The Empty Cell Was Built Into the Questionnaire: Dissociative PTSD and Dissociative Complex PTSD in 57,984 Chinese Adolescents

A group at Deyang People's Hospital, Guizhou Normal University and the Chinese Academy of Sciences re-analysed a database of 57,984 trauma-exposed adolescents to ask whether the dissociative subtype of ICD-11 complex PTSD is the same population as the long-established dissociative subtype of DSM-5 PTSD. The answer they report is near-total overlap: Kappa = 0.90, and zero cases of the complex variant outside the DSM-5 variant. The interesting question is how much of that overlap the questionnaire produced before anyone was screened.

Where the agreement comes from

Everything here was measured with the Global Psychotrauma Screen for Teenagers, a self-report instrument completed by adolescents, with no clinician-administered interview at any point. Its PTSD subscale is five yes-or-no items with a cut-off of 3. Its complex PTSD subscale is seven yes-or-no items – the same five, plus two on disturbances in self-organisation – with a cut-off of 5. Dissociation is two yes-or-no items, depersonalisation and derealisation, with a cut-off of 1, and the same two items feed both subtypes.

Work the arithmetic. To reach 5 on the complex subscale, an adolescent can draw at most 2 points from the self-organisation items, so at least 3 must come from the PTSD items. Three is the PTSD cut-off. A positive complex screen therefore cannot fail to be a positive PTSD screen, and since the dissociation gate is identical, a positive D-CPTSD screen cannot fail to be a positive D-PTSD screen. The 0 in Table 1 is an arithmetic property of the scoring rule. The Kappa of 0.90 is inflated by the same nesting.

We checked the published figures rather than assuming them. The 2×2 table sums exactly to 57,984 (53,111 + 0 + 856 + 4,017); all four percentages, both prevalence estimates and all four comorbidity confidence intervals reproduce from the tabled counts; Kappa recomputes to 0.896; the reported z values of 9.45, 5.14 and 4.42 and the effect sizes of 0.056, 0.11 and 0.094 all reproduce. Nothing is arithmetically incompatible. Where the running text and the tables could be read differently we followed the tables and the reported test statistics – the one-decimal functioning means of 4.6 and 4.9 imply a gap of 0.3, whereas d = 0.099 and t = 4.62 correspond to an unrounded gap nearer 0.24, so the effect size is the figure to quote, not the subtraction.

What this changes at the desk

The clinical claim that survives is modest and worth having. Among adolescents who screen positive for trauma-related dissociation, those who also endorse self-organisation disturbance report worse functioning and carry more anxiety and depression. But the magnitudes are small – d = 0.099, h = 0.11, h = 0.094, every one of them below the conventional threshold for a small effect – and the comparison is between a subset and the superset containing it, tested as though the two were independent samples. The 4,017 complex cases are 4,017 of the 4,873 PTSD cases. Only 856 adolescents are unique to one side.

Two habits follow. First, when a study reports that two diagnostic constructs agree almost perfectly, check whether the instrument scores them from shared items before treating the agreement as evidence about the disorders. Nested subscales manufacture concordance. Second, do not carry 8.4% into a case discussion as a prevalence of the dissociative subtype of PTSD in adolescents. It is the proportion of a Chinese pandemic-era sample that endorsed at least 3 of 5 screening items and at least 1 of 2 dissociation items on a self-report form whose Chinese version has not been systematically validated. A patient who says yes to one question about feeling unreal has told you something worth asking about. They have not told you they have a dissociative subtype.