A Russian complex-PTSD dataset, and the dissociation it never measured
- Cross-sectional online survey, snowball recruitment, 2024 to 2025: 614 adults aged 18 to 77 (mean 34, SD 12), of whom 205 lived in Berdyansk, Donetsk or Mariupol and 409 did not. Women made up 89% of the sample.
- k-means clustering on the International Trauma Questionnaire scales plus post-traumatic growth scores produced four profiles: dysfunctional 115 people, normative 126, transformational 223, dialectical 150.
- The share in the dysfunctional profile was almost identical across the exposure split: 38 of 205 directly involved (18.54%) and 77 of 409 indirectly involved (18.83%). The difference sat in the dialectical profile: 64 of 205 (31.22%) against 86 of 409 (21.03%); the overall association reached significance, chi-square = 8.74, df = 3, p = 0.033.
- Age separated the profiles more sharply than exposure did: 74 of 267 respondents aged 30 and under (27.72%) fell into the dysfunctional profile against 41 of 347 aged 31 and over (11.82%); chi-square = 42.07, df = 3, p < 0.001. The design is cross-sectional and self-report, so it ranks groups rather than showing that age protects anyone.
The International Trauma Questionnaire is the instrument that operationalises the ICD-11 diagnosis of complex PTSD. A team from Moscow State University of Psychology and Education has now run it on 614 adults, a third of whom live in Berdyansk, Donetsk or Mariupol. The paper is worth reading for what it counts, and worth reading twice for what it does not.
Who filled in what, and who administered it
Collection ran online through web forms by snowball sampling across 2024 and 2025. The 614 respondents were aged 18 to 77, mean 34, SD 12. Of them 205 lived in Berdyansk, Donetsk or Mariupol and were classified as directly involved in the conflict; 409 were classified as indirectly involved. Women made up 89% of the sample, 84.9% in the direct group and 91.7% in the indirect group. The two groups did not differ in age, t(612) = 1.85, p = 0.06.
Everyone completed the Russian-language version of the International Trauma Questionnaire, adapted and validated by Maria Padun, Yulia Bykhovets, Nadezhda Kazymova and Yulia Chentsova-Dutton in 2022 on a non-clinical sample. Alongside it came a post-traumatic growth inventory, a hardiness test, an identity resilience scale, an authenticity questionnaire and COPE-30. No clinician saw any of these people. There was no clinical interview, no threshold applied by an assessor and no diagnosis. That is worth holding onto, because the resulting percentages will be quoted without it.
k-means clustering on the questionnaire scales (PTSD, disturbances in self-organisation, complex PTSD) together with the growth scores split the sample into four profiles. The authors named them dysfunctional (115 people), normative (126), transformational (223) and dialectical (150). Written by Maria Odintsova, Natalia Stovbun and Nataly Radchikova of Moscow State University of Psychology and Education with Veronika Grebenshchikova of Azov State Pedagogical University named after P.D. Osipenko, the paper is open access and its dataset is deposited.
Proximity was not the dividing line
The share falling into the dysfunctional profile was almost identical on both sides of the exposure split: 38 of 205 directly involved (18.54%) and 77 of 409 indirectly involved (18.83%). The overall association between exposure and profile did reach significance, chi-square = 8.74, df = 3, p = 0.033, but a different cell carried it. The dialectical profile, in which high growth scores sit alongside high PTSD, disturbance-of-self and complex-PTSD scores, held 64 of 205 directly involved (31.22%) against 86 of 409 indirectly involved (21.03%).
Age separated the profiles far more sharply. Among the 267 respondents aged 30 and under, 74 (27.72%) fell into the dysfunctional profile; among the 347 aged 31 and over, 41 (11.82%) did. Chi-square = 42.07, df = 3, p < 0.001. That is a cross-sectional ranking of groups, not evidence that age protects anyone.
One result deserves a clinician's attention. 130 respondents could not name an external event at all and wrote instead about "losing myself", "emptiness inside myself", "a sense of otherness". Of those 130, 35 landed in the dysfunctional profile. Read that figure carefully, because the paper's own text does not. The 26.92% printed in the table is 35 of the 130 who named no external event; the running text describes it instead as the share of the dysfunctional profile, which would be 35 of 115, or 30.43%. The table is the honest document here, and this is the largest share the dysfunctional profile drew from any trauma category. The dysfunctional group also scored lowest on hardiness, 30.2 ± 13.1 against 53.4 ± 9.4 in the transformational group, F = 133.4, p = 0.0001.
The items that are not in the questionnaire
ICD-11 complex PTSD is PTSD plus three clusters of disturbance in self-organisation: affect dysregulation, negative self-concept and relational difficulties. Depersonalisation, derealisation and dissociative amnesia are not among them, and the International Trauma Questionnaire carries no items that ask about them. So this dataset, carefully built and openly reported, says nothing about how common dissociative symptoms are in a war-exposed Russian population. Nor will any other dataset assembled the same way.
That is the practical point for a reader who will be shown a percentage this year. When a colleague quotes a complex-PTSD figure from a Russian sample, the instrument is almost certainly the International Trauma Questionnaire, the administration is almost certainly self-report, and dissociation is almost certainly absent from the measurement. None of that makes the 115 people in the dysfunctional profile less real. They reported low hardiness, low self-esteem and low awareness, and they were disproportionately young and disproportionately unable to say what had happened to them. A screening score is not a diagnosis. It is also not nothing.
In practice this means one thing. If you need to know whether a trauma patient dissociates, the trauma questionnaire in front of you will probably not tell you. Ask in the session, directly: about lost time, about watching yourself from outside, about the room going flat and unreal. Then decide what the answer changes.
A questionnaire with no items on depersonalisation cannot tell you how common depersonalisation is.
The sample was an online convenience sample recruited by snowball, 89% of it women, and the authors note that they recorded neither the severity of the traumatic event nor the subjective emotional experience attached to it. The four profiles are a k-means partition of this particular sample rather than diagnostic categories, and the cross-sectional design cannot show how anyone moved between them.