I psychiatric categories, one Axis II disorder and for suicidality. Its diagnostic algorithms are consistent with DSM-IV and ICD-10 diagnostic algorithms.
The MINI was validated by a cross-national study involving more than 600 subjects (Sheehan, et al., 1998). The instrument has been found to be scientifically valid and reliable overall when measured against the much longer Composite International Diagnostic Interview (C.I.D.I.) and the Structured Clinical Interview for DSM-III-R Patients (S.C.I.D.-P) (Sheehan, et al., 1998). The reliability of the interview has been shown in various cultures (Kadri, et al., 2005; Lecrubier, et al., 1997; Otsubo, et al., 2005; Rossi, et al., 2004).
The MINI has two to four screening questions per disorder and additional symptom questions within each disorder are asked only if the screening questions are passed. The questions are precise and require only ‘yes’ or ‘no’ answers, reducing interpretational error during the interview (Bhui, et al., 2006). The measure has been found to be sensitive to changes over time allowing the identification of minor modifications of the patients’ clinical conditions (Rossi, et al., 2004).
In the present study diagnoses assessed with the MINI included major depression (with or without melancholic features), dysthymia, suicidality, hypomania, mania, panic disorder, agoraphobia, social phobia, obsessive-compulsive disorder, PTSD, alcohol abuse or dependence, drug abuse or dependence, psychotic disorder, mood disorder
153 with psychotic features, and antisocial personality disorder. Of note, the original MINI was supplemented by an additional module on somatisation disorder from the MINI Plus (Sheehan, et al., 1998), since a high prevalence of somatisation has been previously reported in refugees and other war-affected populations (Klaric, Klaric, Stevanovic, Grkovic, & Jonovska, 2007; Cardozo, et al., 2000; Salama, Spiegel, Van Dyke, Phelps, & Wilkinson, 2000).
The modules on eating disorders (anorexia nervosa and bulimia nervosa) were excluded from the interview schedule, since these disorders were found to have low prevalence in the population from former Yugoslavia (Broers, et al., 2006). Furthermore, a modified version of the PTSD module was used to allow for assessment of past (since the critical war-related traumatic event) as well as current (a month prior to the interview) symptoms of PTSD. Both current and lifetime diagnoses of mania, hypomania, panic disorder, PTSD, psychotic disorder, mood disorder with psychotic features, antisocial personality disorder and somatisation disorder were assessed, whilst only current diagnosis was assessed for the remaining disorders.
The utility of the MINI as a screening tool to identify those at risk in the population from former Yugoslavia (e.g. Eytan, Guthmiller, Durieux-Paillard, Loutan, & Gex-Fabry, 2011; von Lersner, et al., 2008) and other ethnic groups (Bhui, et al., 2006) has been demonstrated previously. When deciding whether to use the MINI or the MINI Plus, a more detailed version that helps mainly with the diagnosis of Psychotic and Mood DSM-IV Disorders, length of the interview was taken into account. Although the MINI Plus covers a much wider spectrum of disorders, it is also significantly longer to administer.