Nirakar Man Shrestha, Bhogendra Sharma, Mark Van Ommeren, Shyam Regmi, Ramesh Makaju, Ivan Komproe, Ganesh B. Shrestha, Joop T. V. M. de Jong. Impact of Torture on Refugees Displaced Within the Developing WorldSymptomatology Among Bhutanese Refugees in Nepal. JAMA. 1998;280(5):443–448. doi:10.1001/jama.280.5.443
From the Center for Victims of Torture, Kathmandu, Nepal (Drs N. Shrestha, Sharma, Regmi, Makaju, and G. Shrestha and Mr Van Ommeren); and the Free University (Drs Sharma, Komproe, and de Jong and Mr Van Ommeren) and the Transcultural Psychosocial Organization (Mr Van Ommeren and Drs Komproe and de Jong), Amsterdam, the Netherlands.
Context.— Most of the world's refugees are displaced within the developing world.
The impact of torture on such refugees is unknown.
Objective.— To examine the impact of torture on Bhutanese refugees in Nepal.
Design.— Case-control survey. Interviews were conducted by local physicians and
included demographics, questions related to the torture experienced, a checklist
of 40 medical complaints, and measures of posttraumatic stress disorder (PTSD),
anxiety, and depression.
Setting.— Bhutanese refugee community in the United Nations refugee camps in the
Terai in eastern Nepal.
Participants.— A random sample of 526 tortured refugees and a control group of 526
nontortured refugees matched for age and sex.
Main Outcome Measures.— The Diagnostic and Statistical Manual of Mental Disorders,
Revised Third Edition (DSM-III-R) criteria
for PTSD and the Hopkins Symptom Checklist-25 (HSCL-25) for depression and
Results.— The 2 groups were similar on most demographic variables. The tortured
refugees, as a group, suffered more on 15 of 17 DSM-III-R PTSD symptoms (P<.005) and had higher
HSCL-25 anxiety and depression scores (P<.001)
than nontortured refugees. Logistic regression analysis showed that history
of torture predicted PTSD symptoms (odds ratio [OR], 4.6; 95% confidence interval
[CI], 2.7-8.0), depression symptoms (OR, 1.9; 95% CI, 1.4-2.6), and anxiety
symptoms (OR, 1.5; 95% CI, 1.1-1.9). Torture survivors who were Buddhist were
less likely to be depressed (OR, 0.5; 95% CI, 0.3-0.9) or anxious (OR, 0.7;
95% CI, 0.4-1.0). Those who were male were less likely to experience anxiety
(OR, 0.66; 95% CI, 0.44-1.00). Tortured refugees also presented more musculoskeletal
system– and respiratory system–related complaints (P<.001 for both).
Conclusion.— Torture plays a significant role in the development of PTSD, depression,
and anxiety symptoms among refugees from Bhutan living in the developing world.
SINCE 1990 a significant proportion of the southern Bhutanese population
has sought refuge in Nepal and India.1 These
refugees, called Lhotsampas, are ethnically Nepali, the majority group in
southern Bhutan.2,3 Lhotsampa
refugees were told to leave Bhutan by government officials and left fearing
harassment and torture by the country's security forces. The Bhutanese Drukpa
government appears to have sought to reduce the number of Lhotsampas in Bhutan
and to weaken a prodemocracy movement.1 The
total number of refugees is estimated to be 103000, which is about one sixth
of Bhutan's official population.1 By the end
of 1994, 85078 of these refugees were living in United Nations refugee camps
in Nepal. Like the majority of refugees in the world, these refugees are displaced
within the developing world.4,5
Most of the research on refugee mental health has taken place in the
West.6 Such research has pointed to high levels
of traumatic experiences and psychiatric sequelae among help-seeking refugees
presenting at clinics7- 11
as well as among refugees sampled from the community.12- 18
Symptoms of posttraumatic stress disorder (PTSD), depression, and anxiety,
as well as multiple somatic complaints, are common.7- 16,18- 21
Many refugees report exposure to torture,9,10,14,19,21- 27
which is likely an independent risk factor for symptomatology.9,19,22,24- 26,28- 36
Studies of torture survivors, however, usually involve small, selected samples.
Moreover, most of these studies have been inadequately controlled because
of difficulties in soliciting nontortured respondents with similar life history,
age, sex, social status, asylum status, and ethnicity, as well as health and
functioning before exposure to torture.37- 39
In the developing world refugees often face an uncertain future with
respect to food, shelter, and physical security. In contrast, refugees living
in the West are more likely to face problems related to asylum status32,40 and acculturation.41,42
As research on refugees settled in the West may not generalize to the majority
of the world's refugees, calls have been made to conduct more research in
the developing world.4,6,43- 46
The few published epidemiological surveys on refugees and ex-refugees
living in the developing world indicate that emotional sequelae are common.21,47 Yet the extent of the impact of torture
on refugees' mental health is unknown. Considering the increasing number of
specialized services for torture survivors around the world,48,49
there is a need to assess the impact of torture on refugees displaced within
the developing world. We sought to do this through a case-control study using
a large, random community sample of tortured Bhutanese refugees. The findings
should assist in determining the need for care of torture survivors residing
in refugee camps in the developing world.
The sample of the study was taken from the Lhotsampa refugee community
living in the United Nations refugee camps in the Terai in eastern Nepal (Figure 1). The Center for the Victims of
Torture, Kathmandu, Nepal, a Nepali nongovernmental organization, had been
helping tortured Bhutanese refugees since 1991.50
By the end of 1994, in cooperation with political parties, human rights organizations,
collaborating agencies, and ex-patients, the center had identified and registered
2331 survivors of physical torture living in the camps. Because the identification
process included a hut-to-hut survey, it is likely that virtually all physical
torture survivors in the camps had been identified. However, those refugees
who were raped by members of security forces may not have come forward because
of the stigma involved with this form of violence.
A sample size of 600 was determined to provide sufficient statistical
power for analyses. Thus, in 1995, 600 (26%) of 2331 identified tortured refugees
were drawn using simple random sampling. Of those 600 refugees, 585 (98%)
were available for interviewing. Interviewers visited the huts of the other
2% of sampled refugees twice but did not find them at home. To increase the
total sample to 600, an additional 20 refugees were sampled. Of those 20,
16 (80%) were available, creating a total sample of 601 (97%) of 620 approached
A matched control group of Bhutanese refugees was created by interviewing
1 neighbor of each of the 601 tortured refugees. Because it was known that
many elderly men had been tortured, controls were chosen on the basis of matching
sex and age to account for sex-related and age-related responses to torture.
A difference in age of 10 years or fewer was accepted as an age match. Although
many elderly respondents did not know their dates of birth, they knew their
age in relation to historical events in Bhutan. The interviewers, familiar
with the dates of these events, determined respondents' ages accordingly.
Of 601 pairs, 526 matched pairs (88%) remained after excluding those younger
than 19 years and incorrectly matched pairs.
Procedure and Instrumentation. The interview schedule included questions on demographics, charges for
torture, the torturer, place of torture, duration of torture, duration in
custody as well as in prison, and a checklist of 52 types of torture believed
to occur in Bhutan. The interview schedule also contained a checklist of 40
medical complaints, covering cognitive, vegetative, psychotic, central nervous,
musculoskeletal, gastrointestinal, genitourinary, cardiovascular, and respiratory
system symptoms, and questions on diminished vision and hearing. This checklist
was followed by questions on the history of respondents' and their family
members' mental and physical health. The interview schedule concluded with
questions covering PTSD criteria from the Diagnostic and
Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R) and the Hopkins Symptom Checklist-25 (HSCL-25),51 translated into Nepali. A PTSD diagnosis was determined
by fitting the positive answers to the symptom questions into DSM-III-R criteria.
Respondents were asked about traumatic events that may have occurred
either during their flight from Bhutan or during the preflight government
repression, as a reference point in administering the PTSD criteria. Examples
of such traumatic events were provided, including torture, murder of relatives
or friends, loss of property, destruction of one's village, and lack of food
In our study an average score of 1.75 or more on the HSCL-25 anxiety
or depression subscales was used to identify people with high depression and
anxiety scores. A cutoff score of 1.75 has been used to identify depression
and anxiety in several populations, including multiethnic samples of torture
survivors and refugees living in the industrialized world.8,9,27,40,52,53
The 1.75 cutoff score has also been applied in a study of Cambodian refugees
residing in United Nations border camps.21
However, because this score was not validated for Bhutanese refugees, depression
and anxiety diagnoses were not assigned. Because of illiteracy in our sample,
the HSCL-25 was administered by the interviewer. The Cronbach α internal
consistency coefficient in this sample was .87 for the HSCL-25 anxiety subscale,
.87 for the HSCL-25 depression subscale, and .88 for the DSM-III-R PTSD items. A subsequent study will examine the diagnostic
validity of the Nepali translation of the DSM-III-R
criteria and the HSCL-25 through the administration of the anxiety, depression,
and PTSD sections of the Composite International Diagnostic Interview.54
Participants were interviewed by local physicians in the refugees' own
huts. Because of illiteracy and mistrust toward written contracts, verbal
consent, rather than written consent, was obtained from each refugee after
a description of the study had been provided. The interviewers were all men,
and they were not blinded with regard to torture status. Because the questions
covering the DSM-III-R criteria do not lead to simple
yes or no responses, respondents were allowed enough time to think before
answering each question and encouraged to talk freely about their responses.
Impressions thus drawn by the interviewers were marked as responses. The interviewers
had received a 1-day training session by a psychiatrist in differential diagnosis
of PTSD, who observed 2 interviews conducted by each interviewer during pretesting.
Statistics. Because the study involved comparisons between 2 matched groups, the
McNemar χ2 test and paired t tests
were used to analyze the discrete and continuous variables. If fewer than
25 matched cases had different values, the binomial distribution was used
to compute significance levels.55 All comparisons
were 2-tailed. Considering large sample sizes as well as a high number of
comparisons, statistical significance for these comparisons was set at .01,
reducing the Meehl effect56 and the likelihood
of chance significance.
Hierarchical logistic regression analyses were performed on the total
sample of 1052 tortured and nontortured refugees to evaluate the impact of
torture and other group differences on the outcome variables. Sequential logistic
regression analyses were performed on the sample of 526 tortured refugees
to investigate the predictors of psychological status within the tortured
group. Statistical significance for the regression analyses was set at .05.
All analyses were performed with SPSS, Version 6.13 (SPSS Inc, Chicago, Ill).55
Differences between the tortured and nontortured refugees with respect
to their year of arrival in Nepal and their age were not significant. Refugees
who reported experiencing torture had been in the refugee camps for an average
of 3.4 years (SD, 0.7) and were an average of 41 years old (SD, 13; range,
21-87 years). Similarly, on average, nontortured refugees had been in the
camps for an average of 3.4 years (SD, 0.6) and were an average of 41 years
old (SD, 13; range, 21-83 years). No significant differences for average age
verifies that the 2 groups were matched properly.
Table 1 displays a comparison
of other demographic variables. Because the 2 groups were matched on sex,
they consisted of an equal number of men (n=404) and women (n=122). The groups
also appeared similar on most other variables. However, a significantly higher
proportion of refugees in the torture group had been a member of a political
or human rights organization in Bhutan (14% vs 8%). Also, a significantly
higher proportion of the nontortured refugees was illiterate (55% vs 47%)
and Buddhist (14% vs 9%). In addition, the nontortured refugees more often
had a history of significant physical illness (18% vs 12%). Analyses of the
impact of these group differences will be provided herein.
The Torture. The interviews took place, on average, 4 years after the start of the
torture (SD, 0.75; range, 0-8 years). The correlation between the year of
the start of the torture in Bhutan and the year of arrival in Nepal was significant
(r =0.35; P<.001).
The majority of tortured refugees (76%) reported that charges against them
were political; 22% reported that the charges were ethnic. The torture took
place in custody or prison (59%), at home (27%), or in another place (14%).
The refugees were generally unable to distinguish between custody and prison.
The majority of tortured refugees reported that their torturers were soldiers
or police (97%), prison guards (1%), or unidentifiable (2%). The refugees
were generally unable to distinguish between police and soldiers. Ninety-three
percent of the refugees reported that the ethnicity of their torturers was
Drukpa, 3% reported that the torturers were Nepali, and 4% did not know the
ethnicity of the torturers. Tortured refugees reported that the duration of
the torture was 1 day in 45%, 2 to 7 days in 32%, 8 to 30 days in 14%, 31
to 365 days in 8%, and 1 to 3 years in 1% of the cases (mean [SD], 21 
days; range, 1-1095 days).
All refugees in the tortured group reported that they had been physically
tortured (this was the criterion to be included in this group). In addition,
90% of these respondents reported experiencing torture techniques that do
not involve the body. The total number of experienced torture techniques was
1 in 4%, 2 to 10 in 46%, 11 to 20 in 43%, and 21 to 42 in 7% of the tortured
refugees (mean [SD], 10 ; range, 1-42). Of 21 refugees who experienced
1 torture technique only, 18 received severe beatings, 1 was made to do force
labor, and 2 were raped.
The most commonly reported torture techniques were severe beatings (97%),
threats (89%), humiliations (80%), verbal sexual humiliations (77%), forced
incongruent acts (66%), social isolation (54%), hygienic deprivation (53%),
being tied down (52%), nutritional deprivation (52%), sleep deprivation (52%),
sensory deprivation (43%), health service deprivation (41%), forced labor
(26%), hair torture (17%), prolonged standing (15%), undressing in front of
others (15%), cold torture (14%), chepuwa (14%), and ear torture (11%). Incongruent
acts are acts that are a violation of one's cultural or religious norms. For
example, being forced to eat beef or pork is spiritual torture for most higher-caste
Hindus. Sensory deprivation involves deprivation of various sensory stimuli
such as light or sound (eg, being blindfolded, hooded, or kept in a dark room).
Hair torture involves being dragged by the hair or having hair pulled out
or burned. Cold torture involves forced exposure to extreme cold, such as
snow. Chepuwa is a Bhutanese torture technique that involves tight clamping
of the thighs or legs with bamboo, sometimes for a number of days. During
chepuwa the torturer may press the 2 sides of the clamps with his legs or
may stand on the 2 sides of the clamps.57
PTSD, Depression, and Anxiety Symptoms. With the exception of sleep disturbances and recurrent intrusive distressing
recollections of the event, the tortured refugees, as a group, suffered significantly
more on each of the DSM-III-R PTSD symptoms (Table 2). A diagnosis of PTSD was significantly
more common in the tortured group than in the nontortured group (14% vs 3%;
McNemar χ21=40.6; P<.001).
In addition, the tortured refugees as a group had significantly higher
cumulative HSCL-25 anxiety scores (17.9 [SD, 6.1] vs 16.4 [SD, 4.3]; paired t  = 4.8; P<.001) and
cumulative HSCL-25 depression scores (22.6 [SD, 7.0] vs 21.3 [SD, 4.9]; paired t =3.7; P<.001). Using
a mean HSCL-25 item score of 1.75 as the cutoff value, significantly more
tortured refugees had high anxiety scores (43% vs 34%; McNemar χ21=8.1; P=.004) and high depression
scores (25% vs 14%; McNemar χ21=19.6; P<.001).
The impact of demographic group differences on PTSD, depression, and
anxiety symptoms was assessed through 3 separate hierarchical logistic regression
analyses. The predictors in these 3 analyses were the variables on which the
2 groups differed. These variables were history of torture, Buddhist religion,
illiteracy, member of political or human rights organization in Bhutan, and
history of physical illness. History of torture was entered in the first step.
The other predictors were entered in the second step. The 3 outcome variables
were DSM-III-R PTSD diagnosis, high HSCL-25 depression
scores, and high HSCL-25 anxiety scores.
Results of the hierarchical logistic regression analyses are displayed
in Table 3. The analyses show
that torture, adjusted for other group differences, predicts symptoms of PTSD,
depression, and anxiety. Moreover, Buddhist religion predicted the absence
of high HSCL-25 depression scores. Because the tortured group included fewer
Buddhists, this group difference partly explains the torture survivors' elevated
depression scores. In addition, illiteracy predicted anxiety scores. Yet,
because illiteracy was more common in the control group, these group differences
do not account for the elevated anxiety scores.
Sequential logistic regression analyses were performed on the sample
of 526 tortured refugees to investigate the predictors of psychological status
within the tortured group. The outcome variables were DSM-III-R PTSD diagnosis, high HSCL-25 depression scores, and high HSCL-25 anxiety
scores. Eighteen separate univariate logistic regressions were performed on
each of the 3 outcome variables. Predictors were time since torture, political
charge for torture, ethnicity of torturer, torture during incarceration in
prison or custody, total number of torture techniques experienced, duration
of torture, survivor of physical sexual torture, age, sex, marital status
(not married), Buddhist, illiteracy, level of education, member of a political
or human rights organization in Bhutan, history of physical illness, history
of mental illness, physical illness in the family, and mental illness in the
family (data not shown). Eight of 54 univariate logistic regressions were
significant: 2 for PTSD, 4 for depression, and 2 for anxiety. Significant
predictors were subsequently entered into a multivariate logistic regression
analysis on each of the 3 outcome variables (Table 4). The total number of torture techniques experienced was
the only significant predictor for PTSD. The total number of torture techniques
experienced was also a risk factor for depression. Female sex was a risk factor
for anxiety. Buddhist religion was a protective factor for both depression
Table 5 displays statistically
significant differences in terms of medical complaints between the tortured
and nontortured groups. The tortured refugees had more medical complaints
than the nontortured refugees. The largest significant differences were observed
in musculoskeletal system– and respiratory system–related complaints.
Forty-one percent of the tortured refugees complained of backache, compared
with 12% of nontortured refugees. Consistent significant differences were
found for 3 vegetative complaints: loss of appetite, sleep disturbance, and
loss of sexual desire. In addition, tortured refugees complained more of diminished
vision and diminished hearing.
The design of this study involves a large, random community sample of
tortured and matched nontortured refugees displaced within the developing
world. Both the tortured and nontortured groups experienced stress due to
displacement, and both groups continue to experience comparable stress as
refugees living in a camp. The 2 groups are highly similar. Differences between
the groups are most likely associated with the self-reported experience of
Yet, as Basoglu37 has discussed, the
design of this type of study does not control for a possible interaction between
torture and refugee trauma. That would require a design involving tortured
refugees, nontortured refugees, tortured nonrefugees, and nontortured nonrefugees.
As the experience of one traumatic event may increase one's vulnerability
to subsequent stressors, it is likely that the greater symptomatology observed
in the tortured group is not only due to torture per se but also due to an
interaction between torture and refugee trauma.
Symptomatology among Bhutanese refugees who were not physically tortured
may be explained as follows. These refugees may have witnessed atrocities
or may have experienced psychological torture, murder of relatives, loss of
property, loss of employment, destruction of their homes, or lack of food
and shelter. Moreover, their future is highly uncertain because they have
not yet been offered asylum status in any country.
The traumatization of the torture survivors in this study appears low
compared with other survivors of violence living in the West.44
Desjaralais et al4 suggest that the long-term
health status of torture survivors can be positively influenced by safety
in a different country, availability of personal social support, availability
of community and mental health resources, and commitment to a political ideology.
Indeed, these 4 factors may have a positive impact on the health of the Bhutanese
refugee torture survivors. The Bhutanese refugee camps are fairly safe. Personal
support is available to many survivors as villages and families have often
been displaced as a whole. Many refugees strongly support political movements
that aim for repatriation. Moreover, community and mental health resources
are available, including a specialized service for torture survivors.50 In addition to these factors, religion provides a
positive way of coping for both Buddhist and Hindu refugees. Many see their
misfortune as a result of bad Karma, that is, the effect of past deeds. They
believe that performing regular rituals will convince offended gods to be
more favorable and to help them return to a safe Bhutan.
Several limitations apply to this study. First, experiences of torture
were self-reported and could not be validated, and interviewers were not blinded
to the torture status of respondents. Second, as medical examinations were
brief and did not involve radiographs, it is unclear whether certain somatic
complaints, such as backaches and chest pains, could possibly have been direct
results of physical torture. If such examinations had been carried out, it
would still have been difficult to establish whether these complaints were
direct physical results of torture or whether these symptoms resulted from
psychogenic processes. Third, without laboratory tests, torture survivors'
affective symptoms could be explained by the possible presence of infectious
diseases caught while in prison. Fourth, although both groups are similar
in terms of age, sex, refugee status, ethnicity, and current stressors, the
groups may be dissimilar in terms of social status and history of persecution
(other than torture), as well as health and functioning before the torture.
Fifth, the checklist of medical complaints as well as the DSM-III-R PTSD symptoms were not ascertained through validated interviews.
Finally, the validity of the PTSD diagnosis has not been validated for the
Bhutanese context. This limitation is true for most research on refugees,
and is important because of likely ethnocultural variations in PTSD.58
Despite these methodological limitations, the significant differences
on all but 2 of the DSM-III-R PTSD symptoms are striking.
In a closely matched case-control study55 of
tortured political activists involving multiple comparisons in Turkey, Basoglu
and colleagues28 found significant elevated
prevalence rates on all but 3 of the DSM-III-R PTSD
symptoms. El Sarraj and colleagues31 found
prevalence rates of 20% or more for all but 1 of the DSM-III-R PTSD criteria among 550 torture survivors in Gaza. Such findings among
diverse cultures indicate that PTSD symptoms may be helpful in describing
at least part of a survivors' reaction to torture.
Although PTSD symptoms may be common among torture survivors, it has
not been established that the DSM-III-R PTSD symptoms
capture the essence of the impact of torture on these Bhutanese survivors.
This study confirms previous research that torture survivors are more likely
to experience somatic complaints as well as depression and anxiety symptomatology.19,24,28 However, the classification
"disorders of extreme stress—not otherwise specified"59
is perhaps more useful in describing the sequelae of torture. This classification
includes many symptoms that frequently have been presented by Bhutanese torture
survivors at the Center for Victims of Torture Nepal. These symptoms include
shame, mistrust, conversion, medically unexplained pain, and conviction of
being permanently damaged. The structured interview for disorders of extreme
stress—not otherwise specified60 has
been included in our subsequent study of Bhutanese torture survivors, currently
Young61 has argued that the PTSD construct
has often been applied incorrectly when Western patients (and clinicians)
mistakenly use trauma as an explanatory model for depression and anxiety symptoms.
However, Young's argument may not apply in the context of South Asia, where
people typically do not see a relationship between trauma and psychological
problems. Rather, patients' explanatory models for distress usually involve
supernatural processes. In our experience, the tortured Bhutanese refugees
almost always explain psychiatric problems as resulting from bad Karma, spirits,
witchcraft, or an offended god. These Bhutanese refugees come from a closed
country and are relatively uneducated. They have not been exposed to the Western
idea that trauma can lead to mental illness. Nevertheless, significantly more
PTSD symptoms were observed among tortured refugees than in a highly similar
control group of nontortured refugees. Thus, we conclude that torture plays
a significant role in the development of PTSD symptoms among Bhutanese refugees.