War exposure, post-traumatic stress symptoms and hair cortisol concentrations in Syrian refugee children

The study investigates how war-related trauma and current living conditions relate to stress physiology and mental health in Syrian refugee minors. The central questions are: (1) Is exposure to war-related events associated with alterations in hair cortisol concentrations (HCC) in children and adolescents? (2) Do current refugee living conditions relate to HCC? (3) Are PTSD symptoms associated with HCC, and does HCC mediate links between war exposure and PTSD? The context is the large global population of displaced youth exposed to conflict, with cortisol proposed as a biomarker of adversity but with inconsistent findings. Understanding these links can clarify biological signatures of trauma and inform mental health risk in refugee children and adolescents.

Prior work in adults often shows positive associations between war exposure and cumulative cortisol measures (hair, nails), though effects can vary with timing, type of trauma, and confounders. Some studies find elevated HCC shortly after trauma but lowered levels with distant or repeated exposures. Evidence on refugees shows mixed results; factors like displacement phase and ongoing insecurity matter. In adolescents, several studies reported no direct association between war exposure and HCC unless PTSD symptoms are present; some found elevated HCC with PTSD, while others found lower HCC with cumulative lifetime trauma in refugee minors. Few studies focus on children or those still living in camps, and samples have often been small. Developmental stage and age of trauma onset may influence HPA-axis responses, highlighting a need for large, developmentally diverse cohorts.

Design: Longitudinal observational study with two waves one year apart (late 2017–early 2019) using an accelerated longitudinal design. Setting: Informal tented settlements in the Beqaa region of Lebanon. Participants: 1591 Syrian refugee child–caregiver dyads (child target age 8–16 at enrolment; some outside range retained due to birthdate uncertainty). Eligibility: Left Syria within previous four years; child in target age range; primary caregiver available. Follow-up: 1000 dyads followed at Year 2; reasons for non-follow-up included relocation or inability to contact. Ethics: Approved by University of Balamand/Saint George Hospital UMC IRB and Lebanese National Consultative Committee on Ethics; caregiver consent and child assent obtained. Measures: Sociodemographics (age, sex, nationality, smoking), time since leaving Syria (categorical 12-month intervals), computed age at end of war exposure (age minus time-since-leaving midpoint), pubertal stage, BMI, recent illness, endocrinological illness/medication. War exposure: War Events Questionnaire (WEQ), 25-item checklist; events counted if reported by child or caregiver; main score = number of different war-related events; additional scores for life-threatening events and categorization of most severe event type. Current living conditions: Perceived Refugee Environment Index (PREI), caregiver-reported multidimensional scale (livelihood, basic needs, housing, services, family/community environment, work, learning), items rated 1–5; total score used (excluding future mobility), with key subscales examined (livelihood, basic needs, housing). PTSD symptoms: Child PTSD Symptom Scale (CPSS), adapted linguistically; validated in a cohort subsample. Hair cortisol: Hair sampled from posterior vertex; 0–2 cm segments assayed via ELISA; normalized to ng/g; log10-transformed for analyses. Hair washing frequency, hair alterations (dye/henna/chemical), and month recorded. Statistical analysis: Initial associations via Mann–Whitney U, Kruskal–Wallis, and correlations. Primary models: Linear mixed-effects models (lme4) with log10 HCC as dependent variable; fixed effects included war exposure, time since leaving Syria, PREI, and PTSD; batch as fixed factor; participant as random intercept. Minimal adequate models retained covariates significantly contributing: age, sex, pubertal stage, smoking, hair colour; month/hair factors considered. Mediation analyses (mediation R package) tested HCC and PTSD as mediators using Year 1 data with 5000 bootstraps; models included batch and sex. Sensitivity analyses excluded atypical hair samples (e.g., coloured extracts), short samples (<2 cm), low weight (<20 mg), and participants with endocrine illness/medication; BMI z-score controlled in a subset.

• Sample: HCC available for 1574 children at Year 1 and 923 at Year 2; age 6–19; 47.4% male at Y1; within-individual HCC correlation across waves r=0.42, p<0.001.
• War exposure association: Small positive association between number of war-related events and HCC. In batch-only model B=0.006 (SE=0.002), p=0.002 (~1.3% per additional event). In full model including covariates effect attenuated (B=0.002, p=0.232). Excluding age from covariates yielded B=0.004, p=0.010 (~1.0% per event), implying age is a key confounder.
• Event type: Indications that life-threatening events (to self/others) might relate more strongly to HCC; each additional observed life-threatening event associated with up to 2.3% higher HCC before full adjustment; effects largely confounded by age.
• Time since leaving Syria: Compared to 0–12 months, HCC was lower for 12–24 months (−12.2%), 36–48 months (−11.5%), and 48+ months (−15.3%) in full models; 24–36 months not significantly different. Pattern persisted when excluding age (e.g., 48+ months ~−18.1%). No interaction pattern indicating differential war-exposure–HCC slopes by time-since group.
• Current living conditions: PREI total and key subscales (livelihood, basic needs, housing) were not associated with HCC in mixed models; results unchanged when adjusting for war exposure.
• PTSD symptoms: PTSD score positively associated with HCC in full model (B=0.001, SE=0.001, p=0.041), ~0.3% increase per point; corresponds to ~17.7% higher HCC from minimum to maximum PTSD scores. Excluding age increased effect slightly (B=0.002, p=0.003; ~0.4% per point).
• Age at end of exposure: Stratified models showed the association between war exposure and HCC was concentrated in those aged ≥12 at the end of war exposure (~2.2% increase in HCC per additional event), while younger groups showed null associations. The war exposure–PTSD link was also stronger in the ≥12 group; PTSD–HCC association was not significant within individual age strata.
• Mediation (Year 1): Partial mediation observed. PTSD partially mediated war exposure→HCC (ACME=0.002, 95% CI 0.0007–0.0031; proportion mediated ~22.9%). HCC partially mediated war exposure→PTSD (ACME=0.018, 95% CI 0.006–0.033; proportion mediated ~2.7%).
• Covariates: Females had substantially higher HCC than males (e.g., ~73% higher in full models); HCC increased with age and pubertal stage; smoking associated with lower HCC; brown hair colour associated with slightly lower HCC versus black.

The findings indicate that exposure to a greater number of war-related events is associated with modest elevations in hair cortisol among Syrian refugee minors, aligning more closely with adult literature than with several prior adolescent studies. The observed attenuation by age highlights developmental confounding: older participants both experienced more events and had higher HCC. When considering developmental timing, associations were most evident among youth who were at least 12 years old at the end of exposure, supporting adolescence as a sensitive period for HPA-axis alterations following trauma. Decreasing HCC with greater time since leaving Syria suggests that elevated activity may diminish over time, consistent with HCC sensitivity to more recent stressors or a transition from hyper- to hypo-secretion post-trauma. Contrary to expectations, caregiver-reported current living conditions did not relate to HCC; this may reflect measurement limitations (caregiver perceptions vs. child experience), restricted variability, or overshadowing by past trauma effects. The positive association between PTSD symptoms and HCC supports a link between current trauma-related symptomatology and elevated basal cortisol output, though effects are small relative to demographic factors. Mediation analyses suggest bidirectional partial pathways: PTSD may partly account for the link between war exposure and HCC, and HCC may modestly mediate the link between exposure and PTSD, though causality cannot be inferred from cross-sectional mediation. Overall, results underscore limited utility of HCC as a standalone biomarker given small effect sizes and strong demographic influences, but they point to biologically meaningful, developmentally contingent associations between war exposure, PTSD burden, and HPA-axis activity.

This large, longitudinal study of Syrian refugee children and adolescents shows that war-related experiences and current PTSD symptoms are associated with elevated hair cortisol concentrations, with effects most evident among those exposed during early adolescence and diminishing with time since displacement. The work highlights adolescence as a potential sensitive window for trauma-related HPA-axis alterations and suggests that while HCC reflects biological embedding of war trauma, its predictive value is modest and strongly influenced by age and sex. Future research should: include true unexposed control groups; capture event chronicity and recency; incorporate child-reported proximal stressors and caregiver factors; examine trajectories to determine persistence or transition of hyper/hypo-secretion; and assess additional moderators (e.g., physical activity) to clarify mechanisms linking trauma, HPA-axis function, and mental health outcomes.

• Retrospective recall of war-related events by child and caregiver, with potential recall bias and limited detail on event chronicity/repetition.
• Temporal ambiguity for causality: HCC reflects recent months, complicating inference about direct long-term effects of war exposure versus mediation by current factors.
• Current living conditions measured via caregiver report may not represent the child’s lived experience; other unmeasured mediators (e.g., parent mental health, caregiving quality, physical activity) may influence HCC.
• Lack of a true unexposed control group; even those with zero reported war events experienced forced displacement and camp-related stressors.
• Small effect sizes and strong associations with demographics (age, sex) limit utility of HCC as a predictive biomarker without careful covariate control.
• Potential unmeasured confounding despite extensive covariate adjustment (e.g., objective physical activity).
• Some follow-up attrition and sample exclusions (e.g., outliers, hair sample constraints), though sensitivity analyses supported robustness.