Reliability, validity and dimensionality of the GHQ-12 in South African populations: Structural equation modelling (SEM)

This is the authors' abstract. We don't add key points for this paper.

Introduction Health Care Workers (HCWs) were among the high-risk groups for SARS-CoV-2 infection and suffer a high burden of poor mental health including depression, anxiety, traumatic stress, avoidance and burnout. The 12-Item General Health Questionnaire (GHQ-12) has showed best fit in both a one-factor structure and a multidimensional structure for the screening of common mental disorders and psychiatric well-being. The aim was to test for the reliability and validity and ascertain the factor structure of the GHQ-12 in a South African HCW population. Methods Data was collected from 832 public hospital and clinic staff during the COVID-19 pandemic in Gauteng, South Africa. The factor structure of the GHQ12 in this professional population was examined by exploratory factor analysis (EFA) to identify factors, confirmatory factor analysis (CFA) for construct validity and structural equation modelling (SEM). Results The GHQ-12 median score was higher (25) in women than in men (24), p=0.044. The determinant for the correlation matrix was=0.047, the Barlett test of sphericity was p<0.001, Chi square=2086.9 and Kaiser-Meyer-Olkin (KMO) of sampling adequacy was 0.86. The four factors identified were labelled as Social-Dysfunction (37.8%), Anxiety-Depression (35.4%) Capable (24.9%) and Self-Efficacy (22.7%). The entire sample had a Cronbach’s alpha of 0.85, with 0.69 for Social-Dysfunction, 0.74 for Anxiety-Depression, 0.64 for Capable and 0.52 for Self-Efficacy in orthogonal (varimax) factor loadings. Conclusions The GHQ-12 tool displayed adequate reliability and validity in measuring psychological distress in a professional group with a four-factor model suggesting multidimensionality in this group rather than a unidimensional construct.