The factor structure of the general health questionnaire (GHQ12) in Saudi Arabia

Psychological health is a major component contributing to overall well-being. The World Health Organization (WHO) [1] defined mental health “as a state of well-being in which every person realizes his or her own potential, can cope with the normal stresses of life, can work productively and fruitfully, and is able to contribute to her or his community.” By tradition, healthcare clinicians can successfully assess the state of an individual’s well-being by gauging substance abuse, anxiety, distress, and depression [2]. Therefore, mental health is not merely an absence of declared negative issues, but it is defined as a state of complete physical, mental, and social well-being [3]. Several instruments are being applied to examine the general health of an individual, but one of the most extensively used self-reported questionnaire consists of the general health questionnaire (GHQ) formulated by Goldberg in the 1970s and noted for being a reliable measure of mental health. The GHQ helps to screen psychological disorders in primary healthcare and outpatient settings [4‐6] and has been utilized in cross-cultural settings [7, 8]. The GHQ has been used for both population based studies and health assessment surveys [9]. It is straightforward to administer, and can be completed in less than 10 min by a single participant [10].

The original GHQ is comprised of 60 items and is composed of a number of different versions such as GHQ-1, GHQ-12, GHQ-20, GHQ-28 and GHQ-30. From these listed versions, GHQ-12 is one of the most commonly used because of its ease of use [11, 12]. The GHQ-12 self-reported questionnaire includes 12 items, each of which is assessed through 4 indexes. Two of the most common types of scoring includes Likert scoring technique (0–1–2-3) and the bi-modal (0–0–1-1) [10]. Banks [13] suggested its usage for the GHQ-12 to compare levels of psychiatric impairment within and between samples. In different segments of the populations across countries, psychometric properties of this questionnaire in a variety of studies has also been appraised [14‐18]. The GHQ-12 tends to have good specificity, reliability, and reasonably high sensitivity [19, 20]. Since its development by Goldberg [9], the GHQ has now being used in many different countries/cultures and has been translated into 38 languages; a testament to the validity and reliability of the questionnaire [21‐25]. Other features that have not been fully explored on the GHQ-12 is the factor structure that underlies the instrument responses. Even though, this questionnaire was initially developed as a unidimensional scale, only limited studies were conducted using the GHQ-12 which provide empirical support to its one-factor latent structure [13, 26]. Alternatively, other multidimensional models such as 2 or 3 factors, are projected to be more suitable. As a result, a three-factor model, projected by Graetz in 1991, has much empirical support behind it [27‐31]. These three factors include GPH-1 (social dysfunction & anhedonia; 6 items), GPH-2 (depression & anxiety; 4 items), and GPH-3 (loss of confidence; 2 items). A typical argument in favor and against the use of these models relate to the unidimensional solution that has repeatedly demonstrated a higher correlation between these factors. For instance, the correlation between the 3 factors varied from 0.72–0.84 in Padrón et al. [29], from 0.76–0.89 in Campbell and Knowles [28], and from 0.83–0.90 in Gao et al. [31]. Conversely, another argument applied against the model proposed by Graetz’s demonstrated a low discriminant validity concerning the factor scores i.e. a derivative of this model.

Considering the fact that factor analysis of the GHQ-12 has yielded two or three factor solutions. Two factor structures in an Iranian study [10] were similar to findings reported in the WHO study on psychological disorders in general healthcare [32]. Another study in New Zealand [33] supported the two-factor structure property of the 12 items GHQ, whereas, in a Spanish population [34] a three-factor structure was shown demonstrating successful stress, self-esteem, and coping. Rajabi et al. [35] indicated that three and two factor models of GHQ, fitted the data better than the one-dimensional model. Similarly, Gao [36] found that Graetz 3- factor model including Anxiety and Depression, Social Dysfunction, and Loss of Confidence fit the data better than other models. The aforementioned studies show the sample was demographically diverse including people from different education levels, general populations, industrial workers, and 18-year-old youth.

Despite being common, mental illness is underdiagnosed by health professionals. The reported prevalence of psychiatric or mental health disorders in Saudi Arabia vary from study to study [37]. Several studies have used GHQ-12 in a community setting in Saudi Arabia [38] however, there appears to be no evidence on the usefulness of GHQ-12 factor or if they exist in revealing between-patient difference in health-related quality of life and clinical states. Even though there are 3 identified distinct factors, it is usually difficult to differentiate them in clinical practice because all these factors are somehow correlated, which has been addressed by Gao et al. [36]. The study indicated that three factor domains of GHQ-12 fails to provide any supplementary information on psychological functioning of people in accordance to the health-related quality of life and clinical variables as compared with one dimensional measure. Whereas, homogeneity of the study population and small sample size, primarily coming from the clinical cases, may limit the generalizability of the study results. An in-depth study of factor structures would be valuable [39] because certain psychological domains can be a focus for targeting interventions to prevent further psychological deterioration within the population. Furthermore, in Saudi Arabia, to date, only one study successfully assessed the psychometric properties of the Arabic version GHQ-12 in university students [20]. However, no exploratory factor analysis (EFA) has been performed to assess the GHQ-12 factor structure. Therefore, verification and assessment of GHQ-12 factor structure is essential. Considering the different outcomes of the GHQ-12 factor structures in previous studies, this study was designed with the main objective to assess the factorial structure of GHQ-12 in a large sample recruited from Al Kharj central region in Saudi Arabia.

To our best knowledge, this is the first study to assess the factorial structure and psychometric properties of the GHQ-12 in males and females using a study sample from Al Kharj central region. Exploratory Factor Analysis (EFA) was used to identify the potential factor structure of GHQ-12. However, EFA did not allow model assessment as identifications of factors are based on the arbitrary cut-points, for which the criteria for goodness-of-fit cannot be used. Li et al. [42] found similar findings in the context of the GHQ-12.

In our study, the PCA extracted 3 factors (Factor 1, Factor 2, and Factor 3) from the GHQ-12. Factor one was labeled as social dysfunction (items 1–6), factor two was labeled as anxiety (items 7–10) and factor 3 as loss of confidence (items 11–12); based on previous literature review and absolute value of 0.5 factor loading for each GHQ item represents the factor labeling. The findings from exploring the factor structure of GHQ-12 were similar to those of several other international versions [29, 43]. These 3 factors explained 55% of the overall variance. Our findings were similar to that of previous factor analytic studies of GHQ-12 that have found three-factor solution [20, 35, 44]. In conformity, Martin and Newell [45] in their factor analysis of Graetz three-factor solution utilizing likert scoring revealed three factors: Anxiety, social dysfunction, and loss of confidence. As shown in table three, the factor loadings were represented to be high in our study.

By studying a large sample of 6151 young Australians, aged 16–25 years in a longitudinal and cross-sectional analysis, Graetz [27] acknowledged that GHQ-12 structure included 3 distinct dimensions, anxiety/depression, low social function and loss of confidence. The corresponding elucidated variances were roughly about 37.0, 12.0 and 8.0%. The findings were supported and verified by several other studies. Some of the research studies for example, Martin [44], and Daradkeh [20] revealed that the GHQ-12 was 3-dimensional, nonetheless, it utilized wide-ranging names from those proposed and used by Graetz [27]. In our study, factors designated as anxiety, social dysfunction, and loss of confidence, were subject to variation possibly due to being more under the influence of the socio-cultural norms and values that govern the population under study. Our study was conducted in Al Kharj and not in other regions, generalizing our results to Saudi Arabia might be of question, however, the characteristics of the population of Al Kharj is comparable to that of all Saudi Arabia, which is mixed between urban and rural population. The association between GHQ scores and education was reported to be non-significant in our study which is consistent with the study conducted by Banks, 1982 [46] - except for intermediate class and loss of confidence. This was because of the opposite trends as people having higher level of education were more prone towards social dysfunction and anxiety, while being less prone to lack of confidence.

This study included respondents mostly within an age range of 18–44 age group, single, females, highly educated, and non-smokers. Our results showed that married participants had better mental health than individuals who were single. This finding corresponds with the findings of the Saki and Kaikhavani [47] study. On the other hand, Maleki et al. [48] demonstrated that mental disorders among women increase with age. Therefore, the variation between the mean total GHQ-12 score was considerably non-significant between age groups and other two factors, while it was statistically significant for social dysfunction. In our study, females were significantly associated with scores obtained for social dysfunction, but not for anxiety or loss of confidence factors as compared to males. In our study, females were comparably related to the presence of higher rates of negative mental health items.

Our results help us to confirm the factorial structure of GHQ-12 in a population based sample of Al Kharj, Saudi Arabia. This tool has been validated for several population thus, being translated into different languages. We conclude that in the Saudi citizens, (age range from 18 and above) from Al Kharj, Saudi Arabia, the GHQ-12 has shown very good psychometric characteristics and supports the scale applicability in this population. Furthermore, the study revealed that item factors were loaded into three main components using principal components analysis including social dysfunction, anxiety, and loss of confidence presenting an overall variance of 55%.