Mental Health Symptoms Following War and Repression in Eastern Afghanistan

Context Decades of armed conflict, suppression, and displacement resulted in a high prevalence of mental health symptoms throughout Afghanistan. Its Eastern province of Nangarhar is part of the region that originated the Taliban movement. This may have had a distinct impact on the living circumstances and mental health condition of the province's population.

Objectives To determine the rate of exposure to traumatic events; estimate prevalence rates of symptoms of posttraumatic stress disorder (PTSD), depression, and anxiety; identify resources used for emotional support and risk factors for mental health symptoms; and assess the present coverage of basic needs in Nangarhar province, Afghanistan.

Design, Setting, and Participants A cross-sectional multicluster sample survey of 1011 respondents aged 15 years or older, conducted in Nangarhar province during January and March 2003; 362 households were represented with a mean of 2.8 respondents per household (72% participation rate).

Main Outcome Measures Posttraumatic stress disorder symptoms and traumatic events using the Harvard Trauma Questionnaire; depression and general anxiety symptoms using the Hopkins Symptom Checklist; and resources for emotional support through a locally informed questionnaire.

Results During the past 10 years, 432 respondents (43.7%) experienced between 8 and 10 traumatic events; 141 respondents (14.1%) experienced 11 or more. High rates of symptoms of depression were reported by 391 respondents (38.5%); anxiety, 524 (51.8%); and PTSD, 207 (20.4%). Symptoms were more prevalent in women than in men (depression: odds ratio [OR], 7.3 [95% confidence interval {CI}, 5.4-9.8]; anxiety: OR, 12.8 [95% CI, 9.0-18.1]; PTSD: OR, 5.8 [95% CI, 3.8-8.9]). Higher rates of symptoms were associated with higher numbers of traumas experienced. The main resources for emotional support were religion and family. Medical care was reported to be insufficient by 228 respondents (22.6%).

Conclusions In this survey of inhabitants of Nangarhar province, Afghanistan, prevalence rates of having experienced multiple traumatic events and having symptoms of anxiety, depression, and PTSD were high. These findings suggest that mental health symptoms in this region should be addressed at the population and primary health care level.

Nangarhar province is part of the Pashtun belt that covers southern and eastern Afghanistan and Pakistan's North West Frontier province. The Taliban movement is rooted in Pashtun tribal culture and in the ideology of the radical Deobandi-sect of Sunni islam, blending both into a rigid social and religious system with strict seclusion of women from public life and harsh punishment of any violation of social rules.¹ The Taliban took the Pashtunwali (the Pashtun code of conduct) far beyond the tribal norm and was uncompromising in its aim to return society to the "purity" of an idealized seventh century.²

In the 1980s, the Nangarhar province was the scene of heavy fighting between the former Soviet Union army and the mujahideen forces. The cave complexes of Tora Bora, situated in Nangarhar's district Pachir wa Agam, used to be a center of mujahideen forces, where prisoners were interrogated and many were killed. Later it was used by Al Qaeda. The bombing raids launched by the United States on Afghanistan from October to December 2001 had a large impact on the region and triggered an exodus from Jalalabad city to neighboring districts and Pakistan.

The fall of the Taliban regime ended the extreme conservatism, but did not lead to an overall liberalization in Nangarhar province. Individuals doubt the stability of the new government, and fear that current liberal behavior could be punished in the future. Other potential stressors in the actual situation are unemployment, general poverty, and an ongoing lack of security in the region.

Given the country's past and present sociopolitical and economic situation and its recent history of violence and persecution, the prevalence of mental health disorders is expected to be high. In a survey conducted during the Taliban regime in 1998 among a community sample of women living in Kabul or in refugee camps in Pakistan, 97% reported symptoms of major depression and 86% reported significant anxiety symptoms.³ Of 310 children and adolescents aged 8 to 18 years interviewed during a community survey in Kabul, 80% said they were sad, frightened, and unable to cope with life; 40% had lost a parent; and 67% had seen dead bodies or part of bodies on the street.⁴ A qualitative study in the Herat province reported a general increase in psychosomatic problems, anxiety, depression, and domestic violence.⁵ In a study using the General Health Questionnaire among a community sample of Afghan refugees in southern Iran, 34.5% of respondents reported mental health symptoms.⁶

We conducted a survey among the general population of Nangarhar province to determine the rate of exposure to traumatic events; estimate prevalence rates of symptoms of posttraumatic stress disorder (PTSD), depression, and anxiety; identify resources used for emotional support and risk factors for mental health symptoms; and assess the present coverage of basic needs.

From January 27 to March 18, 2003, we conducted a 2-stage, 40-cluster sample survey. The study population included all individuals aged 15 years or older, who were residing within the recognized borders of Nangarhar province (Figure 1). Because no accurate list of villages and their population sizes existed, a new list was assembled from district information obtained through the United Nations Children's Fund (UNICEF) Expanded Programme on Immunization. Our estimation of total population size was based on the number of children aged 5 years or younger, vaccinated in the Oral Polio Vaccination Program, and assuming that these children formed 20% of the population. District coordinators for UNICEF were asked to list all villages and their population size. If population figures were unavailable for specific villages, we asked for an indication of the relative size (large, medium, or small) of the settlement. Water and sanitation records of the Danish Committee for Aid to Afghan Refugees were used to complete district lists. Our final list consisted of 1606 villages and settlements. This list included UNICEF's division of the city of Jalalabad into 4 segments. Using the primary sampling frame, we estimated the total population of Nangarhar province to be slightly more than 1.6 million individuals, which corresponds with UNICEF's estimations.⁷

To determine the sample size for our study, we assumed a prevalence rate of 50% of mental health–related problems. We estimated that a minimum of 770 participants would be required for a 95% confidence interval (CI) to detect a prevalence rate between 45% and 55%. The required minimum was increased to 1100 because we anticipated nonresponse to be 30%. Based on available information on household size and age distribution, we further assumed an average of 4 adults per household. Therefore, a minimum of 275 households would need to be included. However, we planned to include a larger sample: 400 households, a trade-off between the desired numbers of clusters and of households. With probability proportional to population size, we selected 40 clusters in the first sampling stage: 33 in rural areas and 7 in the city of Jalalabad (Figure 1 and Figure 2).

In the second stage of sampling, 10 households were selected within each cluster. Identification of cluster samples differed for urban areas and rural villages. No maps of the selected villages were available. In small settlements, we first asked the village leader to list all families and then selected 10 households using a random number table. In larger communities, we asked a village leader to list all mosques, and then selected 1 mosque using a random number table; next, we asked the mullah to list all families and we randomly selected 10. Maps were available for the city of Jalalabad and Nangarhar province. By blindly throwing a pen onto a map, a spot was selected as a starting point for the survey. The first house on the left was selected for the first interviews. The next house was selected to be the closest house to the left when leaving the house just surveyed. This procedure was repeated until 10 households within the cluster had been surveyed. All members of the selected households aged 15 years or older were requested to participate.

We selected 9 male and 6 female interviewers who were fluent in the Pashtu language and were able to read and write. They were trained over a 5-day period, which included a field test. Supervision occurred on a day-to-day basis throughout the survey. To ensure privacy, we encouraged interviewers and participants to complete the questionnaires in private places. Participants were paired up with same-sex interviewers. If household members were not at home, interviewers and household members agreed on a day to complete the interview. If potential participants were still absent or unwilling to respond at the second visit, background information and reasons of nonresponse were noted.

Formal review and approval of this survey has been given by the medical ethical committee of the University of Amsterdam, Amsterdam, the Netherlands. Because of the high illiteracy of the Afghan population (UN 1999 estimation: 64%),⁸ informed consent was obtained from each respondent by reading aloud an explanatory text and then asking for participation.

All instruments in this study were designed as self-reported questionnaires. Due to the high illiteracy rate, we used the questionnaires as a structured interview in which questions were read aloud to each participant. We collected demographic information on sex, age, marital status, education level, religion, and ethnicity. Ethnicity was defined by respondents who chose from a preselected list. All questionnaires were translated into Pashtu with the help of a bilingual mental health expert and backtranslated by another who was blinded to the first translation.

To assess mental health symptoms, we used the Harvard Trauma Questionnaire (HTQ) and the Hopkins Symptom Checklist (HSCL-25). In addition, we asked questions about physical, social, and mental well-being. We chose these instruments to obtain information on common, nonspecific psychiatric problems and to gather information on symptoms of specific disorders such as PTSD, anxiety, and depression, and related life events.

The HTQ combines the measurement of PTSD symptoms over the past 4 weeks and traumatic events experienced over the past 10 years.⁹ Trauma event questions were adapted for specific events among the Afghan population in a similar way as was done in a national mental health survey in Afghanistan in 2002.¹⁰ Because rape appeared to be a delicate issue to address, the interviewers often asked participants if they had ever "heard of" or "knew" someone who had been raped—this being the only wording sufficiently acceptable to elicit a response. Consequently, this question about "witnessing" rape may also have covered "experiencing" it. The PTSD items are derived from the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV).¹¹ We determined if an individual met symptom criteria for the occurrence of PTSD according to a scoring algorithm proposed by the Harvard Refugee Trauma Group on the basis of DSM-IV diagnostic criteria.¹² This definition of PTSD requires a score of 3 or 4 on at least 1 of 4 reoccurring symptoms, at least 3 of 7 avoidance and numbing symptoms, and at least 2 of 5 arousal symptoms.

The HSCL-25 is a widely used screening instrument measuring symptoms of anxiety and depression among individuals during the past 30 days.¹³ Symptoms are scored on a 4-point Likert scale. The HSCL-25 comprises 2 subscales for anxiety and depression (score range, 1-4). It has been consistently shown in several populations that the total score is correlated with severe emotional distress of unspecified diagnosis, and the depression score is correlated with major depression as defined by the DSM-IV.^14,15 Both instruments have been validated in various countries and cultures,^14,16 although at the time not yet in Afghanistan, and were previously used in postconflict settings.¹⁷

To assess resources used for emotional support, we used the following procedure based on the outcomes of focus group interviews and field tests: we asked respondents to think about a situation or event that once made them sad, worried, or tense. We made it clear that they did not need to reveal that situation. We then asked with whom they had talked for emotional support in that specific situation. Respondents could choose from 11 listed options (which also included places): direct family, family in law, friends, neighbors, mullah, shire, or holy place, Allah, village health volunteer/traditional birth attendent, physician, herbalist, or other. To assess the present sufficiency of basic needs, we asked respondents if shelter, food, drinking water, and medical care were sufficiently, reasonably, or not sufficiently available.

Statistical analyses were performed using SUDAAN statistical software (Research Triangle Institute, Research Triangle Park, NC), which accounts for complex sampling designs. All presented data were adjusted for clustering and assigned a population-based weighting factor, based on the population size of each cluster in our final listing of all 1606 villages and settlements. Data were not weighted for nonresponse. Multivariate linear regression models were used to assess the effects of demographic variables and exposure variables on continuous variables (anxiety and depression). To analyze dichotomous outcomes, such as PTSD symptoms, we used multivariate logistic regression models. The results obtained from the regression models were based on partial, not sequential analyses. All P values were derived from adjusted Wald F tests based on these regression models, except for those derived from the analysis of resources used for emotional support, which were based on adjusted Wald χ² tests. When a characteristic had a natural ordering (eg, age, number of traumatic events), a test for linear trend was performed. For the analysis of the effect of exposure variables on mental health outcomes, P values were based on the comparison between those having experienced the event and those who had not. P<.05 was considered statistically significant. Bonferroni corrections for multiple comparisons were applied when comparing traumatic events (P<.002), the number of traumatic events (P<.01), and resources for emotional support (P<.005).

A total of 351 households were surveyed. From these, 1013 individuals aged 15 years or older were interviewed. In the households surveyed, there were 382 (27%) nonresponders, mostly because of practical reasons such as the respondent being absent; 6 persons refused to be interviewed. Due to the loss of stratification data of 2 respondents, the data of 1011 respondents could eventually be used for analysis (representing a 72% participation rate). Three clusters that could not be visited for security reasons were not replaced by newly selected clusters because at the time the imminent attack on Iraq called for evacuation preparedness.

Demographic sample characteristics are summarized in Table 1. Nearly all respondents belong to the Pashtun ethnic group (their national proportion is 44%)⁸ and most live in rural areas. Fifty-five percent were women. Most were married. Eighty-eight percent of female participants and 44% of male participants had not received any education. A majority of men (87%) reported having jobs, most of them being a farmer or stockbreeder; practically all women were housewives. Ninety-six percent of the men and 69% of the women reported being in good physical health. Twenty-two percent of female respondents and 16% of male respondents indicated they had ever been told by a physician, (mental) health professional, or healer that they had a mental illness.

At the time of the survey, 228 (22.6%) respondents indicated that access to medical care was not sufficient. Food was available for 98.0% of individuals; drinking water, 92.9%; and shelter, 94.0%.

A modification of the list of potential trauma events as recited in the HTQ is shown in Table 2. High percentages of the participants reported having experienced multiple traumatic events over the past 10 years. Fourteen percent reported experiencing 11 or more traumatic events. Seventy-one percent experienced a lack of access to medical care and 69% experienced a lack of food or water. Sixty-seven percent indicated that they have been close to death during the previous 10 years. Sixty-two percent experienced or witnessed the Coalition-led bombardments in 2001 and 61% experienced other shelling or rocket attacks from mujahideen or former Soviet Union forces. Sixty-one percent of all participants had to suddenly flee at some point and 50% had lived in a refugee camp.

Table 3 shows estimated mean scores on the HSCL-25 and the HTQ, along with 95% CIs. For the HSCL-25, the estimated mean total score is 1.79 (1.44 for men and 2.10 for women). Mean scores for the HSCL-25 subscales show high levels of symptoms of depression and anxiety, especially among women. When using a standard cut-off score of 1.75,¹⁶ the depression symptom scale scores yield estimated prevalence rates of 38.5% (16.1% in men and 58.4% in women). On the anxiety symptom scale, estimated prevalence rates were 51.8% (21.9% in men and 78.2% in women). The HTQ yielded an estimated total prevalence of 20.4% for PTSD symtoms (7.5% in men and 31.9% in women).

We performed multivariate analyses of the effect of selected demographic factors to mental health outcomes. Table 4 shows mean scores of the HSCL-25 scales for symptoms of anxiety and depression and estimated prevalence rates of participants who met PTSD symptom criteria in relation to separate demographic variables and adjusted for all other listed demographic variables. For all mental health outcomes, higher symptom scores were associated with being female, experiencing poor physical health, and reporting previous mental illness. Higher scores of depression were accociated with being older and having received less eduction. Education was also associated with high scores of anxiety. Symptoms of PTSD were associated with marital status and ethnicity, that is, being single and belonging to the Tajik ethnic minority group.

We also performed multivariate analyses of the effect of war-related traumatic events to mental health outcomes. Table 5 shows mean scores of the HSCL-25 scales for symptoms of anxiety and depression and ORs (95% CIs) for participants who met PTSD symptom criteria in relation to the number of traumatic events experienced, as well as to separate traumatic events. All demographic variables listed in Table 1 4 were controlled for in the analysis. There was a significant linear increase in all selected mental health outcomes with increasing numbers of traumatic events. All trauma exposure variables were significant at the P<.002 level for anxiety and depression scores, except having been injured by a landmine, separated from the family, rape, missing family, recent bombardments, or being kidnapped. Trauma events that were associated with high PTSD symptom scores were having experienced a lack of food or water, or a lack of shelter, having been tortured, having had to flee suddenly, having loss of property, having been kidnapped, and having been close to death.

Ninety-eight percent (989) of the respondents mention "Allah" as the main resource for emotional support when feeling sad, worried, or tense. The second preferred resource was direct family members (812; 81.0%). Family-in-law was mentioned more by women (348; 34.9% compared with 21.1% [206 men]); married women generally live with the husband's direct family. Males scoring high on symptoms of depression and anxiety (scale score ≥1.75) reported seeking support from village health volunteers or traditional health attendants more often than those men with lower symptom scores (P<.001). Females with high depression symptom scores reported seeking less support from their direct family (P<.001), family-in-law (P = .009), friends (P<.001), and neighbors (P<.001) than did females with lower scores.

This survey, conducted in early 2003 among the population of Nangarhar province, Afghanistan, shows a high prevalence of symptoms of anxiety, depression, and PTSD. Anxiety and depression symptom scores were even higher than usually found in postwar situations,^18,19 but not PTSD symptoms.²⁰ However, studies of community samples of Afghan refugees living in Holland and Iran, respectively, reported similar findings.^6,21 This may be related to the country's tragic recent history. During the past 25 years, individuals in Afghanistan have continuously experienced war and civil unrest. The Soviet occupation was followed by violence subsequently from the mujahideen forces, the Taliban regime, and a Coalition-led military campaign. In addition, a 4-year regional drought forced many Afghans to leave their homes in search of food and water.^22,23

Our study had a number of limitations. First, we did not ask respondents when during the previous 10 years they had experienced traumatic events and what were the period of onset and the course of their symptoms. As a consequence, we cannot draw conclusions about the chronicity of mental health symptoms and their relation to traumas experienced. The existence of a relationship is plausible because there is a linear increase of symptom prevalence rates with growing numbers of traumas experienced. Another limitation to this study is the fact that our main measurement instruments have not yet been validated in Afghanistan. Validity has been proven, however, in various languages and cultures. In addition, these instruments only provide outcomes on symptom levels, not diagnoses.

While women generally show higher levels of mental health symptoms than men do, scores in female participants of this survey were extremely high. Previous studies have provided insight in the mental health consequences of the subordination of women in social life in Afghanistan, particularly under the Taliban regime but also before and after.^3,5,24 The differences in outcomes also may reflect differences in coping patterns as preferred by, or available to, women compared with men.

The overall prevalence rates of mental health symptoms found in this survey are lower than those reported from a national survey conducted in Afghanistan in 2002.¹⁰ Slight differences between methods applied during the national and this survey may have contributed to this. For example, in this survey we attempted to include all adult household members (excluding those who were disabled). The national survey included 1 nondisabled member and 1 disabled member (if any) from each household.

The variance in outcomes between this and the national survey may also be explained by cultural and geographic differences of the participants. First, during the Taliban regime repression and restrictions were much harsher in the country's central and northern part with its non-Pashtun population than in Nangarhar, which is a conservative Pashtun area. Second, there has been less continuous fighting in this province than in other regions. The city of Jalababad suffered heavily in the 1980s during the Russian occupation, but was more or less spared since. Kabul, in contrast, was targeted for years by various factions. The front between Taliban and the Northern Alliance has been shifting alternately to the North and to the South for a long time, bringing violence to the country's central and northern regions. Third, this survey sample contains a larger urban proportion than the national survey. Jalalabad's population may have been protected by a greater social connectedness and infrastructure. This may be more relevant in light of the relatively intact state of the city of Jalalabad compared with Kabul. Finally, the population of Jalalabad may have experienced less insecurity because the city is located close to the Pakistan border, therefore providing an easier fleeing route than from Kabul.

These hypotheses are not entirely in concurrence with the experienced numbers of traumatic events as reported. Although it is suggested that circumstances have generally been less harsh in Nangarhar than elsewhere, most traumatic events were reported more frequently than in the national survey. Some of these, however, such as lack of shelter, food, water, and medical care, and death of family members or friends due to illness or lack of food, may be associated with economic and social decline rather than to repression or war. Others, like shelling or rocket attacks, having had to flee, living in refugee camps, and the Coalition-led bombings, may relate to armed violence that took place outside the period of Taliban regime; belonging to the Pashtun belt then did not provide protection, and many temporarily fled to Pakistan. Events like beatings, interrogations, and harrassments probably have taken place under the Taliban regime, and these are reported more often in the national survey.

In this study, religion and the family were reported as the the main resources for emotional support; women do not or cannot frequently make use of any other resources. There is an indication that women with mental health symptoms withdrew from social resources. Both men and women in distress did not report seeking support from physicians. This may relate to a general access problem; although our findings suggest that the present coverage of basic needs is reasonably sufficient, a quarter of the population report this is not the case for medical care.

Among the population of Nangarhar province, Afghanistan, many have experienced traumatic events during a long history of armed conflict, repression, and insufficiency of needs. Mental health symptoms are highly prevalent, especially in those who experienced multiple traumas and in women. The capacity of primary health care workers to raise awareness of basic options for support or treatment and to address mental health needs should be strengthened.