The research was performed following the ethical standards of the 1964 Declaration of Helsinki and was approved by the Ethical Committee of the University Hospital of Verona on 24/10/2022 (protocol number 63939).

This is a cross-sectional validation study. It was carried out in the province of Verona. The reception system in Italy for Ukrainian refugees is built on two different services provided by the governmental authorities, under the Home Office: the Reception and Integration System (RIS), managed at the local level and the Special Reception Centres (SRC), centrally managed [12]. Alongside these systems is the extended network of reception consisting of nonprofit organizations, social service centers, religious organizations, and co-housing measures with families or accommodation provided by other private entities. In Verona, the reception network supporting Ukrainian refugees is coordinated among all 98 municipalities in the province and includes about 117 SRC and four projects related to the RIS [13, 14]. As of April 2023, the number of Ukrainian refugees in the province of Verona reached 2265, of whom 1623 (71.7%) were females [15].

All persons who arrived in Italy from Ukraine after 24 February 2022, following the outbreak of the Russian-Ukrainian conflict, were considered eligible for this study. Refugees older than 14 years old whose native language was Ukrainian were included.

According to Mundfrom et colleagues [16] considering a ratio of variables to factors (p/f) of 6 and a two-factor solution, as in the original questionnaire [17], in a level of communality set as low, the minimum sample size to obtain an excellent-level criterion (0.98) was 120. Accounting for a drop-out rate of 15%, the target sample of participants was set at 146 for this study.

Data was collected between November and February 2023, progressively including all persons meeting the inclusion criteria until the computed sample size was reached.

Ukrainian refugees were recruited in the province of Verona through the local refugee reception network (i.e., regional and local authorities, SRC, RIS, and non-profit organizations).

A written disclosure about the study was first given and those who agreed to participate signed an informed consent form. Both documents were written in Ukrainian, the participants’ mother language. For those under the age of 18, informed consent was signed by their parents or legal guardian.

Each participant was asked to complete the Ukrainian translation of the GHQ-12 together with a short sociodemographic questionnaire (i.e., age, sex, education level, and marital status) and the subscale for PTSD of the International Trauma Questionnaire (ITQ) to serve as external validation. At all phases of the study, the research team was supported by a cultural mediator.

The original GHQ-12 consists of 12 items to be answered by the participant according to the variation, compared to his or her habitual standard, in the frequency of scenarios or behaviors described in the specific statement of the items (Table 1). The GHQ-12 has 6 positive items (answers options: “Better than usual”, “Same as usual”, “Less than usual”, “Much less than usual”) and 6 negative items (answers options: “Not at all”, “No more than usual”, “Rather more than usual”, “Much more than usual”).

In the present study, both scoring methods, bimodal and Likert, were evaluated. In the bimodal scoring method, the response categories have a score of 0, 0, 1, 1 for the positive items, while the negative items are scored the other way round (1,1,0,0). Therefore, the score ranges from 0 to 12 points. In the Likert scoring method, the positive items scored from 0 to 3 and the negative ones from 3 to 0, with a score range between 0 and 36 [18]. The most used cut-offs are between 2 and 4 for the bimodal method and ranged between 10 and 15 for the Likert one [18].

The ITQ is a self-report measure that allows a simple and concise assessment of key aspects of PTSD, according to the ICD-11 diagnostic criteria. The ITQ has two main subscales: the first (9 items), concerns PTSD and assesses three symptom domains, namely re-experiencing, avoidance, and sense of threat; the second (9 items), used to assess the complex PTSD, investigates the symptoms of self-organization disorder and the functional impairment caused by them. Each item is answered on a Likert scale from 0 (not at all) to 4 (very much). The cut-off for PTSD is given by a score > 2 in at least one of the two items of each of the three symptom domains (re-experiencing, items 1 and 2; avoidance, items 3 and 4; hyperarousal, items 5 and 6) plus at least one of the three indicators of functional impairment (items 7, 8 and 9). The ITQ is available in the Ukrainian language-validated version [19].

The PTSD subscale was used in the present study. Previous studies have analyzed psychological distress by combining the PTSD symptom score from the ITQ and the mental health problem risk score from the GHQ-12 to test the links between mental health, well-being, and conflict exposure [20].

The translation process followed the WHO guidelines, which include a forward translation into the target language, i.e. Ukrainian, followed by a backward translation into the original language, i.e., English (Fig. 1) [21].

Flowchart of the translation, pilot test and validation process of the Ukrainian translation of the General Health Questionnaire 12 (GHQ-12) adopted in the present study

After obtaining permission from the Author to translate and the license to use the questionnaire, a professional translator provided the first Ukrainian version of the GHQ-12 from the original English questionnaire. This version was then revised with a third party fluent in both languages. The back-translation was carried out independently by a second professional translator who had not seen the original questionnaire in English. Both the authors and a third person reviewed the translation and revised it consensually. To avoid any conceptual losses during the translation process, the consensual retranslation was then compared with the original GHQ-12.

The translated questionnaire was initially administered to a sample of 28 refugees to test the acceptability and comprehensibility of the Ukrainian version. After completing the questionnaire, a cognitive interview was conducted to assess the clarity of the questions, any problems or difficulties in answering, and possible improvement actions. The pilot-sample was recruited based on sociodemographic criteria in order to be representative of both genders and different age groups (adolescents, adults, and elderly). Refugees who participated in the pre-test were not included in the final study sample.

The original English GHQ-12 and the Ukrainian GHQ-12 are available in the Supplementary material.

A descriptive statistic was first conducted on sociodemographic data using frequencies and proportions for categorical variables and means and standard deviations (SD) or medians and interquartile ranges (IQRs) for continuous ones. Sample distribution was tested via χ2 and Fisher exact test or Mann-Whitney-U non-parametric, as appropriate.

GHQ-12 internal consistency was assessed through Cronbach’s alpha and McDonald’s omega coefficient testing the reliability and considering satisfactory a coefficient greater than 0.70. A tetrachoric correlation matrix was generated to assess the correlation between all the items of the GHQ-12 scored with a bimodal method.

A confirmatory factor analysis (CFA) was carried out to examine the factor structure of the Ukrainian version of the GHQ-12. First, a single-factor structure that contained all the GHQ-12 items was assessed. Secondly, a two-factor structure was tested encompassing two correlated latent factors: “Anxiety/Depression” (items: q1, q3, q4, q7, q8, q12) and “Social Dysfunction” (items: q2, q5, q6, q9, q10, q11). The two-factor structure was the one suggested by the author of the original English version of the GHQ-12 [16].

The models were tested for both the scoring method; for the bimodal method, the diagonally weighted least squares estimator was used and all variables were considered as ordered (ordinal) variables, for the Likert method, the maximum likelihood estimator was used with the Satorra-Bentler adjustment accounting for non-normality and heteroscedasticity of the data [22]. Model fit was evaluated using the χ2 test, the comparative fit index (CFI), the Tucker-Lewis index (TLI), the root-mean square error of approximation (RMSE), and the standardized root-mean-square residual (SRMR). Variance explained by latent variables was assessed through Average Variance Extracted (AVE). Criteria for acceptable model fit indices were based on Hooper et al. [23].

Pearson product moment statistic (Pearson’s correlation coefficient = “ρ”) was used to assess the concurrent validity of the GHQ-12 as the correlation with the ITQ subscale for PTSD. It was expected that the GHQ-12 would positively correlate with the ITQ subscale. A coefficient “ρ” above 0.40 was considered satisfactory. Association between single item score of the GHQ-12 and being screened positive for PTSD at the ITQ was conducted via z-test and t-test for bimodal and Likert scoring methods, respectively.

A p-value < 0.05 was considered significant. All analyses were performed using the R software (version 4.3.0).