Anxiety, posttraumatic stress, insomnia and quality of life in Ukrainian schoolchildren during wartime

Abstract

Background

Children who have remained in Ukraine during the ongoing war are continuously exposed to threat and disruption of services, which places them at high risk of internalizing and trauma-related symptoms.

Methods

In a school-based cross-sectional survey conducted in Ukraine (January-February 2024), 781 students aged 10–17 years completed standardized measures of quality of life (QoL) (PedsQL 4.0), depressive symptoms (CDI 2:SR), state and trait anxiety (STAIC), posttraumatic stress symptoms (CRIES-8), insomnia (Athens Insomnia Scale), resilience (CYRM-R) and coping (JSR). Four multiple linear regression models with forward stepwise selection were estimated.

Results

All models were significant and explained 25–50% of the variance. Emotional functioning (R² = 0.50) was inversely associated with trait anxiety, PTSD symptoms, state anxiety and insomnia, while caregiver/relational resilience and situational emotion-focused coping made small positive contributions. Social functioning (R² = 0.33) was negatively related to trait anxiety and PTSD and positively related to personal resilience. School functioning (R² = 0.26) was negatively related to PTSD, depressive symptoms and insomnia, and positively related to situational resilience and caregiver/relational resilience. Physical functioning (R² = 0.25) was inversely related to state anxiety, PTSD and insomnia, and positively related to caregiver/relational resilience and active coping.

Conclusions

Among war-exposed Ukrainian children who stayed in the country, QoL is mainly determined by symptom burden, especially anxiety, PTSD and sleep disturbance. Resilience and coping have smaller and domain-specific effects. School-based, trauma-informed and sleep-focused interventions should be prioritised in ongoing conflict settings.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13031-026-00803-3.

Introduction

The full-scale Russian invasion of Ukraine has exposed children who remained in the country to chronic threat, intermittent displacement and disruption of basic services, leading to elevated levels of anxiety, posttraumatic stress and sleep disturbance reported across Ukrainian child and adolescent samples. Recent studies involving Ukrainian youth consistently show substantial increases in internalising symptoms, including depression, anxiety and trauma-related distress; however, the functioning of children who remain in active conflict zones is still insufficiently described compared to refugee or displaced peers [33, 53]. Even less is known about how this symptom burden translates into everyday functioning and quality of life (QoL) under ongoing armed conflict.

Unlike refugee populations, children who remain in Ukraine continue to experience ongoing exposure to attacks, chronic instability, prolonged educational disruption and significant strain on family and community resources. Recent field research indicates that these cumulative stressors intensify emotional and functional difficulties among children living in war-affected regions [32, 33, 53]. While the mental health consequences of war have been widely studied [43, 58] emerging evidence suggests that children remaining inside Ukraine present distinct patterns of symptoms and adjustment compared to displaced groups, reinforcing the need to understand how multiple psychological symptoms jointly affect functioning. Previous studies have identified high levels of posttraumatic stress disorder (PTSD), anxiety, depression, and sleep disturbances among Ukrainian children during the war [33, 41, 53]. These symptoms frequently co-occur and interact through mutually reinforcing pathways that can heighten overall impairment [26]. For instance, insomnia has been shown to function not only as a symptom but also as a transdiagnostic contributor to worsening trauma and mood disturbances [59] and recent evidence highlights the role of sleep disruption as a driver of emotional dysregulation and reduced daily functioning among war-exposed youth [58]. While individual symptoms such as PTSD or depression have been linked to diminished QoL, few studies have examined their combined or relative predictive power, particularly in conjunction with protective factors such as resilience and coping.

Children living in the Ternopil region have experienced repeated air-raid alerts, disruptions to daily routines, and varying degrees of threat associated with missile strikes in nearby areas. Although the region is located outside the most intensive combat zones, ongoing alarms, infrastructural strain, and the arrival of families displaced from more heavily affected parts of Ukraine have shaped children’s daily experiences. These conditions influence their sense of safety, school continuity, and emotional stability, and provide essential context for interpreting patterns of mental health and QoL observed in this population [33, 53, 59].

To better capture the complex ways in which mental health difficulties affect everyday functioning, multidimensional instruments such as the Pediatric Quality of Life Inventory (PedsQL) have been developed and can be viewed as a child-friendly functional outcome complementary to symptom measures commonly used in child psychiatry. This tool assesses child functioning across four key domains: physical, emotional, social, and school. Despite increasing attention to trauma symptoms in conflict-affected youth, the cumulative effects of psychological distress and buffering mechanisms on QoL remain underexplored, particularly in children who remain in war zones without access to structured psychosocial support. Newer studies emphasise that PTSD, depression, anxiety, and insomnia each exert unique and overlapping effects on physical symptoms, school engagement, peer functioning and emotional regulation, underscoring the need for models that integrate multiple forms of distress when examining QoL [21, 24].

Prior research among Ukrainian children displaced to Poland following the Russo-Ukrainian war demonstrated the pivotal role of relational resilience, bereavement, and coping mechanisms in modulating depression and anxiety amidst war-related adversities [58]. However, as Chen et al., [9] argue, war and conflict drastically transform the sociocultural, economic, and emotional landscapes in which children develop, thereby disrupting and redefining the role of traditional protective factors, such as family support or socioeconomic resources. These factors may lose their buffering potential under chronic wartime conditions. Research across conflict-affected settings highlights the importance of relational resilience—such as supportive caregiving, peer connectedness and stable adult relationships—as a buffer against trauma-related symptoms, including among Ukrainian youth [5, 13, 25, 40].

Emerging literature supports the critical role of resilience, particularly relational resilience, including supportive caregiving and connectedness to others, as a protective factor in war-affected youth [16, 25, 40, 55]. Likewise, coping strategies such as seeking social support or engaging in meaningful activities may reduce the likelihood of emotional dysregulation and academic disengagement in the face of traumatic stress [14, 42, 48]. As emphasized in prior reviews, empirical studies often isolate these variables rather than examine them in integrated multivariate models that include both risk and resilience. Contemporary findings from war-exposed children in different cultural contexts, including Ukrainian samples, suggest that coping flexibility and relational resources may moderate the association between trauma symptoms and functional outcomes, although their protective strength may vary under prolonged adversity [25, 38].

Recent findings from Ukrainian samples reinforce the need for such models. Children with limited adaptive resources, particularly in the domain of interpersonal communication, have been shown to be more vulnerable to emotional withdrawal and maladjustment [34]. Nevertheless, most evidence still comes from displaced populations, while children who remain in Ukraine continue to live under conditions of persistent danger, disrupted schooling, family separation, and the breakdown of community networks [49, 57]. These conditions demand a deeper understanding of how psychological vulnerability and protective mechanisms interact to shape children’s QoL. Integrating both symptom burden and adaptive resources into a single model offers a more comprehensive framework for understanding how children function under continuous wartime stress [5, 25, 38, 55].

This study aimed to pursue two interrelated objectives. First, it aimed to assess the current psychological status of school-aged children and adolescents living in Ukrainian areas exposed to wartime conditions of varying intensity by examining levels of depression, anxiety, posttraumatic stress symptoms and insomnia. Second, it aimed to determine which psychological and protective factors were most strongly associated with QoL in this population. Specifically, the study examined how mental health symptoms, together with protective resources such as resilience and coping strategies, related to functioning across four QoL domains (physical, emotional, social and school). Using correlational analyses and multiple linear regression with forward stepwise selection, the study integrated both risk and resilience factors into a model of child functioning under conditions of ongoing armed conflict.

Methods

Study design and participants

The cross-sectional survey was conducted during the Russian invasion of Ukraine, between January 21 and February 27, 2024. Lviv State University of Physical Culture approved the study protocol (No. 16, February 13, 2024). Approval was also obtained from the education departments and school administrations to implement and conduct the study. The study began with the completion of an informed consent form by parents and children, and participation in the survey was only possible if both the adult and the child provided informed consent.

Given the wartime context, additional ethical and safety considerations were implemented to minimize potential psychological burden. Participation was voluntary, and children were informed that they could discontinue the survey at any time without providing a reason. Data collection was conducted in familiar school settings under the supervision of trained school staff and research team members. During questionnaire administration, participants were monitored for visible signs of distress, and no time pressure was imposed. In cases where a child expressed discomfort or emotional difficulty, participation was discontinued and the child was referred to the school psychologist or counselor available within the educational system, in accordance with standard school support procedures. No acute adverse events requiring emergency intervention were reported during data collection. The study did not involve diagnostic procedures or clinical interventions.

Inclusion criteria were: (a) approval of the school administration; (b) written informed consent from the parent/guardian and the child; and (c) current enrolment in grade 5 or higher. Of 790 individuals who agreed to participate, 9 were excluded at the consent stage; 781 participants were included in the final sample.

Measures

All measures that were already available in Ukrainian and had been used in previous studies with Ukrainian youth were applied in their existing versions [32, 43, 58].

Quality of life was measured with the Pediatric Quality of Life Inventory (PedsQL™) 4.0 Generic Core Scales, suitable for both healthy and clinical pediatric populations [8]. The 1-month recall form uses a 5-point scale (0 = never to 4 = almost always); items are reverse-scored and linearly transformed to a 0–100 scale, with higher scores indicating better quality of life. Four domain scores were computed (physical, emotional, social, and school functioning), as well as the Psychosocial Summary Score (emotional, social, and school functioning) and the Total Scale Score (all 23 items). The PedsQL demonstrates good reliability across all subscales (α > 0.80) and has a validated Ukrainian version [41]. Internal consistency was α = 0.87 for Physical Functioning, α = 0.84 for Emotional Functioning, α = 0.83 for Social Functioning, and α = 0.80 for School Functioning.

The Children’s Depression Inventory 2nd Edition Self-Report Short Form (CDI 2:SR) [31]. Depressive symptoms were assessed with the Children’s Depression Inventory 2 – Self-Report Short Form (CDI-2:SR) for ages 7–17 (12 items, 3-point responses, past 2 weeks); higher scores indicate greater symptom burden (range 0–24). The CDI-2 has well-documented psychometric properties and is widely used in research and clinical settings [28–30]. Internal consistency in the present sample was α = 0.82.

The State-Trait Anxiety Inventory, Child Version (STAIC) [54] Anxiety was measured with the State–Trait Anxiety Inventory for Children (STAIC), originally developed for 9-12-year-olds, but widely applied in older child and early adolescent samples, including school settings [30, 38]. The instrument comprises 40 items forming two 20-item subscales assessing state (“how I feel right now”) and trait anxiety (“how I generally feel”). Items are rated on a 3-point scale (“hardly ever”, “sometimes”, “often”) and scored 1–3; after reverse-scoring anxiety-absent items according to the manual, item scores were summed to yield state and trait scores ranging from 20 to 60, with higher scores indicating greater anxiety. The STAIC has shown good internal consistency and construct validity across paediatric and adolescent populations [16, 47]. Internal consistency in the present sample was α = 0.92 for the state anxiety scale and α = 0.93 for the trait anxiety scale.

How Do You Cope? (JSR) questionnaire [25] was used to assess coping styles including dispositional (general tendency) and situational coping. The tool consists of two 9-item parts, with responses recoded to a 0–4 scale, where higher values indicate more frequent use of a given strategy. Each part yields three coping styles: active coping, emotion-focused coping, and seeking social support. Active coping refers to efforts aimed at directly addressing or managing the stressor, emotion-focused coping involves regulating emotional responses to stress, and seeking social support reflects attempts to obtain help, comfort, or advice from others. For each coping style we computed a mean item score (not a sum score) from the relevant items, resulting in subscale scores ranging from 0 to 4. The JSR has demonstrated satisfactory reliability and construct validity in empirical studies [40, 55, 58].

The Child and Youth Resilience Measure-Revised (CYRM-R) was used to assess resilience. It consists of a 5-point Likert scale with 17 positively-worded items. Scores were derived for the two subscales: personal and caregiver/relational resilience. Personal resilience (PRES) encompasses both intrapersonal and interpersonal aspects, which are interconnected as both dimensions rely on individuals’ social environments to strengthen their resilience. Caregiver/relational resilience (CRES) pertains to qualities associated with significant relationships, such as perceived emotional support, caregiver availability, and a sense of safety and belonging, which support children’s adaptation under adversity. The measure is suitable for assessing resilience in children and youth from diverse cultural backgrounds [14, 42, 48]. In the current sample, Cronbach’s α was 0.96 for the total scale, 0.95 for personal resilience (PRES), and 0.93 for caregiver/relational resilience (CRES).

The Children’s Revised Impact of Event Scale – 8 items (CRIES-8) [45] was used for assessing PTSD symptoms in children and adolescents. This self-report questionnaire consists of 8 items that are divided into two subscales: intrusion, avoidance. Individual items are rated according to the frequency of their occurrence during the past week and in relation to a specific traumatic event written at the top of the scale. The scale’s ability to efficiently identify children and adolescents at risk for PTSD makes it an invaluable tool in both posttrauma assessment and intervention planning [45]. In the current sample, Cronbach α was 0.89.

The Athens Insomnia Scale (AIS) [52] assessed insomnia symptoms. It is a self-report questionnaire designed to assess the severity of insomnia, in line with ICD-10 criteria. It includes eight items, with the first five focusing on sleep induction, maintenance, and quality, and the remaining three on daytime well-being and functioning. Respondents rate each item on a 0–3 scale based on the severity of symptoms over the past month. A total score ranges from 0 to 24, a higher score means more severe insomnia symptoms. The AIS has been validated in a diverse sample, including the Ukrainian population [43]. Although originally developed in adults, the AIS has been used and psychometrically evaluated in adolescent school-based samples, demonstrating acceptable reliability and validity for assessing insomnia symptoms in youth [11, 61]. The reliability of the AIS was Cronbach α = 0.84 in the current sample.

Demographic items included age, gender, and grade. In addition to the measures above, students completed a set of questions assessing their personal and family situation. These items captured war-related displacement and living/learning changes (whether the child changed place of residence/study due to the war: relocated within Ukraine, relocated abroad, relocated and later returned to the pre-war location, or did not relocate), and, if relocated, time at the new location (months) and pre-war city of residence. Family structure and household composition included number of siblings (numeric) and household members living with the child. Current housing was recorded as own home, staying with friends/relatives, rented apartment, hotel/hostel, or other.

Statistical analysis

All statistical analyses were conducted using IBM SPSS Statistics (version 21; IBM Corp., Chicago, IL, USA). Prior to the main analyses, data were screened for missing values and outliers. Missing data were minimal (less than 5% across variables) and were handled using listwise deletion. Given the low proportion of missingness, no imputation procedures were applied. Descriptive statistics, including sample size (n), means (M), and standard deviations (SD), are presented in Table 1. The normality of distributions was assessed using the Kolmogorov–Smirnov test.

Table 1.

Descriptive statistics of psychological variables, n=781

Variable	Range	M±SD
Depression	0-21	5.9±4.22
Dispositional active coping	0-4	2.0±0.96
Dispositional emotion coping	0-4	1.3±0.95
Dispositional seeking social support	0-4	1.4±0.81
Situational active coping	0-4	2.2±1.03
Situational emotion coping	0-4	1.6±1.00
Situational seeking social support	0-4	1.7±0.90
State anxiety	20-60	33.6±9.24
Trait anxiety	20-60	34.5±9.63
Personal resilience	10-50	32.6±9.75
Caregiver/relational resilience	7-35	25.8±8.17
Physical functioning	0-100	77.7±18.43
Emotional functioning	0-100	65.4±21.97
Social functioning	0-100	78.1±19.85
School functioning	0-100	68.5±19.75
PTSD	0-40	15.3±10.63
Insomnia	0-24	5.7±4.54

M – mean, SD – standard deviation

Associations between study variables and health-related quality of life were examined using Spearman’s rank-order correlation coefficients. To test the study hypotheses, multiple linear regression analyses with forward stepwise selection were performed. Four separate models were estimated, with the PedsQL subscales (Physical, Emotional, Social, and School Functioning) as dependent variables. Independent variables included depression (CDI-2:SR), state and trait anxiety, posttraumatic stress symptoms, insomnia, dispositional and situational coping strategies (active coping, emotion-focused coping, and seeking social support), and personal as well as caregiver/relational resilience. Only variables that showed significant bivariate associations with a given PedsQL domain (p < .05) were entered into the corresponding regression model. Explained variance (R² and ΔR²) was reported for each successive step of the analysis.

Socio-demographic or contextual variables (e.g., age, gender, grade level, family situation or displacement status) were not included as predictors because the analytic focus of the study was on the unique and combined associations of psychological distress and protective factors with QoL. Including a larger set of covariates in a sample of this size would also increase the risk of model overfitting. Regression assumptions were verified prior to model interpretation, including linearity, normality of residuals, homoscedasticity, and independence of errors. Multicollinearity was assessed using variance inflation factors (VIF), with values below 5 indicating acceptable levels. Cohen et al., [12], Ernst, [18] The significance level was set at p ≤ .05. All data were de-identified and stored on encrypted institutional servers. Given the multidimensional nature of QoL, separate regression models were specified for each PedsQL domain, reflecting prior theoretical and empirical evidence for domain-specific associations between psychological symptoms, coping strategies, and functioning in war-affected children.

Results

Participants

The sample was recruited through participating schools based on feasibility and administrative agreement and therefore represents a school-based convenience sample rather than a stratified, probability-based sample. Participating schools included both urban and non-urban settings within the Ternopil oblast. Urban schools were located in Ternopil city (e.g., School No. 7; Classical Lyceum/School No. 16; Schools No. 4, 8, 9, 20 and 24; Gymnasium No. 1; Gymnasium “Harmonia”), while additional schools were located in district towns, settlements and rural communities (e.g., Kozova, Husyatyn, Zaliztsi, Pidvolochysk, Myshkovytskyi, Kozliv, Pidlypetskyi, Pyliava and Kremenets). The final sample included 781 schoolchildren (41.1% girls) with a mean age of 13 years (range: 10–17; M = 12.9, SD = 1.58). Most participants were in the seventh, eighth and ninth grades (19.72%, 27.02% and 22.54%, respectively), with additional representation from the fifth (12.42%), sixth (13.79%), tenth (3.33%) and eleventh grades (1.28%).

War-related exposure, displacement and family context

Children were assessed during ongoing wartime conditions in the Ternopil region, characterised by frequent air-raid alerts and disruptions to daily life. In the sample (N = 781), 25.1% reported having been in a zone of shelling or explosions, and 85.7% reported sheltering during air-raid alerts at least sometimes, with 50.2% reporting frequent sheltering. Approximately 34.3% reported leaving home or school for multiple days due to hostilities.

Indirect exposure was also common: 65.4% reported at least one relative or friend who was fighting, 38.7% reported someone hospitalised due to the war, and 65.4% reported at least one displaced relative or friend. War-related material or economic loss was reported by 27.0% of participants (4.5% substantial). Most children (71.8%) reported no change in place of residence or study, while 15.9% experienced displacement followed by return, 8.6% were currently displaced within Ukraine, and 3.7% were living or studying abroad. Regarding family context, the mean number of siblings was 1.54 (SD = 1.45); 94.6% lived with their mother, 78.7% with their father, and 36.4% with at least one grandparent. Most participants lived in their own home (84.0%), with smaller proportions in rented accommodation (8.3%) or staying with friends or relatives (4.2%).

Descriptive statistics

Descriptive statistics showed mean scores of 33.6 (SD = 9.24) for state anxiety and 34.5 (SD = 9.63) for trait anxiety. Mean depression and insomnia scores were 5.9 (SD = 4.22) and 5.7 (SD = 4.54), respectively. The mean PTSD score was 15.3 (SD = 10.63), indicating substantial variability. Active coping strategies were reported more frequently than emotion-focused or social support–seeking strategies at both dispositional and situational levels. Mean scores for personal resilience and caregiver/relational resilience were 32.6 (SD = 9.75) and 25.8 (SD = 8.17), respectively. PedsQL scores varied across domains, with mean values ranging from 65.4 to 78.1.

General description of the regression models

Four multiple linear regression models were estimated using the same forward stepwise selection logic applied separately to each PedsQL domain (Physical, Emotional, Social and School Functioning). For each model, only predictors that showed significant bivariate associations were entered (p < .05) with the corresponding domain (Table 2). All models were statistically significant and explained between 25% and 50% of the variance in QoL across domains (Table 3).

Table 2.

Bivariate associations between independent variables and PedsQL domains

Variable	Physical functioning	Emotional functioning	Social functioning	School functioning
Age [years]a	-0.07 p = .054	-0.10 p = .007	-0.02 p = .541	-0.05 p = .133
Sexb	2.71 p = .007	5.58 p ≤ .001	1.20 p = .229	2.76 p = .006
Depressiona	-0.33 p ≤ .001	-0.46 p ≤ .001	-0.43 p ≤ .001	-0.31 p ≤ .001
Dispositional active copinga	0.09 p = .011	0.01 p = .828	0.11 p = .002	0.02 p = .527
Dispositional emotion copinga	-0.27 p ≤ .001	-0.41 p ≤ .001	-0.30 p ≤ .001	-0.22 p ≤ .001
Dispositional seeking social supporta	-0.08 p = .026	-0.05 p = .149	0.02 p = .532	-0.06 p = .106
Situational active copinga	0.14 p ≤ .001	-0.01 p = .724	0.13 p ≤ .001	0.01 p = .760
Situational emotion copinga	-0.23 p ≤ .001	-0.45 p ≤ .001	-0.24 p ≤ .001	-0.28 p ≤ .001
Situational seeking social supporta	-0.08 p = .022	-0.09 p = .013	0.01 p = .850	-0.12 p = .001
State anxietya	-0.38 p ≤ .001	-0.49 p ≤ .001	-0.45 p ≤ .001	-0.30 p ≤ .001
Trait anxietya	-0.40 p ≤ .001	-0.58 p ≤ .001	-0.50 p ≤ .001	-0.31 p ≤ .001
Personal resiliencea	0.32 p ≤ .001	0.16 p = .001	0.39 p ≤ .001	0.11 p = .002
Caregiver/relational resiliencea	0.31 p ≤ .001	0.15 p ≤ .001	0.31 p ≤ .001	0.05 p = .140
PTSDa	-0.33 p ≤ .001	-0.53 p ≤ .001	-0.40 p ≤ .001	-0.42 p ≤ .001
Insomniaa	-0.31 p ≤ .001	-0.42 p ≤ .001	-0.36 p ≤ .001	-0.28 p ≤ .001

Correlations were considered significant at p < .05

^aSpearman rank

^bcorrelations and Mann-Whitney U test were used

Table 3.

Forward stepwise multiple linear regression models predicting PedsQL QoL domains

Variable		R 2	ΔR2	β	F	P value
Model 1 – Physical Functioning		0.25			43.02	≤ 0.001
	State anxiety			-0.19		<0.001
	PTSD		0.04	-0.21		<0.001
	Caregiver/relational resilience		0.03	0.17		<0.001
	Situational seeking social support		0.02	-0.19		<0.001
	Insomnia		0.01	-0.12		0.001
	Situational active coping		0.01	0.11		0.005
Model 2 – Emotional Functioning		0.50			131.10	≤ 0.001
	Trait anxiety			-0.24		<0.001
	PTSD		0.12	-0.28		<0.001
	State anxiety		0.03	-0.18		<0.001
	Situational emotion coping		0.01	-0.15		<0.001
	Insomnia		0.01	-0.16		<0.001
	Caregiver/relational resilience		0.01	-0.09		0.003
Model 3 – Social Functioning		0.33			64.24	≤ 0.001
	Trait anxiety			-0.12		0.004
	PTSD		0.06	-0.25		<0.001
	Personal resilience		0.04	0.18		<0.001
	State anxiety		0.01	-0.12		0.006
	Depression		0.01	-0.10		0.018
	Dispositional emotion coping		0.01	-0.07		0.032
Model 4 – School Functioning		0.26			39.68	<0.001
	PTSD			-0.29		<0.001
	Depression		0.03	-0.14		0.002
	Situational seeking social support		0.02	-0.14		<0.001
	Insomnia		0.01	-0.13		0.001
	Caregiver/relational resilience		0.01	-0.27		<0.001
	Personal resilience		0.01	0.20		0.002
	State anxiety		0.01	-0.11		0.009

R² = coefficient of determination; ΔR² = change in R² at each step; β = standardized regression coefficient; F = model F statistic

Regression analysis

Model 1: PedsQL – physical functioning

In the first model, QoL in the Physical Functioning domain was associated with six variables selected and entered sequentially using a forward stepwise procedure: state anxiety, PTSD, caregiver/relational resilience, situational seeking social support, insomnia, and situational active coping. The model was statistically significant (F(6, 774) = 43.02; p < .001) and explained 25% of the variance in overall PedsQL scores for this domain.

Model 2: PedsQL – emotional functioning

The second model showed that QoL in the Emotional Functioning domain was associated with six variables selected and entered sequentially using a forward stepwise procedure, in the following order: trait anxiety, PTSD, state anxiety, situational emotion-focused coping, insomnia, and caregiver/relational resilience. This model explained 50% of the variance (F(6, 774) = 131.10; p < .001).

Model 3: PedsQL – social functioning

The third model indicated that the Social Functioning domain of QoL was associated with six variables selected and entered sequentially using a forward stepwise procedure: trait anxiety, PTSD, personal resilience, state anxiety, depression, and dispositional emotion-focused coping. The model accounted for 33% of the variance (F(6, 774) = 64.24; p < .001).

Model 4: PedsQL – school functioning

The final model addressed the School Functioning domain. It was statistically significant (F(7, 773) = 39.68; p < .001) and explained 26% of the variance in PedsQL scores. The correlates selected and entered sequentially using a forward stepwise procedure were: PTSD, depression, situational seeking social support, insomnia, caregiver/relational resilience, personal resilience, and state anxiety.

Discussion

Children who remain in war-affected regions are exposed to chronic, overlapping stressors, including ongoing danger, disrupted schooling and family separation/dislocation ([8, 41]; UNICEF, [57]). Our findings extend this evidence by showing elevated symptoms of depression, anxiety, PTSD symptoms and insomnia in school-aged children who stayed in Ukraine, indicating a substantial symptom burden in this population [17, 34, 36, 39]. Additionally, with regard to posttraumatic stress, previous studies using the CRIES-8 indicate that mean symptom levels in trauma-exposed child populations often reach or exceed values considered clinically relevant; the severity observed in the present study fell within this range or exceeded that reported in other samples of trauma-affected youth [15, 45]. Insomnia assessed with the Athens Insomnia Scale, for which a cut-off score of ≥ 6 indicates probable insomnia, has been shown to reach near-clinical or clinical levels in populations exposed to chronic stress; levels observed in our sample were comparable to those reported in other studies of children and adolescents exposed to trauma or prolonged threat [51]. Similarly, anxiety assessed with the STAIC and depressive symptoms measured using the CDI-2 were comparable to those reported in several empirical studies of children exposed to heightened psychosocial stress, and consistent with findings from other non-clinical but trauma-exposed samples [1, 10]. Consistent with the study’s dual aim, our findings reveal that mental health difficulties represent a substantial burden among children exposed to war and are associated with their daily functioning across multiple domains of QoL. By focusing on both psychopathological symptoms and protective factors within multivariate models, we offer a portrait of how risk and resilience operate under conditions of chronic instability.

Domain-specific impacts of psychological correlates

Among the four QoL domains, emotional functioning was the most strongly affected, with trait anxiety and PTSD symptoms emerging as the main negative associations, consistent with earlier reports on war-exposed children showing links between internalizing symptoms and emotional dysregulation [7, 28]. Situational emotion-focused coping showed a small but significant inverse association with emotional functioning, in line with evidence that under chronic or uncontrollable stressors (war, protracted insecurity) emotion-focused strategies tend to co-occur with higher distress rather than buffer it. In such contexts children may turn to emotion-focused coping because the situation is not changeable, so the strategy is reactive/need-driven rather than resource-driven, and thus becomes a marker of higher burden. Caregiver/relational resilience also remained positively, though modestly, related to emotional functioning, suggesting that supportive relationships provide only partial buffering when anxiety and PTSD levels are high [19, 27, 29, 46].

Social functioning was the second most strongly explained QoL domain. Trait anxiety was the strongest single predictor, accounting for about one fifth of the variance. This underscores that heightened emotional arousal in unpredictable and threatening environments can impair children’s ability to initiate and sustain peer relationships. This pattern is consistent with studies showing that anxiety in war-affected children interferes with social engagement and may lead to withdrawal or relational hypervigilance [4, 50]. Adding PTSD and personal resilience clarified mechanisms of social functioning under chronic threat. PTSD was linked to poorer social functioning, likely via avoidance, emotional numbing and impaired trust [24] whereas higher personal resilience slightly improved social outcomes, supporting evidence that self-regulatory capacities can buffer social withdrawal in high-stress contexts [44].

Other psychological variables: state anxiety, depressive symptoms and a tendency to focus on emotions – also made small but significant contributions, indicating additional pathways to social withdrawal and reduced peer engagement. Such strategies may provide short-term relief but are insufficient to maintain social participation under chronic insecurity. Taken together, these results indicate that anxiety and trauma-related arousal undermine children’s capacity to maintain social bonds, whereas resilience offers only partial protection, which is consistent with models of stress-related social withdrawal in prolonged adversity [35, 44, 50].

School functioning was moderately explained by psychological symptoms and protective factors. The strongest predictor was PTSD, accounting for about 17% of the variance. This supports prior evidence that trauma-related distress disrupts concentration, academic persistence and school engagement in war-affected children [2, 22, 24]. Depressive symptoms and insomnia also contributed significantly, reflecting the cognitive and motivational impairments typical of mood and sleep dysregulation. Situational support seeking made a small but significant positive contribution, suggesting that mobilizing help from others can support school engagement under stress [24]. Caregiver/relational and personal resilience were linked to slightly better school outcomes, indicating partial buffering by internal and interpersonal resources. State anxiety, despite being included in the model, showed limited independent association with school functioning, likely due to collinearity with other internalizing symptoms.

Physical functioning had the lowest proportion of variance explained. State anxiety was the strongest predictor, followed by PTSD, indicating that somatic complaints and perceived physical limitations are linked to internal distress, consistent with psychosomatic pathways described in stressed children [6, 21]. Caregiver/relational resilience and situational active coping were associated with slightly better physical functioning, whereas situational support seeking showed a small but significant negative association, suggesting that in this context greater help-seeking co-occurred with poorer perceived functioning. These results align with previous findings suggesting that while physical functioning may not be directly impaired by psychological symptoms to the same extent as emotional or academic domains, it can still reflect the bodily burden of chronic threat exposure [6, 37]. In high-adversity settings, anxiety and trauma often somatize as fatigue, headaches or other discomforts, especially when routines and external regulation are missing. The modest effects of protective factors indicate that resilience and coping can buffer somatic responses only partially and do not eliminate the consequences of sustained war-related stress.

General correlates of QoL across domains

Looking across the four domains of QoL, the most consistent correlates of diminished functioning were internalizing symptoms, particularly anxiety (both trait and state), PTSD symptoms, and insomnia. This confirms that war-related distress manifests not only as clinical symptoms but also as disruption of learning, emotion regulation, social bonding and physical activity. Insomnia, in particular, emerged as a transdiagnostic contributor, with independent effects across several domains, supporting the view that it is not merely secondary but actively exacerbates emotional and cognitive vulnerability [59, 60].

While depressive symptoms were present and statistically significant in models associated with school and social functioning, their influence appeared less pronounced than that of anxiety and trauma-related symptoms. This contrasts somewhat with earlier studies emphasizing the broad functional impact of depression in war-affected children [20, 23, 41]. A likely explanation is methodological: earlier studies often used diagnostic interviews or clinical samples, whereas we used self-report in a community sample. In addition, paediatric depression may have a more protracted course, so its functional impact appears later, whereas anxiety and PTSD are more tightly coupled with immediate threat and therefore more salient in cross-sectional data. This underscores the need for longitudinal designs to capture symptom functioning dynamics under prolonged adversity.

Protective factors played a more limited, domain-specific role: caregiver/relational resilience was most helpful for school and social functioning, whereas personal resilience contributed more modestly, especially to social functioning. This pattern supports dynamic views of resilience as emerging from the interplay of internal regulation and external supports [44, 56]. Situational support seeking was linked to better academic functioning but to poorer physical functioning, highlighting that social mobilisation has domain-specific effects under persistent stress. By contrast, dispositional emotion-focused coping showed only marginal or inconsistent effects, pointing to the need for context-sensitive, culturally grounded coping interventions in conflict settings [3, 29, 35].

Studies on refugee and displaced Ukrainian samples have reported stronger links between resilience, coping and mental health, whereas our findings from children who remained in a setting of ongoing danger, unpredictability and infrastructural fragility – suggest that these mechanisms operate more weakly, supporting Chen’s [12] view that chronic instability can attenuate or even reverse their expected benefits.

Practical implications

This study suggests domain-specific intervention priorities for children remaining in war-affected settings. Emotional functioning, most strongly associated with trait anxiety, PTSD and insomnia, points to the need for trauma-informed approaches that integrate anxiety-regulation components with sleep-focused strategies. In the social domain, higher state anxiety and PTSD indicate that structured, supervised peer activities may help limit withdrawal and promote safe interactions. School functioning, adversely related to PTSD, depression and insomnia, but positively related to support seeking and caregiver/relational resilience, could be supported through interventions that normalise help-seeking and strengthen teacher-student relationships. Although resilience- and coping-related resources showed only modest, domain-bound effects, their purposeful reinforcement may still produce incremental gains under chronic instability. These implications should, however, be interpreted in light of the cross-sectional design. In practice, such measures can be delivered mainly by school-based psychologists and counsellors, in coordination with teachers and local community/NGO mental-health teams already operating in Ukrainian schools.

Limitations

Several limitations should be noted. First, the cross-sectional design precludes causal inference; lower QoL could also exacerbate symptoms or constrain coping, so longitudinal studies are needed to establish directionality. Second, exclusive reliance on self-report may have introduced bias (social desirability, recall, age-related limits), so future work should use multi-informant data (parents, teachers, clinicians). Third, data were collected in active wartime conditions in the Ternopil region, which may have affected participation and school-based procedures; therefore, generalisation to other Ukrainian regions or settings with different levels of conflict intensity should be made with caution. Fourth, the regression models did not include socio-demographic or contextual covariates, which may also influence QoL and should be examined in future studies. Finally, the regression models included many predictors, so individual βs should be read as conservative estimates, likely attenuated by shared variance.

Conclusion

This study describes psychological well-being and QoL in a large sample of children who remained in war-affected areas of Ukraine. Across emotional, social, school and physical domains, internalising symptoms, especially anxiety, PTSD and insomnia, were the variables most consistently associated with poorer functioning. Protective factors such as caregiver/relational resilience and support seeking showed smaller, domain-specific effects, but appeared to provide some buffering, particularly for school and social functioning. These patterns are consistent with a two-track approach in practice, in which symptom-focused care (anxiety, trauma, sleep) could be delivered alongside measures that enhance social and family supports.

Author contributions

Conceptualization: MW, IP, PKU; Methodology: MW, PKU; Formal analysis: PKU; Data curation: IP, PKU; Investigation: IP, AO, VN; Writing – original draft: MW, PKU, IP, AN; Writing – review & editing: all afuthors. All authors read and approved the final version of the manuscript.

Funding

This research received no specific grant from any funding agency, commercial or not-for-profit sectors.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of Lviv State University of Physical Culture (No. 16, February 13, 2024). Permissions were obtained from local education departments and participating schools. Written informed consent was obtained from all participants and their parents or legal guardians prior to participation. All procedures complied with the Declaration of Helsinki and institutional ethical standards.

Consent for publication

Not applicable, as no identifiable personal data are included.

Declaration of generative AI

Not applicable.

Competing interests

The authors declare no competing interests.