Women in post-trafficking services in moldova: diagnostic interviews over two time periods to assess returning women's mental health

The study took place in the Republic of Moldova in collaboration with the International Organization for Migration (IOM) in the Center for Assistance and Protection for Victims and Potential Victims of Trafficking in Human Beings. At the time of the study, IOM, as part of their Assistance, Prevention and Protection Programme, provided travel and reintegration assistance to individuals who chose to return or who were unable to remain in the country of destination http://​www.​iom.​md/​index.​php/​en/​programs/​counter-trafficking/​assistance-a-protection-programme. Trafficked women in this study may have initially come into contact with IOM in destination settings or may have been referred to IOM in Moldova by a range of sources in destination countries, including police, immigration services, lawyers, health services and governmental and non-governmental anti-trafficking and support organizations.

As part of their Assistance and Protection Programme (APP), IOM meets returning women who have been referred into their care at their port of entry. Between 2000 and 2008, IOM assisted 2340 women returned from a trafficking experience outside Moldova. IOM's support package for trafficked persons in Moldova is led by specially trained social workers, supported by an experienced psychologist (LG). Assistance generally consists of crisis intervention care, including a medical, psychological, legal and social needs assessment, and residential care of up to 1 month, which can be extended. This is followed by a 12-month community-based rehabilitation program, which often includes social assistance and vocational training. Approximately 80% of returning women accept the acute crisis intervention and/or the rehabilitation program. Mental health interventions include counseling, cognitive-behavioural therapy, antidepressant drug treatment, alcohol detoxification services and treatment for substance abuse and dependence. During rehabilitation, on-going antidepressant treatment relies on women being willing to attend a psychiatric hospital out-patient clinic.

We sampled consecutive women who registered with IOM between December 2007 and October 2008 and had participated in a crisis intervention psychiatric assessment within 5 days of registering. We approached them between February 2008 and December 2008, at least two months after their registration.

Women survivors of trafficking were approached by an IOM social worker two to twelve months after return, either face to face or by phone, generally during their monthly follow-up appointment in the rehabilitation programme. Women were informed of the study aims, subject matter and the voluntary nature of participation was emphasised. Women who gave permission to be approached by the research team and who consented to be interviewed were interviewed at the Rehabilitation Centre or at another place of their choosing. We followed the World Health Organization Ethical and Safety Recommendations for Interviewing Trafficked Women [11]. The process complied with the IOM Data Protection Principles [12]. Ethical approval was obtained from Kings College Research Ethics Committee (CREC/07/08-56) and from N.Testemitanu State Medical and Pharmaceutical University Institutional Review Board.

Women were included if they were aged 18 or over, originally resident in Moldova, had returned to Moldova in the past 2-12 months following a trafficking experience outside of Moldova, registered with IOM in Moldova as a survivor of trafficking and participated in the crisis-intervention assessment (described above) 2-12 months prior. IOM defined all survivors of trafficking according to UN Protocol to Prevent, Suppress and Punish Trafficking in Persons, especially Women and Children [1].

Women were not invited to participate or were excluded if the social worker or research psychiatrist considered them to be too distressed or unwell to take part.

Clinical assessments of women's initial psychiatric conditions were made available to the researchers, for women who consented. These initial psychiatric assessments were conducted within two days of women's arrival at the IOM centre (and within 5 days of registering with IOM as a returning trafficked person) by a senior consultant psychiatrist (MH), based on the International Classification of Diseases (ICD-10) [13]. A semi-structured assessment was carried out using a locally developed checklist, which included items such as, history of feelings or symptoms of emotional distress or mental health or medical problems, current and past psychiatric and medical treatment, history of illicit drug use, nicotine and alcohol use, current medication, family background and family history of illness, personal history (which included information on early development, school/work, relationships, current life circumstances and interests), premorbid personality, a mental state examination.

Socio-demographic variables were available from recorded data, which included marital status before trafficking, employment status prior to trafficking, country trafficked to; age on return from trafficking. IOM provided restricted access to limited data in anonymised aggregated form on women who did not take part in order to enable broad comparisons to be made between participants and non-participants. The process fulfilled to the IOM Data Protection Principles [14].

Participants were interviewed by a Moldovan psychiatrist using the operational criteria of the DSM-IV two to twelve months after having returned to Moldova. This timeframe was selected based on staff's assessment that by this time, women's urgent needs had been met, and they were generally beginning to adjust and resettle and 12 months was the normal maximum period of support provided by the Rehabilitation Centre.

We used the Romanian version of the Non-Patient version of the Structured Clinical Interview for DSM-IV Axis I Disorders [15]. The diagnoses for the rehabilitation period interviews were made by a Moldavian psychiatrist (N.O.), familiar with the operational criteria of the DSM-IV and trained in the SCID (Video course). We assessed current (last month) mental illnesses according to the Diagnostic and Statistical Manual of Mental Disorders, fourth edition [16]. We defined having a psychiatric illness as fulfilling criteria for at least one DSM-IV Axis I psychiatric disorder. We considered translating the Schedules for Clinical Assessment in Neuropsychiatry [17] in order to be able to make ICD diagnoses, but determined that this would have been prohibitively expensive and time consuming. The SCID translated version was pre-tested by two Moldavian psychiatrists trained in standardised interviews (MH, NO) and found to be understandable to the women and of good face validity.

Alcohol use was assessed with the Alcohol Use Disorders Identification Test (AUDIT) [18], a 10-item screening instrument measuring hazardous and harmful alcohol consumption which covers consumption, drinking behaviour and alcohol related problems. Total scores of 8 or more were considered as indicators of hazardous and harmful alcohol use, as well as possible alcohol dependence [19]. Substance abuse/dependence was measured using five questions based on the Diagnostic Interview Schedule about the frequency of drug use, stated dependence, inability to cut down, need for larger amounts, and withdrawal symptoms. A list of drugs used in Moldova was provided, including cannabis, amphetamines, opiates, hallucinogens, ecstasy and solvents. Those who reported the use of any of the listed substance within the previous month were regarded as having a substance abuse. Those who answered positively to one of the questions about inability to cut down need for larger amounts, and withdrawal symptoms were classified as dependent. As the method of first and second data collection approaches differed slightly, the disorders are grouped at a higher level to be comparable - see Analysis.

Data with a convenience sample of 200 trafficked women located across seven European countries and in touch with services suggests that approximately 50% of women recently returned from being trafficked may have a depressive or anxiety disorder [8]. To estimate this prevalence with 95% confidence intervals, +- 7.5%, at 90% power, we needed a sample of 120 women.

All analyses were carried out in Stata version 10 [20]. We described the socio-demographic characteristics of the women in the sample including both pre and post-trafficking information (See Table 1). We then described the distribution of mental health diagnoses recorded during the crisis intervention phase (1-5 days after return) and the re-integration phase (2-12 months after return. We compared diagnoses at the patient level between the two time points by tabulating the diagnoses and carrying out a kappa test of agreement and the Stuart-Maxwell test for marginal homogeneity (an extension of the McNemar test to kxk table). These will test for the level of agreement in diagnosis of the women at the two time points and for symmetry in the cross-tabulation of diagnoses to determine if there is any systematic change in diagnoses or if changes occur evenly. Since the diagnostic tools used in the two time periods were different, in order to compare between them we grouped specific conditions into four broad syndrome clusters; PTSD; anxiety disorders excluding PTSD (panic disorder, generalized anxiety disorder and somatisation disorder); mood disorders (depression and dysthymia); and substance use disorders (substance abuse, substance dependence, alcohol use and alcohol dependence). Those with a diagnosis of adjustment or acute stress disorder during the crisis intervention phase were not included in the statistical test for change in diagnosis because these time-limited diagnoses would be very unlikely to be diagnosed at the later time point and would therefore be inappropriate for comparison.

As shown in Table 1, most women were single before being trafficked and most had completed compulsory lower secondary education (nine years), although 12% had only attended primary school. The mean age for the sample at return was 25 years and nearly one fifth was younger than 21. Forty-seven percent of women had at least a child. Before being trafficked, 68% had been unemployed, however, this decreased to 37% post-trafficking, mostly due to opportunities for study or training organized through the IOM rehabilitation program. For 68%, the trafficking situation lasted for more than six months. Since return, a fifth were not in contact with either of their parents, and less than a third of women reported having someone in whom they thought they could confide.. Most were living rurally after return.

Table 2 shows the total prevalence of all psychiatric illnesses, allowing for the possibility that the total number of diagnoses may be more than the number of women. In the crisis intervention period, the most common diagnoses were anxiety disorders (especially PTSD, diagnosed in 48% of women), mood disorders, adjustment and acute stress disorders and substance use disorders. Two to twelve months later during the rehabilitation period, the most common diagnoses were PTSD, mood disorder or harmful alcohol use.

Table 3 shows the principal single or co-morbid diagnosis for each woman. It shows that the level of co-morbidity is high in the sample at both time periods. During the rehabilitation period, 16% of women had 'pure' PTSD, while 20% had PTSD plus at least one secondary diagnosis. In more detail, (not shown in the table), 7% of women at rehabilitation had PTSD co-morbid with depression or dysthymia, 3% of women had PTSD co-morbid with another anxiety disorder, 5% had PTSD co-morbid with substance misuse (alcohol or drug misuse), and 6% had PTSD co-morbid with both substance misuse and a mood or other anxiety disorder.

As shown in Table 4, of those women having co-morbid PTSD or another anxiety or mood disorder at the early crisis intervention phase, 85% had a psychiatric illness diagnosed later, during rehabilitation. However, for those with pure PTSD or with adjustment or acute stress disorder only 1-5 days after return, only 40% had a diagnosable psychiatric illness at rehabilitation. The kappa test for agreement and Stuart-Maxwell test for symmetry were run on 92 patients who did not have a diagnosis of adjustment or acute distress disorder at the crisis intervention phase. The reported diagnoses gave a kappa level of 0.27 and 41% agreement indicating low agreement between the time points. The test of homogeneity was significant (chi2 = 27.17, df = 5, p < 0.001) indicating that the changes in the diagnoses of the women are not equally distributed but systematically differ between the two time points, for example the large proportion that have moved from PTSD to no-diagnosis.

Study participants were all registered with the International Organisation for Migration (IOM) and recruited via contact with their social worker. This may mean that women with a higher level of need are represented versus if we had attempted to include women not in contact with services, biasing the sample towards overestimating mental distress. However, women not included in the study may have been those who were too distressed to identify and access services or experienced other barriers to service access such as poor education. Given the highly sensitive nature of trafficking, we did not consider it appropriate to approach women who were not contactable through a known service [11]. It is worth noting, however, that very few differences were found between women who participated versus those who did not. We were able to recruit 68% of consecutive women returning through IOM's APP program, supporting our view that our sample is reasonably representative of women participating in IOM's rehabilitation program.

It is unclear to what extent findings from this study may be generalisable to the broader population of trafficked women. Participants had each been referred into care and therefore may differ from the larger population of trafficked women that do not access assistance. It is likely to be only the smallest portion of women who are trafficked who receive post-trafficking support, and there is, to date, no data to compare this sample with what might be considered a 'general population of trafficked persons. For this study, we also do not have information on the differences in mental health support that women received during the rehabilitation period, so we can not account for any interventions that may have influenced women's conditions over two time periods.

A further limitation may be the different instruments used at the two time periods evaluated. The ICD-10 [13] classification was used at when women entered in the crisis phase, in line with normal practice in the centre, and DSM-IV [16] was used at the re-integration phase. Although there are at least minor differences between two classifications in almost every category of disorder, these two widely used classifications are generally found to be functionally equivalent [33] and are largely comparable[34] with a high level of concordance reported for depression, dysthymia, substance dependence and generalised anxiety disorder [33]. For PTSD, concordance between ICD and DSM is reported to range from 35% to 75% [33, 35]. One key difference is that ICD has a potentially lower threshold for experiencing a severely threatening event (such as being in a war zone whereas DSM requires the person to have experienced threat to their life of themselves or others to which they reacted with helplessness and/or horror. The second key difference is that ICD does not require impairment to be present whereas DSM does. The third difference is that DSM requires three symptoms of avoidance whereas ICD requires only one. Overall, the effect o the differences is that using the ICD classification is recognised to lead to an increase in prevalence estimates of PTSD, compared to if DSM criteria are used, with up to a doubling of prevalence of PTSD being reported from ICD versus DSM. Given this, our methodological differences are more likely to explain any apparent recovery by the second time period [33]. As our key finding is the extent of on-going psychiatric illness in the second time period for those with co-morbid PTSD, anxiety or depression at baseline, we can be confident that women with diagnoses at the rehabilitation phase had significant levels of mental distress and impairment, and that any difference between the two methods cannot account for women remaining or becoming more mentally unwell at follow-up. We used DSM at follow-up because a standardised interview for DSM was available in the local languages and also because DSM is well-accepted as portraying the current theoretical construct of PTSD [33, 35].

There is currently limited knowledge about effective treatment for PTSD co-morbid with depression, especially in victims of extreme trauma. Studies are needed to evaluate treatments that show promise, such as narrative exposure therapy [32, 36] and trauma-focussed cognitive behaviour therapy with and without pharmacotherapy, [31] particularly in populations of trafficked persons.