Evaluating the global psychotrauma screen total score using item response theory in an Italian community sample.

Background

The Global Psychotrauma Screen (GPS) is a brief transdiagnostic screener for trauma-related symptoms. Although the Italian version has previously been examined using factor-analytic methods, evidence on item-level functioning and on the interpretability of the symptom total remains limited.

Objective

To determine whether the 17 GPS symptom items support the use of a single total score as an efficient indicator of global trauma-related burden.Method: Data from an age- and gender-balanced analytic sample of 6930 Italian adults recruited online during the COVID-19 outbreak were analysed. Essential unidimensionality was examined using parallel analysis and bifactor indices (ECVg and ωH). One- and two-parameter logistic IRT models were compared, and item fit, local dependence, test information, conditional reliability, and sensitivity to aberrant responding were evaluated.

Results

Bifactor indices supported essential unidimensionality (ECVg = 0.73; ωH = 0.77). The 2PL model outperformed the 1PL model and showed acceptable global fit (RMSEA = 0.058, SRMSR = 0.04, TLI = 0.94, CFI = 0.94). Item discrimination ranged from 0.68 to 3.03, and item difficulty ranged from -0.95 to 2.71. The test information function peaked at θ = -0.24 and indicated useful precision across a broad range of moderate-to-high trauma burden. Item parameters remained stable after removing of respondents with aberrant response patterns.

Conclusions

In this community sample, the Italian GPS symptom total can be interpreted as a psychometrically defensible indicator of global trauma-related burden. The scale was especially informative from slightly below-average to elevated levels of severity, supporting its use as a brief first-line screening measure while still requiring replication in independent and clinical samples.