It is 4 years since civil unrest and fighting broke out in Syria in the Middle East. The conflict has developed into a humanitarian and public health catastrophe, resulting in extraordinary levels of violence, human rights violations and millions being uprooted from their homes and forced into severe deprivation where resources are limited and living conditions poor. Over 12 million people need humanitarian assistance inside Syria (7.6 million are internally displaced) and nearly four million have been externally displaced to neighbouring countries Turkey, Jordan and Lebanon (UNHCR, 2015). Half of those affected are children. The risk of children becoming the ‘lost generation’ is recognised as a serious concern (UNICEF, 2013).
It is well documented that children exposed to war and political violence are more vulnerable to mental health problems as a consequence, in particular, posttraumatic stress disorder (PTSD) and depression (Fazel & Stein, Reference Fazel and Stein2002; Quota et al. Reference Quota, Punamäki and El Sarraj2003; Lustig et al. Reference Lustig, Kia-Keating, Knight, Geltman, Ellis, Kinzie, Keane and Saxe2004; Fazel et al. Reference Fazel, Wheeler and Danesh2005; Attanayake et al. Reference Attanayake, McKay, Joffres, Singh, Burkle and Mills2009; Bronstein & Montgomery, Reference Bronstein and Montgomery2011; Dimitry, Reference Dimitry2011) with refugee children living in camps as opposed to urban or rural areas more likely to suffer (Khamis, Reference Khamis2005). Children are also at a higher risk of developing behavioural problems, especially aggression, as a result of the traumatic events that they have witnessed or been a victim to (Quota et al. Reference Quota, Punamaki, Miller and El-Sarraj2008). The onset of these problems in childhood has a long-term impact on the child, their family, friends, society and future generations. Building resilience and optimising child and adolescent mental health, especially in humanitarian settings, has high priority on international research agendas for prevention and treatment (Mollica et al. Reference Mollica, Cardozo, Osofsky, Raphael, Ager and Salama2004; Tol et al. Reference Tol, Barbui, Galappatti, Silove, Betancourt, Souza, Golaz and van Ommeren2011b; Tol & van Ommeren, Reference Tol and van Ommeren2012), as poor mental health accounts for a significant proportion of disease burden and disability among children and adolescents worldwide (Patel et al. Reference Patel, Flisher, Hetrick and McGorry2007, Reference Patel, Flisher, Nikapota and Malhotra2008; UNICEF, 2005).
In conflict zones, recruiting participants into research studies and collecting reliable and valid data are complex, challenging and can be near impossible in some settings. The specific nature, intensity and length of particular conflicts, the challenges this presents and impact it has on infrastructure, routine and normal lifestyle together with different cultures across war-affected countries makes it difficult to transfer data collection approaches from one conflict to another. A recent systematic review undertaken by the United Nations High Commission for Refugees found that there are no peer-reviewed published studies on the impact of the conflict in Syria on the mental health of refugees and those internally displaced (Quosh et al. Reference Quosh, Eloul and Ajlani2013). Thirteen studies, five quantitative and eight qualitative, were identified from the grey literature which included a diversity of document types produced by humanitarian and non-governmental agencies and organisations, academics and governments, ‘where publishing is not the primary activity of the producing body’. Less than half of the studies that met the inclusion criteria for the review focused specifically on children. Insecurity was reported as causing difficulties in conducting assessments inside Syria. Furthermore, contexts of conflict can pose serious risks to the safety of researchers and participants. It is critical that feasible and ethical recruitment and data collection approaches that can be used safely in the Syrian conflict are identified rapidly as part of the essential process of reliably identifying levels of mental health difficulties among children, so that screening and pre- and post-intervention measures can be administered to identify children's psychological needs and evaluate the effectiveness of interventions used to address them. The primary aim of this study therefore was to test the feasibility of recruiting refugee caregivers of children affected by the Syrian conflict living in and near refugee camps on the Syrian–Turkish border into a questionnaire-based study and of collecting reliable data from them.
The second aim of the study was to examine the extent to which complete data could be obtained from refugee caregivers under field conditions such as those experienced in and near refugee camps, and to identify the level of help parents needed to complete measures. Extreme levels of stress, poor physical and/or mental health, literacy difficulties, poor living conditions and insecurity may contribute to refugees finding it difficult to complete questionnaires. Collecting data in contexts of conflict require psychometrically sound, culturally sensitive and developmentally appropriate screening instruments that are inexpensive, can be quickly and accurately measured, are non-burdensome and require limited training to administer. Two measures which fit these criteria were chosen for the current study.
The Pediatric Emotional Distress Scale (PEDS; Saylor et al. Reference Saylor, Swenson, Reynolds and Taylor1999), which was designed specifically to measure trauma-related symptoms in children as young as 2 years of age, can be completed in 5–8 minutes and has acceptable psychometric properties as found in independent, geographically and developmentally diverse samples of children exposed or unexposed to traumatic events, including natural disasters and sexual abuse (Stokes et al. Reference Stokes, Saylor, Swenson and Daugherty1995; Saylor et al. Reference Saylor, Swenson, Reynolds and Taylor1999). We are not aware of any published data on the PEDS used in conflict.
The Strengths and Difficulties Questionnaire (SDQ; Goodman, Reference Goodman1997) is an inexpensive, brief, user-friendly and effective screening measure of child behavioural, emotional and social problems. It is well established, translated into over 60 languages with normative data available in 10 high-income countries and has been validated in many contexts internationally allowing comparisons to be made between different populations. It is the only widely used mental health scale to have been validated in Arabic. While not trauma specific, it has been used with mixed success in studies to assess the impact of conflict-related trauma on behaviour in children resettled in high-income countries (Papageorgiou et al. Reference Papageorgiou, Frangou-Garunovic, Iordanidou, Yule, Smith and Vostanis2000; Fazel & Stein, Reference Fazel and Stein2003; Derulyn & Broekaert, Reference Derulyn and Broekaert2007; Nielsen et al. Reference Nielsen, Norredam, Christianse, Obel, Hilden and Krasnik2008; Vaage et al. Reference Vaage, Tingvold, Hauff, Van Ta, Wentzel-Larsen, Clench-Aas and Thomsen2009; Durà-Vilà et al. Reference Durà-Vilà, Klasen, Makatini, Rahimi and Hodes2012) and during the flight phase in displaced children exposed to war in the Middle East ( Thabet et al. Reference Thabet, Stretch and Vostanis2000, Reference Thabet, Karim and Vostanis2006) and Afghanistan (Panter-Brick et al. Reference Panter-Brick, Eggerman, Gonzalez and Safdar2009). It has also been used to measure the effect of psychological intervention on post-traumatic stress in Palestinian children in randomized controlled trials (Quota et al. Reference Quota, Palosaari, Diab and Punamaki2012; Barron et al. Reference Barron, Abdallah and Smith2013).
The third aim of the study was to gain insight into emotional and behavioural difficulties among refugee children living in and near refugee camps on the Syrian–Turkish border. The Bahçeşehir Study, a cross-sectional survey conducted in Islahiye Refugee Camp, Turkey (housing over 4000 children) found that in a sample of 185 girls and 103 boys aged 9–18 years 45% were experiencing parent-reported symptoms of PTSD, 60% depression, 22% aggression and 65% psychosomatic symptoms that seriously reduced their level of functioning (Özer et al. Reference Özer, Sirin and Oppedal2013). Seventy four per cent of these children had experienced the death of somebody they cared about and 50% had been exposed to six or more traumatic events. Mobayed (Reference Mobayed2014) (cited in Abou-Saleh & Mobayed, Reference Abou-Saleh and Mobayed2013) found that in two refugee camps the prevalence of PTSD among 129 Syrian children aged 10–16 years was 41%–76%; these rates were higher than those for adults (36%–62%).
This study therefore aims to contribute to programmes of research in the Syrian conflict identifying refugee children's mental health needs accurately and reliably in order to evaluate interventions used to address them.
Materials and methods
Participants were adults caring for children aged 4–10 years and who had been forced to flee their homes due to armed conflict in Syria. Table 1 shows their characteristics. They were recruited from primary schools run by Generation Freedom in two different refugee camps, Qah Refugee Camp and Bab-Al-Salam Refugee Camp, in two locations in the northern part of Syria, and near a refugee camp in Reyhanli in southern Turkey away from immediate armed conflict. Qah Refugee Camp was established in 2012 and hosts more than 20 000 refugees. The continual movement of families on a day-to-day basis means that these figures have likely changed since this was written.
a Missing value for one participant.
b 1 step-mother, 5 aunts.
c 1 aunt was a widow.
d 1 spouse absent, 2 separated/divorced, 11 widowed.
e Child care, sewing, internet/phone-based work etc.
f includes stay-at-home parents, being retired.
The primary caregiver was the adult who had primary responsibility for caring the children and was also the respondent who completed the questionnaire.
Generation Freedom is a sub-organisation of Watan who specialise in educating children aged between 7 and 17. Watan is a Syrian non-governmental organisation (NGO) working internationally and locally in Turkey, Syria, Jordan and Lebanon to provide humanitarian assistance to people affected by the Syrian conflict. The locations for the current study were chosen for practical and security reasons. Generation Freedom has extensive knowledge and experience of working in these locations having provided humanitarian assistance here since the conflict started. This allowed the researcher to safely cross the border and access the camps and schools.
Family background questionnaire
An abbreviated, adapted version of the Family Background Questionnaire (FBQ) (Sanders et al. Reference Sanders, Markie-Dadds and Turner2000) was used to obtain demographic information, both general (e.g. child age, gender and relationship to caregiver) and context-specific (e.g. length of time displaced and current living situation).
The PEDS (Saylor et al. Reference Saylor, Swenson, Reynolds and Taylor1999), completed by parents, comprises 17 general behaviour items rated on a Likert scale from 1 (‘Almost never’) to 4 (‘Very often’) which contribute to three subscales: Anxious/Withdrawn (six items) (e.g. ‘Seems sad and withdrawn’), Fearful (five items) (e.g. ‘Refuses to sleep alone’) and Acting out (six items) (e.g. ‘Has temper tantrums’). There are an additional four trauma-specific items. Total scores, based on the first 17 items, range from 21 to 84. Total PEDS and Acting Out scores have been found to correlate with the Eyberg Child Behaviour Inventory (r = 0.62, p < 0.001 and r = 0.86, p < 0.001, respectively) (Frederick, Reference Frederick and Pynoos1985a). Sensitivity to change and convergent and discriminant validity in clinical samples (based on a cut-offs determined through use of the DSM-IV-TR) have been found (Saylor et al. Reference Saylor, Swenson, Reynolds and Taylor1999). Cronbach's alphas in the current study were 0.72 for Anxious/Withdrawn, 0.75 for Fearful, 0.72 for Acting Out and 0.83 for the total score.
Strengths and difficulties questionnaire
The parent version of the SDQ (Goodman, Reference Goodman1997) for 4–16-year olds screens for behavioural, emotional and social problems based on ICD-10 and DSM-IV criteria using 25 items rated on a 3-point Likert scale ranging from 0 (‘Not True’) to 2 (‘Certainly True’). Single informant ratings have shown satisfactory reliability and validity in high-income countries (Goodman, Reference Goodman2001) and across culturally diverse settings including Bangladesh (Goodman et al. Reference Goodman, Renfrew and Mullick2000; Mullick & Goodman, Reference Mullick and Goodman2001), Pakistan (Samad et al. Reference Samad, Hollis, Prince and Goodman2005; Syed et al. Reference Syed, Hussein and Mahmud2007) and Yemen (Alyahri & Goodman, Reference Alyahri and Goodman2006) where the Arabic version of the questionnaire demonstrated adequate discrimination between community and clinic 5–10-year-old children. Internal consistency in the current study was α = 0.57 for Emotional Symptoms, α = 0.59 for Conduct Problems, α = 0.60 for Hyperactivity, α = 0.38 for Peer Problems, α = 0.82 for Prosocial Behaviour and α = 0.66 for Total Difficulties.
Ethics approval for the study was obtained from The University of Manchester. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. This included very detailed risk assessments. Participant information sheets, consent forms and the PEDS were translated into Arabic with back-translation and printed in the UK before being taken by the researcher (A–E) to Generation Freedom run makeshift primary schools in or near refugee tented camps on the Syria–Turkey border. Translation and back-translation of the PEDS were carried out by two translators at a professional translation service. One translator conducted the translation and the other the back-translation. Discrepancies were discussed and resolved between the translators. Generation Freedom staff supported the researcher with border crossings, access to the refugee camps and schools during data collection and maintained the researcher's safety. Data were collected over a period of 23 days between March and April 2013 with 1 week spent in each location. Following permission to approach parents from the Director of Generation Freedom, parents were approached by the researcher on a one-to-one or group basis. The school in the Qah Refugee Camp was located in a large building and in five purpose-built tents in the Bab As-Salam Camp. In Reyhanli, data were collected in a building housing 700 refugees across the road from a school and near to a refugee camp. Primary caregivers of children were given a participant information sheet and the opportunity to ask questions. Those who wished to take part were given a consent form and the questionnaires and either completed both immediately at the school or took them away and returned them the following day.
Data were analysed using IBM SPSS Statistics for Windows, Version 22 (IBM Corp, 2013). First, descriptive statistics for complete and missing data on the FBQ, PEDS and SDQ were obtained including the proportion of: (1) participants with complete and missing values on at least one item of the questionnaires, (2) items with complete and missing values on each questionnaire; (3) complete and missing values across all items of the questionnaires. Descriptives of completeness of data were obtained for participants who were not helped by the researcher to complete the questionnaires v. those who were helped by the researcher. Second, internal consistency of the PEDS and SDQ was established using Cronbach's alpha coefficients for the scales as a whole as well as the individual subscales. Items on the PEDS and SDQ with less than 10% incomplete data were replaced with the mean for that item before Cronbach's alpha coefficients were calculated. One participant was excluded from the reliability analysis for the SDQ due to 15 incomplete items. The reliability analysis for the SDQ Total Difficulties and Hyperactivity scales excluded item 25 due to the amount of incomplete data for this item (27.4%). Finally, means, standard deviations and range of scores on each of the subscales of the PEDS and SDQ, their total scores and the proportion of children with scores at or above clinical cut-offs were calculated. The frequencies of responses to each of the four trauma-specific items of the PEDS were also calculated.
Feasibility of recruitment and distributing and receiving back questionnaires
The researcher was able to distribute written participant information about the study and provide a verbal explanation of the study to caregivers of children at primary schools in and near refugee camps without causing apparent additional distress to parents. It was possible to obtain written informed consent for participation. Questionnaires were distributed to 144 parents and completed and returned by 106 parents (response rate 74%).
Extent of completion of measures
The researcher helped 65 parents to complete the questionnaires and the remaining 41 parents completed the questionnaires without the researcher's assistance.
For the FBQ, 83 (78.3%) caregivers had completed all items of the questionnaire. Four (23.5%) of the 17 items of the questionnaire had complete data from every participant. 41(2.3%) values were missing across all items of the questionnaire. Items with the most omitted responses related to primary caregiver relationship to child, partner education, partner previous and current occupational status and questions responded to with ‘other’ which required further elaboration.
For the PEDS, 82 (77.4%) parents had complete data; 24 (22.6%) had omitted a response to at least one item (Table 2). Data were complete for 8 (38.1%) of the 21 items. Twenty six values (1.2%) were missing across the 21 items of the scale across all participants. Responses to item 13 ‘Gets frustrated too easily’ were the most frequently omitted with six parents not completing this item. This was followed by item 11 ‘Seems hyperactive’ (missing n = 4). These items are both included in the Acting Out scale. Twelve parents did not respond to at least one of the items on this scale, eight did not for the Anxious/Withdrawn scale and five for the Fearful scale.
Note. Items 1–17 are included in scale calculations.
Incomplete data on PEDS individual items led to there being 20 participants with incomplete total scores for the Total score, nine for the Anxious/Withdrawn scale and Acting Out scales and three for the Fearful scale.
Data were more complete for participants who completed the PEDS without the help of the researcher than those who were helped by the researcher. Of the 41 parents who completed the PEDS without the help of the researcher, 31 (75.6%) had complete data; ten (24.4%) had missing data including eight with one omitted item and two with two omitted items. Of the 65 parents who were helped by the researcher, 51 (78.5%) had complete data; 14 (21.5%) had missing data on one item.
For the SDQ, 61 (57.5%) caregivers had complete data; 45 (42.5%) did not respond to at least one question (Table 3). Four items (16%) out of the 25 items were complete across all participants. There were 77 (2.9%) missing values in total across the 25 items of the questionnaire. Responses to item 25 ‘Sees tasks through to the end, good attention span’ were most frequently omitted with 29 parents not completing this item. This was followed by item 3 ‘Often complains of headaches, stomach-aches or sickness’ (missing n = 6), item 20 ‘Often volunteers to help others (parents, teachers, other children)’ (missing n = 5), item 6 ‘Rather solitary, tends to play alone’, item 12 ‘Often fights with other children or bullies them’ and item 13 ‘Often unhappy, down-hearted or tearful’ (missing n = 4 for all three items). Items from the Hyperactivity scale, which item 25 is included in, were the least complete; 33 parents did not respond to one or more of the five items that make up this scale. Items from the Emotional Symptoms scale had the second most incomplete data with 16 respondents omitting responses to one or more of the five items, followed by Peer Problems (n = 11), Conduct Problems (n = 10) and Prosocial Behavior (n = 7).
Note. The Prosocial scale is excluded from the total difficulties score.
Incomplete data on SDQ individual items caused one participant to have incomplete total scores for the Emotional Symptoms, Conduct Problems, Peer Problems and Total Difficulties scales to be missing and two participants to have a missing total score for the Hyperactivity scale. Although there was a sizeable amount of missing data on individual items of the SDQ, the algorithm used to calculate total scores allows responses on up to two out of the five items for each scale to be missing.
SDQ data were more complete for parents who were not helped by the researcher than those who were helped by the researcher. Of the 41 parents who completed the SDQ without the help of the researcher, 26 (63.4%) had complete data and 15 (36.6%) had incomplete data; 7 (46.7%) of these 15 participants omitted one item, 3 (20%) omitted two items, 4 (26.7%) omitted two items, 4 (26.7%) did not complete three items and 1 (6.7%) did not complete fifteen items. Of the 65 parents who were helped by the researcher to complete the SDQ, 35 (53.8%) had complete data and 30 (46.2%); 25 (83.3%) of the 30 omitted one item, 3 (10%) omitted two items and 2 (6.7%) did not complete three items. The majority (90%) of participants who omitted responding to item 25 (for which the most data was missing) were helped by the researcher to complete the questionnaire.
Levels of behavioural and emotional problems
Table 4 displays means, standard deviations and range of scores for each of the subscales and total scores on the PEDS and SDQ and the proportion of children with total and subscale scores at or above clinical cut-offs. Overall, children had high scores on total scale and subscale scores on the PEDS and SDQ and the proportion of children meeting clinical cut-offs on both measures was highest for emotional difficulties. Forty nine per cent of children met the cut-off on the PEDS Anxious/Withdrawn subscale, 62.3% on the Fearful subscale and the total for the instrument. Fewer children met the cut-off for behavioural problems as measured with the Acting Out subscale; however, the proportion still amounted to a third of children. Analysis of the frequencies of responses to each of the PEDS trauma-specific items (Table 5) revealed that 50% of children ‘often’ or ‘very often’ brought up the trauma they had experienced in conversation, 54.7% ‘often’ or ‘very often’ appeared fearful of things that were a reminder of the trauma and 42.4% were reported to ‘often’ or ‘very often’ create pictures, stories or games about the traumatic event. In contrast, 81.1% of children were rated as ‘almost never’ or ‘sometimes’ avoiding talking about the traumatic event even when asked. On the SDQ, parents reported that 45.3% of children met the clinical cut-off for emotional symptoms, 37.7% for conduct problems and 38.7% for total difficulties.
Note. 18 = Creates games, stories or pictures about [traumatic event (TE)]; 19 = brings up (TE) in conversation; 20 = avoids talking about (TE) even when asked; 21 = seems fearful of things that are reminders of (TE).
a data missing for one participant.
This study aimed to test the feasibility of recruitment and using two well-validated and brief questionnaires to collect mental health data about refugee children living in and near refugee camps on the Syrian–Turkish border. The results demonstrated that it is feasible to reach and recruit refugees caring for children with research participation information via NGO run primary schools in and near refugee camps using informed consent procedures while at the same time maintaining the safety of researchers and participants. Further, it proved feasible to achieve a good return rate for completed questionnaires for a reasonable sample size of caregivers. This may have been accounted for by the researcher assisting more than half of caregivers with completion of the questionnaires. Although under significant stress, parents appeared able to concentrate and complete measures, and the procedure did not cause apparent distress. The ethics of conducting research in contexts of conflict has been raised (Ford et al. Reference Ford, Mills, Zachariah and Upshur2009). Recently, El-Khani et al. (Reference El-Khani, Ulph, Redmond and Calam2013) discussed the ethics concerning asking recently displaced families about psychological need, particularly the risk of reactivating distressing accounts of trauma. The authors reported parents were able to talk about parenting experiences, were keen to take part in discussions, stressed the normality of talking about such experiences among themselves and valued the opportunity to share their experiences.
The strong and good working relationship between the research team and Generation Freedom played a key role in making this possible. The safety of the researcher and refugees was a significant concern at all times but was maintained due to expert support from Generation Freedom whose volunteers and employees have extensive experience of working in conflict as well as good local knowledge of the Syrian conflict. Their role was essential in enabling the researcher to cross the Syrian–Turkish border, providing the researcher with access to refugee camps and schools, logistical support to facilitate data collection and ensuring safety of the researcher and refugees at all times. This highlights that while it is possible to conduct research in the midst of a conflict, careful planning and support from expert agencies is vital.
The second aim of the study was to assess the extent to which complete data could be obtained. Firstly, and surprisingly, data for both scales was more complete for participants who were not helped by the researcher compared with those who were helped by the researcher. This was particularly the case for the last item of the SDQ. Participants informed the researcher that they did not wish to respond to particular items or that the items were not applicable to their child. Although the specific reasons for this require further exploration it seems caregivers did not find all items on the SDQ relevant to their situation. Second, while most caregivers completed all items of the PEDS, almost half the caregivers had incomplete data on the SDQ. This may reflect that the PEDS has better specificity for this population and in this context; unlike the SDQ it is designed specifically to measure trauma-related symptoms. It appeared that caregivers found items on the SDQ less applicable in their context. Reliability for the SDQ was poor in this context. This is consistent with studies that have used the SDQ as a secondary outcome measure in evaluations of psychological interventions to reduce post-traumatic stress in Palestinian children (Quota et al. Reference Quota, Palosaari, Diab and Punamaki2012; Barron et al. Reference Barron, Abdallah and Smith2013). However, the reliability analysis was based solely on Cronbach's alphas and for Total Difficulties and the Hyperactivity scales omitted item 25 due to the large amount of missing data for this item. While the internal consistency for the PEDS was good but not excellent, it offers promise in being a potentially reliable screening tool of child mental health in this context. These findings suggest further research is needed to rigorously test the reliability and validity of these scales in this context. Third, although for individual items there was a significant amount of data missing, the calculation of total scores was not affected for the SDQ as the algorithm for calculating total scores allows up to two items per scale to be missing.
While it was not possible to aim for a large or representative sample in this context, the third aim of the study was to gain insight into the levels of emotional and behavioural difficulties among refugee children living in or near refugee camps on the Turkish–Syrian border. Consistent with previous research on the impact of conflict in the Middle East on internally displaced and refugee children (Dimitry, Reference Dimitry2011), results indicated high levels of emotional and behavioural problems measured using both the PEDS and SDQ; elevated mean levels of difficulties for all subscales were found compared with existing normative data apart from SDQ Prosocial Behaviour. Nearly half of children were reported as having clinically significant levels of PEDS rated anxiety and being withdrawn and close to two thirds were fearful. These are higher levels than have been reported using the PEDS in samples of children exposed to natural disaster (Saylor et al. Reference Saylor, Swenson, Reynolds and Taylor1999). This highlights the need for evidence-based interventions providing psychological support to children and families that are tailored to the nature of the trauma that children have been exposed to. Examining the frequency of scores on trauma-specific PED items, parents reported that the majority of children did not avoid talking about traumatic events. This may have important therapeutic implications on helping children cope effectively with trauma.
A similar percentage of children meeting the cut-off for PEDS anxiety symptoms was found for emotional symptoms reported by parents using the SDQ. The percentage of children (38%) meeting the clinical cut-off for conduct problems rated using the SDQ was comparable with the rate (37%) found in Syrian children using the same measure in the IMC and UNICEF (2013) conducted mental health and psychosocial support (MHPSS) assessment in the Za'atari refugee camp in Jordan. In the same MHPSS assessment, a greater number of children (60% v. 45%) had clinically significant scores for SDQ rated emotional symptoms and fewer children (6% v. 19%) had prosocial behaviour scores that met the clinical cut-off. Mean levels of emotional symptoms and conduct problems were higher than average levels reported in other studies of refugee children exposed to conflict and resettled in high-income countries (Papageorgiou et al. Reference Papageorgiou, Frangou-Garunovic, Iordanidou, Yule, Smith and Vostanis2000; Fazel & Stein, Reference Fazel and Stein2003; Durà-Vilà et al. Reference Durà-Vilà, Klasen, Makatini, Rahimi and Hodes2012). One might expect to see increased severity of mental health outcomes in children still living in the challenging environment of a refugee camp within close proximity to the conflict compared with resettled children. Prosocial behaviour measured on the SDQ was lower in comparison with norms. Although children may experience a traumatic event, they are still likely to show or increase acts of prosocial behaviour (Thabet et al. Reference Thabet, Karim and Vostanis2006; Betancourt & Khan, Reference Betancourt and Khan2008; Masten, Reference Masten2011). Camps are also believed to become more community-oriented when families live in such close quarters (Dimitry, Reference Dimitry2011).
Some methodological considerations must be considered in interpreting the findings. First, opportunistic rather than random sampling was utilized. This is common in refugee camps where the possibility of using random sampling is limited for practical reasons. The extent to which families who participated are representative of the population from where they were recruited is not known. It is possible that the results may underestimate mental health problems in children residing in or near to camps on the Syrian-Turkish border given that those who have better mental health may be more likely to participate in research than those who are less well.
Second, this was a single informant design. Although parents are typically viewed as a prime source of data about their children's emotional and behavioural status, school-age children and their parents may not agree about posttraumatic event exposure symptoms (Belter et al. Reference Belter, Dunn and Jeney1991). Further, parents whose own distress is high and who may be experiencing mental health difficulties tend to rate their children's behaviours higher (Saylor et al. Reference Saylor, Belter, Stokes, Wolchik and Sandler1997). However, some evidence suggests that parents tend to underestimate their children's responses to traumatic events (Almqvist & Broberg, Reference Almqvist and Broberg1997; Awadh et al. Reference Awadh, Vance, El-Beblawi and Pumariega1998) or on some measures, in general, parent ratings appear to be less reliable than children's self-reports of their own behaviour (Barron et al. Reference Barron, Abdallah and Smith2013). Obtaining additional ratings from teachers or other key people in the child's environment may give a more reliable picture. This raises another feasibility question in this context, whether it is possible to obtain proxy reported data on child behaviour and is a question that we could address in our future studies given that we have demonstrated an effective recruitment route via primary schools for refugee children where access to teachers is possible.
Third, evaluation of the reliability of both questionnaires was limited due to the researcher helping many, but not all participants and participants completing questionnaires in different locations (e.g. schools, their tents). However, this study stresses the complexity of being able to reliably collect data in this challenging and chaotic environment. Reliance on Cronbach's alphas only as the sole index of reliability of the scales is not ideal.
Fourth, there are currently no normative data available for the PEDS or SDQ in Syrian or Middle Eastern populations. Therefore cut-offs were based on those used in the USA for the PEDS and on British norms for the SDQ. More research is required to establish cut-offs specific to the culture and context within which refugee children in the Syrian conflict live.
This study demonstrated an effective method of accessing and recruiting caregivers into a study from primary schools in and near refugee camps on the Syrian–Turkish border and collecting child mental health data in this context, with support provided by an NGO with expert local knowledge and experience of working in the Syrian conflict and while maintaining researcher and caregiver safety. Findings indicated that the PEDS appeared a more reliable tool than the SDQ for assessing refugee children's mental health difficulties. The high levels of emotional and behavioural problems stresses the immediate need for providing evidence-based psychological support to refugee children and families affected by the Syrian conflict. Further feasibility studies are required to identify effective approaches to improve data completeness in the challenging conditions and rigorously test the reliability and also the validity of the measures used.
We would like to thank Generation Freedom and Watan for their support as it would not have been possible to conduct the study without it and the families who took part.
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflicts of Interest