Major depressive disorder following terrorist attacks: A systematic review of prevalence, course and correlates

Figure 1 presents the flow chart for the selection of the included studies. A four-step procedure was used. First, a search of the peer-reviewed literature in the PsycINFO and Medline databases was conducted (without time limit) using the following keywords: depression, terrorist, terrorism, mental health, disaster and trauma. Searches were undertaken between January 12 and 16, 2009. The initial database search identified 567 potentially eligible studies for this review. Second, two independent reviewers analyzed the title and abstracts of all retrieved studies and excluded those which did not meet the selection criteria. The majority of studies excluded in this step were papers that analyzed psychological consequences different from MDD, other kinds of disasters not categorized as terrorist attacks or other populations that were not either direct victims nor general population. Third, full manuscripts were obtained for all publications included after the second step. We examined the complete text of the articles and once again eliminated those which did not comply with the selection criteria. The majority of studies excluded in this step were papers that analyzed the psychological consequences of terrorist attacks without assessing MDD with diagnostic criteria. Fourth, to verify that our final sample was comprehensive and that our search was appropriate, we compared it with previous review papers [1, 3, 6].

Our search identified 11 studies of MDD following terrorist attacks: 6 were carried out with victims, 4 with general population samples, and one with victims and general population. Of these, 9 were cross-sectional studies and 2 were cohort studies. The most relevant information of each reviewed study is summarized in Tables 1 (studies with victims) and 2 (general population studies). In these, information about the terrorist attacks, assessment time, sample size, method and main results (MDD prevalence) is shown.

In our review, most studies examine the impact of terrorist attacks in Madrid (March 11, 2004) and New York (September 11, 2001). Nevertheless, one study assesses the impact of different terrorist attacks occurred in France (between 1982 and 1987), another one assesses the consequences of the Oklahoma City Bombing (1995), and yet another one compares the consequences of the Oklahoma City Bombing with the attack on the US embassy in Nairobi, Kenya (1998).

The measures used to establish MDD prevalence were the Diagnostic Interview Schedule (DIS)/Disaster Supplement, based on the DSM-III-R criteria [24] (used in 2 studies); the Diagnostic Interview Schedule (DIS) based on the DSM-IV criteria, with adjustments for cultural fit [25] (used in 1 study); The Mini International Neuropsychiatric Interview (MINI), based on the DSM-IV criteria [26] (used in 2 studies); and the SCID's major depressive disorder (MDD) interview [27], based on the DSM-IV-TR criteria (used in 5 studies). In one study [28] the researches assessed MDD with 15 items created ad hoc and based on the DSM-III criteria.

The correlates of MDD reported in the reviewed studies were classified as pretraumatic, peri-traumatic, posttraumatic, and sociodemographic factors.

Excluded studies that assess the prevalence of other psychological problems after terrorist attacks have generally been focused on PTSD. The research literature suggests that the burden of PTSD in persons exposed to disasters is significant. Specifically, the prevalence of PTSD among direct victims ranges between 30% and 40%, while the range in the general population is between 5% and 10% [see 3 for a review]. Furthermore, a common result is that the prevalence of PSTD in the aftermath of a natural disaster is often lower than the rates documented after human-made disasters (such as terrorist attacks) [1, 3, 6].

Other studies not included in our review examined the prevalence of MDD after natural disasters or chronic exposure to trauma. Some of the natural disasters evaluated have been the 1999 Turkey earthquakes [38], the 2004 Asian Tsunami [39], the 2004 hurricane in Florida [40] or the 2005 hurricane Katrina [41]. Natural disasters affect broad geographic areas, leading investigators to study populations that often include both direct and indirect victims [3]. Consequently, reports of MDD prevalence rates after natural disasters vary widely. For instance, a study on the Turkey earthquakes showed a higher prevalence of MDD closer to the epicentre (16%) compared to 100 km away (8%) [38]. Another study on the Asian tsunami found a higher prevalence of MDD in displaced people (30%) than in two other samples of non-displaced people (21% and 10%, respectively) [39]. Focused on chronic exposure to trauma, different studies have examined the significant impact of terrorism in the Israeli population since the beginning of the Al Aqsa intifada in September 2000. In a study conducted in April-May 2002, Bleich et al. [42] showed that over half of a national representative sample of Israel reported feeling depressed. In another population-based study carried out between January 2002 and December 2005 [43], 15.4% of participants reported a MDD, with rates of MDD being 2.4 times higher among Arab Israelis than among Jews. This difference between Arabs and Jews has been shown in other studies, and is consistent with both MDD as well as other psychological problems [44, 45], suggesting that the mental health impact of terrorism differs among diverse groups living in Israel. Along the same line, different authors have documented the psychological impact of impending forced settler disengagement in Gaza. Hall et al. [46] assessed a sample of Israeli settlers who, after having been exposed to ongoing terrorism, were forced to leave their homes. The prevalence of MDD was 16.8%, 5 times greater than that of settlers living in the occupied territories before the Gaza disengagement. Other papers have assessed the predictors of depressive symptoms in population-based cohort studies [11, 47].

Together with victims and general population, specific population sub-groups (e.g., emergency personnel or children and adolescents) have also been evaluated. In contrast to the findings in the assessment of PTSD, where the prevalence of PTSD among rescue workers is higher than in the general population [4], prevalence of MDD in rescue workers seems to be lower than in victims or in general populations, as different studies carried out after March-11 terrorist attacks have clearly shown [48]. On the other hand, the impact of terrorism in children and adolescents reveals that a substantial proportion of youth reports a wide array of clinical needs and functional impairments months after an attack [see 49 for a review]. The role of protective factors for depression in adolescents has also received attention in the literature. For example, two prospective studies documenting changes in depressive symptoms (as measured by CES-D) in Israeli adolescents exposed to missile attacks [50] and suicide bombings [51] showed that social support (mainly friendly social support) buffers the effect of terrorism-related perceived stress in predicting changes in depression.

First, although we only included studies that assessed MDD based on diagnostic criteria, most of them used instruments that did not include an assessment of either manic or psychotic symptoms, therefore we could not classify the disorder beyond probable MDD [8]. Although the prevalence of bipolar affective disorder is not much higher than 1% [14] it could be inflating the MDD percentages. Future investigations should take this into account in order to help improve our understanding of the psychopathological processes involved. Second, as we mentioned in the selection criteria section, the reviewed studies were not designed to ensure that persons were free from psychopathology before the occurrence of the terrorist attacks, which means that prevalence, instead of incidence, was assessed. Moreover, none of them included a control group that would enable the comparison between exposed and non-exposed populations. We overcame this challenge by comparing the prevalence of MDD following terrorism with the prevalence reported in other general population surveys. However, it is important to be cautious when interpreting this comparison because the majority of these epidemiologic studies are referred to a whole nation's population (e.g. Spain or United States) [15, 58] and not to the city where terrorist attacks occurred (e.g. Madrid or New York), hence there could be differences between the two. Future studies should attempt to analyze the incidence of MDD by establishing baseline psychopathological assessments that may be used as population cohorts to document MDD incidence in the event of terrorism exposure. Third, there are several challenges facing longitudinal studies that aim to document the course of MDD. Several studies suffer from attrition, that is, the reduction in the number of people who participated in the follow-up studies. This may have biased the prevalence estimates of MDD, especially in the case of small samples [33]. Fourth, we have to be careful when extracting conclusions with respect to some correlates of MDD, mainly the pre-traumatic factors. In the reviewed studies, the assessments documented always took place after the terrorist attack in question. It is possible that pre-event reports are biased in the sense that depressed persons may selectively recall stressful situations that occurred before the attack to a greater extent than non-depressed persons.

In relation to the characteristics of our review, we only considered studies that had assessed samples of direct or indirect adult victims. Even though we did not include studies that assessed children or adolescents, further work with this age group is clearly warranted. We also limited our search to studies that analyzed the consequences of terrorist attacks and not other kinds of disasters; knowing and comparing the prevalence and course of MDD after natural, technological, or other disasters linked to interpersonal violence (such chronic exposure to trauma) could help us understand the onset of mental disorders after mass traumatic events. Finally, we have focused our review on the examination of MDD using diagnostic criteria. This enables us to compare prevalence rates of MDD with previous epidemiological surveys and between studies carried out after different terrorist attacks. However, MDD is not the sole disorder within the unipolar spectrum and extant research after terrorism has also highlighted the high prevalence and impairment associated with other forms of depression, such as mild or minor depression. Including these other forms of depression, together with the risk factors associated with it, could be of research and public health interest.

Our review highlights some key areas that are important for future research and may serve to guide intervention. First, the course of MDD after terrorist attacks remains unclear. That is why greater efforts are needed to elucidate the course of MDD after terrorist attacks. Second, there is very limited literature about psychological constructs that may be associated with MDD after terrorist attacks [3, 9]. It would be interesting, in this context, to analyze the role played by other variables that have been shown to be related to MDD, such as attributional style [59], self-esteem [60] or response styles to depression [61], and to examine the way in which certain psychological variables interact with other sociodemographic variables to predict the onset of MDD. For example, it is possible to analyze which psychological factors mediate the relationship between MDD and gender. This line of research will be useful in helping to identify the persons with higher probability of developing MDD following a terrorist attack and to improve the efficacy of the interventions from which they will benefit. Third, more research is needed on the role of MDD in psychiatric comorbidity after terrorist attacks. Although some reviewed studies have reported high rates of comorbidity between MDD and PTSD, more works are needed to have a better understanding of this relationship. For example, an interesting objective would be to examine the form in which both pathologies vary over the time after terrorism. In this line, some authors have recently documented the important role that depressive symptoms plays in the development and persistence of stress post-traumatic symptoms after different traumatic events [62]. Fourth, some of the studies in this revision included victims who had been bereaved [33, 37]. Although not reported in these papers, differences in the prevalence of MDD may exist between victims who have been directly injured by a terrorist attack and those who have been bereaved. Moreover, bereaved people could develop other psychological problems, such as complicated grief syndrome. A number of studies support the differentiation between complicated grief and MDD [63, 64], and some authors have shown that it is a usual reaction in bereaved people after terrorism [65]. A clear definition of victims in future works could provide us with a better understanding of the psychological consequences in people directly and strongly exposed to terrorism.