This cross-cultural study pooled data from 407 children and adolescents (ages 6–17) exposed to acute trauma, recruited from emergency departments in South Africa (n=222) and the UK (n=185). About one month post-trauma, it compared three narrative-based approaches (manual coding, LIWC word-count, NLP/TAACO) versus self-report questionnaires (TMQQ/ACTMQQ) for measuring trauma memory disorganisation and sensory qualities, and their ability to predict concurrent PTSD severity.
Self-reported trauma memory disorganisation and sensory vividness consistently predicted PTSD severity in both countries (SA: sensory β=0.35, f²=0.27; disorganisation β=0.22, f²=0.04; UK: sensory β=0.37, f²=0.24; disorganisation β=0.29, f²=0.11). By contrast, all narrative-based indices (manual coding, LIWC, NLP) failed to predict PTSD in either sample and showed no meaningful convergence with self-report measures. Self-report predicted PTSD across all UK age groups but only in adolescents ≥12 years in the South African sample.
- isiXhosa narratives were translated into English before analysis, potentially distorting the linguistic features that NLP and LIWC tools are designed to detect. - Cross-sectional design limits conclusions about long-term PTSD trajectories; longitudinal replication is needed. - South African participants' broader cumulative trauma histories were not captured, which may have influenced memory processing and self-report validity in younger children.
Brief self-report tools (e.g., TMQQ/ACTMQQ) are reliable, cross-culturally valid screens for PTSD risk in trauma-exposed youth and should be preferred over narrative coding in clinical and research settings. Clinicians in LMIC contexts should note that self-report may underestimate PTSD risk in children under 12, where culturally adapted, developmentally tailored tools are still needed.
Explore related topics
