Assessing Trauma in Sierra Leone
11 January 1999
Target Population & Sample The survey was conducted after receiving the permission of the appropriate authorities, during the first two weeks of May 1999, four months after the atrocities in Freetown. Because everyone in Freetown had been subjected to traumatic experiences, both Internally Displaced Persons (IDP's) and residents were included in the sample. A two-stage cluster sampling method was used, a methodology based on vaccination surveys. The methodology is extensively described in the various handbooks of WHO. The sampling method entails a first phase where 30 clusters are chosen. In the second phase a pre-set number of individuals are chosen per cluster. The sampling technique itself ensures that every individual has an equal chance to be chosen. The result obtained through sampling techniques is an approximation of the real value in the entire population. The real population value is in a range around the value obtained by the sampling method. The narrower the range, the more precise is the estimation. The precision depends on the sample size and the inter-cluster variation and the intra-cluster variation of the specific survey. The precision of the results with this two-stage sampling technique is less than the precision one would get with a random sampling technique. The sample consisted of 30 clusters of 8 respondents, as the intra-cluster variation was thought to be reasonably small, since most traumatic events take place on a community level and not on an individual level. The sampling frame is based on the 1997 census of the Ministry of Health and UNICEF, which gives a population of 600,000. The rural part of the Western area (encompassing Freetown and its peninsula) was excluded because most of the area was not accessible during the survey for security reasons. The areas (clusters) were chosen with a chance proportional to the population size. The teams went to the center point of these areas; a pen was spun to determine the direction and every tenth house to the right was selected until the eight necessary for the cluster had been identified. The most senior member of the household present was interviewed. Any refusals were noted and the selection process continued to the next tenth house. There was a note made on each questionnaire of the displaced or resident status of the interviewee. Where the cluster was in a displaced camp one person from each section of the camp was interviewed, depending on the layout of the camp. Four survey teams were selected. Each team had to conduct eight interviews each day. All interviews were scheduled in the first two weeks. Eight interviews per day per team were the maximum due to the difficult nature of the information gathered. Training The survey teams consisted of two trained local counselors who did the interviews and a support team of one expatriate staff member and a driver. The training consisted of the following elements: introduction to Doctors Without Borders, the nature and purpose of the survey, confidentiality of the data and information, survey technique, data registration and task division among crews. Some survey questions might have provoked strong emotions, so the counselors received special training on how to deal with them. They were also informed on referral possibilities for those in need of follow-up psychosocial support. Counselors practiced interviewing skills on each other. The items of the questionnaire were discussed in depth until a final interpretation was agreed on each question. A pilot study of eight interviews was carried out by the teams in the National Stadium IDP site, Kingtom area, Aberdeen Junction and Murray Town. After the pilot interviews, problems of interviewing, sampling and approaching people were discussed. Ambiguities in the questionnaire were addressed. The training (including the pilot study) lasted two days. The Interview The counselors worked in pairs. After the counselors introduced themselves and MSF, the purpose of the survey was explained to the potential participant. In the introduction it was clearly stated that the participant would not receive any compensation, that the data were treated confidentially and that the interview would last for a maximum of 40 minutes. After the introduction the participant could decide whether to participate. The timing of the interviews was crucial, since people had to be at home and be available. It was important that the participants completed the survey. To avoid exceeding the interview time it was explained that direct and short answers were necessary. Extra discussions or conversations were avoided. However, the counselors were permitted to stop or interrupt the interview when they deemed the questions to be too emotionally upsetting for the participant. When the counselor believed that the participant needed follow-up support, referral to professional counselors was facilitated. All teams had a daily technical and emotional debriefing. Further emotional support for the counselors was provided through the MSF psychosocial peer support system for national staff, which was trained by the MSF Amsterdam Public Health Department and Psychosocial Care Unit. The Psychosocial Questionnaire The structured interview was based on a questionnaire consisting of 35 questions with subdivisions. To control the time of the interview most questions offered a limited number of alternatives from which the participant could choose. Only two questions in the health section of the questionnaire were open ended. To limit the emotional burden the questions were put as factually and simply as possible. When unclear, a short explanation was allowed. Participants were not allowed to fill the questionnaire later nor were they permitted to study the questionnaire in advance. Interviewers had to respect confidentiality at all times. No trans-cultural tools to measure traumatic stress are available. To assess the level of trauma, three important indicators of traumatic stress were measured. The first indicator is the presence of a potential traumatic event. The second indicator is the impact of event scale, which expresses the extent of traumatic stress response. The third indicator appraises physical complaints, which likely are correlated to traumatic stress. When all three indicators of traumatic stress were positive, at least strong circumstantial evidence for the prevalence of traumatic stress was found. The psychosocial questionnaire was composed of four sections. The first section assessed the demographics and personal background of the participant. A second section appraised traumatic events such as exposure to violent situations, who was lost and the traumatic events witnessed. Both the number of traumatic experiences and their length are important risk factors in the development of PTSD (Kleber & Brom, 1992). The third section measured the impact of these events. To measure the prevalence of traumatic stress responses the Impact of Event Scale was used (Horowitz, Wilner & Alvarez, 1979). This psychometric instrument assesses two central dimensions of coping with drastic life events: intrusion and denial. It has been used worldwide and generally consistent structures have been found across samples and situations (Dyregrov, Kuterovac & Barath, 1996; Joseph, Williams, Yule & Walker, 1992; Robbins & Hunt, 1996; Schwarzwald, Solomon, Weisenberg & Mikulincer, 1987; Silver & Iacono, 1984; Zilberg, Weiss & Horowitz, 1982). Despite its wide use, interpretations of the outcomes should be done with appropriate care since the Impact of Event Scale is not validated either for Western Africa or for Sierra Leone. The final section of the questionnaire evaluated current physical health complaints and needs. PTSD is frequently associated with somatization. Physical symptoms like headaches, stomach problems, general body pain, dizziness or palpitations are often expressed by people suffering from traumatic stress. A high prevalence indicates a possible high level of traumatic stress or PTSD. Physicals are registered by means of open-ended questions. The access to health care and the perceived health levels were registered using the Lickerd scale. Data registration The forms were registered anonymously. Data were entered in a spreadsheet in EXCEL, and data were analyzed by EXCEL and EPIINFO-6. Results All four teams contributed equally to the survey (each 25%). The fixed number of interviews in each cluster (N=80) was extended in four clusters (Old Warf, Aberdeen, Approved School/Kuntoloh, National Workshop). The total number of respondents was 248 (N= 248), of which three respondents were excluded because they were younger than 15 years. Demographics (First section) In total 91 (37%) respondents were recently displaced; only 66 (27%) were residents. The others (37%) could not be placed in one of these categories. A possible explanation is that many people had been displaced in earlier years. About half (52%) of the respondents were female (Confidence interval 95% level: 46.4 Ã? 56.8). The age of the respondents varies from 15 up to 81 years with a majority of the respondents in the middle age group of 35-44 years (29%). The majority has attended primary school, also in the older age groups; on average 30% have not had formal education.