Evaluation of the efficacy of a South African psychosocial framework for the rehabilitation of torture survivors

Page created by Christopher Castro
 
CONTINUE READING
34

                                                                                                                          SCIENTIFIC ARTICLE

                                                    Evaluation of the efficacy of a South
                                                    African psychosocial framework for
                                                    the rehabilitation of torture
                                                    survivors
                                                    Dominique Dix-Peek*, Merle Werbeloff**

                                                                                                           Abstract
                                                         Key issues box:                                   Introduction: To address the consequences
                                                         • When treating torture survivors                 of past torture experiences as well as current
                                                            in developing countries, it is                 traumas and daily stressors, the Centre for
                                                            essential to consider the inherent             the Study of Violence and Reconciliation
                                                            stressors and daily traumas in these           (CSVR) developed a contextually
                                                            environments.                                  appropriate psychosocial framework for
                                                         • To assist with the psychosocial                 the rehabilitation of individuals who have
                                                            impacts due to torture, an evidence-           been affected by torture. Method: To test
                                                            based, contextually appropriate                the efficacy of this framework, a quasi-
                                                            psychosocial rehabilitation                    experimental study was conducted with
                                                            framework has been developed by                torture survivor clients of the CSVR who
                                                            the Centre for the Study of Violence           met the 1985 United Nations Convention
                                                            and Reconciliation.                            Against Torture (UNCAT) definition. A
                                                         • Following a 3-month intervention                comparison group of clients (n=38) was
                                                            based on this psychosocial                     initially included on a waiting list and
                                                            framework, torture survivors’                  thereafter received treatment, whilst the
                                                            anxiety and functioning improve,               treatment group of clients (n=44) entered
                                                            despite harsh contextual realities.            straight into treatment. Results: Baseline
                                                         • However, ethical complexities                   t-test comparisons conducted on 13
                                                            and high levels of participant                 outcome indicators revealed significantly
                                                            attrition during longer therapeutic            better initial psychological health and
                                                            intervention present challenges                functioning of clients in the treatment group
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                            to experimental endeavours                     than those in the comparison group, with
                                                            designed to test the efficacy of the           moderately large differences on PTSD,
                                                            rehabilitation framework.                      trauma and anxiety, and strong difference in
                                                                                                           depression scores. Three-month follow-up
                                                                                                           comparisons using the conservative
                                                                                                           Wilcoxon test revealed significantly greater
                                                    *)  Centre for the Study of Violence and Reconcilia-   improvement on the functioning and anxiety
                                                        tion (CSVR), South Africa                          indicators of the treatment group relative
                                                    **) Wits School of Governance, University of the
                                                                                                           to the waiting-list comparison group (odds
                                                        Witwatersrand, South Africa
                                                    Correspondence to: ddixpeek@csvr.org.za                ratios = 2.49 and 2.61 respectively). After a
35

SCIENTIFIC ARTICLE

further three months, when treatment was          presents this framework and describes a
based on the CSVR framework for both              quantitative study designed to examine
groups, fewer than half the respondents           the effectiveness of its use for torture
remained in the study (n=20 in the                rehabilitation.1
treatment group; n=16 in the comparison
group), and the Wilcoxon repeated measures        Torture and its consequences
test results on changes since baseline were       Torture is a gross human rights violation
counter-intuitive: for these remaining clients,   that distresses the individual, family,
there were now more significant outcome           community and society at large. Despite
improvements for the comparison group             a number of international conventions
than for the treatment group. However,            prohibiting torture and cruel, inhuman and
the relative odds ratios for the groups           degrading treatment (CIDT), torture is
were not significant for these indicators.        still practised in more than 140 countries
Furthermore, the clients who dropped out          internationally, and still widely practised in
from the treatment group had shown overall        Africa where it has been criminalised in only
improvement in their psychological health         10 countries (Amnesty International, 2014).
and functioning in the initial three months       Sub-Saharan Africa hosts the largest number
of the study, whereas those who dropped           of refugees internationally, with South
out from the comparison group had shown           Africa receiving most of the asylum seekers
improvements on fewer indicators. Thus,           in Africa (McColl et al., 2010; UNHCR,
the research findings on the efficacy of the      2015). Many of these people are likely to
framework are inconclusive. Discussion:           have been tortured in their country of origin
We suggest that this inconclusiveness can         or in transit (McColl et al., 2010). Within
be explained by the severe challenges             South Africa, current victims of torture
and ethical complexities of psychosocial          include youth in conflict with the law, non-
research on vulnerable groups. The study          nationals, and people who are in the wrong
highlights the serious problem of attrition       place at the wrong time (Bantjes, Langa, &
of participants in the treatment programme        Jensen, 2012; Langa, 2013).
which affected the overall study, and which            The consequences of torture may be
may explain findings that at first appear         categorised broadly as physical, psychological
counter-intuitive.                                and social. The physical consequences
                                                  may be complicated by the psychological
Introduction                                      consequences (Quiroga & Jaranson,
Continuing traumas, daily stressors and           2005). Similarly, torture-related mental
                                                                                                           T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

trauma reactions from past traumas are            health problems can cause physical and
characteristic of torture survivors in            social problems which impact on both the
South Africa. Consequently, in 2013,              personal and social functioning of survivors
the Centre for the Study of Violence
and Reconciliation (CSVR) based in
Johannesburg, designed a contextually
appropriate, evidence-based psychosocial          1
                                                      While the intervention was named a “model”,
framework for guiding the long-term                   due to the multi-modal nature of the document,
                                                      it was felt to be more representative to call it a
rehabilitation of individual survivors of
                                                      multi-modal framework for the rehabilitation of
torture (Bandeira et al., 2013). This paper           torture survivors.
36

                                                                                                                  SCIENTIFIC ARTICLE

                                                    (Baird, Williams, Hearn, & Amris, 2016).           Although the South African Constitution
                                                    The physical sequelae of torture include       (Republic of South Africa, 1996) and the
                                                    pain - often chronic, disability and medical   Refugee Act (Republic of South Africa,
                                                    conditions (Harlacher, Nordin, & Polatin,      1998) provide basic civil and political
                                                    2016; Jørgensen, Auning-Hansen, & Elklit,      rights such as free basic healthcare,
                                                    2017; Patel, Kellezi, & Williams, 2014).       access to education and employment to
                                                        The psychological consequences             refugees and asylum seekers regardless
                                                    of torture are often viewed in terms of        of nationality and legal status, in reality
                                                    posttraumatic stress disorder (PTSD)           these constitutional rights may be denied
                                                    and major depressive disorder (MDD)            (Bandeira, 2013; Higson-Smith, 2013;
                                                    (Harlacher et al., 2016; Higson-Smith,         Langa, 2013; Patel et al., 2014; Quiroga
                                                    2013). Lesser acknowledged psychological       & Jaranson, 2005). Refugees and asylum
                                                    consequences include cognitive impairment,     seekers experience high levels of crime
                                                    decreased functioning, sleep disturbances,     and violence, xenophobia and exploitation
                                                    memory problems, attentional deficits and      (Bandeira, Higson-Smith, Bantjes, & Polatin,
                                                    anger. Co-morbid psychiatric conditions        2010; Higson-Smith, 2013; Langa, 2013;
                                                    include depression, suicide, psychosis and     Mohamed, Dix-Peek, & Kater, 2016), and
                                                    substance abuse (Bandeira, 2013; Quiroga,      the South African asylum-seeking process
                                                    2017; Quiroga & Jaranson, 2005).               has been associated with inconvenience,
                                                        The social consequences of torture         cost and distress (Higson-Smith & Bro,
                                                    impact on the social wellbeing of the          2010; Langa, 2013). Continuing traumas
                                                    survivor. These effects include the loss of    threaten the survival of many refugees and
                                                    “employment, status, family and identity”      asylum seekers through ongoing threats
                                                    (Higson-Smith, 2013, p.165), separation        from the police, government officials and
                                                    from loved ones (Higson-Smith, 2013;           community members, and through domestic
                                                    Quiroga & Jaranson, 2005), the loss of         violence, sexual violence and xenophobia
                                                    functioning and safety, and loss of cultural   (Higson-Smith, 2013; Mohamed et al.,
                                                    and community connections (Higson-             2016). Such traumas “may influence the
                                                    Smith, 2013).                                  way that survivors respond or adapt to
                                                                                                   their precarious circumstances, but it is the
                                                    Living in contexts of continuing threat        circumstances themselves that produce and
                                                    and daily stress                               maintain the client’s psychological state”
                                                    Vulnerability to mental health challenges      (Higson-Smith, 2013, p. 166). Fear, anger
                                                    has been linked to pre-migratory exposure      and distress, emotional collapse, helplessness
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    to war trauma and torture, the migration       and hopelessness (Bandeira, 2013; Kaminer
                                                    process itself, and the post-migratory         & Eagle, 2010) are associated with daily
                                                    stressors of host countries which include      stressors of documentation problems,
                                                    poverty and inadequate housing, and            concerns over the health and schooling of
                                                    difficulties with asylum procedures            family members, accommodation problems,
                                                    (Bogic, Njoku, & Priebe, 2015; Buhmann,        unemployment, poverty, loss of social and
                                                    Mortensen, Nordentoft, Ryberg, &               material support, ostracism and lack of
                                                    Ekstrøm, 2015; Jaranson & Quiroga,             security (Bandeira et al., 2010; Higson-
                                                    2011; Jørgensen et al., 2017; Stammel et       Smith, Mulder, & Masitha, 2007; Miller &
                                                    al., 2017).                                    Rasmussen, 2010).
37

SCIENTIFIC ARTICLE

Research on treatment approaches for              and educational groups (Phaneth, Panha,
torture survivors                                 Sopheap, Harlacher, & Polatin, 2014).
There are few scientific research studies         Furthermore, the literature reviewed indicates
on the treatment of torture survivors as          a strong prevalence of European and North
clinicians are inclined to prioritise direct      American studies on the rehabilitation of
services to clients over research processes.      torture survivors, with few outcome studies
Ethical concerns of differential treatment        based in Africa. There is clear need for
of clients in control trials (Jaranson &          scientific/ experimental research in this area
Quiroga, 2011; Stammel et al., 2017), the         of work in an African context.
expense of scarce resources of rehabilitation
programmes (Bandeira, 2013), and                  The CSVR multi-modal framework for
generalisability of study results (Pérez-sales,   the psychosocial rehabilitation of
Witcombe, & Otero Oyague, 2017) are               individual survivors of torture
among the challenges of outcome studies.          Given the dire contextual realities in which
    The few experimental or quasi-                many torture survivors live, research on
experimental designs on torture survivors         the effectiveness of treatment frameworks
are usually affected by small sample sizes,       needs to include psychological and social
the lack of control groups (Jaranson &            health as consequences of torture (Patel,
Quiroga, 2011) and instruments with               Kellezi, & Williams, 2014) in addition to the
limited validity and reliability often focused    usual consequences of PTSD and MDD
only on improvement in PTSD and MDD               (Bandeira, 2013; Higson-Smith, 2013).
(Jaranson & Quiroga, 2011). Meta-analyses         Accordingly, the Centre for the Study
of randomised control trials illustrate           of Violence and Reconciliation (CSVR)
improvements of generally small effect            developed a multi-modal guiding framework
sizes in PTSD and depression when using           for the rehabilitation of individual torture
Cognitive Behavioural Therapy (CBT)               survivors that takes into account their lived
and Narrative Exposure Therapy (NET),             realities of continuing traumas and ongoing
with moderate improvement on follow-up            daily stressors (Bandeira et al., 2013).
(Patel et al., 2014). However, exceptional            The development of the multi-modal
experimental studies do exist, with substantial   CSVR model, hereafter referred to as the
improvements found using the Common               CSVR framework, was based on a literature
Elements Treatment Approach, compared             search, analysis of intervention process notes
to the less effective Cognitive Processing        of CSVR clinical staff, and a Delphi process
Therapy (Weiss et al., 2015). Quasi-              for expert consensus on the most severe
                                                                                                   T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

experimental and outcome research on the          impacts of torture and the most appropriate
rehabilitation of torture survivors provides      methods of torture intervention in a South
further support for CBT (Halvorsen &              African context. There was consultation
Stenmark, 2010; Neuner et al., 2010) and          between the research and clinical teams
NET (Dibaj, Overaas Halvorsen, Edward             on the categorisation of the impacts, their
Ottesen Kennair, & Inge Stenmark, 2017;           assessment and the most appropriate
Hansen, Hansen-Nord, Smier, Engelkes-             intervention approaches. As a result, the
Heby, & Modvig, 2017), culturally tailored        CSVR framework comprises the 18 most
health promotion intervention (Berkson,           severe impacts of torture (14 after grouping
Tor, Mollica, Lavelle, & Cosenza, 2014)           the impacts) in contexts such as those found
38

                                                                                                                  SCIENTIFIC ARTICLE

                                                    in South Africa, and the most appropriate          Given the reality of unsafe
                                                    intervention strategies associated with each       circumstances, it may be appropriate that
                                                    (Table A - 1, numbered alphabetically per          clients’ reactions include paranoia and
                                                    area). Aspects of trauma-focused CBT               hypervigilance. The clinical approach
                                                    (TFCBT), NET, dialectical behavioural              outlined in the framework uses reality
                                                    therapy, supportive therapy, problem solving       testing, dealing with perceived threats,
                                                    and solution-focused therapy underpin              safety planning, skills development
                                                    therapeutic interventions in the framework,        and symptom management to assist
                                                    with an emphasis on empowerment. As such,          clients to reduce repeated victimisation
                                                    the framework is considered to be multi-           and increase their skills to ensure their
                                                    modal and informed by mutiple theories.            continued safety.
                                                        Many centres adopt a multi-disciplinary,   •   Continuing traumas while dealing with
                                                    multi-modal, or “common-sense” approach            past traumas and torture experiences:
                                                    to psychosocial interventions (Pérez-sales         The framework provides guidelines
                                                    et al., 2017) in which the therapist chooses       on how clinicians may help clients
                                                    from different modalities according to             understand their reactions to past
                                                    the needs of the torture survivors. These          traumas and how these reactions relate
                                                    approaches provide a more meaningful               to their responses to the ongoing traumas
                                                    personalisation of the survivors’ needs,           that may affect them.
                                                    which together with a clear therapeutic        •   Managing relationships with service
                                                    relationship and culturally tailored goals,        providers, family members and
                                                    may be more appropriate than a rigid               community members:
                                                    “one-size-fits-all” therapeutic model.             The framework provides guidelines
                                                    While there are several multi-modal                on how clinicians may assist clients to
                                                    treatment approaches for assisting torture         deal with their emotional reactions to
                                                    survivors (Drozdek, 2015; Stammel et al.,          traumas, and to build the social capital
                                                    2017), the CSVR framework offers clearly           necessary to maintain relationships and
                                                    articulated therapeutic guidance relevant          obtain services, for example, medical and
                                                    to South African and developing contexts.          legal assistance or documentation from
                                                    For example, it provides clear outlines            the Department of Home Affairs.
                                                    for the role of clinicians in attending to
                                                    the therapeutic needs of clients, and for      Research design
                                                    the “core business” of the CSVR clinical       The design of the current research is
                                                    team, both essential therapeutic elements      described as quasi-experimental. It
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    for maintaining therapeutic boundaries,        comprises two groups of participants,
                                                    empowering clients and increasing their        both groups observed over six months
                                                    resilience. Although it is assumed that what   but differentiated according to when their
                                                    the client brings to therapy will be the       participants approached the Centre, when
                                                    focus of the session, the CSVR framework       they commenced treatment with the CSVR
                                                    provides guidelines on how to assist clients   framework, and the duration of treatment
                                                    when certain needs arise, for example:         based on the CSVR framework (Figure 1).
                                                    • Lack of safety, repeated victimisation       The ‘treatment’ group comprised clients
                                                        and high levels of violence in a South     who came for counselling between July 2014
                                                        African context:                           and December 2015 and were treated based
39

SCIENTIFIC ARTICLE

on the CSVR framework for six months. The        for the comparison group. Secondly, it was
‘comparison’ group comprised clients who         hypothesised that there would be greater
came for counselling at the Centre between       improvement in the psychological and
January and June 2014 and were placed on         functioning measurements from T1 to T3 for
a 3-month ‘Waiting list’ condition before        the treatment group than for the comparison
commencing three months of treatment             group, as the treatment group received
based on the CSVR framework.                     treatment based on the CSVR framework
    Psychological wellbeing and functioning      for six months, while the comparison group
were measured three times: participants          received this treatment for three months
in the treatment group were measured             only following the 3-month waiting list
before the start of the CSVR intervention        condition.
treatment (T1), three months after the start
of CSVR framework treatment (T2), and            Methods
after a further three months of the CSVR         Procedure
framework treatment (T3); participants in        The treatment group received treatment
the comparison group were measured before        based on the CSVR framework for six
the start of a 3-month waiting period (T1),      months; the comparison group received an
at the end of the 3-month waiting period         initial 3-month waiting condition before
(T2), and then three months after the start      receiving treatment based on the CSVR
of the CSVR framework treatment (T3) (see        framework for three months.
Figure 1).                                           In the 3-month waiting condition, a
    Written informed consent was obtained        clear waiting list management took place. A
from all participants. Ethical clearance         trained trauma professional phoned clients
was provided by an internal CSVR ethics          in the comparison group every two weeks
committee as well as by external experts in      to check that they still wanted to come
the violence and/or torture field, including a   for counselling, and to refer clients for
lecturer at a Johannesburg-based university      medical, legal and humanitarian assistance
and two partners working at international        if necessary. This condition compared with
NGOs providing services to survivors of          clients who went straight into treatment
torture. This approach to ethical clearance      using the CSVR framework (see The
was necessary as there is no overarching         CSVR multi-modal framework for the
national ethics body in South Africa, and        psychosocial rehabilitation of individual
CSVR is not affiliated with a university.        survivors of torture, on page 37).
                                                 Emergency cases in both groups (clients
                                                                                                 T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

Hypotheses                                       experiencing psychosis, who were suicidal
As the treatment group received specialised      or other emergencies) were contained
counselling for the first three months           and referred for psychiatric assistance at a
and the comparison group received less           local hospital, or for medication from the
specialised communication (as part of the        consultant psychiatrist at the CSVR.
waiting list protocol) for their first three         Baseline measures were carried out
months, it was hypothesised that there           using the same instrument at the start of the
would be greater improvement in the              CSVR treatment condition of the treatment
psychological and functioning measurements       group, and at the start of the waiting time
from T1 to T2 for the treatment group than       condition of the comparison group (T1). A
40

                                                                                                                  SCIENTIFIC ARTICLE

                                                    second set of measures was carried out three   i.e. six months after commencement of the
                                                    months after commencement of the CSVR          CSVR condition of the treatment group, and
                                                    condition of the treatment group, and three    three months after commencement of the
                                                    months after the start of the waiting time     CSVR condition of the comparison group
                                                    condition of the comparison group (T2). A      (T3) - see Figure 1.
                                                    third set of measures was carried out three         Every two weeks, clinicians participated
                                                    months after the second set of measures,       in supervision sessions which focused
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    Figure 1: Study procedure
41

SCIENTIFIC ARTICLE

both on fidelity to the framework as well           (waiting list) baseline (T1), were included in
as guidance with therapeutic concerns.              the waiting list group, and had also
Supervisors ensured that clinicians followed        completed a treatment baseline (T2). At T3,
the framework and concurrently documented           16 participants from this group remained,
which aspects of the framework were utilised        having completed the waiting list baseline,
using a fidelity checklist. Any difficulties with   treatment baseline and 3-month follow-up
the implementation of the framework were            assessment, and these 16 were included in
discussed with the researchers and indicated        the follow-up study (Figure 1).
in the fidelity checklist.                              Assessment measures: The assessment
                                                    tools used for all clients at T1, T2 and T3
Participants                                        included the Harvard Trauma
All participants were selected according            Questionnaire (HTQ) measuring
to the 1985 UNCAT definition of torture             posttraumatic stress disorder (PTSD),
(United Nations General Assembly, 1985)             clients’ self-perception of functioning and
and were over 18 years old. No substantial          overall trauma (Harvard Program in
historical events were noted for the 6-month        Refugee Trauma, 1992), the Hospital
period that separated the groups.                   Anxiety and Depression Scale (HADS)
    Participants in the treatment group: The        measuring anxiety and depression
participants in the treatment group were            (Zigmond & Snaith, 1983), the De Jong
selected from clients who approached the            Gierveld Loneliness Scale measuring
CSVR between 1 July 2014 and 30 December            clients’ emotional loneliness and social
2015. Over this period, 197 clients were            loneliness (de Jong Gierveld & Van Tilburg,
screened, and 24 clients excluded based on the      2006), the number of areas of pain in the
UN torture definition, and a further 129            body,2 and management of aspects of
clients excluded due to incomplete T1 or T2         functioning.3 Reliability measures are
outcome measures. Thus 44 clients had               presented in Table A - 2.4
completed both their baseline assessment and a          All assessments were administered by
three-month follow-up and were included in          trained psychologists or social workers.
the treatment group. Of this group, 20              Translation through interpretation was
completed their six-month follow-up and were
included in the follow-up study (Figure 1).
    Participants in the comparison group: The       2
                                                        The total number of areas of pain was based on
participants in the comparison group were               the areas of pain in the body pointed out or indi-
selected from clients who approached the                cated verbally.
                                                                                                              T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

                                                    3
                                                        For management of aspects of functioning, four
CSVR from January to June 2014. Over this
                                                        Likert-type scaled questions were adapted from
period, 87 clients were screened, 66 of whom            the International Classification of Pain and
met the inclusion criteria for the waiting list         Disability (ICF) (WHO, 2001): Family-related
group, with 21 excluded as they had not                 stressors, External stressors (excluding family-
                                                        related stressors), Managing situations that made
experienced torture according to the UN
                                                        the client angry, and Psychological or emotional
torture definition. A further 28 clients were           functioning. Scale responses ranged from manag-
then excluded as they had not completed a               ing very poorly, to managing very well.
follow-up assessment, leaving 38 clients as
                                                    4
                                                        As Cronbach’s coefficient α is correlated with the
                                                        number of items in the scale, average inter-item
participants in the comparison group. These
                                                        correlations are also presented as their values are
38 clients had completed a comparison                   independent of scale length (Table A - 2).
42

                                                                                                                              SCIENTIFIC ARTICLE

                                                    provided to clients in the main languages of                    Finally, post hoc analyses were used
                                                    clients who come to the clinic. These include              to examine the effects of the higher than
                                                    French, Swahili, Amharic, Somali and                       expected attrition rate of respondents of the
                                                    Lingala. All interpreters were trained on the              two groups who dropped out of the study
                                                    assessment tools.                                          after T2, i.e. during the 3-month counselling
                                                         Analysis methods: Initially the baseline              period between T2 and T3. This comparison
                                                    (T1) psychological and functioning                         was deemed necessary to examine whether
                                                    measurement means of the treatment and                     a systematic bias may have been added to
                                                    comparison groups were compared using                      the data at T3, should the participants who
                                                    t tests and Cohen’s d effect sizes, and                    left the comparison group after T2 differ
                                                    categorical variables compared via the Chi                 from the participants who left the treatment
                                                    square test. Thereafter the mean changes                   group after T2. Such a situation would bias
                                                    of the two groups were compared for the                    the samples and the results of the T1-T3
                                                    first three months (T1 to T2) and then                     comparison relevant to the second hypothesis.
                                                    for the six month period (T1 to T3) via                    This analysis was a simple examination of
                                                    mixed analysis of variance (ANOVA).5 In                    the demographics, as well as of the direction
                                                    both cases, the ANOVA interaction effects                  of the mean differences for each scale from
                                                    were examined for evidence of a significant                T1 to T2 for the participants in each group
                                                    differential pattern in the mean scores across             who dropped out after T2. No other post hoc
                                                    the time periods depending on the group,                   analyses on the results are presented.
                                                    and partial eta-squared effect sizes were used
                                                    as measures of the strength of differences.                Results
                                                    Where necessary, the Scheffé post hoc test                 Table 1 shows details of the participant
                                                    was used to locate significant differences                 characteristics for both groups.
                                                    between means. In view of the multiple                     Approximately half of the participants were
                                                    violations of the assumptions of the mixed                 male, fewer than 10% were South Africans,
                                                    ANOVA analyses, the conservative non-                      and the dominant groups were from Congo
                                                    parametric Wilcoxon signed-ranks test was                  (DRC), Ethiopia and Somalia. Over half
                                                    used on the separate groups. In addition, the              were living with a partner, spouse, family
                                                    odds ratio was used to indicate effect sizes               member or friend, approximately 80% were
                                                    for the individual groups, and for assessing               either asylum seekers or refugees, and fewer
                                                    the odds of improvement in the treatment                   than half of the participants of each group
                                                    group relative to the comparison group.6                   had experienced torture within the past year
                                                                                                               in each group (a third of the comparison
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                                                                               versus 39% of the treatment group). Before
                                                    5
                                                        Ideally, mixed multivariate Analyses of Variance       the torture, 5% or fewer were unemployed
                                                        (MANOVA) would be computed on the compo-               whereas approximately three-quarters or
                                                        nents of the scales to detect multivariate response    more were unemployed at the start of the
                                                        patterns, while controlling the family-wise Type 1
                                                                                                               study. The ages of the participants ranged
                                                        error rate with greater power to detect differences.
                                                        However, the sample sizes were small, and in all       from 18 to 72 years, with mean age in the
                                                        instances, assumption violations of the multivari-     mid-thirties (M=36.20 years, SD=10.35
                                                        ate analyses mitigated against the MANOVA tests.       years for the treatment group, and M=34.82
                                                    6
                                                        This exploratory approach was undertaken whilst
                                                                                                               years, SD=8.68 years for the comparison
                                                        recognising that the family-wise Type 1 error in-
                                                        creases with multiple statistical tests.               group). The treatment group had a higher
43

SCIENTIFIC ARTICLE

Table 1: Demographics of treatment and comparison groups

                                                            Treatment      Comparison
Demographics
                                                           group (n=44)   group (n=38)
                                                           n       %       n      %      p^
Gender                   Female                            22     50%      16    42%     0.47
                         Male                              22     50%      22    58%
Nationality              Burundi                           2      5%       1      3%     0.98
                         Congolese (DRC)                   15     34%      13    34%
                         Eritrean                          0      0%       4     11%
                         Ethiopian                          8     18%      13    34%
                         Somali                            12     27%      3      8%
                         South African                     1      2%       3      8%
                         Zimbabwean                        2      5%       0      0%
                         Other *                           4      9%       1      3%
Marital status           Currently married                 24     55%      8     21%     0.01
                         Divorced/separated                3      7%       4     11%
                         Never married                     13     30%      18    47%
                         Widowed                           4      9%       7     18%
                         Missing                            0     0%       1      3%
Currently living         Living alone/strangers/shelter    14     32%      7     19%     0.14
                         Living with family/ children      16     36%      5     13%
                         Living with friends               7      16%      10    26%
                         Living with spouse/ partner       4      9%       4     11%
                         Other                             0      0%       4     11%
                         Missing                           3      7%       8     21%
Legal SA status          Asylum seeker/ refugee            37     84%      30    79%     0.78
                         Citizen                           2      5%       3      8%
                         Undocumented                      4      9%       1      3%
                         Missing                           1      2%       4     11%
Pre-torture employment   Minor/ student                    5      11%      5     13%     0.83
                         Unskilled labour/ semi-skilled    22     50%      21    56%
                                                                                                T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

                         Skilled/ professional             14     32%      10    26%
                         Unemployed                        2      5%       1      3%
                         Missing                           1      2%       1      3%
Current employment       Minor/ student                    0      0%       2      5%     0.39
                         Unskilled labour/ semi-skilled    8      18%      3      8%
                         Skilled/ professional             3      7%       2      6%
                         Unemployed                        32     73%      31    82%
                         Missing                           1      2%       0      0%
44

                                                                                                                               SCIENTIFIC ARTICLE

                                                                                                                       Treatment           Comparison
                                                    Demographics
                                                                                                                      group (n=44)        group (n=38)
                                                                                                                       n         %          n        %        p^
                                                    Education                   No schooling                           4        9%          1       3%
45

SCIENTIFIC ARTICLE

Table 2: Trauma types experienced by respondents of the comparison and treatment groups

                                               Treatment group Comparison group Odds ratio: Treatment/
                                                    (n=44)          (n=38)          Comparison
                                                                                  OR        Fisher exact
                                                  n       %      n        %                     test
                                                                                                (p)
Torture/ CIDT                                    44     100%     38     100%        -             -
Witness to trauma                                31       70%     30       79%        0.64        0.27
Assault                                          25       57%     28       74%        0.47        0.09
Bereavement/ traumatic bereavement               16       36%     21       55%        0.46        0.07
Armed robbery                                    18       41%     18       47%        0.77        0.36
Xenophobia                                       12       27%     17       45%        0.46        0.08
Mugging                                           5       11%     13       34%        0.25        0.02
War                                              15       34%     13       34%        0.99        0.59
Hostage/ kidnapping/ abduction                    9       20%      8       21%        0.96        0.58
Rape/ attempted rape/ sodomy                     14       32%      6       16%        2.49        0.08
Motor vehicle accident                            2       5%       5       13%        0.31        0.16
Hijacking                                         5       11%      5       13%        0.85        0.53
Relationship/ domestic violence                   2       5%       5       13%        0.31        0.16

that three of the four initial coping or                   Comparison of the changes in the psychological
managing scores of the treatment group                     and functioning measures across the first three
are significantly better than those of the                 months (T1 to T2) for each group: Based on
comparison group, and the PTSD, trauma,                    the direction of the differences between each
anxiety and depression scores are lower                    of the T1-T2 pairwise means for the
(α=.05). The effect sizes of these differences             treatment and the comparison groups
are moderately large (Cohen’s d with                       considered separately, all 13 of the pairwise
magnitude at least .5), with the exception                 differences of the treatment group are in the
of the strong difference in depression scores              ideal direction of improved psychological
of the two groups (t(79)=3.36, p
46

                                                                                                                      SCIENTIFIC ARTICLE

                                                    assumption violations for the T1 or T2 scores     2.61 for anxiety means that the clients in
                                                    of 7 of the 13 ANOVA comparisons, the             the treatment group reduced their anxiety
                                                    T1-T2 comparisons of the separate groups          2.61 times more than the clients in the
                                                    were re-examined using the conservative           comparison group did.
                                                    non-parametric Wilcoxon signed-ranks test             There thus appears to be partial support
                                                    with Z score conversion for large sample          for the expectation of greater improvement
                                                    sizes and asymptotic significance, and the        in psychological health and functioning
                                                    odds ratio (OR) used to indicate effect size      from T1 to T2 for the treatment group
                                                    (Table 3).                                        relative to the comparison group. We
                                                        The Wilcoxon results of the treatment         therefore conclude, conservatively, that
                                                    group show seven significance changes             there is insufficient statistical evidence
                                                    (emotional and total loneliness, external         for overall support of the first hypothesis
                                                    stressors, functioning, trauma, anxiety and       of the research and we instead discuss
                                                    areas of pain), all with odds ratios greater      the significant T1-T2 differences on the
                                                    than 1 (Table 3). By contrast, there are two      individual indicators, specifically differences
                                                    significant T1-T2 differences (depression         on the indicators of functioning and anxiety.
                                                    and areas of pain) in the comparison                  Comparison of the changes in the
                                                    group. For the treatment group, the odds of       psychological and functioning measures across
                                                    improvement relative to non-improvement           the three time periods (T1-T3) for each group:
                                                    are greater (odds > 1) for 9 of the indicators    The comparisons from T1 to T3 are based
                                                    of psychological health and functioning,          on substantially reduced sample sizes owing
                                                    compared to 5 indicators for the comparison       to attrition of participants in the two groups
                                                    group. Furthermore, ratios of the odds            after the second set of measures. Under half
                                                    for the treatment group relative to the           the respondents remained after T2 (45% or
                                                    comparison group computed per indicator           20 of the 44 participants in the treatment
                                                    variable show OR values greater than 1            group, and 42% or 16 of the 38 participants
                                                    for most indicators (emotional and total          in the comparison group).
                                                    loneliness, the coping indicators of external         Based on the mixed ANOVA tests, there
                                                    stressors and psychological difficulties, the     is no evidence of significant interaction
                                                    HTQ scales of PTSD, functioning and               effects that would show a difference in the
                                                    trauma, the HADS anxiety scale, and areas         changes of scores over time depending on
                                                    of pain - Table 3). Thus, for these indicators,   the group, and the magnitude of all effect
                                                    members in the treatment group improved           sizes, as measured by partial eta squared
                                                    their psychological health and functioning        values, are negligible to weak. However,
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    from T1-T2 more than members in the               significant main effects across time, in the
                                                    comparison group did. However, based on           direction of increased psychological health
                                                    the Fisher Exact Probability test (1-tailed,      and functioning, were found for coping
                                                    α=.05), the odds ratios are only significant      with anger, the HTQ, the HADS, as well as
                                                    for the HTQ functioning scale and for the         for number of areas of pain. In particular,
                                                    HADS anxiety scale. The OR value of 2.49          based on post hoc Scheffé comparisons,
                                                    for functioning means that the clients in the     measures of self-perception of functioning,
                                                    treatment group improved their functioning        depression and number of areas of pain
                                                    2.49 times more than the clients in the           improved significantly from one time to the
                                                    comparison group did; the OR value of             next. As for the T1-T2 comparisons, the
47

SCIENTIFIC ARTICLE

Table 3: Wilcoxon signed-rank test and odds ratio effect sizes for T1-T2 repeated measures for
treatment and comparison groups

                                                                                               Odds ratio:
                                            Treatment group           Comparison group
                                                                                               Treatment/
                                                (n=44)                    (n=38)
                                                                                               Comparison
Indicators
                                                                                                       Fisher
                                                            Odds                      Odds             exact
                                           Z     p^                   Z     p^                 OR
                                                             ^^                        ^^               test
                                                                                                        (p)
Connection to     Emotional               2.06 0.04 *        1.32    0.10 0.92         1.06     1.25    0.40
others            loneliness

                  Social loneliness       0.98 0.33          1.16    0.13 0.89         1.24     0.94    0.53

                  Total loneliness        2.11 0.03 *        1.29    0.03 0.98         1.09     1.18    0.44

Coping or         Angry situation         1.43 0.15          0.83    0.49 0.63         0.84     0.98    0.58
managing
                  Family related          0.59 0.56          0.67    1.59 0.11         0.87     0.77    0.38
                  stressors

                  External stressors      2.06 0.04 *        1.06    0.07 0.94         0.69     1.54    0.23

                  Psychological           1.34 0.18          0.97    0.58 0.56         0.79     1.23    0.42
                  difficulties

Harvard           PTSD                    1.70 0.09          1.21    0.27 0.79         1.10     1.10    0.52
trauma
questionnaire     Functioning             2.57 0.01 *        1.71    0.39 0.69         0.68     2.49   0.04 *

                  Trauma                  2.28 0.02 *        1.56    0.02 0.99         0.79     1.97    0.10

Hospital          Anxiety                 2.83 0.01 **       1.56    0.48 0.63         0.60     2.61   0.04 *
anxiety and
depression        Depression              0.99 0.32          0.87    3.17 0.01 **      1.79     0.49    0.10
scale

Pain              Number of areas         2.44 0.01 *        1.23    2.26 0.02 *       0.94     1.31    0.36
                  of pain
                                                                                                                 T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

* p < .05; ** p < .01
^ denotes the asymptotic significance value (2-tailed) for Wilcoxon Z for large samples
^^ tied changes were randomly distributed in equal proportions to positive and negative changes, rendering the
   odds ratio more conservative than the Z score

T1-T3 comparisons were re-examined using                 psychological health and functioning,
the conservative non-parametric Wilcoxon                 exceptions being social loneliness and
signed-ranks test on the separate groups,                coping with external stressors. However,
owing to violations of ANOVA assumptions.                counter-intuitively, the results of the
    For both groups, most of the outcome                 Wilcoxon signed-ranks tests on the T1-T3
means changed in the direction of improved               comparisons computed on the separate
48

                                                                                                                       SCIENTIFIC ARTICLE

                                                    groups revealed more significant T1-T3              In Table 4, the means for each scale
                                                    improvements for the comparison group           are presented for those members of the
                                                    than for the treatment group. For the           treatment group and the comparison group
                                                    comparison group, 7 of the 13 outcome           who left the programme after the second set
                                                    measures improved significantly from T1         of measures.9 The outcome measures with
                                                    to T3 (α=.05), (coping with anger, the          T1-T2 changes in the direction of improved
                                                    HTQ measures of PTSD, functioning and           psychological health and functioning are
                                                    trauma, the HADS measures of anxiety and        indicated with a tick ().10
                                                    depression, and areas of pain), compared to         The members of the treatment group
                                                    only 1 of the 13 outcome measures (areas of     who left the programme after T2 had
                                                    pain) for the treatment group. Nevertheless,    improved their mean scores on all 13
                                                    none of the odds ratios computed via            measures at T2. By contrast, the members
                                                    the ratio of odds for the treatment and         of the comparison group who left the
                                                    comparison groups per indicator variable        programme after T2 had improved their
                                                    are significant based on the Fisher             mean scores on fewer than half (6 out of 13)
                                                    Exact Probability test (α=.05). Thus,           of the measures (social and total loneliness,
                                                    although there are more significant T1-T3       family related stressors, psychological
                                                    improvements in the comparison group            difficulties, depression and areas of pain),
                                                    than in the treatment group, the magnitude      a significant difference in percentages
                                                    of these improvements is not sufficient to      (p=.002). Thus, participants in the treatment
                                                    render the odds of the improvements in          group who left the programme at T2 had
                                                    the comparison group significantly greater      improved their overall psychological health
                                                    than the odds of the improvements in the        and functioning, whereas the participants
                                                    treatment group.                                in the comparison group who left the
                                                        However, the counter-intuitive Wilcoxon     programme at T2 had not shown this
                                                    test results on the T1-T3 comparisons,          overall level of improvement. After T2,
                                                    whatever their magnitude, need to be            the treatment group appears to have been
                                                    addressed. A possible explanation is            weakened by the loss of members who
                                                    presented in a post hoc set of comparisons      had improved health after three months,
                                                    using a basic analysis of attrition after T2.   while the comparison group had not lost
                                                    Comparison of participants in each group        overall improvers. This differential pattern
                                                    who dropped out after T2: Based on the          of attrition between the groups may explain
                                                    Chi square test for between-group               the counter-intuitive findings of greater
                                                    comparisons, there is no evidence of            improvement from T1-T3 of the remaining
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    significant differences in the demographic      members of the comparison group
                                                    characteristics of participants who             compared to the remaining members of the
                                                    dropped out of the treatment group after        treatment group.
                                                    T2 versus those who dropped out of the
                                                    comparison group after T2. However,
                                                    comparison of the T1-T2 changes in the          9
                                                                                                         As tests of significance are reserved for a future
                                                    psychological health and functioning                 in-depth article on attrition, this analysis does
                                                    measures of these two groups suggests                not indicate significance or effect size.
                                                                                                    10
                                                                                                         In each comparison, the direction of the means is
                                                    that there may be a differential pattern of
                                                                                                         based on the difference in the pairwise means of
                                                    attrition between the groups.                        individuals.
49

SCIENTIFIC ARTICLE

Table 4: Mean scores from T1-T2 of participants who left the comparison and treatment groups
after T2

                                                               Participants who left after T2:

                                                      those who left from       those who left from
                                                      the treatment group         the comparison
                                                             (n=24)                group (n=22)

Indicators                                             Mean          T1-T2       Mean        T1-T2

                                              T1        2.83           ü          2.48
                Emotional loneliness          T2        2.52                      2.50

Connection to                                 T1        2.87           ü          2.52           ü
                Social loneliness             T2        2.75                      2.32
others

                                              T1        5.70           ü          5.00           ü
                Total loneliness              T2        5.17                      4.82

                                              T1        2.61           ü          2.14
                Angry situation               T2        2.67                      2.00

                                              T1        2.23           ü          1.60           ü
                Family related stressors      T2        2.43                      2.21
Coping or
managing                                      T1        1.87           ü          1.90
                External stressors            T2        2.33                      1.86

                                              T1        2.35           ü          2.10           ü
                Psychological difficulties    T2        2.74                      2.24

                                              T1        2.54           ü          2.76
                PTSD                          T2        2.33                      2.91

Harvard                                       T1        2.48           ü          2.60
trauma          Functioning                   T2        2.25                      2.82
questionnaire
                                              T1        2.50           ü          2.67
                                                                                                      T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

                Trauma                        T2        2.28                      2.83

                                              T1        1.71           ü          1.86
Hospital        Anxiety                       T2        1.31                      1.89
anxiety and
depression                                    T1        1.57           ü          1.89           ü
scale           Depression                    T2        1.28                      1.63

                                              T1        1.96           ü          3.23           ü
Pain            Areas of pain                 T2        1.83                      2.32
50

                                                                                                                   SCIENTIFIC ARTICLE

                                                    Discussion                                      assumptions required for parametric
                                                    The research was designed to test the           statistical analyses. Attrition created a
                                                    efficacy of the CSVR psychosocial               further problematic effect in our study as
                                                    intervention framework using a sample of        the treatment group clients who dropped
                                                    torture survivors, almost all of whom were      out of the study after T2 showed greater
                                                    asylum seekers, refugees and undocumented       overall improvement in their psychological
                                                    persons in a South African environment          health and coping from T1 to T2 than the
                                                    filled with daily stressors and continuing      comparison group clients who dropped
                                                    traumas. These individuals were found to        out of the study after T2. This differential
                                                    have clinical baseline scores on PTSD,          attrition effect may have biased the longer-
                                                    depression, anxiety and low levels of           term comparison of the mean scores
                                                    functioning, which may be explained, at         from T1 to T3 (the second hypothesis of
                                                    least in part, by their history of torture      the study), creating paradoxical results.
                                                    which severely impacts psychological and        Thus, our main finding pertaining to the
                                                    psychiatric wellbeing (Bandeira et al., 2010;   hypotheses on the efficacy of the CSVR
                                                    Higson-Smith, 2013; McColl et al., 2010;        framework is that we have not found
                                                    Patel et al., 2014; Quiroga & Jaranson,         sufficient evidence to warrant the efficacy
                                                    2005). The CSVR framework is designed           of the framework, owing mainly to several
                                                    to address the complexity of the past           severe challenges inherent in research of this
                                                    traumas exacerbated by contextual realities     nature. It may be fairer to conclude that
                                                    of a hostile South African environment          the research results on the efficacy of the
                                                    (Bandeira, 2013; Bandeira et al., 2013).        framework are inconclusive.
                                                                                                        However, there are several interesting
                                                    Efficacy of the CSVR framework                  results in the research. Firstly, improvement
                                                    The assessment of the efficacy of the CSVR      in the psychological wellbeing and
                                                    framework was based on the expectation          functioning of the clients of both groups
                                                    of a long-term differential effect when an      from T1 to T3 occurred despite the context
                                                    initial three-month waiting list condition      of the lack of safety and continuing daily
                                                    was included, compared to when clients          traumas of the torture survivor clients.
                                                    entered straight into treatment. The findings   This result is consistent with the CSVR
                                                    of the research did not provide clear support   framework’s prioritisation of empowerment,
                                                    for the hypotheses of greater improvement       problem solving and trauma-related therapy.
                                                    in the psychological health and coping          Secondly, after three months of treatment
                                                    indicators of the CSVR treatment group          under the CSVR framework, there are
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    than the waiting list comparison group.         indications of clients’ improved connection
                                                    However, the results are complex and less       to others. This result is encouraging within
                                                    than clear.                                     the inherent mistrust, isolation and fear
                                                        The data gathered reflect the challenges    as consequences of torture (Quiroga
                                                    of attempting rigorous quasi-experimental       & Jaranson, 2005). Building trusting
                                                    quantitative research in the context of         relationships after a torture experience
                                                    torture, using established measurement          should be viewed as a long-term therapeutic
                                                    scales that were found to have lower than       goal (as per the approach of the CSVR),
                                                    desirable reliability for our sample, and       as torture often results in the annihilation
                                                    data that do not meet all the stringent         of interpersonal trust (Bandeira, 2013;
51

SCIENTIFIC ARTICLE

Bandeira et al., 2013). Further research is         CSVR treatment framework for the two
warranted on the prioritisation of a family         groups, we found that the psychological
and group model or framework in the South           health and functioning of members of the
African context to mitigate the isolation felt      treatment group was somewhat better than
by the torture survivor, and the best way to        that of the comparison group at baseline
reintegrate the torture survivor into trusting      testing. The groups also differed on marital
familial, group and community relationships.        status and education levels.
Thirdly, after three months in therapy,                 In addition to these baseline differences
clients appear to deal better with their past       between the groups, the nature of the
traumas and torture experiences in the              comparison group condition, i.e. the
context of unsafe situations, specifically in       waiting list protocol, was not a true placebo
terms of increased functioning and reduced          comparison as it offered a level of support
anxiety levels.                                     to clients in the comparison group. The
    We move now to discuss the main                 comparison group clients received support in
challenges of scientific research on trauma,        telephone calls every second week, referrals
based on the lessons of this study.                 offered for legal, medical or humanitarian aid,
                                                    and assistance offered if they were psychotic,
Challenges of research on trauma rehabilitation     suicidal or required other emergency
It is essential to conduct thorough research        assistance. After three months of receiving
on the outcomes of torture rehabilitation           this level of support, these waiting list clients
programmes that could improve the quality           may well have felt somewhat better on a
of the services and care to torture survivors,      psychological and functioning level than they
contribute to professional development in           had at baseline. Indeed, the results of the
the field, and empower torture survivors            comparison group clients showed significant
(Jaranson & Quiroga, 2011). Although                improvement in their depression scores.
quantitative studies involving the principles       However, despite the possible confounding
of randomised experimental design, sound            effect of the waiting list condition, it is
sampling techniques, objective measures and         an ethical necessity in the context of the
rigid statistical analysis are scientific ideals,   vulnerability of torture survivors.
such criteria are, in reality, unattainable             High levels of overall attrition are
in the complex and challenging world of             expected in this population due to the
torture rehabilitation with its thorny ethical      transient nature of tortured refugees in
considerations and limitations. Ethical             South Africa who often need to move
compliance precludes randomised control             to find employment or accommodation,
                                                                                                        T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

group designs; instead, researchers of torture      or due to safety concerns. Furthermore,
attempting scientific studies are restricted to     many refugees and asylum seekers use the
quasi-experimental or outcome designs with          CSVR for introduction to legal, medical
compromised internal validity.                      or humanitarian aid, and then terminate
     We present the consequences of these           contact with the centre. Although previous
research design restrictions in our research.       CSVR reports indicate an attrition rate of
Although there was no apparent reason               21-44% of clients annually (Dix-Peek, 2012a,
for the groups to differ at baseline, and no        2012b), the current research indicates a
outstanding historical events occurring in          much higher rate (75% or 129 of 173 clients
the time interval between the start of the          in the treatment group, and 42% or 28 of
52

                                                                                                                                SCIENTIFIC ARTICLE

                                                    66 clients in the comparison group).11 The               who improved their health may be evidence
                                                    high rate of attrition from screening to T1,             of the preliminary success of the CSVR
                                                    referred to here as overall attrition, for both          framework, although the sustainability of the
                                                    groups, but particularly for the treatment               improvement is unknown.
                                                    group, is a clear limitation of the current                  The topic of attrition is planned as part
                                                    research. Contextual challenges of staffing              of a future journal article.
                                                    changes within CSVR may have contributed
                                                    to clients not completing their baseline T1 or           Conclusion
                                                    follow-up T2 assessments and the consequent              The CSVR psychosocial framework for the
                                                    stark overall attrition for the treatment                rehabilitation of torture survivors provides a
                                                    group. CSVR has attempted to mitigate the                set of therapeutic guidelines in the context
                                                    high overall attrition through ensuring that             of daily stressors, continuing traumas,
                                                    clients receive transport money to and from              past torture and trauma events. It is also a
                                                    the office, conducting drop-out reports for              framework designed for developing country
                                                    a better understanding of why clients stop               settings where survivors live with lack of
                                                    coming for counselling, providing ongoing                safety and extreme socio-economic concerns.
                                                    training with CSVR clinicians and monitoring             Thus evidence-based research on the
                                                    and evaluation staff, and ensuring adherence             efficacy of this framework is essential as a
                                                    to the monitoring and evaluation system.                 first step towards investigating its legitimacy
                                                         The high rate of attrition of clients after         as “best practice” for torture survivors in
                                                    the initial three months of the programme                developing countries. However, conducting
                                                    is also of great concern for the CSVR                    such research is often not prioritised due
                                                    rehabilitation endeavours as over half of                to scarce financial and human resources,
                                                    the clients left the study after T2 (55% of              practitioners who may not recognise the
                                                    the treatment group members and 58%                      need to prioritise outcome research over the
                                                    of the comparison group members). A                      “core” therapeutic and psychosocial work
                                                    rudimentary post hoc attrition analysis of the           (Bandeira, 2013; Montgomery & Patel,
                                                    means of the outcome measures suggests                   2011), attrition and small samples sizes,
                                                    that different patterns of attrition in the              limited academic and research expertise,
                                                    comparison and treatment groups may                      and ethical concerns (Jaranson & Quiroga,
                                                    explain the counter-intuitive results from               2011). Our study has aimed to contribute to
                                                    T1-T3. The comparison group clients who                  a detailed understanding of the intricacies
                                                    left the programme at T2 showed limited                  and challenges of undertaking psychosocial
                                                    improvement in their psychological health                research in the torture field. It highlights
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                    and functioning, while the treatment group               the serious problem of attrition and offers
                                                    clients who left the programme at T2                     several lessons for future studies.
                                                    showed improvement on all indicators. While
                                                    attrition of participants was not planned,               References
                                                    losing participants from the treatment group             Amnesty International. (2014). Torture in 2014: 30
                                                                                                                 years of broken promises.
                                                                                                             Baird, E., Williams, A. C. de C., Hearn, L., &
                                                                                                                 Amris, K. (2016). Interventions for treating
                                                                                                                 persistent pain in survivors of torture. Cochrane
                                                    11
                                                         These numbers exclude clients who were ex-              Database of Systematic Reviews, (1). http://doi.
                                                         cluded as they were not tortured according to the       org/10.1002/14651858.CD012051
                                                         UNCAT definition of torture.
53

SCIENTIFIC ARTICLE

Bandeira, M. (2013). Developing an African Torture              evaluation of combined narrative exposure
    Rehabilitation Model: A contextually-informed,              therapy and physiotherapy for comorbid PTSD
    evidence-based psychosocial model for the                   and chronic pain in torture survivors. Torture,
    rehabilitation of victims of torture. Part 1. Setting       27(1), 13–27.
    the foundations of an African Torture Rehabilitation    Dix-Peek, D. (2012a). Profiling torture II: Addressing
    Model through research. Johannesburg: Centre for            torture and its consequences in South Africa: Towards
    the Study of Violence and Reconciliation.                   accessible, effective and holistic rehabilitation
Bandeira, M., Garcia, M., Kekana, B., Naicker,                  services for victims of torture in Southern Africa.
    P., Uwizeye, G., Smith, E., & Laveta, V.                    Monitoring and Evaluation progress report January
    (2013). Developing an African Torture Model: A              to December 2012. Johannes: Unpublished report
    contextually-informed, evidence based psychosocial          by the Centre for the Study of Violence and
    model for the rehabilitation of victims of torture.         Reconciliation.
    Part 2: Detailing an African Torture Rehabilitation     Dix-Peek, D. (2012b). Profiling torture II: Addressing
    Model through engagement with the clinical team.            torture and its consequences in South Africa.
    Johannesburg: Centre for the Study of Violence              Monitoring and Evaluation Report: 2009-2011.
    and Reconciliation.                                         Johannesburg: Unpublished report by the Centre
Bandeira, M., Higson-Smith, C., Bantjes, M., &                  for the Study of Violence and Reconciliation.
    Polatin, P. (2010). The land of milk and honey:         Drozdek, B. (2015). Challenges in treatment of
    a picture of refugee torture survivors presenting           posttraumatic stress disorder in refugees: towards
    for treatment in a South African trauma centre.             integration of evidence-based treatments with
    Torture, 20(2), 92–103.                                     contextual and culture-sensitive perspectives.
Bantjes, M., Langa, M., & Jensen, S. (2012). Finding            European Journal of Psychotraumatology, 6(24750).
    our way: Developing a community model for                   http://doi.org/10.3402/ejpt.v6.24759
    addressing torture. Johannesburg: Unpublished           Halvorsen, J., & Stenmark, H. (2010). Narrative
    report by the Centre for the Study of Violence              Exposure Therapy for posttraumatic stress
    and Reconciliation.                                         disorder in tortured refugees: A preliminary
Başoğlu, M. (2009). A multivariate contextual                   uncontrolled trial. Scandinavian Journal of
    analysis of torture and cruel, inhuman, and                 Psychology, 495–502.
    degrading treatments: Implications for an               Hansen, A. K. V., Hansen-Nord, N. S., Smier, I.,
    evidence-based definition of torture. American              Engelkes-Heby, L., & Modvig, J. (2017). Impact of
    Journal of Orthopsychiatry, 79(2), 135–145. http://         Narrative Exposure Therapy on torture survivors in
    doi.org/10.1037/a0015681                                    the MENA region. Torture, 27(3), 49–63.
Berkson, S. Y., Tor, S., Mollica, R., Lavelle, J., &        Harlacher, U., Nordin, L., & Polatin, P. (2016).
    Cosenza, C. (2014). An innovative model of                  Torture survivors’ symptom load compared to
    culturally tailored health promotion groups for             chronic pain and psychiatric in-patients. Torture,
    Cambodian survivors of torture. Torture Journal,            26(2), 74–84.
    24(1), 1–16. http://doi.org/2012-14 [pii]               Harvard Program in Refugee Trauma. (1992).
Bogic, M., Njoku, A., & Priebe, S. (2015). Long-                Measuring trauma, measuring torture: Healing the
    term mental health of war-refugees: a systematic            wounds of mass violence. Harvard.
    literature review. BMC International Health             Higson-Smith, C. (2013). Counseling torture
    and Human Rights, 15(29), 2–41. http://doi.                 survivors in contexts of ongoing threat:
    org/10.1186/s12914-015-0064-9                               Narratives from sub-Saharan Africa. Peace and
Buhmann, C., Mortensen, E. L., Nordentoft, M.,                  Conflict: Journal of Peace Psychology, 19(2), 164–
                                                                                                                        T O RTU R E Vol u m e 2 8 , N um be r 1, 2 01 8

    Ryberg, J., & Ekstrøm, M. (2015). Follow-up                 179. http://doi.org/10.1037/a0032531
    study of the treatment outcomes at a psychiatric        Higson-Smith, C., & Bro, F. (2010). Tortured
    trauma clinic for refugees. Torture, 25(1), 1–16.           exiles on the streets: a research agenda
    Retrieved from http://www.ncbi.nlm.nih.gov/                 and methodological challenge. Intervention,
    pubmed/26021344                                             8(1), 14–28. http://doi.org/10.1097/
de Jong Gierveld, J., & Van Tilburg, T. (2006). A               WTF.0b013e3283387948
    6-Item Scale for Overall, Emotional, and Social         Higson-Smith, C., Mulder, B., & Masitha, S. (2007).
    Loneliness: Confirmatory Tests on Survey Data.              Human dignity has no nationality: A situational
    Research on Aging, 28(5), 582–598. http://doi.              analysis of the health needs of exiled torture
    org/10.1177/0164027506289723                                survivors living in Johannesburg, South Africa.
Dibaj, I., Overaas Halvorsen, J., Edward Ottesen                Johannesburg: Centre for the Study of Violence
    Kennair, L., & Inge Stenmark, H. (2017). An                 and Reconciliation.
54

                                                                                                                                    SCIENTIFIC ARTICLE

                                                    Jaranson, J. M., & Quiroga, J. (2011). Evaluating                 Syst Rev, (11). http://doi.org/10.1002/14651858.
                                                        the services of torture rehabilitation                        CD009317.pub2.Copyright
                                                        programmes: History and recommendations.                  Patel, N., Kellezi, B., & Williams, A. C. de C.
                                                        Torture, 21(2), 98–140.                                       (2014). Psychological, social and welfare
                                                    Jørgensen, S. F., Auning-Hansen, M. A., & Elklit, A.              interventions for psychological health and
                                                        (2017). Can disability predict treatment outcome              well-being of torture survivors. The Cochrane
                                                        among traumatized refugees? Torture, 27, 12–26.               Database of Systematic Reviews, 11(11). http://doi.
                                                    Kaminer, D., & Eagle, G. (2010). Traumatic stress                 org/10.1002/14651858.CD009317.pub2
                                                        in South Africa (1st ed.). Johannesburg: Wits             Pérez-sales, P., Witcombe, N., & Otero Oyague, D.
                                                        University Press.                                             (2017). Rehabilitation of torture survivors and
                                                    Kline, R. B. (1998). Principles and practices of structural       prevention of torture: Priorities for research
                                                        equation modeling (3rd Ed). New York: The                     through a modified Delphi Study. Torture,
                                                        Guilford Press.                                               27(3), 3–48.
                                                    Langa, M. (2013). Exploring experiences of torture            Phaneth, S., Panha, P., Sopheap, T., Harlacher, U.,
                                                        and CIDT that occurred in South Africa                        & Polatin, P. (2014). Education as treatment for
                                                        amongst Non- Nationals living in Johannesburg.                chronic pain in survivors of torture and other
                                                        Johannesburg: Centre for the Study of Violence                violent events in Cambodia: Experiences with
                                                        and Reconciliation.                                           implementation of a group-based “pain school”
                                                    McColl, H., Higson-Smith, C., Gjerding, S., Omar,                 and evaluation of its effect in a pilot study.
                                                        M. H., Rahman, B. A., Hamed, M., El Dawla,                    Journal of Applied Biobehavioral Research, 19(1),
                                                        AS., Fredericks, M., Paulsen, N., Shabalala,                  53–69. http://doi.org/10.1111/jabr.12015
                                                        G., Low-Shang, C., Valadez Perez, F., Colin,              Quiroga, J. (2017). Comment 1: Torture as a chronic
                                                        LS., Hernandez, AD., Lavaire, E., Zuñiga,                     disease. Torture, 27(3), 38–40.
                                                        APA., Calidonio, L., Martinez, CL., & Jamei,              Quiroga, J., & Jaranson, J. (2005). Politically-
                                                        YJ. Awad, Z. (2010). Rehabilitation of torture                motivated torture and its survivors. Torture,
                                                        survivors in five countries: common themes and                16(2–3), 1–112.
                                                        challenges. International Journal of Mental Health        Republic of South Africa. (1996). The Consitution of
                                                        Systems, 4, 16. http://doi.org/10.1186/1752-                  the Republic of South Africa. Pretoria.
                                                        4458-4-16                                                 Republic of South Africa. (1998). South African
                                                    Miller, K. E., & Rasmussen, A. (2010). War                        refugee Act. Cape Town.
                                                        exposure, daily stressors, and mental health              Stammel, N., Knaevelsrud, C., Schock, K., Walther, L.
                                                        in conflict and post-conflict settings: Bridging              C. S., Wenk-Ansohn, M., & Böttche, M. (2017).
                                                        the divide between trauma-focused and                         Multidisciplinary treatment for traumatized
                                                        psychosocial frameworks. Social Science and                   refugees in a naturalistic setting: symptom
                                                        Medicine, 70(1), 7–16. http://doi.org/10.1016/j.              courses and predictors. European Journal of
                                                        socscimed.2009.09.029                                         Psychotraumatology, 8(sup2). http://doi.org/http://
                                                    Mohamed, S., Dix-Peek, D., & Kater, A. (2016).                    dx.doi.org/10.1080/20008198.2017.1377552
                                                        The psychological survival of a refugee in South          UNHCR. (2015). Mid-year trends 2015. New York:
                                                        Africa — the impact of war and the on-going                   United Nations.
                                                        challenge to survive as a refugee in South Africa         United Nations General Assembly. (1985). United
                                                        on mental health and resilience. International                Nations Convention against Torture and other
                                                        Journal of Emergency Mental Health, 18(4), 1–11.              Cruel, Inhuman and Degrading Treatment.
                                                    Montgomery, E., & Patel, N. (2011). Torture                       March 16, 2016.
T O RTU R E Vo lu m e 2 8 , Nu m be r 1 , 2 0 1 8

                                                        rehabilitation: Reflections on treatment outcome          Weiss, WM., Murray, LK., Zangana, GA.,
                                                        studies. Torture, 21(2), 141–145.                             Mahmooth, Z., Kaysen, D., Dorsey, S., Lindgren,
                                                    Neuner, F., Kurrek, S., Ruf, M., Odenwald, M.,                    K., Gross, A., McIvor Murray, S., Bass, JK.,
                                                        Elbert, T., & Schauer, M. (2010). Can asylum-                 & Bolton, P. (2015). Community-based mental
                                                        seekers with post-traumatic stress disorder be                health treatments for survivors of torture and
                                                        successfully treated? A randomised controlled pilot           militant attacks in Southern Iraq: a randomised
                                                        study. Cognitive Behaviour Therapy, 39(3), 81–91.             control trial. BMC Psychiatry, 249–265.
                                                    Patel, N., Kellezi, B., & Williams, A. (2014).                WHO. (2001). International Classification of
                                                        Psychological, social and welfare interventions               functioning, disability and health. Geneva.
                                                        for psychological health and well-being of torture        Zigmond, A. S., & Snaith, P. (1983). Hospital
                                                        survivors SO-: Cochrane Database of Systematic                Anxiety and Depression Scale. Acta Psychiatrica
                                                        Reviews YR-: 2014 NO-: 11. Cochrane Database                  Scandinavica, 67, 361–370.
You can also read