Abstract

This study aimed to gather evidence on the adequacy of the General Health Questionnaire (GHQ-12) in Brazil, considering a random sample of Brazilian physicians. Specifically aimed: (1) to test the GHQ-12 bifactor structure compared to alternative models, (2) to check its factorial invariance regarding to gender and the diagnosis of mental and behavioral disorders, and (3) to know the association of this measure with indicators of poor health (e.g., suicidal thoughts, decreased libido, medication use). The study included 1,085 physicians with a mean age of 45.7 (SD = 10.6), mostly male (61.5%), married (72.6%) and Catholic (59.2%). They answered the GHQ-12, the Positive and Negative Suicidal Ideation Inventory, and demographic questions. The best fit model was the bifactor structure composed of anxiety and depression, in addition to a general dimension, which presented Cronbach’s alpha, McDonald’s ω and composite reliability higher than 0.70 just for a general fact. Psychological distress scores correlated with suicidal ideation and indicators of health and sexual satisfaction. This is a psychometrically suitable instrument that can be used in terms of its total, but its specific factors need to be used with caution.

Key words:
Physicians; Health; Psychological distress; Anxiety; Depression

Resumo

Este estudo teve como objetivo reunir evidências sobre a adequação do Questionário de Saúde Geral (QSG-12) no Brasil, considerando uma amostra aleatória de médicos brasileiros. Especificamente, objetivou-se: (1) testar uma estrutura bifactor em comparação com modelos alternativos, (2) verificar invariância fatorial em relação ao gênero e ao diagnóstico de transtornos mentais e comportamentais, e (3) conhecer a associação com indicadores de saúde precária (por exemplo: pensamentos suicidas, diminuição da libido e uso de medicamentos). Participaram do estudo 1.085 médicos, com média de idade de 45,7 (DP = 10,6), maioria de sexo masculino (61,5%), casados ​​(72,6%) e católicos (59,2%). Eles responderam ao QSG-12, ao Inventário de Ideação Suicida Positiva e Negativa e a questões demográficas. O modelo de melhor ajuste foi a estrutura bifactor (composta por ansiedade, depressão e uma dimensão geral), que apresentou alfa de Cronbach, ômega de McDonald e confiabilidade composta superior a 0,70 apenas para o fato geral. Pontuações de sofrimento psicológico se correlacionaram a ideação suicida e indicadores negativos de saúde e satisfação sexual. O instrumento foi psicometricamente adequado e pode ser utilizado em termos de sua pontuação total, mas seus fatores específicos precisam ser utilizados com cautela.

Palavras-chave:
Médicos; Saúde; Sofrimento psíquico; Ansiedade; Depressão

Introduction

The General Health Questionnaire (GHQ) is a response to a recurring problem in health care, the correct diagnosis of mental disorders enabling immediate and appropriate treatment. The full version of this self-administered measure, containing 60 items, was proposed by Goldberg in 1972¹1 Goldberg DP. The detection of psychiatric illness by questionnaire. London: Oxford University Press; 1972.. Shorter versions were later published, including the 12-item version (GHQ-12), which is the object of interest in this study²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384.. This shorter version is easy to administer and contains evenly distributed items written as positive (e.g., “Have you felt capable of making decisions?”) and negative (e.g., “Have you been thinking of yourself as a worthless person?”)²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384.. Thus, some studies have investigated the psychometric adequacy of GHQ-12 at the national and international levels, considering the general population, young people, elderly people, and groups of professionals.

In Brazil, some studies have already been conducted on this measure. For example, authors³3 Borges LO, Argolo JCT. Adaptação e validação de uma escala de bem-estar psicológico para uso em estudos ocupacionais. Avali Psicol 2002; 1(1):17-27. analyzed a sample of 446 people (158 unemployed and 288 employed) and reported exploratory factor analysis of two items: self-efficacy (α = 0.85), and depression and emotional exhaustion (α = 0.75). Gouveia et al.⁴4 Gouveia VV, Chaves SSS, Oliveira IP, Dias MR, Gouveia RSV, Andrade PR. A utilização do QSG-12 na população geral: estudo de sua validade de construto. Psic Teor Pesq 2003; 19(3):241-248. examined a general population sample of 306 people with confirmatory factor analysis. Their findings showed that a bifactor model was more appropriate [depression (α = 0.81) and anxiety (α = 0.66)] than a unifactorial model (psychological distress, α = 0.84). Gouveia, Barbosa, and Andrade⁵5 Gouveia VV, Barbosa GA, Andrade EO, Carneiro MB. Factorial validity and reliability of the General Health Questionnaire (GHQ-12) in the Brazilian physician population. Cad Saude Publica 2010; 26(7):1439-1445. analyzed a sample of 7,512 physicians and tested their factorial structure with one and two factors [depression (α = 0.85) and social dysfunction (α = 0.82)]. They reported that the latter emerged in exploratory factor analysis and showed the best fit indicators in confirmatory factor analysis. Nevertheless, Gouveia and colleagues²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384. analyzed a sample of 1,180 people (university students, elementary school teachers, military police officers, and people from the general population) and, controlling the effect of negative items, observed that a general factor structure could be more appropriate, which they named psychological distress (M _α = 0.84).

At the international level, over a hundred studies have been conducted to explore or prove the GHQ-12 factorial structure⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194.^,77 Guan M, Han B. Factor structures of General Health Questionnaire-12 within the number of kins among the rural residents in China. Front Psychol 2019; 10:1774. in countries as diverse as Germany⁸8 Hinz A, Zenger M, Brähler E, Spitzer S, Scheuch K, Seibt R. Effort-reward imbalance and mental health problems in 1074 German teachers, compared with those in the general population. Stress Health 2003; 32(3):224-230., Saudi Arabia⁹9 El-Metwally A, Javed S, Razzak HA, Aldossari KK, Aldiab A, Al-Ghamdi SH, Al-Zahrani JM. The factor structure of the general health questionnaire (GHQ12) in Saudi Arabia. BMC Health Serv Res 2018; 18(1):595., Austria¹⁰10 Friedrich F, Alexandrowicz R, Benda N, Cerny G, Wancata J. The criterion validity of different versions of the General Health Questionnaire among non-psychiatric inpatients. Soc Psychiatry Psychiatr Epidemiol 2011; 46(7):635-641., China¹¹11 Liang Y, Wang L, Yin X. The factor structure of the 12-item general health questionnaire (GHQ-12) in young Chinese civil servants. Health Qual Life Outcomes 2016; 14(1):136-145., Colombia¹²12 Ruiz FJ, García-Beltrán DM, Suárez-Falcón JC. General Health Questionnaire-12 validity in Colombia and factorial equivalence between clinical and nonclinical participants. Psychiatry Res 2017; 256:53-58., Spain¹³13 Molina JG, Rodrigo MF, Losilla JM, Vives J. Wording effects and the factor structure of the 12-Item General Health Questionnaire (GHQ-12). Psychol Assess 2014; 26(3):1031-1037., India¹⁴14 Kashyap GC, Singh SK (2017). Reliability and validity of general health questionnaire (GHQ-12) for male tannery workers: a study carried out in Kanpur, India. BMC Psychiatry 2017; 17(1):102., Iran¹⁵15 Namjoo S, Shaghaghi A, Sarbaksh P, Allahverdipour H, Pakpour AH. Psychometric properties of the General Health Questionnaire (GHQ-12) to be applied for the Iranian elder population. Aging Ment Health 2017; 21(10):1047-1051. and Japan¹⁶16 Ohno S, Takahashi K, Inoue A, Takada K, Ishihara Y, Tanigawa M, Hirao K. Smallest detectable change and test-retest reliability of a self-reported outcome measure: Results of the Center for Epidemiologic Studies Depression Scales General Self Efficacy Scale, and 12 item General Health Questionnaire. J Eval Clin Pract 2017; 23(6):1348-1354.. Meta-analyses conducted by Gnambs and Staufenbie⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194. address the factorial structure of this measure. For example, exploratory factor analysis (K = 38, N = 76,473) identified two factors that showed clusters of negative and positive items. A second study (K = 84, N = 410,640) using confirmatory factor analysis identified a bifactorial structure, with the most of the variance was explained by the general factor.

Despite the evidence on the GHQ-12 factorial structure, multiple factors can recognizably affect it, including translation bias, how items were written, and cultural and even clinical biases⁷7 Guan M, Han B. Factor structures of General Health Questionnaire-12 within the number of kins among the rural residents in China. Front Psychol 2019; 10:1774.. In this sense, some studies have been developed to prove the measure invariance of this instrument. For example, using an eight-item version in Brazil, Fernandes and Vasconcelos-Raposo¹⁷17 Fernandes HM, Vasconcelos-Raposo J. Factorial validity and invariance of the GHQ12 among clinical and nonclinical samples. Assessment 2013; 20(2):219-229. noted that the unifactorial solution was invariant in clinical and nonclinical samples. Romppel et al.¹⁸18 Romppel M, Hinz A, Finck C, Young J, Brähler E, Claesmer H. Cross-cultural measurement invariance of the General Health Questionnaire-12 in a German and a Colombian population sample. Int J Methods Psychiatric Res 2017; 26(4):e-1532. used the German and Spanish versions of the full GHQ-12 with a unifactorial solution and reported a configurational invariance between Germany and Colombia, but no metric and scalar invariance. However, little is known about the invariance of this measure in Brazil regarding the sex of the participants5. The absence of studies with this objective weakens the security of GHQ-12 comparisons between sex in Brazil, a variable that can affect mental health indicators⁹9 El-Metwally A, Javed S, Razzak HA, Aldossari KK, Aldiab A, Al-Ghamdi SH, Al-Zahrani JM. The factor structure of the general health questionnaire (GHQ12) in Saudi Arabia. BMC Health Serv Res 2018; 18(1):595..

Even though there is still no consensus on the GHQ-12 structure, and despite little evidence of its factorial invariance, there is evidence of its suitability as a screening tool¹⁹19 Endsley P, Weobong B, Nadkarni A. The psychometric properties of GHQ for detecting common mental disorder among community dwelling men in Goa, India. Asian J Psychiatr 2017; 28:106-110.. Also, as the scores of this measure correlate with or are good explainers of subjective well-being²⁰20 Qin M, Vlachantoni A, Evandrou M, Falkingham J. General Health Questionnaire-12 reliability, factor structure, and external validity among older adults in India. Indian J Psychiatry 2018; 60(1):56-59., quality of life²¹21 Serrano-Aguilar P, Ramallo-Fariña Y, Trujillo-Martín MDM, Muñoz-Navarro SR, Perestelo-Pereza L, Cuevas-Castresana CDL. The relationship among mental health status (GHQ-12), health related quality of life (EQ-5D) and health-state utilities in a general population. Epidemiol Psychiatr Sci 2009; 18(3):229-239., burnout²²22 Higuchi Y, Inagaki M, Koyama T, Kitamura Y, Sendo T, Fujimori M, Yamada N. A cross-sectional study of psychological distress, burnout, and the associated risk factors in hospital pharmacists in Japan. BMC Public Health 2016; 16:534., suicidal ideation and suicide attempted²³23 Eskin M, Voracek M, Tran T, Sun J, Janghorbani M, Giovanni CM, Harlak H. Suicidal behavior and psychological distress in university students: a 12-nation study. Arch Suicide Res 2016 20(3):369-388., and general health²⁴24 Lindkvist M, Feldman I. Assessing outcomes for cost-utility analysis in mental health interventions: mapping mental health specific outcome measure GHQ-12 onto EQ-5D-3L. Health Qual Life Outcomes 2016; 14(1):134., it is a useful measure in the context of mental health, even helping to identify mental disorders in workers seeking outpatient medical care²⁵25 Makowska Z, Merecz D, Moscicka A, Kolasa W. The validity of General Health Questionnaires, GHQ-12 AND GHQ-28, in mental health studies of working people. Int J Occup Med Environ Health 2002; 15(4):353-362..

Hence, it seems justifiable to analyze additional evidence about the GHQ-12’s psychometric adequacy. The specific objective of the present study is to test the presumably more suitable bifactor structure⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194. by comparing it with a simple one-factor model, a one-factor model but with control for the effects of negative items, and a two-factor model. Furthermore, we aim to verify GHQ-12 measure invariance in relation to sex and to the presence (or not) of a diagnosis of mental and behavioral disorders. Finally, we aim to understand to what extent its scores correlate with indicators of sexual satisfaction, general health, and suicidal ideation, which are critical mental health elements²⁰20 Qin M, Vlachantoni A, Evandrou M, Falkingham J. General Health Questionnaire-12 reliability, factor structure, and external validity among older adults in India. Indian J Psychiatry 2018; 60(1):56-59..

Method

Procedure

This was a survey research with an ex post facto design. It was a random sample drawn from the universe of physicians registered in the Medical Council system. The questionnaires were sent via regular mail to the addresses of the randomly chosen doctors, and all physicians participated voluntarily with no material reward and had the anonymous nature of their collaboration ensured according to the ethical principles of human subject research. The sample was composed exclusively of physicians, as this study is part of a larger survey to assess the Health of Brazilian Doctors. A minimum sample of 1,065 physicians was estimated for a population of 500,000 physicians (available on the Federal Council of Medicine website), an error (e) of 3%, a confidence margin of 95% and a z-score of 1.96. Only completely filled in were accepted for this research. Of the 1,115, 2.33% were excluded for this reason, resulting in 1,084 participants.

Sample calculation. Sample size = {[z2 * p(1-p)]/e2}/1+[ z2 * p(1-p)/e2N]. N = population size; e = margin of error (percentage in decimal format); z = z score.

Instruments

The participants answered a booklet consisting of demographic questions (e.g., age, sex, marital status, religion, and employment status), whether they used any prescription drugs, whether they experienced decreased libido, and whether they considered their sex life satisfactory (the answers for these three questions were classified as 0 = No or 1 = Yes). In addition, all participants answered the following instruments:

General Health Questionnaire. Proposed by Goldberg¹1 Goldberg DP. The detection of psychiatric illness by questionnaire. London: Oxford University Press; 1972., the shorter version (GHQ-12) considered here has been widely used as a screening tool to detect non-severe (non-psychotic) psychiatric disorders. Although different factor structures and internal consistency coefficients are used in Brazil, their psychometric indicators are often acceptable²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384.

3 Borges LO, Argolo JCT. Adaptação e validação de uma escala de bem-estar psicológico para uso em estudos ocupacionais. Avali Psicol 2002; 1(1):17-27.^-44 Gouveia VV, Chaves SSS, Oliveira IP, Dias MR, Gouveia RSV, Andrade PR. A utilização do QSG-12 na população geral: estudo de sua validade de construto. Psic Teor Pesq 2003; 19(3):241-248.^,2626 Damásio BF, Machado WDL, Silva JPD. Estrutura fatorial do Questionário de Saúde Geral (QSG-12) em uma amostra de professores escolares. Aval Psicol 2011; 10(1):99-105.. The 12 items are scored in terms of how much the person has experienced a given symptom on a four-point scale. The alternative answers for negative items regarding mental health (e.g., Have you lost much sleep due to worrying?) range from 1 (Absolutely not) to 4 (Much more than usual), while the alternative answers for positive affirmative items (e.g., Have you felt capable of making decisions?) range from 1 (More than usual) to 4 (Much less than usual). The scoring system used by Goldberg (1972) implies that each of these items is recoded as 0 (Absence of psychiatric disorder; scores 1 and 2) or 1 (Presence of psychiatric disorder; scores 3 and 4). Therefore, the higher the total score of this measure, the higher the level of psychological distress.

Positive and Negative Suicidal Ideation Inventory. This version was initially proposed by Osman et al.²⁷27 Osman A, Gutierrez PM, Jiandani J, Kopper BA, Barrios FX, Linden, SC, Truelove RS. A preliminary validation of the Positive and Negative Suicide Ideation (PANSI) Inventory with normal adolescent samples. J Clin Psychol 2003; 59(4):493-512. and includes 14 items that measure two main oblique factors (r = -0.49, p < 0.001): negative suicidal ideation (eight items, e.g., seriously considering the possibility of ending your life for not being able to meet the expectations of others; feeling hopeless about the future and considering the possibility of ending your life), and positive suicidal ideation (six items, e.g., feeling enthusiastic about your success at school and/or work; feeling confident about being able to deal with most problems in your life). The authors demonstrated that this factorial structure best fit the data (Robust comparative fit index = 0.96, non-normed fit index = 0.95, and root mean square error of approximation = 0.03), with each factor showing Cronbach’s alpha above 0.80.

Data analysis

The data were tabulated and analyzed using SPSS (version 21). First, descriptive statistics (mean, standard deviation, frequency) were calculated for sample characterization. The Cronbach’s alpha and McDonald’s ω were calculated to evaluate the internal consistency of the GHQ-12 dimensions. AMOS software (version 21) was then used for confirmatory factor analyses (CFA) to prove the best factor structure of this instrument. In this case, the covariance matrix was considered as input using Maximum Likelihood (ML) estimation and the following adjustment indicators: GFI (Goodness-of-Fit Index), AGFI (Adjusted Goodness-of-Fit Index), and TLI (Tucker-Lewis Index), with values ​​equal to or above 0.90 taken as acceptable, and RMSEA (Root Mean Square Error of Approximation) with recommended values ​​close to 0.05 and accepting up to 0.10. Three indicators were also used to compare the models (ECVI, CAIC, and Δχ²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384.). Finally, Pearson correlations were calculated to determine the adequacy of GHQ-12 to explain physical health variables and suicidal ideation.

The R software (version 3.3.2)²⁸28 R Development Core Team. R: A language and environment for statistical computing. 2015. was used to confirm the factorial invariance of GHQ-12 with a multigroup confirmatory factor analysis (MGCFA)²⁹29 semTools Contributors. semTools: Useful tools for structural equation modeling. R package version 0.4-12. 2016.. Hierarchical models were tested for the invariance types: (1) configural: the same factorial (two-dimensional) model for the groups; (2) metric: equivalent structure and factor loadings (λ); and (3) scalar: in addition to the other equivalent parameters, it considers equivalent thresholds. For interpretative purposes, the invariance was evaluated based on the difference between ΔCFI (if lower than 0.01, invariant model)³⁰30 Cheung GW, Rensvold RB. Evaluating goodness-of-fit indexes for testing measurement invariance. Struct Equation Model 2002; 9(2):233-255. and ΔRMSEA indicators (if lower than or equal to 0.015, invariant model)³¹31 Chen FF. Sensitivity of goodness of fit indexes to lack of measurement invariance. Struct Equation Model 2007; 14(3):464-504..

Results

Description of participants

Of a total of 1,089 physicians, 547 reported having been diagnosed with mental and behavioral disorders and 542 reported not having been diagnosed. Most respondents were male (61.5%), married (72.2%), Catholic (59.2%) and employed (88.8%), with a mean age of 45.7 years (SD = 10.6).

GHQ-12 factorial structure evidence

Confirmatory factor analyses were used to test the GHQ-12 factorial structure in the group studied. Four models were tested: the unifactorial model (M1), in which a general factor of psychological distress was explained by the 12 items of the measure; the bifactorial model (M2), which combined one depression and one anxiety factor; a general factor with control of negative items (M3); and finally the bifactor model (M4), whose items are saturated in both specific components and which represents psychological distress in the general factor. The results of these analyses are shown in Table 1.

The findings for the unifactorial model (M1) showed unsatisfactory adjustment indicators (GFI = 0.89, CFI = 0.89, TLI = 0.87, and RMSEA = 0.11). The most promising indicators were seen for the model (M2) (GFI = 0.90, CFI = 0.91, TLI = 0.89, and RMSEA = 0.10) and for model 03 (M3) (GFI = 0.91, CFI = 0.92, TLI = 0.89, and RMSEA = 0.10), but they were even better for the bifactor model (M3) (GFI = 0.95, CFI = 0.95, TLI = 0.92, and RMSEA = 0.08). Direct comparison of these models showed the M₃ model to be superior to M₂ [Δχ² (5) = 59.00, p < 0.001] and, although the M₃ and M₄ models were both satisfactory, the latter was statistically more appropriate [Δχ² (7) = 352.95, p < 0.001].

Table 2 shows that the depression and anxiety factors had lower factor loadings when compared to the general factor.

Cronbach’s alphas, McDonald’s ω, composite reliability (CR) and average variance extracted (AVE) were calculated specifically for the general factor (α = 0.92, ω = 0.92, CR = 0.92, AVE = 0.53), but also for the factors of depression (α = 0.89; ω = 0.82, CR = 0.27, AVE = 0.07) and anxiety (α = 0.80; ω = 0.81, CR = 0.23, AVE = 0.06). Therefore, this psychometric parameter seems to confirm the adequacy of the general factor only.

Factorial invariance

Table 3 shows that the GHQ-12 configural, metric, and scalar invariance parameters were corroborated regarding participants’ sex and presence/absence of diagnosis of mental and behavioral disorders considering the previously described criteria (ΔCFI < 0.01 and ΔRMSEA ≤ 0.015).

Thus, this instrument has full factorial invariance and can be used to compare men and women, as well as clinical and nonclinical groups (general population).

Correlational validity

After demonstrating the adequacy of the GHQ-12 structure and factorial invariance, we tried to validate its usefulness to understand other constructs. In this case, general and specific scores for anxiety and depression were calculated and correlated with suicidal ideation factors and sexual satisfaction and general health indicators (prescription drug use). The results can be seen in Table 4.

This table shows that the general factor (psychological distress) correlates with the general dimension of suicidal ideation (r = 0.71, p < 0.001) and their specific negative ideation factors (r = 0.46, p < 0.001), but most importantly with positive ideation (r = −0.77, p < 0.001). There was also a correlation between psychological distress and decreased libido (r = 0.38, p < 0.001), satisfaction with sexual life (r = -0.37, p < 0.001), and continuous use of some type of prescription drug (r = 0.20, p < 0.001).

Discussion

The main objective of this study was to test the bifactor structure of the GHQ-12, which proved to be more adequate in studies outside Brazil6, comparing it with the unifactorial and bifactorial models. We also aimed to verify the measure invariance of this instrument regarding sex and the presence (or not) of a diagnosis of mental and behavioral disorders and to find evidence of a correlational validity of this measure, correlating it with indicators of sexual satisfaction, general health, and suicidal ideation.

Of the main findings, multiple confirmatory factor analyses evidenced that the best fit was the bifactor model with two specific factors (anxiety and depression) and a general dimension of psychological distress. This finding corroborates reports of participants from several countries and samples⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194.^,3232 Tomás JM, Gutiérrez M, Sancho P. Factorial validity of the General Health Questionnaire 12 in an Angolan sample. Euro J Psychol Assess 2017; 33(2):116-122. This elucidates divergences found in the national literature regarding the factorial structure of the GHQ-12²2 Gouveia VV, Lima TJSD, Gouveia RSV, Freires LA, Barbosa LHGM. Questionário de Saúde Geral (QSG-12): o efeito de itens negativos em sua estrutura fatorial. Cad Saude Publica 2012; 28(2):375-384.^,33 Borges LO, Argolo JCT. Adaptação e validação de uma escala de bem-estar psicológico para uso em estudos ocupacionais. Avali Psicol 2002; 1(1):17-27.^,2626 Damásio BF, Machado WDL, Silva JPD. Estrutura fatorial do Questionário de Saúde Geral (QSG-12) em uma amostra de professores escolares. Aval Psicol 2011; 10(1):99-105.^,3333 Tabachnick BG, Fidell LS. Using multivariate statistics. London: Pearson; 2013..

The bifactor structure is commonly used when there are factor models in the literature, which assume acceptable adjustments and, therefore, can compete with each other. The results of this study showed this competition between two structures (general factor versus specific factors), but point to a solution between the divergences of the general and specific structure when it presents factor loadings and better reliability indicators for the general factor.

The emphasis on the general dimension of the GHQ is in line with the original objective of the scale, which aims to make a pre-clinical assessment of people’s general mental health status⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194.. After that, a professional would more specifically assess dimensions such as depression and anxiety. Possibly due to the complexity and the various aspects that make up mental health, dimensions such as depression and anxiety emerge from the structure of the scale, but as the data show, these specific factors are not reliable. Thus, as highlighted in international meta-analysis studies, specific factor scores reflect a very limited variation, demonstrating that the GHQ-12 is essentially one-dimensional⁶6 Gnambs T, Staufenbiel T. The structure of the General Health Questionnaire (GHQ-12): two meta-analytic factor analyses. Health Psychol Rev 2018; 12(2):179-194..

Regarding the adjustment of the Bifactor model, we would like to highlight the value of the RMSEA, as it was within the acceptable limit. The discussion in the literature about these reference points is wide and complex, for this reason it is highly recommended to use more than one index. This happens because the indices are affected by different elements that vary between studies, such as model complexity, sample size, number of latent and observable variables, degrees of freedom, among others. Specifically, the RMSEA tends to increase with the addition of more variables to the model. The use of this or other types of reference to interpret the RMSEA depends on model specifications, degrees of freedom and sample value and not on a single analysis criterion. For this reason, the interpretation of the model fit needs to be done in a way that considers the other indicators together³⁴34 Marôco J. Análise de equações estruturais: fundamentos teóricos, software e aplicações. Pêro Pinheiro: ReportNumber; 2014..

The bifactor structure also showed full measure invariance for sex and diagnosis (presence or absence of mental and behavioral disorders). These findings are even more encouraging than those reported in previous studies¹⁷17 Fernandes HM, Vasconcelos-Raposo J. Factorial validity and invariance of the GHQ12 among clinical and nonclinical samples. Assessment 2013; 20(2):219-229.^,1818 Romppel M, Hinz A, Finck C, Young J, Brähler E, Claesmer H. Cross-cultural measurement invariance of the General Health Questionnaire-12 in a German and a Colombian population sample. Int J Methods Psychiatric Res 2017; 26(4):e-1532.. The relevance of these two variables must be shown and the measure has to demonstrate invariance. The sex of the participants explains significant variations in the level of psychological distress experienced, evidencing that females are more negatively affected⁹9 El-Metwally A, Javed S, Razzak HA, Aldossari KK, Aldiab A, Al-Ghamdi SH, Al-Zahrani JM. The factor structure of the general health questionnaire (GHQ12) in Saudi Arabia. BMC Health Serv Res 2018; 18(1):595.^,3535 Sánchez-Fuentes M, Santos-Iglesias P, Sierra JC. A systematic review of sexual satisfaction. Int J Clin Health Psychol 2014; 14(1):67-75.^,3636 Van Droogenbroeck F, Spruyt B, Keppens G. Gender differences in mental health problems among adolescents and the role of social support: results from the Belgian health interview surveys 2008 and 2013. BMC Psychiatry 2018; 18(1):6-12.. On the other hand, the condition of being part of a clinical group or the general population may be relevant because different types of symptoms and magnitude levels may be experienced. Thus, measure invariance should be demonstrated under these conditions, corroborating a previous study that used a shorter GHQ-12 version¹⁷17 Fernandes HM, Vasconcelos-Raposo J. Factorial validity and invariance of the GHQ12 among clinical and nonclinical samples. Assessment 2013; 20(2):219-229..

As for the correlation between psychological distress and its specific anxiety and depression factors and external variables, our results corroborate what has been theoretically expected and verified in previous studies²³23 Eskin M, Voracek M, Tran T, Sun J, Janghorbani M, Giovanni CM, Harlak H. Suicidal behavior and psychological distress in university students: a 12-nation study. Arch Suicide Res 2016 20(3):369-388.^,2424 Lindkvist M, Feldman I. Assessing outcomes for cost-utility analysis in mental health interventions: mapping mental health specific outcome measure GHQ-12 onto EQ-5D-3L. Health Qual Life Outcomes 2016; 14(1):134.^,3737 De Ryck I, Van Laeken D, Nöstlinger C, Platteau T, Colebunders R. Sexual satisfaction among men living with HIV in Europe. AIDS Behav 2012; 16(1):225-230.

38 Mosack V, Steinke EE, Wright DW, Walker C, Medina M, Moser DK, Chung, ML. Effects of depression on sexual activity and sexual satisfaction in heart failure. Dimens Crit Care Nurs 2011; 30(4):218-225.^-3939 Santos WSD, Ulisses SM, Costa TMD, Farias MG, Moura DPFD. A influência de fatores de risco e proteção frente à ideação suicida. Psicol Saude Doenças 2016; 17(3):515-526.. Psychological distress and suicidal ideation are directly correlated, such that participants with greater distress experience an inability to participate in pleasurable activities and feelings of sadness, apprehension, irritability, and tension⁴⁰40 Rock PL, Roiser JP, Riedel WJ, Blackwell AD. Cognitive impairment in depression: A systematic review and meta-analysis. Psychol Med 2014; 44(10):2029-2040.. Thus, mental disorders seem to be a risk factor for suicidal ideation and even suicide⁴¹41 Botega NJ, Werlang BSG, Cais CDS, Macedo MMK. Prevenção do comportamento suicida. Psico 2006; 37(3):213-220..

The association between psychological distress and continuous use of prescription drugs reinforces the finding that the prevalence of symptoms of non-severe mental health disorders may be related to dysfunctional behavioral indicators³⁸38 Mosack V, Steinke EE, Wright DW, Walker C, Medina M, Moser DK, Chung, ML. Effects of depression on sexual activity and sexual satisfaction in heart failure. Dimens Crit Care Nurs 2011; 30(4):218-225.. Finally, the indicators of sexual satisfaction and decreased libido were correlated with psychological distress as well as anxiety and depression, corroborating the findings by De Ryck et al.³⁷37 De Ryck I, Van Laeken D, Nöstlinger C, Platteau T, Colebunders R. Sexual satisfaction among men living with HIV in Europe. AIDS Behav 2012; 16(1):225-230. These results are potentially applicable by showing the negative impact of mental disorder on sexuality and may serve to guide effective communication about sexual concerns and expectations for partners experiencing psychological distress³⁵35 Sánchez-Fuentes M, Santos-Iglesias P, Sierra JC. A systematic review of sexual satisfaction. Int J Clin Health Psychol 2014; 14(1):67-75..

To summarize, these findings confirm the psychometric adequacy of GHQ-12 to the Brazilian reality, corroborating previous studies³3 Borges LO, Argolo JCT. Adaptação e validação de uma escala de bem-estar psicológico para uso em estudos ocupacionais. Avali Psicol 2002; 1(1):17-27.

4 Gouveia VV, Chaves SSS, Oliveira IP, Dias MR, Gouveia RSV, Andrade PR. A utilização do QSG-12 na população geral: estudo de sua validade de construto. Psic Teor Pesq 2003; 19(3):241-248.^-55 Gouveia VV, Barbosa GA, Andrade EO, Carneiro MB. Factorial validity and reliability of the General Health Questionnaire (GHQ-12) in the Brazilian physician population. Cad Saude Publica 2010; 26(7):1439-1445.. They also show the factorial invariance of this measure as reported by Gouveia et al.⁵5 Gouveia VV, Barbosa GA, Andrade EO, Carneiro MB. Factorial validity and reliability of the General Health Questionnaire (GHQ-12) in the Brazilian physician population. Cad Saude Publica 2010; 26(7):1439-1445., as well as the association of psychological distress with central variables in the context of intimate interpersonal relationships, such as libido and sexual satisfaction. Although this study has positive points, such as the survey of a random sample of a professional category at the national level and being able to indicate a structural solution for a health scale widely used in the national and international context, it is necessary to emphasize that this study cannot offer a generalization to other populations of professionals, demanding attention regarding its use. Future studies are necessary to present more evidence about the predictive power of this measure, including aspects such as work performance or wish to leave the profession. Within the scope of parameters, it is important to present evidence of the temporal stability of GHQ-12 and its adequacy to understand the progress of a given clinical condition.

References

Publication Dates

History