Available online at www.sciencedirect.

com

Comprehensive Psychiatry 54 (2013) 406 – 413

www.elsevier.com/locate/comppsych

What is the General Health Questionnaire-12 assessing?

Dimensionality and psychometric properties of the General Health
Questionnaire-12 in a large scale German population sample
Matthias Romppel a, b , Elmar Braehler a , Marcus Roth c , Heide Glaesmer a,⁎
a
Department of Medical Psychology and Medical Sociology, University of Leipzig, 04103 Leipzig/Germany
b
Department of Applied Social Sciences, Leipzig, University of Applied Sciences, 04251 Leipzig/Germany
c
Department of Psychology, University of Duisburg-Essen, 45127 Essen, Germany

Abstract

Since the dimensionality and the related psychometric properties of the 12-item General Health Questionnaire (GHQ-12) are still under
debate, the present study compares different factor solutions from the literature to determine which shows the best fit and to investigate
reliability and construct validity. The analyses are based on a German population based representative sample (N = 2,041), using face-to-face-
interviews. The confirmatory factor analysis indicates the best fit to the one-factor model including response bias on the negative worded
items according to Hankins. Thus, the importance of methodical aspects for the dimensionality was emphasized. Moreover, the correlations
of the different subscales of the two- and three-factor models with several external criteria (BDI, PHQ-2, SF-36, PHQ-Anxiety, SPIN) do not
substantially differ. The preferred unidimensional model shows good psychometric properties. According to its associations with the external
criteria under study, the GHQ-12 as a unidimensional measure seems to be a useful screening tool for the assessment of mental distress or a
minor psychiatric morbidity with a main focus on depressive symptomatology.
© 2013 Elsevier Inc. All rights reserved.

1. Background Most of the studies interpreted a factor of “anxiety/

depression” and a factor of “social dysfunction” [2-9].
Since its development in 1972, the General Health Beyond these two factors, some studies revealed a third
Questionnaire (GHQ) has been widely used as a screening factor expressing “loss of confidence” [10-14]. Over time
instrument for minor psychiatric morbidity [1]. It has been various names had been provided for these factors [15].
translated in many languages and extensively validated in Martin (1999) developed a correlated three-factor model, in
different populations [2]. Originating from the 60-item which the factors were labelled “cope”, “stress”, and
GHQ, shorter versions have been developed and validated, “depression”; this has unfortunately not been replicated in
like the 12-item version (GHQ-12). The GHQ-12 has other studies so far [11,12,15]. It has repeatedly been stated
recently become the most popular form of the scale because that the GHQ-12 measures psychiatric morbidity in more
of its relatively good psychometric properties and its brevity. than one domain. A three-factor solution is usually
Although the GHQ-12 was designed as a unidimensional considered to be superior [14].
scale, several studies revealed two- or three-factor solutions. Several studies discussed the impact of methodical
aspects on the factorial solutions revealed in the different
studies. As there are different scoring methods available for
the GHQ-12, a study by Campbell has shown that the scoring
⁎ Corresponding author. Department of Medical Psychology and method substantially affects the model estimation [16].
Medical Sociology, University of Leipzig, Philipp-Rosenthal-Str. 55, Especially the importance of positive and negative worded
04103 Leipzig/Germany. Tel.: +49 341 9718811; fax. +49 341 9718809. items is under debate [17,18]. This so called “wording
E-mail address: Heide.Glaesmer@medizin.uni-leipzig.de (H. Glaesmer). effect” means that the positive or negative formulation of
0010-440X/$ – see front matter © 2013 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.comppsych.2012.10.010
 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413 407

items influences the interpretation of these items and the support or discourage the future application of the instrument
response of the participants. In its consequence it has an in population based studies.
effect on the factorial structure. This effect of the wording
on the factorial structure has been confirmed in different
studies on the dimensionality of other psychometric 2. Methods
instruments like the Rosenberg Self-Esteem Scale and the
Life-Orientation-Test [19-21]. Marsh [19] preferred a 2.1. Subjects
unidimensional solution with response bias on the nega- A representative sample of the German general
tively phrased items for the Rosenberg Self-esteem scale population was selected with the assistance of a
[19]. Up to now studies have assumed that the GHQ-12 is demographic consulting company. The area of Germany
free of response bias. Following the finding of Marsh [19], was separated into 201 sample areas representing the
Hankins tested a unidimensional model including response different regions of the country. Households of the
bias for the negatively phrased items for the GHQ-12 in respective area and members of this household fulfilling
comparison with a simple unidimensional model and a the inclusion criteria (age at or above 14, able to read and
three-dimensional model [17]. This unidimensional model understand the German language) were selected randomly
including response bias showed superior fit compared to the by Kish-selection-grid technique. The Kish-selection-grid-
two other models under study. Hankins suggests that this technique targets individuals on the doorstep among
finding underpins that the previous findings for the factorial household residents. The system is devised so that all
structure of the GHQ-12 were based on methodical artefacts individuals in a household have an equal chance of
[22]. A study by Ye [18] replicated the results of Hankins selection. The sample is representative in terms of age,
[17,22] in a Chinese sample. The study of Wang and Lin gender, and education. A first attempt was made for 3,194
[23] showed that the GHQ-12 has a unidimensional addresses, of which 3,108 were valid. If not at home, a
structure after controlling for wording effects. maximum of four attempts were made to contact the
In summary, the factorial structure of the GHQ-12 selected person. 872 subjects (28.1%) refused participa-
remains under debate [15,18,22], and thus, a solid statement tion, 137 subjects (4.4%) were not reached after four
about the reliability of the GHQ-12 is still lacking, because it attempts, and 10 subjects (0.3%) refused participation
depends on the factorial structure. Nevertheless, Hankins because of severe health problems. All subjects were
[17] reported the impact of dimensionality and response bias visited by a study assistant, informed about the investi-
on reliability. As the calculation of reliability by alpha is gation, and self-rating questionnaires were presented. The
based on the assumption of no response bias and the best assistant waited until participants answered all question-
fitting model indicates response bias, the alpha may be over- naires and offered help if the meaning of questions was
estimated [17]. Finally, a comprehensive psychometric not clear. A total of 2,066 people between the ages of 14
evaluation needs statements about the validity of the and 93 years agreed to participate, completing the self-
instrument. A useful procedure is the inclusion of external rating questionnaires in November and December 2002
criteria to clarify what the GHQ-12 assesses. Most of the (participation rate: 66.5%). 25 subjects were excluded
psychometric studies lack this final step, and the designa- from the following analyses because of incomplete data.
tion of the different dimensions is based on a review of the Thus, a dataset of 2,041 people is included in this study.
related items. Table 1 gives an overview of the demographic character-
The two- and three-factor-models distinguish different istics of the sample.
domains like “depression” and “loss of confidence”. A
unidimensional solution refers to a more holistic concept 2.2. Instruments
like “mental distress” or “minor psychiatric morbidity”,
possibly measured by the GHQ-12. A reliable statement In our study the German Version of the 12-item General
about what the GHQ-12 assesses would be very useful Health Questionnaire (GHQ-12) [24] with a four-point
to guide the application of the instrument in research Likert-Scale (0-1-2-3) was used. Thus, the total score ranges
and practice. from 0–36, with higher scores representing higher levels of
Addressing the controversial debate about the dimen- mental distress. Supplementing the GHQ-12, several other
sionality, reliability and construct validity of the GHQ-12, instruments assessing mental distress and self-perceived
we [1] test and compare different dimensional models general health status were used in the study.
discussed in the previous literature, [2] quantifying psycho- The Beck Depression Inventory (BDI) [25,26] is one of
metric properties, and [3] studying associations of the GHQ- the most commonly used measures to assess the severity of
12 with depression, anxiety, social phobia, and self- depressive symptomatology. It is well established in research
perceived health status to determine what the GHQ-12 and clinical practice. Its reliability and validity were proven
assesses in a large scale German representative population in numerous studies [27].
sample. The primary aim of our study is to test psychometric To assess anxiety and depression, the Short form of the
properties of the GHQ-12 in the general population and to Patient Health Questionnaire (PHQ) [28,29] was used. This
408 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413

Table 1
Demographic characteristics of the representative population sample.
Total (N = 2,041) Male (N = 959) Female (N = 1,082)
Age M 48.8 47.1 50.2
SD 18.1 17.5 18.5
Age groups 14–24 years 11.9% (242) 14.0% (134) 10.0% (108)
25–34 years 13.2% (269) 13.4% (128) 13.0% (141)
35–44 years 17.6% (360) 17.3% (166) 17.9% (194)
45–54 years 16.4% (334) 16.9% (162) 15.9% (172)
55–64 years 18.0% (368) 19.0% (182) 17.2% (186)
65–74 years 14.8% (302) 15.8% (151) 14.0% (151)
≥75 years 8.1% (166) 3.7 (36) 12.0% (130)
Urbanity Rural area 28.9% (590) 29.1% (279) 28.7% (311)
Urban area 71.1% (1,451) 70.9% (680) 71.3% (771)
Education No qualifications 2.0% (41) 1.2% (11) 2.8% (30)
Less than 10 years 46.3% (944) 45.0% (431) 47.4% (513)
10 years of education 35.3% (718) 36.5% (350) 34.0% (718)
More than 10 years 16.3% (338) 17.4% (167) 15.8% (171)
Net household income b 750 €/month 10.8% (211) 9.4% (87) 12.0% (124)
750 to 1249 €/month 28.4% (554) 25.2% (232) 31.2% (322)
1250 to 1999 €/month 36.2% (708) 38.6% (352) 34.2% (352)
≥2000 €/month 24.6% (480) 26.8% (247) 22.6% (233)

instrument for psychiatric case definition in primary care ally, it contains a single question that assesses change in
demonstrated good validity and reliability. Compared with health from 1 year ago.
structured clinical interviews, 98% sensitivity and 80%
specificity were shown [30]. Depressive symptoms were 2.3. Statistical analyses
screened with the PHQ-2 [31,32], a short form of the PHQ
depression subscale. This short depression screener con- The models described in the introduction section above
tains two items assessing anhedonia and depressed mood were tested via confirmatory factor analyses using Mplus
over the past two weeks, scoring from 0 (“not at all”) to 3 6.1 [36]. In a confirmatory factor analysis a theoretically
(“nearly every day”). The total score of the PHQ-2 ranges derived factor structure is defined, and the degree to which
from 0 to 6. Compared with structured clinical interviews, the covariance structure estimated by the model fits the
83% sensitivity and 93% specificity was shown [31]. The empirically observed covariance structure is assessed [37].
five items of the anxiety module of the PHQ assess panic Model 1 represents the three-dimensional conception of the
attacks. The first question assesses whether there was a GHQ-12, with three latent variables (social dysfunction,
panic attack within the last four weeks. If there was at least anxiety/depression, and loss of confidence) and six, four
one attack, four further items assess whether there were and two measured variables loading onto them. Since a
previous attacks, if the attacks are unexpected, impairing latent variable, that is represented using only two in-
and if there are typical physical symptoms like tachycardia dicators, is locally under-identified, an equality constraint
or dizziness. Response categories for these items are yes on the two loadings associated with the latent variable can
vs. no. Again, a total anxiety score ranging from 0 to 5 be placed, following the recommendation of Little and
was calculated. colleagues [38]. Model 2 depicts the two-factor model, with
The Social Phobia Inventory (SPIN) [33,34] contains 17 two latent variables (social dysfunction and anxiety/
items referring to fear, avoidance and physiological symp- depression) and six measured variables loading onto each.
toms of social phobia in the previous seven days. Response Model 3 also represents a three-dimensional conception,
categories for these items range from 0 (“not at all”) to 4 but with “cope”, “stress”, and “depression” as latent
(“extremely”). The total score ranges from 0–68. Good variables and four, three and five measured variables.
psychometric properties have been established [33,34]. Model 4 represents the one-dimensional conception of the
A popular measure of self-perceived general health GHQ, with all 12 items defined as indicators of a single
status is the Short Form 36 (SF-36) [35]. It consists of 36 factor. Finally, we tested the unidimensional model
items which cover eight domains: Physical functioning (10 described by Hankins [22] as Model 5. As described
items), Role Limitations – Physical (four items), Bodily above, in this model the GHQ-12 was modelled as a
Pain (two items), Social Functioning (two items), General measure of one construct but with correlated error terms on
Health (five items), Role Limitations – Emotional (three the negative formulated items, modelling response bias.
items), Vitality (four items), and Mental Health (five This model was therefore identical to model 4, but it
items). After transformation, each subscale ranges from 0 to contains correlations between the error terms on the
100; higher scores indicate better health status. Addition- negative items (items 4, 5, 6, 7, 8, 9).
 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413 409

In our analyses the maximum likelihood (ML) method of the Fornell-Larcker-ratio [40]. A Fornell-Larcker-ratio
estimation was used; error covariances were constrained to smaller than 1 indicates good discriminant validity. The
zero in all models except model 5, in order to avoid χ 2 value, the Bayesian Information Criterion (BIC), the
overfitting and capitalizing on chance associations in the comparative fit index (CFI), the Tucker Lewis index (TLI),
data [39]. In addition to the standardized factor loadings, the the root mean square error of approximation (RMSEA), and
construct reliability (the extent to which the indicators of a the standardized root mean square residual (SRMR) are
construct share common variance) and the average variance reported as fit indices. When comparing models, the model
extracted (the extent to which the variance in the indicators with the lowest BIC value is usually the one to be preferred.
is accounted for by the latent construct) were calculated. Although, strictly speaking, the comparison of BIC values
Values greater than .7 (reliability) and .5 (average variance provides only a ranking, a difference of 10 points or more
extracted) indicate good reliability and convergent validity. has been suggested as a nearly certain indicator of a
Discriminant validity (the extent to which a construct in the difference in fit between models [41]. Values larger than
model is distinct from another construct) was tested using 0.95 for TLI and CFI, and values smaller than 0.06 for

Table 2
Standardized factor loadings and goodness-of-fit statistics for five alternative confirmatory factor analysis models of the General Health Questionnaire (GHQ).
Model 1 Model 2 Model 3 Model 4 Model 5
Social Anxiety/ Loss of Social Anxiety/ Cope Stress Depression Global Global
Dysfunction Depression Confidence Dysfunction Depression
1. Able to concentrate .66 .65 .65 .70 .66
2. Capable of making .58 .58 .63 .70 .58
decisions
3. Face up to problems .61 .61 .60 .64 .61
4. Lost sleep over .66 .64 .73 .73 .60 a
worry
5. Constantly under .59 .56 .66 .66 .55 a
strain
6. Could not overcome .75 .74 .74 .81 .66 a
difficulties
7. Unhappy and .80 .81 .81 .85 .65 a
depressed
8. Loss of .88 .79 .79 .87 .56 a
self-confidence
9. Thinking of self .82 .75 .76 .84 .54 a
as worthless
10. Play useful part .52 .54 .52 .57 .52
in things
11. Enjoy day-to-day .64 .63 .59 .69 .63
activities
12. Reasonably happy .75 .74 .62 .78 .75

Construct reliability .79 .80 .84 .80 .86 .69 .70 .86 .89 .88
Average variance .40 .50 .72 .40 .52 .36 .44 .56 .41 .37
extracted
Fornell-Larcker-ratio 1.9 1.5 1.0 1.6 1.2 2.0 1.7 1.2 - -

Correlations between .86 .65 .80 .86 .82

factors
.85 .83
Fit-statistics of the model
Chi² ⁎ 1146.9 1507.7 1658.0 1942.5 752.0
df 51 53 51 54 39
BIC 34963.4 35308.8 35474.5 35736.1 34659.8
CFI 0.90 0.86 0.85 0.82 0.93
TLI 0.87 0.83 0.81 0.78 0.89
RMSEA (90% CI) 0.10 0.12 0.12 0.13 0.10
(0.10-0.11) (0.11-0.12) (0.11-0.12) (0.13-0.14) (0.09-0.10)
SRMR 0.06 0.06 0.07 0.07 0.05
BIC = Bayesian Information Criterion; CFI = comparative fit index; TLI = Tucker Lewis index (non-normed fit index); RMSEA = root mean square error of
approximation; 90% CI = limits of the 90% confidence interval for RMSEA; SRMR = standardized root mean square residual.
a
Error terms are allowed to covary.
⁎ All p's b .001.
410 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413

RMSEA and smaller than 0.08 for SRMR are considered as 3.2. Convergent and discriminant validity
indicators of a good fit [42].
To analyse the convergent and discriminant validity of The average variance extracted is satisfactory only for
the GHQ-12, Pearson's correlations were computed to some of the factors. Due to the high intercorrelations
determine the relations between the scales of the GHQ-12 between the factors, all but one factor (“Loss of Confidence”
according to the best model identified and the BDI, the in model 1) miss the criterion for a good discriminant
PHQ, the SF-36, and the SPIN, measuring mental distress validity of the constructs, namely a Fornell-Larcker-ratio
and self-perceived health status. smaller than 1 (Table 2).
Means, standard deviations, item-total correlations, and Pearson's correlations between the scales of the GHQ and
response-probabilities were calculated as psychometric the scales of several other instruments measuring mental
properties of the items, Cronbach's α was calculated as a distress and self-perceived health status are presented in
measure of the internal consistency. Table 3. Because of the large sample size, the coefficients are
To predict the general GHQ-12-score, a total of 12 all significant at the level of p b .001.
measures (BDI, PHQ-2, PHQ-Anxiety, SPIN, subscales of As presented in Table 3, the highest correlations of
the SF-36) were entered as predictors in a stepwise linear the GHQ are found with the BDI and the SF-36-
regression analysis. Subscale “Mental Health”, followed by the PHQ-2 and
Statistical analyses were conducted with the SPSS 15.0 the SF-36-Subscales social functioning and vitality. The
statistical package and Mplus 6.1 [36]. other correlations show low ranges. From the perspective
of discriminant and convergent validity, the assumption
that only one global factor describes the GHQ-12 seems
to be justified.
3. Results
3.3. Psychometric properties of the GHQ
3.1. Confirmatory Factor Analysis (CFA)
Item and scale characteristics of the unidimensional
As shown in Table 2, all factor loadings are greater than GHQ-12-conception were evaluated on the basis of the
.50, with indicator reliabilities (squared factor loadings) total sample (N = 2.041). As shown in Table 4, item–total
ranging between .27 and.77. In all models item 10 (“Play correlations are in the upper range, and response–
useful part in things”) has the smallest factor loading. In all probabilities are in the medium range between p = 0.37
models the latent variables show good construct reliabilities and 0.52. Internal consistency as a measure of the
(range .69 to .89). When comparing the BIC values of the reliability of the scale can be considered to be very
models, the best model is model 5 followed by model 1, good (Cronbach's α = .89, α = .79, and α = .86 for the full
although all models demonstrate rather unsatisfactory fit scale, the positive worded items, and the negative worded
indices (Tucker-Lewis Index b .90 and root mean square error items, respectively).
of approximation N .08). The unidimensional model incorpo-
3.4. Predicting GHQ-score
rating response bias (model 5) reaches the best overall fit for
the data and a significantly better fit than model 4, the As shown in Table 5, the stepwise procedure results in a
unidimensional model with error covariances constrained to regression model with only six predictors successfully
zero (Δχ 2 = 1190.5, df = 15, p b .001). predicting the total score of the GHQ-12 with an accounted

Table 3
Table 4
Pearson correlations between the General Health Questionnaire (GHQ) and
Psychometric properties of the German adaptation of the General Health
other scales.
Questionnaire (GHQ).
Cronbach's Models 4 and 5
Items M SD rit p
alpha
Global-scale
1. Lost sleep over worry 0.95 0.50 0.44 0.49
BDI .93 .63 2. Constantly under strain 1.02 0.48 0.51 0.51
PHQ-Anxiety .65 .33 3. Able to concentrate 0.77 0.74 0.63 0.44
PHQ-2 (Depression) .78 .55 4. Play useful part in things 0.67 0.72 0.72 0.42
SF-36: Physical Functioning .93 -.26 5. Face up to problems 1.01 0.49 0.56 0.50
SF-36: Role Limitations (physical) .92 -.33 6. Capable of making decisions 0.79 0.73 0.56 0.45
SF-36: Bodily Pain .89 -.38 7. Could not overcome difficulties 0.67 0.65 0.59 0.42
SF-36: General Health. .79 -.43 8. Reasonably happy 0.46 0.63 0.64 0.37
SF-36: Vitality .81 -.52 9. Enjoy day-to-day activities 0.55 0.64 0.67 0.39
SF-36: Social Functioning .79 -.54 10. Unhappy and depressed 0.63 0.67 0.71 0.41
SF-36: Role Limitations (emotional) .92 -.43 11. Loss of self-confidence 1.05 0.53 0.64 0.51
SF-36: Mental Health .82 -.62 12. Thinking of self as worthless 1.10 0.48 0.55 0.53
SPIN .91 .28
M = mean; SD = standard deviation; rit = part-whole corrected item-total
All p's b .001. correlation (related to the total score); p = response probability.
 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413 411

Table 5 methods of the GHQ-12 needs further investigation,

Results of the stepwise multiple regression analysis for the prediction of the although another study by Hankins shows the stability of
global score of the GHQ-12.
the unidimensional solution across different scoring methods
b β T p ΔR 2 [17]. Other authors have realized better fits in models that
BDI 0.11 0.33 13.44 b.001 0.39 treat the responses as ordinal [43,44]. Fortunately, the Likert-
SF 36: Mental Health −0.72 −0.25 −9.65 b.001 0.06 Scoring is very common, and thus our results are applicable
SF 36: Social Functioning −0.30 −0.12 −4.92 b.001 0.01
on a large number of studies. A generalizability theory
PHQ-2 0.41 0.10 4.09 b.001 b0.01
SF 36: Role Limitations (physical) 0.02 0.10 4.44 b.001 b0.01 approach may give additional insights into multiple sources
SF 36: Role Limitations (emotional) −0.02 −0.07 −2.80 .005 b0.01 of variability [45]. In addition, the use of bifactor-models
Multiple R = 0.69, R 2 = 0.47, corrected R 2 = 0.47, F = 306.13, df = 6,
[46,47] may be an alternative promising approach, although
p b .001. these models pose challenges with identification and
convergence. To our knowledge both approaches have not
been tested so far with regard to the GHQ-12.
variance of 47%. However, although six variables are Moreover, Kulenovic et al. [48] mentioned that the
significant predictors, only the BDI and the SF-36 subscale factorial structure of the GHQ-12 depends on the sample
“Mental Health” contribute a relevant amount of explained under study; our results are applicable to the general
variance to the criterion. These two variables alone explain population. Further investigation is needed to clarify its
45% of the variance. The additional variance accounted for stability in other populations (e.g. clinical samples).
by the remaining predictors seems to be negligible. Beside the dimensional interpretation of the GHQ-12, a
categorical interpretation using cut-off-scores is often
applied. In the literature different cut-off-scores are sug-
4. Discussion gested for different populations [49,50], and in different
screening applications [51]. To support this categorical
Although the GHQ-12 is widely used, its dimensionality approach, a reliable cut-off-score is needed. Since our study
is still under debate. Thus our study addresses dimension- does not include clinical interviews, this remains a
ality, psychometric properties and conceptual determination contentious issue.
of the GHQ-12 in a large-scale population based sample. In a Beyond psychometric issues, a conceptual localization of
first step, common factorial models of the GHQ-12 were the GHQ-12 seems useful. Focusing its associations with
tested to determine the dimensionality of the instrument. several established psychometric instruments, the total score
None of the previous models considered (one, two- and of the GHQ-12 is highly correlated with depressive
three-factor solutions) shows sufficient fit indices. Regarding symptomatology, and with impaired mental health and
the debate about the methodical aspects of the dimensionality, vitality (SF-36). On the other hand, it shows only low range
we tested a unidimensional model including response bias on correlations with anxiety (PHQ-anxiety module) and social
the negative formulated items, which was first mentioned by phobia (SPIN). In a stepwise linear regression, only BDI and
Hankins [22]. This model showed sufficient fit indices in the the mental health scale of the SF-36 contributed a substantial
study of Hankins [22] and the best fit of all models examined in amount of explained variance. In summary, the GHQ-12
our study. This result underpins the unidimensionality as it was seems to be a more or less useful screening instrument for the
originally intended in the development of the GHQ-12, yet it assessment of mental distress or a minor psychiatric
emphasizes the importance of the response bias. Furthermore, morbidity with a main focus on depressive symptomatology.
all factors of the suggested two- and three-factors models are With regard to the prime importance of depressive
highly inter-correlated, thus the amount of differential symptomatology among the several mental disorders and
information assessed with the different subscales seems not as a comorbid condition, this seems worthwhile. Neverthe-
to be of particular relevance. Finally, we took a look at the less, several short screening instruments for the different
correlations of the total GHQ-12-score with several estab- aspects of mental distress have been developed and have
lished scales like the BDI, the SF-36 and the PHQ. From a been shown to have better psychometric properties compared
content-related point of view, these findings underpin the to the GHQ-12. Thus we would recommend preferring
unidimensional conceptualization of the GHQ-12. Not least, psychometrically sound instruments like the different
its sufficient internal consistency and its medium-range modules of the Patient Health Questionnaire [28].
response probabilities substantiate the unidimensionality of
the GHQ-12.
Even though our study is based on a large-scale References
population based sample, some shortcomings and future
challenges have to be mentioned. As it was determined in the [1] Goldberg DP, Williams P. A user's guide to the General Health
Questionnaire. Basingstoke: NFER-Nelson; 1988.
study of Campbell et al. [16], factorial structure depends on [2] Werneke U, Goldberg DP, Yalcin I, Ustun BT. The stability of the
the scoring method. Consequently, our results are reliable for factor structure of the General Health Questionnaire. Psychol Med
the Likert-Scoring, but its validity for the alternative scoring 2000;30(4):823-9.
412 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413

[3] Vanheule S, Bogaerts S. Short communication: the factorial structure [24] Schmitz N, Kruse J, Tress W. Psychometric properties of the General
of the GHQ-12. Stress Health 2005;21(4):217-22. Health Questionnaire (GHQ-12) in a German primary care sample.
[4] Toyabe SI, Shioiri T, Kobayashi K, Kuwabara H, Koizumi M, Endo Acta Psychiatr Scand 1999;100(6):462-8.
T, et al. Factor structure of the General Health Questionnaire (GHQ- [25] Hautzinger M, Bailer M, Worall H, Keller F. Beck-Depressions-
12) in subjects who had suffered from the 2004 Niigata-Chuetsu Inventar2nd ed. . Goettingen: Hogrefe; 1995.
earthquake in Japan: a community-based study. BMC Public Health [26] Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory
2007;7:175. for measuring depression. Arch Gen Psychiatry 1961;4:561-71.
[5] Schmitz N, Kruse J, Tress W. Improving screening for mental [27] Richter P, Werner J, Bastine R. Psychometrische Eigenschaften des
disorders in the primary care setting by combining the GHQ-12 and Beck-Depression Inventars (BDI). Eine Überblick. Zeitschrift für
SCL-90-R subscales. Compr Psychiatry 2001;42(2):166-73. Klinische Psychologie 1994;23:3-19.
[6] Politi PL, Piccinelli M, Wilkinson G. Reliability, validity and factor [28] Spitzer RL, Kroenke K, Williams JBW. Validation and utility of a self-
structure of the 12-Item General Health Questionnaire Among Young report version of PRIME-MD – the PHQ primary care study. JAMA
Males in Italy. Acta Psychiatr Scand 1994;90(6):432-7. 1999;282(18):1737-44.
[7] Picardi A, Abeni D, Pasquini P. Assessing psychological distress [29] Löwe B, Spitzer RL, Zipfel S, Herzog W. Gesundheitsfragebogen für
in patients with skin diseases: reliability, validity and factor Patienten (PHQ-D). Maunal und Testunterlagen. Karlsruhe: Pfizer;
structure of the GHQ-12. J Eur Acad Dermatol Venereol 2001; 2002.
15(5):410-7. [30] Lowe B, Spitzer RL, Grafe K, Kroenke K, Quenter A, Zipfel S, et al.
[8] Kalliath TJ, O'Driscoll MP, Brough P. A confirmatory factor analysis Comparative validity of three screening questionnaires for DSM-IV
of the General Health Questionnaire-12. Stress Health 2004;20(1): depressive disorders and physicians' diagnoses. J Affect Disord
11-20. 2004;78(2):131-40.
[9] Gureje O. Reliability and the factor structure of the Yoruba Version of [31] Kroenke K, Spitzer RL, Williams JBW. The Patient Health
the 12-Item General Health Questionnaire. Acta Psychiatr Scand Questionnaire-2 – Validity of a two-item depression screener. Med
1991;84(2):125-9. Care 2003;41(11):1284-92.
[10] Campbell A, Walker J, Farrell G. Confirmatory factor analysis of the [32] Lowe B, Kroenke K, Grafe K. Detecting and monitoring depression
GHQ-12: can I see that again? Aust N Z J Psychiatry 2003;37(4): with a two-item questionnaire (PHQ-2). J Psychosom Res 2005;58(2):
475-83. 163-71.
[11] French DJ, Tait RJ. Measurement invariance in the General Health [33] Sosic Z, Gieler U, Stangier U. Screening for social phobia in medical
Questionnaire-12 in young Australian adolescents. Eur Child Adolesc in- and outpatients with the German version of the Social Phobia
Psychiatry 2004;13(1):1-7. Inventory (SPIN). J Anxiety Disord 2008;22(5):849-59.
[12] Shevlin M, Adamson G. Alternative factor models and factorial [34] Connor KM, Davidson JRT, Churchill LE, Sherwood A, Foa E,
invariance of the GHQ-12: A large sample analysis using confirmatory Weisler RH. Psychometric properties of the Social Phobia
factor analysis. Psychol Assess 2005;17(2):231-6. Inventory (SPIN) – new self-rating scale. Br J Psychiatry 2000;
[13] Cheung YB. A confirmatory factor analysis of the 12-item General 176:379-86.
Health Questionnaire among older people. Int J Geriatr Psychiatry [35] Ware JE, Sherbourne CD. The Mos 36-Item Short-Form Health Survey
2002;17(8):739-44. (Sf-36).1. Conceptual-framework and item selection. Med Care
[14] Graetz B. Multidimensional properties of the General Health 1992;30(6):473-83.
Questionnaire. Soc Psychiatry Psychiatr Epidemiol 1991;26(3): [36] Muthén LK, Muthén BO. Mplus User's Guide. 6th ed. Los Angeles:
132-8. Muthén & Muthén; 2010.
[15] Makikangas A, Feldt T, Kinnunen U, Tolvanen A, Kinnunen ML, [37] Brown TE. Confirmatory factor analysis for applied research. New
Pulkkinen L. The factor structure and factorial invariance of the 12- York: Guilford University Press; 2006.
item General Health Questionnaire (GHQ-12) across time: evidence [38] Little TD, Lindenberger U, Nesselroade JR. On selecting indicators for
from two community-based samples. Psychol Assess 2006;18(4): multivariate measurement and modeling with latent variables: When
444-51. “good” indicators are bad and “bad” indicators are good. Psychol
[16] Campbell A, Knowles S. A confirmatory factor analysis of the Methods 1999;4(2):192-211.
GHQ12 using a large Australian sample. Eur J Psychol Assess 2007; [39] Maccallum RC, Roznowski M, Necowitz LB. Model modifications in
23(1):2-8. covariance structure-analysis – the problem of capitalization on
[17] Hankins M. The reliability of the twelve-item general health chance. Psychol Bull 1992;111(3):490-504.
questionnaire (GHQ-12) under realistic assumptions. BMC Public [40] Fornell C, Larcker DF. Evaluating structural equation models with
Health 2008;8:355. unobservable variables and measurement error. J Mark Res 1981;
[18] Ye SQ. Factor structure of the General Health Questionnaire (GHQ- 18(1):39-50.
12): The role of wording effects. Pers Individ Diff 2009;46(2): [41] Raftery AE. Bayesian model selection in social research. Sociological
197-201. Methodol 1995;25:111-63.
[19] Marsh HW. Positive and negative global self-esteem: a substantively [42] Hu YJ, Bentler PM. Cutoff criteria for fit indexes in covariance
meaningful distinction or artifactors? J Pers Soc Psychol 1996;70(4): structure analysis: Conventional criteria versus new alternatives. Struct
810-9. Equation Model 1999;6(1):1-55.
[20] Mook J, Kleijn WC, Vanderploeg HM. Positively and negatively [43] Wang WC, Cunningham EG. Comparison of alternative estimation
worded items in a self-report measure of dispositional optimism. methods in confirmatory factor analyses of the general health
Psychol Rep 1992;71(1):275-8. questionnaire. Psychol Rep 2005;97(1):3-10.
[21] Roth M, Decker O, Herzberg PY, Brahler E. Dimensionality and norms [44] Aguado J, Campbell A, Ascaso C, Navarro P, Garcia-Esteve L,
of the Rosenberg Self-esteem scale in a German general population Luciano JV. Examining the factor structure and discriminant validity of
sample. Eur J Psychol Assess 2008;24(3):190-7. the 12-item General Health Questionnaire among Spanish postpartum
[22] Hankins M. The factor structure of the twelve item General Health women. Assessment 2012;19(4):517-25.
Questionnaire (GHQ-12): results of negative phrasing? Clin Pract [45] Llabre MM, Fitzpatrick SL. Revisiting measurement models in
Epidemiol Ment Health 2008;4:10. psychosomatic medicine research: a latent variable approach. Psycho-
[23] Wang L, Lin WP. Wording effects and the dimensionality of the som Med 2012;74(2):169-77.
General Health Questionnaire (GHQ-12). Pers Individ Diff 2011; [46] Chen FF, Hayes A, Carver CS, Laurenceau JP, Zhang ZG. Modeling
50(7):1056-61. general and specific variance in multifaceted constructs: a
 M. Romppel et al. / Comprehensive Psychiatry 54 (2013) 406–413 413

comparison of the bifactor model to other approaches. J Pers 2012; [49] Goldberg DP, Oldehinkel T, Ormel J. Why GHQ threshold varies from
80(1):219-51. one place to another. Psychol Med 1998;28(4):915-21.
[47] Chen FF, West SG, Sousa KH. A comparison of bifactor and second- [50] Donath S. The validity of the 12-item General Health Questionnaire in
order models of quality of life. Multivariate Behav Res 2006;41(2): Australia: a comparison between three scoring methods. Aust N Z J
189-225. Psychiatry 2001;35(2):231-5.
[48] Kulenovic M, Kusevic Z, Grba S. Factor analysis of the General Health [51] Clarke DM, Smith GC, Herrman HE. A comparative-study of
Questionnaire (GHQ 12) on the sample of the unemployed and screening instruments for mental-disorders in general-hospital patients.
students. Coll Antropol 1995;19(2):407-11. Int J Psychiatry Med 1993;23(4):323-37.