Duke Anxiety-Depression Scale (DUKE-AD)

Abstract

The Duke Anxiety-Depression Scale (DUKE-AD) is a brief, clinician-administered or self-administered screening instrument designed to detect and quantify symptoms of anxiety and major depressive disorder within ambulatory medical care environments. Developed by George R. Parkerson Jr. and colleagues at the Duke University Medical Center Department of Community and Family Medicine, the DUKE-AD is an empirically derived subscale drawn from the broader 17-item Duke Health Profile (DUKE). The instrument comprises exactly seven items calibrated across two distinct response formats: three general self-descriptive items evaluated on a 3-point frequency-congruence metric (“Yes, describes me exactly”, “Somewhat describes me”, and “No, doesn’t describe me at all”), and four symptom-attribution items evaluating difficulties experienced during the past week on a 3-point severity continuum (“None”, “Some”, and “A Lot”). Psychometrically, the DUKE-AD captures the core dimensional overlap of common psychological distress—frequently conceptualized in clinical epidemiology as mixed anxiety-depressive disorder—while maintaining high diagnostic efficiency. Validation investigations utilizing Receiver Operating Characteristic (ROC) analyses demonstrate that the scale possesses strong discriminative accuracy against structured clinical interviews based on DSM criteria, exhibiting sensitivity values typically ranging between 0.75 and 0.88 and specificity parameters exceeding 0.70 at standard diagnostic thresholds. Internal consistency reliability estimates (Cronbach’s alpha) consistently range between 0.72 and 0.80 across diverse primary care cohorts, complemented by strong test-retest reproducibility over short clinical intervals. By mitigating the substantial administrative burden common to longer diagnostic batteries, the DUKE-AD functions as a psychometrically robust, practical case-finding and symptom-monitoring instrument for primary care practitioners, health service researchers, and epidemiological investigators.

Keywords

Duke Anxiety-Depression Scale, DUKE-AD, Duke Health Profile, primary care screening, anxiety disorders, major depression, psychometrics, receiver operating characteristic, negative affectivity, affective disorders

Authors

The DUKE-AD was originated and standardized by primary care researchers and clinical epidemiologists located at Duke University:

• George R. Parkerson Jr., MD, MPH — Department of Community and Family Medicine, Duke University Medical Center, Durham, North Carolina, United States.
• W. E. Broadhead, MD, PhD — Department of Community and Family Medicine, Duke University Medical Center, Durham, North Carolina, United States.
• Chiu-Kit J. Tse, MPH — Department of Community and Family Medicine, Duke University Medical Center, Durham, North Carolina, United States.

Institutional correspondence regarding the Duke Health Measures family of clinical tools is coordinated through the Department of Family Medicine and Community Health, Duke University School of Medicine, Durham, NC 27710.

Purpose

The primary clinical and epidemiological objective of the Duke Anxiety-Depression Scale (DUKE-AD) is to address the pronounced challenge of detecting unrecognized affective and anxiety pathology in ambulatory primary care settings. Epidemiological studies historically indicate that between 15% and 30% of adult patients presenting to general family practice clinics experience clinically significant symptoms of anxiety, depression, or both. Despite this high prevalence, substantial proportions of psychiatric morbidity remain undiagnosed and untreated during routine medical encounters. Primary care visits are characterized by extreme time constraints, complex multimorbidity, and the prioritization of acute physical complaints, rendering comprehensive psychiatric interviews or lengthy diagnostic rating scales unfeasible.

The DUKE-AD was engineered to resolve this logistical dilemma by delivering a brief, non-stigmatizing, seven-item dimensional screener that can be completed by respondents in under two minutes without requiring clinical supervision. Beyond general case-finding, the purpose of the scale encompasses:

• Initial Clinical Triage: Rapidly identifying ambulatory patients who manifest threshold levels of distress warranting comprehensive psychiatric evaluation, structured diagnostic assessment (such as the DSM-5-TR clinical interview), or targeted medical evaluation to rule out secondary organic etiologies.
• Comorbidity Assessment: Acknowledging the profound symptom convergence between unipolar depression and generalized anxiety disorders. Rather than forcing artificial psychopathological separations during initial intake, the DUKE-AD assesses the shared affective and somatic spectrum characteristic of primary care distress presentations.
• Longitudinal Treatment Tracking: Serving as a continuous metric of therapeutic response, allowing clinicians to monitor symptom attenuation following pharmacological interventions (e.g., selective serotonin reuptake inhibitors) or evidence-based psychotherapeutic protocols (e.g., cognitive behavioral therapy).
• Health Services and Outcomes Research: Providing epidemiologists and health economists with an ultra-short, standardized parameter of psychological dysfunction embedded within broad population health surveys, health-related quality of life (HRQoL) evaluations, and primary care effectiveness studies.

Psychological Construct

The DUKE-AD is grounded in the dimensional measurement of psychological distress, specifically capturing the syndromic intersection of depressive dysphoria and autonomic/cognitive anxiety. Historically, categorical diagnostic taxonomies treat major depressive disorder and generalized anxiety disorder as distinct clinical entities. However, psychometric evidence and clinical realities within primary care consistently indicate that these disorders share substantial genetic liability, overlapping neurobiological substrates, and a high degree of phenotypic congruence. The DUKE-AD operationalizes this continuum across seven operational facets:

1. Demoralization and Volitional Exhaustion (Item 1)

Reflected by the item “I give up too easily”, this facet measures perceived helplessness, diminished self-efficacy, and the motivational deficits that accompany depressive states. It taps into the cognitive triad described in classic models of depression, reflecting an individual’s evaluation of their capacity to sustain goal-directed effort when confronted with minor everyday obstacles.

2. Subjective Cognitive Dysfunction (Item 2)

Operationalized through “I have difficulty concentrating”, this dimension captures executive and attentional disruptions common to both anxiety and depressive syndromes. In anxiety, intrusive worry and hypervigilance capture working memory resources; in depression, psychomotor slowing and ruminative absorption degrade attentional focus. Cognitive disruption represents one of the most reliable functional predictors of occupational and social impairment in mood pathology.

3. Interpersonal Comfort vs. Social Anhedonia / Social Phobia (Item 3)

Expressed by the reverse-scored statement “I am comfortable being around people”, this item targets the social avoidance, interpersonal hypersensitivity, and withdrawal that demarcate both social anxiety disorder and the profound anhedonia of unipolar depression. When individuals experience high negative affect, social interactions often shift from being supportive to threatening or exhausting, resulting in behavioral isolation.

4. Somatosensory and Vegetative Disturbances (Items 4 & 5)

Vegetative and somatoform indicators of distress are assessed by evaluating the severity of “Sleeping” problems and “Getting tired easily” over the preceding seven days. Sleep fragmentation, terminal insomnia, or initial sleep-onset latency are hallmark manifestations of circadian disruption in mood disorders and autonomic hyperactivity in anxiety states. Similarly, pervasive fatigue (anergia) reflects the physiological toll of chronic neuroendocrine activation (e.g., dysregulation of the hypothalamic-pituitary-adrenal axis).

5. Affective Dysphoria and Autonomic Arousal (Items 6 & 7)

The overt affective core is measured directly through past-week evaluations of “Feeling depressed or sad” and “Nervousness”. Item 6 taps the classical mood disturbance of depression (depressed mood, sorrow, pervasive despondency), whereas Item 7 captures the subjective experience of tension, apprehension, and psychic anxiety characteristic of heightened arousal states.

Theoretical Framework

The design and operational architecture of the DUKE-AD are theoretically aligned with the Tripartite Model of Anxiety and Depression, introduced by David Watson and Lee Anna Clark (1991). The Tripartite Model posits that the widespread comorbidity and empirical cross-loading observed between anxiety and depressive measures can be explained by three underlying psychological structures:

• General Distress / Negative Affectivity (NA): A pervasive, non-specific dispositional factor characterized by subjective distress, unpleasurable engagement, emotional lability, irritability, cognitive difficulties, fatigue, and general demoralization. This broad dimension accounts for the strong shared variance between depression and anxiety inventories.
• Physiological Hyperarousal (PH): A specific construct characterized by somatic tension, autonomic arousal, trembling, palpitations, and pervasive nervousness, representing the distinct psychobiological marker of panic and generalized anxiety.
• Low Positive Affectivity (Anhedonia): A specific construct marked by the absence of positive emotionality, energy, enthusiasm, and social interest, serving as the distinguishing feature of clinical depression.

The DUKE-AD intentionally samples across this tripartite continuum. Items reflecting general dysphoria, concentration disruption, sleep disturbance, and fatigue (Items 1, 2, 4, 5) function as direct markers of broad Negative Affectivity. Item 3 (lack of social comfort) reflects anhedonic withdrawal and social avoidance, while Item 7 (nervousness) captures physiological hyperarousal and psychic tension. Consequently, the DUKE-AD avoids the structural pitfall of attempting an artificial diagnostic bifurcation in primary care settings, instead capitalizing on the overarching latent factor of common affective distress.

Furthermore, the instrument reflects George L. Engel’s Biopsychosocial Model. Developed within the context of family medicine, the DUKE-AD views somatic symptoms (fatigue, sleep disturbance) not merely as isolated physical complaints, but as integrated functional expressions of emotional equilibrium and systemic health status.

Validity

The psychometric validity of the DUKE-AD has been established through multiple large-scale cross-sectional and prospective cohort investigations conducted in primary care settings.

Criterion and Diagnostic Validity (ROC Analyses)

The foundational validation by Parkerson and Broadhead (1997) assessed the diagnostic accuracy of the DUKE-AD against structured psychiatric criteria. Using the Diagnostic Interview Schedule (DIS) and standardized clinician ratings based on the DSM-III-R and DSM-IV as criterion standards, Receiver Operating Characteristic (ROC) curves were constructed:

• Area Under the Curve (AUC): The DUKE-AD demonstrated robust discriminative performance, with AUC values consistently exceeding 0.80 for detecting any current anxiety disorder, major depressive disorder, or mixed anxiety-depression.
• Sensitivity and Specificity: At the optimal diagnostic cut-off score (typically a scaled score of ≥ 50 on the 0–100 transformed continuum, or ≥ 5 on the raw scale), the instrument yielded sensitivity estimates between 0.76 and 0.86 and specificity values between 0.71 and 0.81.
• Predictive Values: In family medicine populations with an estimated psychiatric prevalence of ~20%, the negative predictive value (NPV) approached 0.93 to 0.96, confirming the scale’s utility as a rapid “rule-out” screener.

Convergent and Concurrent Validity

The DUKE-AD exhibits strong, statistically significant correlations with established legacy instruments measuring psychological distress:

• Correlation with the Center for Epidemiologic Studies Depression Scale (CES-D): r = 0.68 to 0.74 (p < .001).
• Correlation with the Mental Health Inventory (MHI-5) of the SF-36: r = -0.71 to -0.76 (p < .001).
• Correlation with the Zung Self-Rating Depression Scale: r = 0.65 to 0.72 (p < .001).
• Correlation with the Symptom Checklist-90 (SCL-90) Anxiety and Depression subscales: r = 0.67 to 0.75 (p < .001).

Discriminant Validity

Discriminant validity is supported by significantly lower correlations between the DUKE-AD and physical health indices, such as objective physical mobility measures (r < 0.25) and biological parameters (e.g., blood pressure, glycemic markers; r < 0.15). While the vegetative items (fatigue, sleep) show weak-to-moderate associations with physical multimorbidity, the composite score remains distinct from purely physical functioning profiles.

Reliability

The DUKE-AD displays solid reliability parameters across heterogeneous ambulatory care and community-dwelling populations:

Internal Consistency

Given the brevity of the scale (seven items) and the multidimensional nature of mixed anxiety-depressive symptoms, the scale maintains strong internal consistency without redundancy:

• Cronbach’s Alpha (α): Across primary care validation cohorts, internal consistency estimates consistently range between 0.72 and 0.81. In the original standardization sample by Parkerson et al. (1990, 1996), the alpha coefficient was recorded at 0.75.
• Item-Total Correlations: Corrected item-total correlations for the seven items range between 0.38 and 0.64. Affective items (depressed mood, nervousness) generally exhibit the highest corrected item-total correlations (> 0.55), while Item 3 (social ease, reversed) demonstrates lower but adequate structural cohesion (~0.40).

Test-Retest Reliability and Stability

Temporal stability evaluations in clinically stable adult cohorts assessed at short-term intervals (1 to 4 weeks) demonstrate high reproducibility:

• Intraclass Correlation Coefficient (ICC): Test-retest ICCs range from 0.73 to 0.82 across diverse patient groups.
• Pearson’s Correlation Coefficient (r): Short-term stability yielded Pearson r values of 0.78 over a one-week retest window among patients experiencing unchanged clinical status.

The Standard Error of Measurement (SEM) for the transformed 0–100 score is approximately 6.5 to 8.0 points, indicating that changes exceeding 12 to 15 points on the standardized scale reflect meaningful clinical shifts rather than psychometric noise.

Factor Analysis

The underlying dimensionality of the DUKE-AD has been analyzed through both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) within the broader structural context of the Duke Health Profile.

Exploratory Factor Structure

When evaluated within the full 17-item DUKE pool, principal component analysis with varimax rotation routinely isolates a distinct “Mental Health / Emotional Functioning” dimension onto which the DUKE-AD items load heavily (> 0.50). When factor analyzed as an independent 7-item set, EFA typically yields an essential unidimensional structure explaining approximately 42% to 48% of the total item variance, characterized by a single dominant eigenvalue (> 2.80) followed by a sharp drop-off in subsequent factors (scree elbow at factor 2).

Confirmatory Factor Analysis and Fit Indices

In structural equation modeling examining alternative measurement models (unidimensional vs. correlated two-factor [Anxiety vs. Depression] vs. bi-factor models):

• Bi-Factor Model Architecture: A bi-factor model—featuring one general Psychological Distress factor loading onto all seven items, alongside two small orthogonal group factors (Somatic/Cognitive and Pure Affective)—provides an excellent fit to observed data: χ²/df < 2.5, Comparative Fit Index (CFI) = 0.97, Tucker-Lewis Index (TLI) = 0.95, and Root Mean Square Error of Approximation (RMSEA) = 0.048 (90% CI: 0.032–0.064).
• Standardized Factor Loadings (General Factor):
  • Item 1 (Give up too easily): λ = 0.52
  • Item 2 (Difficulty concentrating): λ = 0.60
  • Item 3 (Comfortable around people – reversed): λ = 0.44
  • Item 4 (Trouble sleeping): λ = 0.56
  • Item 5 (Getting tired easily): λ = 0.58
  • Item 6 (Feeling depressed or sad): λ = 0.78
  • Item 7 (Nervousness): λ = 0.72

These findings substantiate using a single composite sum or transformed index score, as the primary general distress factor accounts for the vast majority of reliable variance.

Instrument / Measurement Tool

• Test Type: Clinical screening questionnaire; standardized patient-reported outcome measure (PROM).
• Administration Format: Self-administered paper-and-pencil questionnaire, digital web-based survey, or clinician-facilitated interview.
• Item Count: 7 items.
• Completion Time: Approximately 1 to 2 minutes.
• Target Population: Adult primary care, outpatient specialty clinic, and community populations (aged 18 years and older).
• Response Scales:
  • Items 1 through 3 (General self-description):
    • (2) Yes, describes me exactly
    • (1) Somewhat describes me
    • (0) No, doesn’t describe me at all
  • Items 4 through 7 (Past week symptom difficulty):
    • (0) None
    • (1) Some
    • (2) A Lot
• Scoring and Transformation Rules:
  • Reverse Coding: Item 3 (“I am comfortable being around people”) is positively phrased and must be reversed prior to summation: “Yes, describes me exactly” = 0; “Somewhat describes me” = 1; “No, doesn’t describe me at all” = 2.
  • Raw Summed Score: Sum the coded values of all seven items (ranging from 0 to 14, where higher scores signify greater distress/symptom severity).
  • Normalized Transformed Metric (0–100 Scale): Following standard Duke Health Profile scoring methodology, raw scores are converted to a standardized 0–100 dysfunction metric using the formula:
    
    Scaled Score = (Raw Score / Maximum Possible Score [14]) × 100
  • Clinical Cutoff Scores: A transformed score ≥ 50 (or raw score ≥ 5) typically represents the optimal diagnostic cut-off point indicating clinically meaningful anxiety or depressive symptoms warranting further evaluation.

Permissions & Fee and Test Year

The DUKE-AD was derived from the Duke Health Profile (published in 1990) and formally validated as a dedicated anxiety-depression screener by George R. Parkerson Jr. and W. E. Broadhead in 1997. The instrument and its parent batteries (the Duke-UNC Health Profile and Duke Health Profile) are copyrighted by Duke University.

Duke Health measures are made broadly accessible for educational, non-commercial clinical, and academic research purposes without licensing fees, provided the questions are reproduced without unauthorized modifications and appropriate academic citations are maintained. Detailed user guides, automated scoring algorithms, and permission request forms can be obtained via the official Duke Health Measures portal at https://cfm.duke.edu/research/duke-health-measures. Commercial health technology applications, pharmaceutical clinical trials, or fee-for-service digital platforms require formal licensing agreements through Duke University’s Office of Licensing and Ventures.

References

Clark, L. A., & Watson, D. (1991). Tripartite model of anxiety and depression: Psychometric evidence and taxonomic implications. Journal of Abnormal Psychology, 100(3), 316–336. https://doi.org/10.1037/0021-843X.100.3.316

McDowell, I. (2006). Measuring health: A guide to rating scales and questionnaires (3rd ed.). Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195165678.001.0001

Parkerson, G. R., Jr. (1994). User’s guide for the Duke Health Profile (DUKE). Department of Community and Family Medicine, Duke University Medical Center.

Parkerson, G. R., Jr. (2002). User’s guide for Duke health measures. Department of Community and Family Medicine, Duke University Medical Center.

Parkerson, G. R., Jr., & Broadhead, W. E. (1997). Screening for anxiety and depression in primary care with the Duke Anxiety-Depression Scale. Family Medicine, 29(3), 177–181.

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. J. (1990). The Duke Health Profile: A 17-item measure of health and dysfunction. Medical Care, 28(11), 1056–1072. https://doi.org/10.1097/00005650-199011000-00007

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. J. (1991a). Development of the 17-item Duke Health Profile. Family Practice, 8(4), 396–401. https://doi.org/10.1093/fampra/8.4.396

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. J. (1991b). Comparison of the Duke Health Profile and the MOS Short-Form in healthy young adults. Medical Care, 29(7), 679–683. https://doi.org/10.1097/00005650-199107000-00007

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. J. (1992). Quality of life and functional health of primary care patients. Journal of Clinical Epidemiology, 45(11), 1303–1313. https://doi.org/10.1016/0895-4356(92)90170-U

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. (1995). Health status and severity of illness as predictors of outcomes in primary care. Medical Care, 33(1), 53–66. https://doi.org/10.1097/00005650-199501000-00005

Parkerson, G. R., Jr., Broadhead, W. E., & Tse, C. K. J. (1996). Anxiety and depressive symptom identification using the Duke Health Profile. Journal of Clinical Epidemiology, 49(1), 85–93. https://doi.org/10.1016/0895-4356(96)00005-7

Parkerson, G. R., Jr., Gehlbach, S. H., Wagner, E. H., Strogatz, D. S., & Broadhead, W. E. (1981). The Duke-UNC Health Profile: An adult health status instrument for primary care. Medical Care, 19(8), 806–828. https://doi.org/10.1097/00005650-198108000-00002

Items of the Scale