Over the past decade, developed Western countries have applied increasingly stringent measures to discourage those seeking asylum from entering their borders (United Nations High Commissioner for Refugees, 2000). Australian policy has shifted in two fundamental ways. In 1992, mandatory detention for the duration of the asylum determination procedure was introduced for persons arriving by boat or without valid entry visas. In 1999, time limits on residency were instituted by establishing a policy of temporary protection. Holders of temporary protection visas have restricted access to healthcare, education and work, no opportunity for overseas travel and are ineligible for migration to family reunions. Although these policies have led to substantial controversy, countries of the European Union are considering following Australia's lead by adopting similar approaches (Reference Milne and TravisMilne & Travis, 2003). The present study applied multilevel modelling (Reference GoldsteinGoldstein, 1995) to test the hypothesis that past detention and ongoing temporary protection visa status each contributes to ongoing post-traumatic stress disorder (PTSD), depression and associated disability in refugees.
The study group comprised Sabaean-Mandaeans (Mandaeans), a small pre-Christian sect of approximately 100 000 people originating mainly from Iran and Iraq (Reference BuckleyBuckley, 2002). The Mandaeans have suffered long-term discrimination in their home countries, with abuses and persecution escalating prior to and during the Iraq war in 2003 (Reference KendalKendal, 2004). Mandaean refugees in Sydney fall into two administrative categories based on their visa status (permanent protection visa (PPV) or temporary protection visa (TPV)). Those arriving before 1999 were granted the former; those arriving after 1999 were issued the latter. Only those shown to be genuine refugees according to the 1951 United Nations Convention are eligible for either visa category. Mandaeans who arrived in Australia without entry documents were initially held in detention centres, with the remainder - those arriving with holiday or student visas - being allowed to live in the community while their refugee applications were processed.
No census or immigration figures were available for the Mandaean minority in Sydney. Community leaders estimated that there were fewer than 400 Arabic-speaking Mandaean adults living in the city. Sampling of very small and dispersed populations such as the Mandaeans creates formidable methodological challenges (Reference Spring, Westermeyer and HalconSpring et al, 2003) with random or small-area based probability sampling frames not being feasible. It was important, nevertheless, that the sampling frame adopted had an equal probability of capturing those with permanent and those with temporary protection visas.
Community leaders provided lists of Arabic-speaking Mandaeans and 76 families (172 adults) living in Sydney were identified. To widen the sample beyond those in contact with community groups, snowball or linkage sampling (Reference Patrick, Pruchno and RosePatrick et al, 1998) was used, in which primary respondents were asked to provide the names and contact details of other Mandaean families. This procedure identified an additional 38 families (96 adults). The final sample comprised 241 Mandaeans (90% individual response rate for those contacted) living in 104 house-holds (91% household response rate). Based on estimates by community leaders, the sample represented 60% of Arabic-speaking Mandaean adults living in Sydney.
The study applied two psychometrically tested symptom measures: the Harvard Trauma Questionnaire (HTQ; Reference Mollica, Caspi-Yavin and BolliniMollica et al, 1992), which assesses exposure to refugee-related trauma and associated PTSD symptoms, and the Hopkins Symptom Checklist - 25 (HSCL; Reference Mollica, Wyshak and de MarneffeMollica et al, 1987), an instrument that records symptoms of depression and anxiety. The measures yield both continuous scores and diagnostic categories (PTSD and major depressive episode, respectively) based on established DSM-IV-derived algorithms (Reference Mollica, Sarajlic and ChernoffMollica et al, 2001). The Medical Outcomes Study - Short Form (SF-12; Reference Gandek, Ware and AaronsonGandek et al, 1998) is a widely used international instrument that provides a measure of physical (Physical Component Score; PCS) and mental (Mental Component Score; MCS) health status and disability. We applied the MCS, which is usually scored according to four disability levels: none (50 or above); mild (40-49); moderate (30-39); and severe (below 29) (Reference Sanderson and AndrewsSanderson & Andrews, 2003).
Three additional measures were developed by the research team: (a) the Post-migration Living Difficulties (PMLD) Checklist, which identifies ongoing stresses that discriminate between refugee and asylum-seeker populations (Reference Silove, Steel and McGorrySilove et al, 1998); (b) the Detention Experiences Checklist, which details 64 common adverse experiences specific to the detention environment and which has been piloted among detained families (Reference Steel, Momartin and BatemanSteel et al, 2004); and (c) the Detention Symptom Checklist, a modification of the HTQ (Reference Mollica, Caspi-Yavin and BolliniMollica et al, 1992), which relates symptoms specifically to the detention experience (Reference Steel, Momartin and BatemanSteel et al, 2004).
Translation of measures into Arabic was undertaken using established translation and masked back-translation procedures (Reference BontempoBontempo, 1993). The original questionnaires were translated by an Arabic-speaking mental health professional and back-translated by a certified Arabic-speaking healthcare interpreter. Two Arabic-speaking mental health professionals reconciled some minor discrepancies identified by the process.
Approval for the study was obtained from the South West Sydney Area Health Service Human Ethics Committee. A research assistant, a nurse practitioner of Mandaean background, made contact with families and individual participants. After consent was obtained, the research assistant visited respondents in their homes and administered the measures.
Preliminary univariate analyses using the Statistical Package for the Social Sciences (SPSS) version 11.5 for Windows were applied to assess for differences across comparison groups in levels of trauma exposure, post-migration stresses, detention experiences and psychiatric morbidity using χ2 analysis for categorical variables and analysis of variance (ANOVA) for numeric variables. Since key predictor variables varied in their frequency across comparison groups and because participants were clustered in family groupings (Reference GoldsteinGoldstein, 1995), we next applied multilevel modelling using MLwiN 1.1 (Reference Rasbash, Browne and GoldsteinRasbash et al, 2001). The purpose of this was to establish whether detention and temporary protection status exerted independent effects on psychiatric symptoms when accounting for family clustering and other predictors and covariates identified by the univariate analysis.
Table 1 presents the major characteristics of the sample and shows that temporary (58%) and permanent protection visas (42%) were each well represented and that more than half of the sample (62%) had been detained. Those with permanent visas had lived in Australia for longer.
|Male gender, n (%)||132(54)|
|Age, years: mean (s.d.)||38(14.2)|
|Country of origin, n (%)|
|Marital status, n (%)|
|Detained on arrival, n (%)||150(62)|
|Families, n (individuals) 1|
|with one member||31(31)|
|with two members||40(80)|
|with three members||15(45)|
|with four members||10(40)|
|with five members||3(15)|
|with six members||5(30)|
|Residential status, n (%)|
|Time in Australia, years: mean (s.d.)|
|Permanent residents||4.7(3.3) *|
* P <0.001.
1. Total number of families=104.
Table 2 shows the extent and nature of pre-migration trauma experiences as measured by the HTQ in holders of temporary and permanent visas. Holders of the former had experienced more traumas than holders of the latter, including death of family and friends, being close to death (i.e. almost dying) and lacking the basic necessities of life. This difference may reflect the increasing persecution of Mandaeans in Iraq leading up to and following the 2003 war.
|Individuals exposed to event, n (%) 1|
|Trauma event||TPV holders||PPV holders|
|Unnatural death of family or friend||110 (79)||63 (62)**|
|Being close to death||106 (76)||29 (29)***|
|Murder of family or friend||105 (75)||62 (61)*|
|Murder of stranger or strangers||68 (49)||32 (32)**|
|Lack of food or water||65 (46)||23 (23)***|
|Ill health without access to medical care||53 (38)||16 (16)***|
|Imprisonment||52 (37)||15 (15)***|
|Forced separation from family members||36 (26)||11 (11)**|
|Lack of shelter||26 (19)||11 (11)|
|Torture||25 (18)||12 (12)|
|Combat situation||21 (15)||8 (8)|
|Serious injury||20 (14)||9 (9)*|
|Forced isolation from others||20 (14)||6 (6)|
|Brainwashing||18 (13)||6 (6)|
|Lost or kidnapped||16 (11)||6 (6)|
TPV, temporary protection visa; PPV, permanent protection visa.
* P<0.05; ** P<0.01; *** P<0.001.
1. Mean number (s.d.) of trauma events for TPV holders=5.3 (2.8) v. 3.1 (3.1) for PPV holders, P<0.001 (range=0-16).
Post-migration living difficulties
Table 3 shows the causes of serious stress in the temporary and permanent visa holders. Holders of temporary visas reported greater stress for the majority of items. Consistent with their status, they continued to live in fear of repatriation, were unable to resolve family separations and struggled more with issues of day-to-day living.
|Living difficulty||Individuals exposed to difficulty, n (%) 1|
|TPV holders||PPV holders|
|Unable to return home in an emergency||134 (96)||8 (8)***|
|Fearful of repatriation||133 (96)||2 (2)***|
|Concern for family in country of origin||121 (87)||62 (61)***|
|Unemployment||85 (61)||22 (22)***|
|Insufficient money to buy food, necessary clothes or pay rent||77 (55)||22 (22)***|
|Loneliness and boredom||76 (55)||14 (14)***|
|Communication difficulties||74 (53)||27 (26)***|
|Separation from family||66 (47)||12 (12)***|
|Long-term health problems||66 (47)||6 (6)***|
|Social isolation||65 (47)||11 (11)***|
|Poor access to treatment for health problems||63 (45)||7 (7)***|
|Poor access to emergency medical care||58 (42)||8 (8)***|
|Poor access to long-term healthcare||65 (47)||7 (7)***|
|Poor access to dental care||47 (34)||5 (5)***|
|Poor access to counselling||41 (29)||5 (5)***|
|Bad work conditions||18 (13)||6 (6)|
|Discrimination||17 (12)||3 (3)*|
|Difficulties obtaining help from charities||15 (11)||1 (1)**|
|Conflict with other ethnic groups in Australia||8 (6)||2 (2)|
|Difficulties obtaining government help with welfare||8 (6)||2 (2)|
|Conflict with immigration officials||4 (3)||0 (0)|
|Interviews with immigration officials||3 (2)||1 (1)|
TPV, temporary protection visa; PPV, permanent protection visa.
* P<0.05; ** P<0.01; *** P<0.001.
1. Mean number (s.d.) of living difficulties for TPV holders=9.2 (3.8) v. 2.3 (2.4) for PPV holders, P<0.001 (range=0-23).
Immigration detention experiences
The majority of temporary visa holders (124 out of 139, 90%) and 30% of permanent-visa holders (30 out of 102) had been held in immigration detention centres on arrival in Australia. Table 4 includes all those held in detention and details the 20 highest-ranking adverse experiences that caused serious or very serious stress while confined. The results are stratified in terms of short detention (0-5 months) and long detention (≥6 months) based on the median time spent in detention (6 months). Although groups with short- and long-term detention both reported substantial stress, the latter group scored higher on almost all items (total score=21.3, s.d.=10.4 v. 9.8, s.d.=10.1).
|Factor||Individuals reporting experience, n (%) 1|
|0- to 5-month detainees (n=57)||≥6-month detainees (n=93)|
|Fears of being sent home||33 (59)||88 (96)***|
|Separation from family||32 (56)||52 (57)|
|Worries about family back home||33 (59)||80 (87)***|
|Delays in processing refugee application||18 (32)||77 (85)***|
|Not being informed about progress of refugee application||11 (19)||75 (81)***|
|Boredom||20 (35)||76 (83)***|
|Isolation||13 (23)||73 (79)***|
|Overcrowding in rooms||9 (16)||44 (48)***|
|Woken during the night/head counts or other disturbances||9 (16)||43 (46)***|
|Poor-quality food||15 (26)||41 (45)*|
|Racist comments by other detainees||28 (49)||76 (82)***|
|Being sworn at by other detainees||26 (46)||75 (81)***|
|Being intentionally humiliated by other detainees||25 (44)||72 (77)***|
|Seeing people making suicide attempts||17 (30)||63 (68)***|
|Seeing people engage in self-harm||12 (21)||60 (65)***|
|Witnessing physical assault||17 (30)||57 (61)***|
|Poor access to dental care||16 (29)||53 (57)**|
|Not getting treatment for health problems||12 (21)||44 (48)**|
|Poor access to emergency medical care||14 (25)||44 (48)**|
|Poor access to long-term medical care||10 (18)||42 (45)**|
* P <0.05; ** P<0.01; *** P<0.001.
1. Mean number of problems (s.d.) for those detained for 0-5 months=9.8 (10.1) v. 21.3 (10.4) for those detained ≥6 months, P <0.0001 (range=0-64).
Immigration detention symptoms
Table 5 shows the traumatic stress symptoms associated with negative detention experiences. Even though the mean time since release from detention was nearly 3 years (35.5 months), the group detained for ≥6 months reported more severe distress for all nine symptoms.
|Traumatic stress symptom||Individuals reporting experience, n (%) 1|
|0-to 5-month detainees (n=57)||≥ 6-month detainees (n=93)|
|Feeling extremely sad and hopeless when thinking about detention||15 (26)||69 (74)***|
|Sudden and upsetting memories of time in detention||14 (25)||68 (73)***|
|Images of the threatening or humiliating events in detention||10 (18)||58 (62)***|
|Avoiding talking about detention because causes distress||10 (18)||54 (58)***|
|Sudden attacks of anger over small things since being in detention||8 (14)||53 (57)***|
|Becoming nervy, sweaty, shaky and/or having rapid heart beat when thinking about detention||10 (18)||52 (56)***|
|Nightmares about things that happened in detention||6 (11)||38 (41)***|
|Avoiding interaction with other people since being in detention||3 (5)||26 (28)**|
|Feeling numb since being in detention||4 (7)||24 (26)**|
** P <0.01; *** P <0.001.
1. Mean number of symptoms causing distress (s.d.) for those detained for 0-5 months=1.4 (2.3) v. 4.8 (3.0) for those detained ≥6 months, P <0.0001 (range=0-9).
Psychiatric status and disability
Figures 1 and 2 display the univariate results for depression, PTSD and mental health-related disability according to residency status and length of detention respectively. The HTQ and HSCL were analysed according to the standard algorithms to yield diagnoses of PTSD and major depressive episode respectively. The standard cut-off on the MCS was applied for moderate-to-severe mental health-related (MCS) disability. Holders of temporary protection visas had higher rates of depression, PTSD and disability than those with permanent visas. Those who had experienced long-term detention also continued to experience greater rates of depression, PTSD and MCS disability.
Calculation of intraclass correlation coefficients indicated substantial family clustering for PTSD (0.42), depression (0.40) and MCS disability (0.27). To ascertain the independent effects of detention and temporary protection status, we applied multilevel modelling, an analysis which included the impact of family clustering and other predictor variables identified in the univariate analyses.
Prediction of psychiatric status and disability
Three multilevel models were calculated using continuous scores for PTSD, depression and MCS disability respectively. Level one predictor variables were age, gender, number of traumas prior to arrival, months in detention (0 for those not detained), residency status (temporary or permanent visa) and months living in the community (i.e. post-detention or post-immigration length of stay in Australia). The level two variable was family clustering.
Variables excluded on the basis of preliminary analyses were number of detention experiences (Table 4) and detention-specific symptoms (Table 5), since these correlated highly (r=0.72 and r=0.70 respectively) with the duration of detention. Similarly, indices of post-migration living difficulties were strongly related to temporary protection status and hence made no additional contribution as predictors.
Table 6 displays the key characteristics of the multilevel regression analysis, with each model emerging as statistically significant. The R 2 estimates indicate that the overall models account for a substantial proportion of the variance for each dependent variable, although more so for PTSD and depression than for MCS disability. Intraclass correlation coefficients show the level of family clustering for each dependent variable, with the family variance partition coefficient indicating that nearly 30% of the symptom measures and 20% of the disability measures were accounted for by family clustering. Calculation of standardised regression weights (β) allowed for direct comparison of the magnitude of the effect of each predictor variable. After including the effects of age, gender, previous trauma exposure, family composition and length of residency, time in detention and temporary protection status made an equal and substantial contribution to psychiatric morbidity and disability.
|PTSD symptoms||Depressive symptoms||MCS disability score|
|Intraclass correlation for families||0.42||0.40||0.27|
|Family variance partition component, % 2||30.1||29.0||19.5|
|R 2 estimate||0.44||0.45||0.29|
|Model significance||χ2=131.4, d.f.=6, P <0.001||χ2=135.4, d.f.=5, P <0.001||χ2=96.48, d.f.=5, P <0.001|
|Level one predictor variables, B (β) 3|
|Gender||0.34 (0.23)||0.43 (0.27)||2.93 (0.14)|
|Age||0.014 (0.26)||0.013 (0.24)||0.16 (0.21)|
|Previous trauma||0.041 (0.18)||0.036 (0.14)||0.77 (0.23)|
|Months in detention||0.023 (0.14)||0.032 (0.18)||0.36 (0.16)|
|Residency status||0.33 (0.22)||0.50 (0.31)||3.84 (0.18)|
|Length of residency in Australia||−0.003 (−0.15)||NS||NS|
NS, not significant, B, regression weight; β, standardised regression weight; SF-12, Medical Outcomes Study - Short Form.
1. All reported parameters are significant at 0.05 or greater unless marked NS.
2. Percentage of explained variance due to family.
3. Standardised regression weight used to determine the relative effect of each parameter.
A subsidiary analysis showed that those who had no family in Australia (a proxy measure for family isolation) were at greater risk for depression and PTSD than those living in families of three or more persons. Families consisting of two available members fell into an intermediate risk category (F 2,238=3.82, P<0.05 for depression and F 2.237=4.09, P<0.05 for PTSD).
Developed Western countries have implemented increasingly stringent policies to deter those seeking asylum (Reference Silove, Steel and WattersSilove et al, 2000). Australia has adopted particularly restrictive policies, including the detention of people seeking asylum but arriving without entry visas and, more recently, limiting those found to be refugees to temporary protection visas only. European countries and North America are considering the adoption of similar policies (Reference CrispCrisp, 2003). It is timely, therefore, to consider the impact of such policies on the mental health of refugees.
The present study suggests that both prolonged detention and temporary protection contribute substantially to the risk of ongoing depression, PTSD and mental health-related disability in refugees. The independent influence of these two risk factors remained robust after controlling for other variables previously identified as risk factors (Reference De Jong, Komproe and Ommerende Jong et al, 2001), including female gender, greater age, extent of past traumas, length of residency and family separation.
Limitations of the study design
Before interpreting the results in greater detail, limitations of the study design require consideration. A population-wide probabilistic sampling method was not feasible because of the minority status and dispersal of the target group. It is possible that community members with the most severe mental disturbances were more likely to respond, although previous research suggests that greater community contact is a protective factor against emotional distress (Reference Steel, Silove and BirdSteel et al, 1999). It is also noteworthy that the rates of PTSD and depression fall within the limits of prevalence data in other groups of asylum seekers and refugees studied in Australia and other resettlement countries (Reference Hobbs, Moor and WansbroughHobbs et al, 2002; Reference Silove, Steel and McGorrySilove et al, 2002). Furthermore, Goodman (Reference Goodman1961) argued that snowball sampling, although not ideal for establishing absolute prevalence rates, is appropriate for comparing subgroups within samples, the key focus of the present study.
Transcultural measurement issues must be considered as a potential source of error. We applied a standard translation and masked back-translation method, reconciling minor semantic and linguistic differences with the assistance of two Arabic-speaking mental health professionals. The relationships yielded between key variables provided indirect support for the validity of the measures. For example, we replicated the well-established dose-response relationship between trauma exposure and mental disorder, a finding that is robust across the majority of published studies (e.g. Reference Hauff and VaglumHauff & Vaglum, 1994; Reference Mollica, McInnes and PooleMollica et al, 1998).
Recall bias is always a potential confounding variable, particularly when reporting past traumatic events. An important recent study (Reference Herlihy, Scragg and TurnerHerlihy et al, 2002) has shown that refugees remain consistent in reporting major traumatic events such as those we recorded, with more variability occurring in recall of minor historical details. The higher levels of trauma reported by holders of temporary protection visas who have arrived more recently in Australia was consistent with a history of escalating violence and persecution directed at the Mandaean group in Iraq in the lead-up to the 2003 war.
At the same time it may be argued that holders of temporary visas are prone to exaggerating their plight, a potentially self-serving bias motivated by the hope of advancing future claims for extension of their visas. Although this cannot be ruled out entirely, the anonymous nature of the study meant that there was no direct personal gain resulting from participation. Holders of temporary visas were not currently applying for extensions visa, so that further documentation at the time of the study was not directly useful. Furthermore, the criteria used by immigration officials in judging the need for further protection focus principally on risk of future persecution, not on past history or ongoing living difficulties, the subject of our study.
Although the shared cultural, historical and political experiences of the permanent and temporary visa holders strengthened the comparisons, the focus on a single refugee population potentially limits the generalisability of the findings, making it important for future studies to replicate our results in other cultural groups.
Longer-term mental health impact of detention
Our study suggests that prolonged detention exerts a long-term impact on the psychological well-being of refugees. Refugees recording adverse conditions in detention centres also reported persistent sadness, hopelessness, intrusive memories, attacks of anger and physiological reactivity, which were related to the length of detention. Previous studies examining the effects of detention concur with our findings (Reference Steel and SiloveSteel & Silove, 2001; Reference Sultan and O'SullivanSultan & O'Sullivan, 2001; Reference Keller, Rosenfeld and Trinh-ShevrinKeller et al, 2003), although our study is the first to show that such mental health effects persist for a prolonged period after detention. These effects were independent of other established predictors of psychiatric morbidity in refugees, such as past exposure to trauma and recency of arrival (Reference Steel, Silove and PhanSteel et al, 2002). The present findings provide systematic support for the observations of successive commissions of inquiry undertaken in Australia (Human Rights and Equal Opportunity Commission, 1998; Office of the High Commissioner for Human Rights, 2002) that have raised repeated concerns about the mental health effects of prolonged detention.
The present study is the first to investigate the specific effects of temporary protection on the mental health of refugees. In the past refugee settlement countries such as Australia have offered permanent residency to refugees according to the 1951 United Nations Convention and Australia continues to administer a separate programme of permanent resettlement for a quota of refugees screened in other countries. Permanent protection means that previously traumatised refugees are given certainty about their futures, allowing them to plan their lives with a substantial level of security. A recent epidemiological study undertaken among Vietnamese refugees in Australia (Reference Steel, Silove and PhanSteel et al, 2002) has provided evidence that permanent residency is associated with improvement in the mental status of previously traumatised individuals.
In contrast, the present study adds to evidence (Reference Silove, Steel and WattersSilove et al, 2000) that insecure residency and associated fears of repatriation contribute to the persistence of psychiatric symptoms and associated disabilities in refugees. Temporary protection status was strongly associated with daily stresses related to financial and work difficulties, and problems in accessing healthcare, language classes and other educational opportunities. Countries considering the adoption of temporary protection regimes therefore need to consider how such provisions may undermine the sense of security that seems to be essential for refugees to recover from trauma-related psychiatric symptoms.
The multilevel modelling analysis indicated that there was a concentration of mental distress within family groups. Risk of mental illness was lower in larger family units, whereas those refugees who were isolated from other family members were more likely to experience severe psychiatric symptoms. Temporary protection status in Australia specifically denies refugees the right to family reunion and prevents holders from re-entering Australia if they travel overseas, making direct contact with families in other countries effectively impossible. The common consequence is prolonged separations that compound the disruptive effects on families of past persecution (Reference Silove, Steel and WattersSilove et al, 2000). Our findings therefore highlight the need to consider carefully the impact of refugee policy changes on family unity and the potential risk that enforced family separations may result in prolonged mental disorder in isolated refugees.
Clinical Implications and Limitations
▪ To prevent further psychological harm to previously traumatised refugees, it is necessary to minimise detention and ensure that conditions in detention are humane.
▪ Certainty of residency to persons recognised as refugees seems to be essential for recovery from trauma-related psychiatric symptoms.
▪ Families and social groups that are not kept together or reunited may be at greater risk of prolonged mental disorder.
▪ The study used non-random sampling.
▪ The study was limited to one refugee group, the Mandaeans.
▪ There was potential for exaggeration bias in reporting.
This study was supported by an Australian National Health and Medical Research Council programme grant (3000403).