Depression in adolescents exposed to war trauma Risk factors for development of depression in adolescents exposed to war trauma: does PTSD matter?

In 2014, after a period of political instability, anti-Ukrainian military troops occupied the Eastern part of Ukraine. Evacuation of the civil population, especially children, did not start immediately. Military activity began in 2014 and has led to thousands of deaths, resettled populations, destroyed houses, and infrastructure. According to official data, over 1.5 million people were internally displaced (IDPs), with at least 300,000 of them being children and adolescents. Many were directly exposed to war-related events, such as bombardments, rifle fire, living in shelters, death of parents and members of primary support group, and so on (UN Report, 2018).

Posttraumatic stress disorder (PTSD) is one of the most severe psychopathologies that can develop as a response to natural or human-made catastrophic events, including war. When untreated, PTSD can persist throughout life and cause severe functional and emotional impairment (Priebe et al., 2009). PTSD is highly comorbid with other mental health pathologies, most commonly with the major depressive disorder. Recent cross-sectional studies of the affected child population showed that up to half of the children and adolescents with PTSD also meet the criteria for a major depressive episode (Adams et al., 2015; Tang et al. 2018; Ayazi et al., 2012).

It is well known that those suffering from PTSD can develop major depressive disorder (MDD) and vice versa. The overlap between the symptoms of both conditions (Flory et al., 2015) can make their diagnosis more complicated. PTSD/MDD comorbidity also influences the severity, course, and outcomes of the illness. A study of adult earthquake survivors found that persons with comorbid PTSD/MDD had significantly higher severity of symptoms and a lower rate of recovery from PTSD (Tural et al., 2012; McLean et al., 2014). Adults with cooccurring PTSD and major depressive episodes had a 2.6 times higher rate of suicidality compared to individuals with PTSD or MDD alone (Ramsawh et al., 2014). Studies of children with PTSD without comorbidities, particularly MDD, documented improved response to psychological treatments (Goldbeck et al., 2016; Jensen et al., 2017; Nanni et al. 2012; Mannarino et al., 2012).

Despite the growing body of research on outcomes of psychological trauma in children, the majority of studies are focused on the cross-sectional prevalence of PTSD, depression, and/or their comorbidity (Perkins et al., 2018; Cenat et al., 2015, Kar et al., 2006; Attanayake et al., 2009; Tang et al. 2018). Majority of studies assess the outcomes of non-war-related trauma (Lai et al., 2013; Valdez et al., 2014; Cheng et al., 2018). Few studies focused on the nature of the relationship between PTSD and MDD (Thabet et al., 2004; Ayazi et al., 2012; Adams et al., 2015; Pejuskovic et al., 2020).

The extreme level of violence is a hallmark of war-related trauma. Studies of Sudanese refugees found that children who were exposed to violent attacks and were injured demonstrated a higher rate of post-traumatic stress disorder (Geltman et al, 2005). Overall, war-related trauma is associated with a higher number of exposed individuals, a higher number of traumatic events, and their greater severity. The study comparing the evacuated youth after Hurricane Katrina with war refugees from Africa and Syria found that war refugees, besides the higher severity of internalizing symptoms, also reported more cases of separation with family and of witnessing more members of a primary support group or strangers injured or killed (Myles et al., 2018).

Besides the lack of data from longitudinal studies on the comorbidity of war-related PTSD and MDD in children, there is also a lack of studies of war-related PTSD and MDD in children from Europe. Most of the published studies of war-affected children have so far assessed children and refugees from Muslim or African countries with unique cultural and religious features.

In the study of war refugees from Africa and Syria who were evacuated to Berlin, these children demonstrated a higher level of distress associated with relocation that was associated with language acquisition and the adaptation to a new culture (Myles et al., 2018). In other instances, refugees have been distressed by racial discrimination and complex legal immigration processes (Fazel et al., 2012).

Several studies outside of Europe found religious coping (such as praying) by youth as the most helpful and frequent coping strategy to alleviate stress and improve stress-related outcomes (Zukerman et al., 2014; Volkl-Kernstock et al., 2014). According to the latest data from sociological studies, only 55% of inhabitants in Eastern Ukraine consider themselves to belong to an organized religion (Razumkov Center, 2016) – a much lower percentage compared to Muslim or African countries.

Overall, there is a clear need for prospective investigations of war-related PTSD and its comorbidity with MDD in diverse cultural settings.

In this study, we aimed to explore the effects of PTSD on development of depression. The primary goal was to estimate the effect of PTSD on the further onset of depression in youth with war trauma. Secondary goals were to identify potential risk and protective factors for development of depression in adolescents with war trauma. We hypothesized that adolescents with war-related PTSD will have a higher incidence of depression compared to adolescents with war-related trauma, but without PTSD. We also hypothesized that adolescents with PTSD will be more vulnerable to post-displacement traumatic events compared to their peers without PTSD.

Over the course of the year-long observation, 11 adolescents were lost to follow-up due to their relocation back to the occupied territory. The loss to follow-up was 11 out of 160, or well below 10%. A Consort Flowchart for the study is presented in a Figure 1.

Figure 1

Among those who completed the study, 26/79 youth with PTSD developed a depressive disorder (32.9%), compared to 6/70 without PTSD (8.5%): RR = 3.83 (95% CI 1.67–8.78), p = 0.0014; NNTH = 4.108 (8.53–2.7). Throughout the year, none of the adolescents involved in the study developed new-onset PTSD.

During the study, 36 (24.1%) had a non-suicidal self-harming behaviour, 27 (18.1%) had suicidal thoughts. Among them, 5 (3.3%) adolescents with PTSD and comorbid depression were hospitalized due to suicidal planning and high suicidal risk. Among adolescents with comorbid PTSD and depression, there were 3 patients with psychotic symptoms that improved with SSRI treatment and were considered to represent psychotic features of depression. Information about mental health care received by adolescents with PTSD in the community is presented in Table 2.

Throughout the year of observation, both cohorts were exposed to multiple post-displacement life events that were regarded as traumatizing. Among those who completed the study, 25 (16%) were exposed to bullying, 31 (20%) complained regarding loss of financial/social status, 11 (7%) had parents divorced, 19 (12%) had a parent being fired at and 2 (1%) had lost their grandparents. Information about types of post-displacement traumatic events is presented in Table 3.

All adolescents were characterized by low school attendance. Adolescents with PTSD were also characterized by low access to psychological and psychiatric care. For example, only 26 of the 79 adolescents with PTSD received Trauma-focused cognitive-behavioural therapy (TF-CBT) during the period of observation.

Before completing the analysis, we examined the demographic differences between the exposed and control groups. No differences were found in adolescents’ gender, age, and the time since trauma (p > 0.05). In spite of the apparent difference in the number of war-related traumatic events (3.0 [2.0–4.0] in PTSD group and 2.0 [1.0–3.0] in the non-PTSD group), it was not statistically significant (p > 0.05). Adolescents in the PTSD group showed a higher level of school non-attending, χ² (1) = 5.294, p < 0.05. Children in both PTSD and non PTSD group were equally exposed to post-displacement traumatic events, χ² (1) = 1.138, p = 0.286.

Adolescents in the PTSD group (Mdn = 32.0) compared to the non-PTSD group (Mdn = 27.0) had a higher baseline (T0) severity of depression scores, U = 953.0, p < 0.001.

The association between estimated co-variables and depression was analysed using binary logistic regression. Results of the regression are presented in Table 4.

We analysed the possible risk factors for the development of depression and found that having PTSD combined with being a female and non-attending school on a regular basis were significant factors that increased the risk of the onset of depression in adolescents with war-related traumatic exposure. Considering that more than 80% of all depression cases were in the PTSD cohort, we also analysed it separately.

The post-hoc logistic analysis estimated that secondary traumatic events and school non-attending were correlated with increased risk for comorbid depression in adolescents with PTSD, while treatment with TF-CBT decreased this risk. Results of the regression are presented in Table 5.

Throughout the year-long observation, adolescents in a PTSD group enrolled in the study demonstrated a significant decline in severity of PTSD symptoms on CAPS-CA. Changes in severity of CAPS-CA scores in both PTSD and non-PTSD groups are presented in Figure 2.

Figure 2

The results of repeated measures ANOVA determined that CAPS-CA severity in PTSD group differed statistically significantly between the time points, Wilks’ Lambda = 0.782, F (2, 77) = 10.708, p = 0.001. Post-hoc tests using the Bonferroni correction revealed that CAPS-CA severity significantly declined from T0 to T1 assessments (41.25 ± 6.54 vs 38.6 ± 8.45, p = 0.001). CAPS-CA severity at T2 assessment demonstrated further decline up to 35.4 ± 12.91, and it was statistically significantly different relative to T0 (p = 0.001) and T1 assessments (p = 0.001). The corrected p-threshold was 0.016.

Despite the decrease in PTSD symptoms, the severity of depressive symptoms on CDRS-R demonstrated an upward trend, in particular due to the increase in severity of depression among patients who did not demonstrate improvement in PTSD severity. Changes in severity of CDRS-R scores in both PTSD and non-PTSD groups are presented in Figure 3.

Figure 3

A Mann-Whitney test indicated that the severity of depressive symptoms on CDRS-R in T1 point was higher in the PTSD group (Mdn = 34.0) than in the non-PTSD group of children (Mdn = 21.0), U = 6.5, p < 0.001.

The results of Mann-Whitney test indicated that the severity of depressive symptoms on CDRS-R in T2 point was higher in the PTSD group (Mdn = 34.0) than in the non-PTSD group of children (Mdn = 24.0), U = 103.5, p < 0.001.

The results of repeated measures ANOVA determined that CDRS-R severity in PTSD group differed statistically significantly between the time points, Wilks’ Lambda = 0.703, F (2, 77) = 16.245, p = 0.001. Post-hoc tests using the Bonferroni correction revealed that CDRS-R severity significantly increased from T0 to T1 assessments (32.6 ± 2.34 vs 39.2 ± 12.03, p = 0.001). CDRS-R severity at T3 assessment increased to 42.17 ± 14.84, that was statistically significantly different relative to T0 (p = 0.001) and T2 assessments (p = 0.002). The corrected p-threshold was equal to 0.016.

This study sought to examine the direct and moderating effects of PTSD and other risk factors on the development of depression in a sample of war-exposed adolescents. To our knowledge, this is the first study assessing the effects of war-related PTSD on adolescents in Ukraine. Due to the limited resources and high prevalence of stigma, the majority of adolescents were deprived of both early interventions and subsequent highly specialized trauma-focused treatment. As a result, this study reflects the onset of MDD among adolescents with PTSD as its natural course. That is particularly valuable because most military actions occur in low and middle-income countries with limited access to resources.

In our study, the prevalence of depression among war-exposed adolescents was about 20% and was consistent with the published literature (Kien et al., 2019). However, due to the methodological features, particularly the exclusion of adolescents with any psychiatric disorders except for PTSD from this study, it likely underestimates the actual prevalence of depression among the affected adolescents in the general population.

The above exclusion may help explain the lower estimated comorbidity rate of PTSD and depression when compared to the recent studies in children and adolescents (Lai et al., 2013; Adams et al., 2015; Tang et al., 2018), with published rates of comorbidity of about 50%.

This exclusion may also explain the lower (32.9%) than the literature average (40%) cross-sectional prevalence of depression after traumatic events among children with PTSD.

Some of the past cross-sectional studies that demonstrated a higher rate of comorbidity (Tang et al., 2018) were made several years after direct traumatization. Since the prospective study of children earthquake survivors estimated that the stability of PTSD tended to decrease across time, while depressive symptoms increased (Cheng et al., 2018), the relatively brief duration of time since war-related trauma may be another reason for the lower cross-sectional prevalence of depression after traumatic events among children with PTSD.

Before the analysis, we did not find any significant differences among adolescents with and without PTSD in their gender distribution, age, number of war-related traumatic events, and time since trauma. We did find that both groups of adolescents were highly exposed to the multiple post-displacement traumatic events, and exhibited high levels of school non-attending – both consistent with other studies of children exposed to war-related trauma (Joshi et al., 2015).

Children in both PTSD and non-PTSD groups had a significant difference in baseline depression severity. It is possible that such a difference may be the result of non-specific PTSD symptoms (mood lability, irritability, sleeping problems) common for both PTSD and MDD (Post et al., 2011; Gros et al., 2012). As a result of this symptom overlap, the children with PTSD had higher baseline depression scores on CDRS-R compared to the children without PTSD. Another important possibility is that some of the children who developed PTSD were already in the process of developing clinical depression.

In our study, depression was more prevalent among girls, consistent with the role of female gender as a significant risk factor for development of depression in traumatized adolescents. These finding were consistent with other studies in traumatized youth (Valdez et al., 2014; Adams et al., 2015) and in the general population (Salk et al. 2017). However, in a subsample of adolescents with PTSD, female gender was nonsignificant.

Post-hoc analysis confirmed the significant effect of PTSD as a risk factor for development of depression. Different theories have been proposed to explain the interaction between PTSD and MDD, including a common genetic mechanism of development and shared risk loci, among others (Smoller, 2016; Duncan et al., 2018).

Another possible model of interaction of PTSD and MDD involves a demoralization mechanism as a response to chronic medical conditions (Mangelli et al., 2005). While most of the children demonstrate improvement in PTSD symptoms with time (Punamaki et al., 2014), those with poor recovery demonstrate stable severity of depressive symptoms (Lai et al., 2013).

We estimated that more than a third of the sample was exposed to significant traumatic life events after displacement, such as bullying, loss of social status, parent's divorce, and so on. Despite the lack of significance of the number of secondary traumatic events with respect to subsequent depression in the full sample, a separate regression analysis in the PTSD cohort of adolescents revealed that secondary traumatic events were a significant risk factor for development of depression.

If meaningful, this finding would be consistent with a ‘double-hit hypothesis’, namely that adolescents who are more vulnerable to developing PTSD are also susceptible to the dose effect of the number of secondary traumatic events, while adolescents who do not develop PTSD are also less vulnerable to developing MDD.

In addition, these findings may reflect an indirect effect of PTSD-related changes in the coping style strategies and cognitive appraisal (Ehlers et al., 2003; Pfefferbaum et al., 2014; Smith et al., 2013) that can lead to both catastrophizing of routine stimuli and chronicity of the stress reactions (Trickey et al., 2012; Greca et al., 2013; Punamäki et al., 2015).

We did not include baseline CDRS-R scores in the final logistic model due to a high correlation between PTSD variable and CDRS-R (T0) scores that we contributed to an overlap between non-specific PTSD and depressive symptoms.

Two factors were estimated to be protective. TF-CBT is a highly effective form of treatment of adolescent PTSD that also has evidence for decreasing the severity of co-occurring depression (Lenz et al., 2015). In our study, TF-CBT appeared to decrease the risk of development of depression that may result from improvement in PTSD severity (Lai et al., 2012).

Social support in general and peer support specifically are vital for well-being of adolescents and can both alleviate PTSD symptoms and serve as a protective factor against depression (van Harmelen et al., 2016; Reavell et al., 2017). School non-attending on a permanent basis was associated with increased risk for development of depression in our study. This finding could be partially related to decreased social and peer support.

Another possible explanation may be that school non-attending is a confounding variable that represents a marker of increased impairment due to PTSD and depression.

Whatever the explanation, this study's findings regarding key role of schools provides strong support for increased resource allocation to enhance school peer and teacher engagement programs as part of early identification and treatment interventions.

In our study, we found that PTSD severity in adolescents with war trauma had declined over time, while depression severity increased, particularly among children with poor recovery. This was consistent with some of the previous findings. The study by Lai and colleagues found a stability of depressive symptoms among children that demonstrated a trajectory of poor recovery from PTSD (Lai et al., 2013). Even though that study did not demonstrate increased depression over time, some of the possible explanations include initially low severity of posttraumatic symptoms and a much younger age of the sample.

In a study of earthquake survivors, Cheng with colleagues found the opposite results – namely that depressive and posttraumatic symptoms had a mutual affect in early period; however, they found in the later periods that depressive symptoms predicted the stability of PTSD. This difference can be a result of the different methodological approaches in our studies, particularly the exclusion criteria and sample features. Cheng and colleagues did not exclude possible pre-exposure mental health disorders. Thus, unlike our study, Cheng and colleagues involved children who had comorbid depression and PTSD at the moment of study enrolment.

This study has a number of limitations. The sample size was medium and limited to a relatively demographically homogenous population of 15–17 year old adolescents. Therefore, results may not generalize to traumatized younger children with other demographic characteristics.

Even though DSM-5 criteria were already available, we applied the DSM-4-TR criteria that corresponded to the routine clinical/research practice and to the available instruments in Ukraine at the moment of the study. This makes comparison of our results to studies using DSM-5 more difficult.

Secondly, PTSD and depression severity was assessed using only data obtained from the adolescent.

Thirdly, all adolescents were recruited through social service agencies; due to the distrust in public services, a significant part of the population did not apply for social assistance. On the other hand, majority of children with PTSD who were recruited into the study had moderate to severe severity of PTSD since those with less severe symptoms did not apply for medical or psychological assistance.

Another limitation was the validation of instruments used in the study. Even though CAPS-CA was translated and back-translated, and K-SADS-PL and CDRS-R were commonly used in clinical trials and routine practice, no studies on their validation or psychometric properties in Ukrainian had been conducted to date. We also significantly modified the existing version of TESI-C to meet our needs without subsequent validation. We aimed to decrease the possible chance of diagnostic error by using several assessment instruments, including DSM-4-TR criteria.

It was not uncommon for displaced adolescents to spend most of their time in war-affected regions, while coming to non-occupied territory only for social and financial assistance and hiding this fact due to the fear of possible loss of assistance and financial payments from the government. Between assessments, adolescents were not observed by a research team, so we cannot exclude that some of these adolescents were traumatized by the ongoing war during the study.

Another limitation was that we were not searching for other mental disorders except PTSD and depression during the T0 to T2 period of observation. Despite the fact that no adolescents received medical assistance due to any other reasons except for depression and PTSD in the community, it was possible that episodes of hypomania or mild forms of anxiety disorders were not detected.

Despite the fact that we did not check for the possible presence of learning or intellectual disability, all children were considered to have normal intelligence due to attendance of regular local schools. However, we could not rule out possible mild intellectual disability and possible learning difficulties that could increase the risk of depression.

We were not able to obtain medical charts of all adolescents from the occupied territories, so our data on the absence of previous mental health disorders was often based on the information received from the adolescents and their parents.

The results of this study need to be considered carefully through the family context. We did not assess the child-parent attachment. Secure attachment is especially important in the development of trauma and depression resilience, as demonstrated by the studies of foster children (Sapienza et al., 2011; Suzuki et al., 2015). Besides the fact that war exposure is extremely traumatizing for children, it may affect their parents’ mental health too (Charlson et al., 2019). Parental psychopathology mediates the amount of perceived social support and even itself becomes a risk factor for child internalizing disorders (Lester et al., 2010; Zalta et al., 2018). A study of refugees in Australia estimated that caregiver's PTSD was associated with higher levels of child conduct problems, hyperactivity, emotional symptoms, and peer problems (Bryant et al., 2018). Future studies will need to address this limitation as well as address importance of religion and other potential contributors to trauma and depression resilience.

This study's findings make important contributions to our understanding of the interactions among war-related traumatic stress, PTSD symptoms, and depression. However, the role of PTSD as a moderator of depression merits further investigation in larger scale longitudinal studies in diverse populations.

While adolescents with war-related PTSD are at high risk for development of depression, many countries, especially those with low resources, have been unable to provide the children in need with robust and evidence-based mental health services, including TF-CBT. It is vital to develop such capacity, to support children in their access to schools, and to build enhanced social supports to more effectively deal with post-displacement traumatic life events.