General Health Questionnaire-28 (GHQ-28)

Abstract

The General Health Questionnaire-28 (GHQ-28) is one of the most extensively utilized and psychometrically validated screening instruments designed to detect current non-psychotic psychiatric morbidity, general psychological distress, and common mental disorders across community and primary care populations. Originally developed by Sir David Goldberg and V. F. Hillier in 1979 as a shortened, 28-item multidimensional variant of the original 60-item General Health Questionnaire (GHQ-60), the scale assesses short-term changes in psychological and physiological functioning rather than enduring personality traits. The instrument comprises four distinct 7-item subscales derived through factor analysis: Somatic Symptoms (Subscale A), Anxiety and Insomnia (Subscale B), Social Dysfunction (Subscale C), and Severe Depression (Subscale D). Respondents evaluate their health over recent weeks on a 4-point response continuum ranging from “not at all” to “much more than usual”. Scoring can be executed via binary GHQ scoring (0-0-1-1), which emphasizes clinically meaningful departures from standard baseline functioning, or via conventional Likert scaling (0-1-2-3 or 1-2-3-4), which yields continuous parametric distributions suitable for structural equation modeling and linear regression analyses.

Across four decades of empirical investigation spanning dozens of languages and cultural contexts, the GHQ-28 has demonstrated exceptional psychometric properties. Internal consistency estimates routinely yield Cronbach’s alpha coefficients between .82 and .93 for individual subscales and exceeding .90 for the total composite score. Criterion-related validity against gold-standard psychiatric interviews—such as the Clinical Interview Schedule (CIS-R), the Structured Clinical Interview for DSM Disorders (SCID), and the Composite International Diagnostic Interview (CIDI)—demonstrates high diagnostic accuracy, with receiver operating characteristic (ROC) analyses frequently reporting areas under the curve (AUC) between .85 and .95, sensitivity estimates between 76% and 88%, and specificity parameters between 78% and 90% at standard threshold cutoffs. Confirmatory factor analytic investigations widely support the four-factor correlated model, although hierarchical and bifactor models featuring a pervasive general psychological distress factor alongside specific orthogonal symptom dimensions are also robustly supported. The GHQ-28 remains a premier screening tool in psychiatric epidemiology, occupational medicine, clinical psychology, and psychosomatic research worldwide.

Keywords

General Health Questionnaire-28, GHQ-28, David Goldberg, psychological distress, psychiatric screening, somatic symptoms, anxiety and insomnia, social dysfunction, severe depression, psychometrics, primary care epidemiology, mental health assessment

Authors

The General Health Questionnaire family of instruments was conceptualized, developed, and empirically validated by Sir David P. Goldberg, Professor Emeritus of Psychiatry at the Institute of Psychiatry, Psychology and Neuroscience (IoPPN), King’s College London, and former Professor of Psychiatry at the University of Manchester, United Kingdom. The specific 28-item scaled version was developed in collaboration with V. F. Hillier, a statistician at the Department of Community Medicine, University of Manchester, who conducted the original factor-analytic derivation published in 1979.

Subsequent psychometric standardization, cross-cultural adaptations, and validation guidelines were coordinated extensively by David P. Goldberg in partnership with Paul Williams (Institute of Psychiatry, London) through their authoritative clinical manual, A User’s Guide to the General Health Questionnaire (1988, 1991). Major international normative studies, including the landmark World Health Organization (WHO) Collaborative Study on Psychological Problems in General Health Care, were conducted under the leadership of David Goldberg, Richard Gater, Norman Sartorius, T. Bedirhan Üstün, and colleagues across 14 countries.

Purpose

The primary clinical and epidemiological purpose of the General Health Questionnaire-28 is to identify short-term psychiatric morbidity and minor, non-psychotic psychological distress in general medical settings, workplaces, and community samples. Unlike comprehensive diagnostic interviews designed to establish lifetime psychiatric taxonomy according to the Diagnostic and Statistical Manual of Mental Disorders (DSM-5-TR) or the International Classification of Diseases (ICD-11), the GHQ-28 functions as a cost-effective, rapid first-stage screening instrument. It differentiates individuals who are currently experiencing acute distress, affective disturbance, or somatic decompensation requiring clinical attention from those who maintain baseline psychological equilibrium.

The theoretical rationale governing the design of the GHQ-28 centers on “breakage” or current deviations from an individual’s habitual state of health. Traditional psychological inventories frequently confound long-standing trait neuroticism or chronic somatic illness with current psychological disorder. Goldberg specifically constructed the GHQ items to assess whether respondents have recently experienced symptoms that are worse or more pronounced than their typical baseline. This focus on current deviation enables clinicians to filter out stable, chronic conditions while capturing emergent psychiatric morbidity, such as generalized anxiety, major depressive episodes, adjustment disorders, and stress-induced somatization.

In clinical practice, the GHQ-28 serves several distinct operational functions:

• Primary Care Triage: General practitioners routinely encounter patients presenting with vague somatic complaints that mask underlying affective or anxiety disorders. The GHQ-28 provides an empirical mechanism to uncover hidden psychological distress without imposing excessive burden on routine clinical workflows.
• Epidemiological Surveillance: In public health and population research, the GHQ-28 provides a standardized metric of community mental health burden, facilitating cross-demographic comparisons and longitudinal tracking of psychological resilience following societal crises, natural disasters, or macroeconomic fluctuations.
• Occupational Health Monitoring: Organizations utilize the instrument to detect occupational burnout, chronic work-related strain, and executive dysfunction, allowing human resources and health units to deploy targeted psychological interventions prior to clinical absenteeism.
• Treatment Evaluation: Because the scale is acutely sensitive to state changes, researchers and clinicians employ the GHQ-28 as an outcome measure to track therapeutic response following pharmacotherapy, cognitive-behavioral interventions, or lifestyle modifications.

Psychological Construct

The GHQ-28 operationalizes psychological distress across four interconnected yet clinically distinct dimensions. Each subscale consists of seven items derived through principal components analysis with varimax rotation, intentionally selected to minimize item overlap while maximizing diagnostic coverage of common mental disorders.

Subscale A: Somatic Symptoms (Items 1–7)

Subscale A evaluates physiological manifestations of psychological strain and subjective somatic wellbeing. Rather than indexing severe structural pathology, this dimension captures functional somatic symptoms frequently associated with autonomic nervous system arousal, stress-induced exhaustion, and affective dysregulation. Items assess perceived general health, the subjective feeling of being “run down and out of sorts,” the perceived need for a restorative tonic, general sensations of physical illness, tension headaches, cranial pressure or tightness, and autonomic instability manifested through alternating hot or cold spells. High scores on this subscale reflect substantial bodily distress, conversion phenomena, or hypochondriacal preoccupation secondary to emotional turmoil.

Subscale B: Anxiety and Insomnia (Items 8–14)

Subscale B measures affective, cognitive, and physiological indicators of generalized anxiety, hyperarousal, and associated sleep disruption. The domain captures the hallmark cognitive-affective features of clinical anxiety, including losing sleep over persistent worry, severe middle and terminal insomnia (difficulty staying asleep once off), feeling constantly under psychological strain, irritability and affective lability (“getting edgy and bad-tempered”), spontaneous episodes of unprovoked panic or acute fear, subjective sensations of cognitive overload (“found everything getting on top of you”), and tonic motor tension characterized by feeling nervous and strung-up all the time. This subscale directly indexes sympathetic hyperactivation and cognitive rumination.

Subscale C: Social Dysfunction (Items 15–21)

Subscale C assesses functional impairment, behavioral competence, and hedonic capacity in daily interpersonal and task-oriented domains. Uniquely framed around everyday activities, this subscale captures disruptions in executive functioning, motivational inertia, and loss of productivity. Items query the individual’s capacity to keep busy and occupied, cognitive slowing during routine tasks (taking longer over things), self-perceptions of competence and effectiveness, task satisfaction, subjective sense of utility and social worth (“playing a useful part in things”), decisional autonomy, and the capacity to derive pleasure from habitual activities (anhedonia). Elevated scores on this subscale signify functional paralysis, social withdrawal, and the erosion of self-efficacy.

Subscale D: Severe Depression (Items 22–28)

Subscale D indexes the most critical, life-threatening dimension of non-psychotic psychiatric morbidity: severe depressive symptomatology, cognitive despair, and suicidal ideation. This subscale transcends mild demoralization to evaluate severe affective collapse. Items assess core cognitive depressive schemas, including pervasive feelings of personal worthlessness, absolute hopelessness regarding the future, nihilistic perceptions that life is not worth living, explicit contemplation of self-harm or suicide (“thought of the possibility that you might make away with yourself”), severe functional collapse due to emotional breakdown, passive death wishes (“wishing you were dead and away from it all”), and intrusive, recurrent suicidal thoughts. Due to its direct interrogation of active suicidality, this subscale functions as an indispensable clinical safety alert.

Theoretical Framework

The architectural foundation of the General Health Questionnaire-28 rests upon Goldberg’s Threshold Model of Psychiatric Morbidity and the wider biopsychosocial paradigm of minor psychiatric disorders. During the 1960s and 1970s, psychiatric nosology was predominantly divided between categorical hospital-based classifications of psychotic illnesses and specialized psychoanalytic formulations of neuroses. Goldberg recognized that this dichotomy fundamentally failed to capture the realities of community healthcare, where millions of individuals experienced debilitating emotional distress that did not meet classical criteria for schizophrenia or bipolar manic-depressive illness, yet impaired their biological, vocational, and social functioning.

Goldberg posited that psychological disorder exists along a continuous quantitative spectrum of severity, running from optimal psychological well-being, through transient situational demoralization, to full-blown clinical disorder. Within this continuum, Goldberg introduced the concept of the “Threshold of Morbidity”. The GHQ was explicitly engineered to identify the precise inflection point where distress transitions from normative human adaptation into clinically significant impairment requiring intervention. This conceptualization anticipated modern dimensional frameworks of psychopathology, including the Hierarchical Taxonomy of Psychopathology (HiTOP).

A second foundational pillar of the GHQ theoretical framework is the “State versus Trait” Distinction. Influenced by psychometric theories separating enduring personality vulnerability (e.g., Eysenckian Neuroticism) from transient clinical states (e.g., Spielberger’s State-Trait Anxiety model), Goldberg asserted that an effective screening instrument must measure acute clinical “breakage”. Chronic neurotics or individuals with long-standing personality disorders may live permanently with elevated trait anxiety. If an instrument measures habitual status, such individuals score high indefinitely, diluting the instrument’s sensitivity to acute psychiatric decompensation. By phrasing items relative to what is normal for the respondent (“no more than usual” versus “rather more than usual”), the GHQ framework isolates active symptomatic episodes from characterological baseline functioning.

Furthermore, the factor-derived 4-subscale architecture of the GHQ-28 reflects the clinical observation that affective decompensation unfolds across multiple interrelated physiological and behavioral domains. Goldberg and Hillier recognized that common mental disorders frequently present as a triad of somatization, hyperarousal/insomnia, and social-occupational dysfunction, which may culminate in severe cognitive depression and suicidality. By decomposing the general distress construct into these four distinct sub-domains, the GHQ-28 provides clinicians with an idiographic psychological profile that elucidates the specific behavioral and somatic pathways through which distress manifests in a given patient.

Validity

The validity of the GHQ-28 has been documented through hundreds of empirical investigations conducted across inpatient, primary care, community, and occupational cohorts worldwide.

Criterion-Related Validity

Criterion validity has been established by benchmarking the GHQ-28 against validated semi-structured diagnostic interviews. In the seminal validation study by Goldberg and Hillier (1979), the GHQ-28 was evaluated against the Clinical Interview Schedule (CIS). Using the traditional GHQ scoring method (0-0-1-1) at a threshold cutoff score of 4/5, the instrument achieved a sensitivity of 88.0% and a specificity of 84.2%, with an overall misclassification rate of only 13.8%. In the landmark multi-center WHO Collaborative Study on Psychological Problems in General Health Care (Goldberg et al., 1997), spanning 15 centers across the globe, the GHQ-28 demonstrated robust diagnostic performance across diverse linguistic, socioeconomic, and cultural environments, yielding an overall sensitivity of 83.4% and a specificity of 82.7% against the Composite International Diagnostic Interview (CIDI).

Construct and Factorial Validity

Construct validity has been verified through extensive exploratory and confirmatory factor analyses. The structural independence and mutual correlation of the four latent factors—Somatic Symptoms, Anxiety/Insomnia, Social Dysfunction, and Severe Depression—have been replicated across multiple nationalities and clinical populations. Research consistently confirms that each 7-item set converges strongly upon its designated latent construct with high standardized factor loadings (typically ranging from .55 to .88), while maintaining moderate inter-factor correlations (typically between .40 and .68). This structure confirms that while the four subscales reflect distinct facets of psychological distress, they contribute coherently to a broader overarching construct of psychological morbidity.

Convergent and Discriminant Validity

Convergent validity is documented through substantial correlations between GHQ-28 subscales and established dimensional measures of affective psychopathology:

• Subscale B (Anxiety/Insomnia) correlates strongly with the State-Trait Anxiety Inventory (STAI; r = .71 to .78) and the Beck Anxiety Inventory (BAI; r = .68 to .75).
• Subscale D (Severe Depression) demonstrates convergent validity with the Beck Depression Inventory (BDI-II; r = .74 to .83) and the Center for Epidemiologic Studies Depression Scale (CES-D; r = .70 to .80).
• Subscale A (Somatic Symptoms) correlates with the Patient Health Questionnaire-15 (PHQ-15; r = .62 to .71).
• Total Score: The overall GHQ-28 score exhibits correlations ranging from .65 to .82 with the Hospital Anxiety and Depression Scale (HADS) total score and the Brief Symptom Inventory (BSI).

Discriminant validity is evidenced by the tool’s capacity to differentiate between non-psychiatric control populations, primary care attenders with transient distress, and patients with established mood or anxiety disorders. Furthermore, longitudinal clinical trials demonstrate significant decreases in total and subscale GHQ-28 scores following successful pharmacotherapy or psychotherapy, confirming the scale’s sensitivity to clinical change.

Reliability

The GHQ-28 demonstrates internal consistency, split-half reliability, and test-retest stability across diverse research settings.

Internal Consistency

Numerous psychometric studies employing Likert scoring have reported Cronbach’s alpha coefficients for the full 28-item scale ranging between .90 and .95, reflecting high measurement precision. When evaluated at the subscale level, internal consistency estimates consistently meet or exceed acceptable psychometric standards:

• Subscale A (Somatic Symptoms): Cronbach’s α typically ranges from .82 to .89.
• Subscale B (Anxiety and Insomnia): Cronbach’s α typically ranges from .85 to .92.
• Subscale C (Social Dysfunction): Cronbach’s α typically ranges from .78 to .86.
• Subscale D (Severe Depression): Cronbach’s α typically ranges from .88 to .94.

Studies evaluating McDonald’s omega (ω_t and ω_h) within structural equation modeling paradigms have corroborated these findings, reporting total omega values above .92 for the full instrument, confirming that the high alpha values reflect true construct saturation rather than item redundancy.

Test-Retest Reliability

Because the GHQ-28 is deliberately calibrated to measure transient “state” changes in psychological distress, test-retest reliability is influenced by the retest interval. Stability coefficients are highest over short intervals where true psychiatric status remains unchanged. Over test-retest intervals of 1 to 2 weeks, Pearson correlation coefficients and intra-class correlation coefficients (ICC) range from .78 to .90 in clinically stable populations. When retest intervals extend beyond 4 to 6 weeks, coefficients drop into the .50 to .65 range, reflecting genuine clinical fluctuations, spontaneous remission, or treatment effects rather than psychometric instability.

Split-Half Reliability

Early evaluations by Goldberg and Hillier (1979) established split-half reliability coefficients using the Spearman-Brown prophecy formula exceeding .90. Subsequent cross-cultural investigations (e.g., in Spanish, Arabic, Chinese, and Turkish validations) have confirmed split-half reliability figures ranging between .85 and .93, demonstrating the structural coherence of the instrument across varying linguistic translations.

Factor Analysis

The factor architecture of the GHQ-28 represents a cornerstone of its psychometric legacy. The scale was developed through an exploratory factor analysis (EFA) of the original 60-item instrument administered to a sample of 523 general medical patients (Goldberg & Hillier, 1979). Using principal components analysis with varimax orthogonal rotation, the authors extracted four robust factors that explained the largest shares of variance. The seven items loading most strongly on each respective factor without substantial cross-loadings (>.30 on competing factors) were selected to form the 28-item scaled instrument.

Exploratory Factor Analytic Findings

In Goldberg and Hillier’s original derivation, the four extracted components accounted for 53.5% of the total variance:

• Factor I (Somatic Symptoms): Explained approximately 25.7% of the total variance, capturing autonomic complaints, feelings of exhaustion, and bodily weakness.
• Factor II (Anxiety and Insomnia): Accounted for roughly 12.8% of the variance, with high loadings on terminal insomnia, cognitive worry, and feelings of constant tension.
• Factor III (Social Dysfunction): Accounted for approximately 8.3% of the variance, indexing disruptions in daily routine execution, anhedonia, and impaired decision-making.
• Factor IV (Severe Depression): Explained roughly 6.7% of the variance, characterized by feelings of worthlessness, hopelessness, and explicit suicidal ideation.

Confirmatory Factor Analysis (CFA) and Structural Equation Modeling

Subsequent modern psychometric studies have tested several alternative structural models using Confirmatory Factor Analysis (CFA):

• Unidimensional Model: A single global distress factor loading onto all 28 items. This model uniformly demonstrates poor fit across structural studies (e.g., Root Mean Square Error of Approximation [RMSEA] > .10; Comparative Fit Index [CFI] < .80), indicating that general psychological distress cannot adequately explain the multi-faceted symptom expression.
• Four-Factor Correlated Model: The classical Goldberg & Hillier model with four distinct, correlated latent factors. This model consistently achieves good to excellent fit across diverse international datasets (e.g., RMSEA between .045 and .062; CFI between .92 and .96; Tucker-Lewis Index [TLI] between .91 and .95). Standardized factor loadings across items consistently range between .50 and .88, confirming the structural validity of the original four-subscale arrangement.
• Bifactor Model: Incorporates a single general distress factor (the “g-factor” of mental morbidity) alongside four orthogonal specific factors. Recent investigations (e.g., testing large community and occupational samples) demonstrate that the bifactor model yields superior global fit indices (RMSEA < .04; CFI > .97), explaining why both the composite total score and the four individual subscale scores provide valuable, non-redundant clinical information.

Instrument / Measurement Tool

The technical parameters and administration characteristics of the General Health Questionnaire-28 are outlined below:

• Instrument Name: General Health Questionnaire-28 (GHQ-28).
• Primary Developer: Sir David P. Goldberg (with V. F. Hillier, 1979).
• Instrument Type: Self-report screening inventory for non-psychotic psychiatric morbidity and general psychological distress.
• Target Population: Adults and adolescents aged 16 years and older; suitable for primary care patients, occupational cohorts, community samples, and medical outpatients.
• Administration Format: Paper-and-pencil questionnaire, digital/web-based assessment, or structured clinical interview administration.
• Completion Time: Approximately 5 to 10 minutes.
• Total Item Count: 28 items, organized into four balanced subscales of 7 items each:
• Subscale A: Somatic Symptoms (Items 1–7)
• Subscale B: Anxiety and Insomnia (Items 8–14)
• Subscale C: Social Dysfunction (Items 15–21)
• Subscale D: Severe Depression (Items 22–28)
• Response Scale Options: A 4-point response continuum indexing departure from habitual baseline:
• 1 = Not at all
• 2 = No more than usual
• 3 = Rather more than usual
• 4 = Much more than usual
• Scoring Protocols:
• Traditional GHQ Scoring (0-0-1-1): Responses 1 and 2 are scored as “0” (absence of morbidity), while responses 3 and 4 are scored as “1” (presence of morbidity). Total scores range from 0 to 28. Standard clinical threshold cutoff is commonly set at 4/5 (scores ≥ 5 identify probable psychiatric “caseness”), though cutoffs of 5/6 or 6/7 are adopted in populations with elevated base-rate somatic morbidity.
• Likert Scoring (0-1-2-3 or 1-2-3-4): Items are scored continuously from 0 to 3 (or 1 to 4), producing a total score range of 0 to 84 (or 28 to 112). Subscale scores range from 0 to 21. Likert scoring produces a wider parametric distribution favored in epidemiological modeling, regression analyses, and psychometric factor modeling. Common cutoff for caseness using 0-1-2-3 Likert scoring is typically 23/24.
• Modified Scoring (C-GHQ): Used occasionally to prevent the under-detection of chronic conditions by re-weighting persistent symptoms.
• Critical Clinical Safety Alert: Items 25 (“Thought of the possibility that you might make away with yourself?”) and 28 (“Found that the idea of taking your own life kept coming into your mind?”) assess active suicidal ideation. Any endorsement of these items (≥ 3 on Likert, or 1 on binary scoring) mandates immediate comprehensive clinical risk evaluation regardless of the overall composite score.

Permissions & Fee and Test Year

The General Health Questionnaire-28 was formally introduced to the scientific literature by David P. Goldberg and V. F. Hillier in their 1979 publication in Psychological Medicine, following the initial release of the broader GHQ family in 1972 and 1978. The commercial copyright for the English-language version and all authorized official translations is held by GL Assessment (formerly NFER-Nelson Publishing Company Ltd., Berkshire, United Kingdom).

The GHQ-28 is a proprietary instrument. Researchers and clinicians wishing to utilize the instrument in formal investigations, commercial health assessments, clinical trials, or digital health platforms are required to purchase authorized test packs, scoring licenses, or digital user permissions directly from GL Assessment (www.gl-assessment.co.uk). A licensing fee per administration is standard practice. While the scale items have been reprinted extensively in public academic dissertations, historical validation papers, and governmental technical reports, official usage in clinical diagnostic protocols and published research requires formal copyright clearance and administrative compliance.

References

• Goldberg, D. P., & Hillier, V. F. (1979). A scaled version of the General Health Questionnaire. Psychological Medicine, 9(1), 139–145. https://doi.org/10.1017/s0033291700021644
• Goldberg, D. P., & Williams, P. (1988). A User’s Guide to the General Health Questionnaire. NFER-Nelson.
• Goldberg, D. P., Gater, R., Sartorius, N., Üstün, T. B., Piccinelli, M., Gureje, O., & Rutter, C. (1997). The validity of two versions of the GHQ in the WHO study of mental illness in general health care. Psychological Medicine, 27(1), 191–197. https://doi.org/10.1017/s0033291796004242
• Hoelterhoff, M. E. (2010). Resilience against death anxiety in relationship to post-traumatic stress disorder and psychiatric co-morbidity (Doctoral dissertation, University of Plymouth). PEARL Research Repository. http://pearl.plymouth.ac.uk/pearl_jspui/bitstream/10026.1/306/4/Hoelterhoff%20M%20E_2010.pdf
• Sterling, M. (2011). General Health Questionnaire–28 (GHQ-28). Journal of Physiotherapy, 57(4), 259. https://doi.org/10.1016/S1836-9553(11)70060-1
• Templer, D. I. (1970). The construction and validation of a death anxiety scale. The Journal of General Psychology, 82(2), 165–177. https://doi.org/10.1080/00221309.1970.9920634
• Willmott, L., Boardman, F. K., & Henshaw, C. A. (2004). Understanding the General Health Questionnaire: Its structure and measurement properties in an antenatal cohort. Social Psychiatry and Psychiatric Epidemiology, 39(8), 613–620. https://doi.org/10.1007/s00127-004-0803-0

Items of the Scale