1. Abstract
The Health Anxiety Inventory (HAI), developed by Paul M. Salkovskis, Katharine A. Rimes, Hilary M. C. Warwick, and David M. Clark (2002), is an empirically validated self-report instrument designed to quantify the severity of health anxiety and hypochondriacal ideation across non-clinical, medically ill, and psychiatric populations. Formulated within the cognitive-behavioral framework of health anxiety, the HAI addresses critical methodological limitations of earlier instruments—such as the Whiteley Index and the Illness Attitudes Scales—by systematically separating authentic physical symptoms associated with genuine medical illness from cognitive misattributions, catastrophic appraisals, and somatic hypervigilance. The 18-item short form of the instrument (frequently designated as the Short Health Anxiety Inventory or SHAI) captures two robust psychometric dimensions: the Perceived Likelihood of Becoming Ill (Items 1–14, assessing core health anxiety, awareness of bodily sensations, and catastrophic misinterpretations) and the Perceived Negative Consequences of Becoming Ill (Items 15–18, assessing conditional beliefs regarding coping inability and anticipated devastation if illness were diagnosed). Each item utilizes a four-point forced-choice response format scored from 0 to 3, yielding total scale scores ranging from 0 to 54. Extensive psychometric evaluations indicate exceptional internal consistency (Cronbach’s α ranging from .89 to .95 for the total scale across diverse cohorts), excellent test-retest reliability (r = .90), clear structural stability via confirmatory factor analyses, and robust convergent, discriminant, and treatment-sensitive predictive validity in cognitive-behavioral intervention trials for Illness Anxiety Disorder and Somatic Symptom Disorder.
2. Keywords
Health Anxiety Inventory, HAI, Short Health Anxiety Inventory, SHAI, health anxiety, hypochondriasis, illness anxiety disorder, cognitive-behavioral therapy, somatic amplification, psychometrics
3. Authors
The Health Anxiety Inventory was formulated and validated through collaborative clinical research conducted across premier British psychological research institutions:
- Paul M. Salkovskis, Ph.D.: Professor of Clinical Psychology and Applied Science, Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience (IoPPN), King’s College London, and Oxford University Department of Psychiatry, United Kingdom.
- Katharine A. Rimes, Ph.D.: Senior Lecturer and Clinical Psychologist, Department of Psychology, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, United Kingdom.
- Hilary M. C. Warwick, M.A., B.M., B.Ch., M.R.C.Psych.: Consultant Psychiatrist and Clinical Researcher, Oxford Cognitive Therapy Centre and Department of Psychiatry, Warneford Hospital, University of Oxford, United Kingdom.
- David M. Clark, D.Phil.: Professor of Experimental Psychology, Department of Experimental Psychology, University of Oxford, United Kingdom.
4. Purpose
The primary clinical and empirical objective of the Health Anxiety Inventory is to provide a psychometrically sensitive, theoretically coherent, and dimensionally sound measure of health-related anxiety that remains valid in both healthy populations and individuals with concurrent physical medical illnesses. Prior to the development of the HAI, prevalent instruments evaluating hypochondriasis, such as the Whiteley Index (Pilowsky, 1967) and the Illness Attitudes Scales (Kellner, 1986), suffered from substantial criterion confounding. Specifically, these legacy instruments incorporated items that queried the presence or frequency of actual somatic symptoms (such as pain, fatigue, or palpitations). As an unintended psychometric consequence, medically ill individuals—such as patients suffering from rheumatoid arthritis, multiple sclerosis, coronary artery disease, or chronic pain syndromes—invariably obtained artificially elevated scores indicative of pathology, even when their psychological appraisal of their physical health was realistic and non-pathological.
To overcome this pervasive clinical assessment confound, Salkovskis and colleagues (2002) operationalized health anxiety strictly in terms of cognitive distortions, hypervigilance, catastrophic appraisals, and selective attention rather than somatic symptom presence. The purpose of the HAI is therefore threefold:
- Dimensional Screening and Diagnostic Evaluation: To screen and differentiate mild, subclinical health worries from clinically debilitating Illness Anxiety Disorder (formerly hypochondriasis in DSM-IV; classified under Somatic Symptom and Related Disorders in DSM-5) and Somatic Symptom Disorder.
- Measurement Across Medical Cohorts: To enable valid assessment of excessive health preoccupation within general practice, outpatient medical clinics, oncology, cardiology, and neurology, separating psychological catastrophizing from genuine organic disease pathology.
- Treatment Outcome Tracking: To function as a highly sensitive, change-responsive instrument within clinical trials evaluating Cognitive Behavioral Therapy (CBT), pharmacological interventions, and psychoeducational treatments for health anxiety.
5. Psychological Construct
The construct measured by the Health Anxiety Inventory is health anxiety, conceptualized not as a discrete somatic defect or categorical psychiatric condition, but rather as an anchored continuous spectrum ranging from adaptive health consciousness to severe, disabling illness anxiety. Within this conceptualization, health anxiety is defined as the persistent, disproportionate apprehension regarding the threat of having or developing a severe, progressive, or life-threatening illness, driven by systematic cognitive misattributions of benign or ambiguous somatic sensations.
The Health Anxiety Inventory operationalizes health anxiety across two fundamental, highly correlated cognitive dimensions:
Perceived Likelihood of Becoming Ill (Core Health Anxiety)
Covering Items 1 through 14 of the scale, this dimension assesses the subjective probability an individual assigns to harboring or acquiring a catastrophic medical illness. The construct is driven by cognitive confirmation biases and somatic hyperawareness. It taps several interlinked behavioral and psychological features:
- Selective Somatic Hypervigilance: Heightened scanning and awareness of autonomic variations, muscle twitches, minor headaches, or benign gastrointestinal sensations (e.g., Items 2, 3, and 9).
- Catastrophic Somatosensory Misinterpretation: The habitual cognitive tendency to interpret ambiguous, normative, or stress-induced somatic fluctuations as definitive indicators of life-threatening pathology, such as brain tumors, heart disease, or malignancies (e.g., Items 6, 12, and 14).
- Intrusive Illness Cognitions and Resistance Impairment: The recurrence of uncontrolled, intrusive images and ruminations centering on death, invalidism, or disease contraction, accompanied by an inability to mentally suppress or dismiss these thoughts (e.g., Items 4, 11, and 13).
- Reassurance-Seeking and Reassurance Resistance: The persistent pursuit of medical consultations, diagnostic scans, and interpersonal reassurance, coupled with the rapid decay or complete rejection of medical validation (Item 7).
Perceived Negative Consequences of Becoming Ill
Covering Items 15 through 18, this secondary dimension assesses the individual’s anticipated inability to tolerate, survive, or psychologically cope with serious illness if it were to occur. Rather than querying the likelihood of illness, these conditional statements evaluate worst-case scenario catastrophizing:
- Anticipated Inability to Cope: Catastrophic expectations of personal helplessness, total psychological collapse, or complete emotional disintegration upon receiving a serious diagnosis (Item 16).
- Global Life Disruption and Anhedonia: The cognitive belief that disease contraction inherently extinguishes all possibilities of subjective life satisfaction, meaning, and engagement (Items 15 and 17).
- Perceived Inefficacy of Medicine: Severe pessimism regarding the competence of medical practitioners and therapeutic treatments to mitigate suffering or control illness progression (Item 18).
6. Theoretical Framework
The Health Anxiety Inventory is theoretically anchored in the cognitive-behavioral model of hypochondriasis and health anxiety formulated by Paul M. Salkovskis (1989), Warwick and Salkovskis (1990), and based upon Aaron T. Beck’s general cognitive formulation of emotional disorders. Under Beckian cognitive architecture, psychological distress is mediated not by events or physical phenomena themselves, but by the dysfunctional cognitive schemas, automated beliefs, and information-processing biases through which individuals appraise those phenomena.
The cognitive model formalizes subjective perceived threat via an overarching mathematical heuristic of anxiety:
Perceived Threat = (Perceived Likelihood of Event × Perceived Severity / Consequences of Event) / (Perceived Coping Ability + Perceived External Rescue Resources)
Within this paradigm, pathological health anxiety arises when an individual systematically overestimates both the likelihood and the terribleness of physical illness, while profoundly underestimating personal coping mechanisms and the curative power of medical intervention. Salkovskis and colleagues embedded this exact formulation into the structural split of the HAI: Items 1–14 measure the numerator’s perceived probability and severity, while Items 15–18 capture the collapse of coping and rescue beliefs.
The model describes a continuous maintenance cycle comprising four dynamic psychological loops:
- Trigger Stimuli: Internal bodily sensations (e.g., transient palpitations, aches, benign lymph nodes), external medical information (e.g., media reports, illness of an acquaintance), or health-related discussions trigger pre-existing dysfunctional assumptions (e.g., “Bodily changes are always a sign of dangerous disease”).
- Cognitive Misinterpretation: The individual appraises benign sensations as imminent catastrophic threats, triggering acute physiological arousal and subjective anxiety.
- Physiological and Attentional Escalation: The sympathetic autonomic activation triggered by anxiety generates genuine secondary somatic sensations (e.g., tachycardia, dizziness, paresthesia, diaphoresis). Attentional focus narrows onto the target bodily region, lowering sensory detection thresholds and confirming the individual’s catastrophic hypothesis.
- Maladaptive Safety-Seeking Behaviors: To manage intolerable distress, the individual executes safety behaviors—including body checking (e.g., continuous palpation, mirror inspection), reassurance-seeking from physicians, online searching (cyberchondria), and avoidance of physical exertion. These behaviors paradoxically amplify bodily awareness, introduce further physical irritation, and prevent the cognitive disconfirmation of threat beliefs.
7. Validity
The Health Anxiety Inventory has been subjected to extensive psychometric evaluations, consistently displaying robust construct, convergent, discriminant, and predictive validity across numerous cultural adaptations and clinical settings.
Construct and Structural Validity
The structural validity of the 18-item inventory was originally demonstrated by Salkovskis et al. (2002) and replicated by independent psychometricians (e.g., Abramowitz et al., 2007; Alberts et al., 2013). Exploratory and confirmatory factor analyses robustly identify a primary factor accounting for core illness anxiety (Likelihood) and a distinct secondary factor reflecting negative illness consequences (Consequences). The construct validity is further underscored by the scale’s sensitivity across the dimensional spectrum: scores incrementally ascend from healthy community volunteers to medical patients, anxiety disorder controls, and hypochondriacal clinical cohorts.
Convergent Validity
The HAI displays strong convergent correlations with concurrent measures of hypochondriasis and health distress. Salkovskis et al. (2002) observed robust correlations between the HAI-18 total score and:
- The Whiteley Index (WI): r = .65 to .78 across diverse samples, demonstrating strong conceptual convergence with standard hypochondriacal symptom indices.
- The Illness Attitudes Scales (IAS): r = .70 to .83, validating its alignment with dimensions of disease conviction, fear of death, and treatment preoccupation.
- The Beck Anxiety Inventory (BAI): moderate correlations (r = .45 to .56), confirming shared variance with autonomic panic and anxiety while preserving clear construct independence.
- The Beck Depression Inventory (BDI-II): modest-to-moderate correlations (r = .38 to .48), reflecting expected emotional comorbidity without psychometric collapse into general negative affect.
Discriminant Validity
The defining psychometric strength of the HAI lies in its discriminant capacity. Salkovskis et al. (2002) demonstrated that the HAI-18 successfully differentiates individuals diagnosed with hypochondriasis / illness anxiety disorder from both healthy controls and patients with verified, severe chronic medical conditions (e.g., clinical cohorts suffering from renal failure undergoing hemodialysis, chronic neurological disorders, or chronic pain). While patients with authentic somatic disease score significantly higher on the somatic items of the Whiteley Index, their scores on the HAI Likelihood and Negative Consequences subscales remain substantially lower than those of hypochondriacal cohorts, confirming that the HAI specifically identifies cognitive-affective distortion rather than genuine physiological illness.
Predictive and Treatment Validity
The HAI demonstrates high sensitivity to therapeutic change. Randomized controlled trials of CBT for health anxiety (e.g., Clark et al., 1998; Greeven et al., 2007; Tyrer et al., 2014) report large pre-to-post treatment effect sizes (Cohen’s d ranging from 1.1 to 1.8 on the HAI-18). Treatment-induced reductions in HAI scores reliably predict long-term remission, reduced medical utilization, and stabilized functional recovery across multi-year follow-up intervals.
8. Reliability
The Health Anxiety Inventory exhibits consistently high internal consistency and temporal stability across diverse clinical, non-clinical, and medical research cohorts.
Internal Consistency
In the initial instrument development study by Salkovskis et al. (2002), the total 18-item scale yielded exceptional internal consistency coefficients:
- Total Scale (18 items): Cronbach’s α = .89 in healthy non-clinical controls, α = .92 in clinical hypochondriacal patients, and α = .94 in mixed medical outpatient cohorts.
- Core Health Anxiety / Likelihood Subscale (Items 1–14): Cronbach’s α consistently ranges between .88 and .93.
- Negative Consequences Subscale (Items 15–18): Cronbach’s α ranges between .72 and .82, acceptable given the brief 4-item parameter.
Subsequent psychometric evaluations across international adaptations—including the American English validation (Abramowitz et al., 2007: α = .89), the French translation (Alberts et al., 2013: α = .91), and the German version (Bailer et al., 2013: α = .90)—have confirmed comparable internal consistency metrics.
Test-Retest Stability
The temporal stability of the instrument over unmanipulated intervals has been repeatedly established. Salkovskis et al. (2002) documented a test-retest correlation coefficient of r = .90 over an interval of 7 to 21 days in non-clinical cohorts. Long-term stability assessments in waitlist control conditions across clinical trials demonstrated intraclass correlation coefficients (ICC) ranging between .82 and .88 across 8 to 12 weeks, verifying that the HAI reliably measures stable cognitive schemata rather than transient emotional fluctuations.
9. Factor Analysis
Extensive factor analytical inquiries have thoroughly investigated both the full 64-item version and the predominant 18-item short form (HAI-18 / SHAI) of the instrument.
Exploratory Factor Analysis (EFA)
During original item construction, Salkovskis et al. (2002) conducted principal components and exploratory factor analyses on patient and non-clinical datasets. The analyses isolated two clear, conceptually coherent factors with eigenvalues exceeding Kaiser’s criterion (λ > 1.0), accounting for a substantial portion of the total variance:
- Factor 1: Perceived Likelihood of Illness / Core Health Anxiety (Items 1–14): Factor loadings for items within this component ranged from .48 to .78, capturing fear of disease, frequency of symptom monitoring, reassurance-seeking decay, and catastrophic interpretations of bodily alterations.
- Factor 2: Perceived Negative Consequences of Illness (Items 15–18): Factor loadings ranged from .54 to .81, cleanly isolating conditional expectations regarding total lifestyle destruction, emotional collapse, and lack of medical assistance upon contracting disease.
Confirmatory Factor Analysis (CFA)
Subsequent independent validation studies (e.g., Abramowitz et al., 2007; Alberts et al., 2013) formally compared competitive factor models via confirmatory structural equation modeling:
- Unidimensional Model: A single-factor model collapsing all 18 items exhibited inadequate fit indices across clinical samples (χ²/df > 3.8, CFI = .86, TLI = .84, RMSEA = .092).
- Correlated Two-Factor Model: The theoretical two-factor model (Likelihood and Negative Consequences) demonstrated superior fit across student, clinical, and medical populations: χ²/df = 1.84 to 2.12, Comparative Fit Index (CFI) = .95 to .97, Tucker-Lewis Index (TLI) = .94 to .96, Root Mean Square Error of Approximation (RMSEA) = .044 to .052 (90% CI [.036, .061]), and Standardized Root Mean Square Residual (SRMR) = .041.
While some investigators (e.g., Wheaton et al., 2010) have identified secondary bifactor structures or a distinct three-factor division separating illness likelihood from body awareness, the two-factor solution remains the most robust, psychometrically verified, and clinically relevant structure.
10. Instrument / Measurement Tool
- Test Type: Standardized psychological self-report rating scale / psychometric inventory.
- Target Population: Adults and adolescents aged 16 and older; validated in healthy populations, medical settings, and psychiatric cohorts.
- Format: 18 forced-choice vignettes, each containing four graduated statements representing incremental severity levels.
- Administration Time: Approximately 5 to 10 minutes.
- Item Count: 18 items (divided into two subscales: Items 1–14 and Items 15–18).
- Response Scale: 4-point forced-choice options per item, scored 0 to 3 (0 = least anxious/normal, 3 = most severe health anxiety). Each item contains four distinct clinical descriptive anchors: (a) = 0, (b) = 1, (c) = 2, (d) = 3.
- Scoring Rules:
- Total HAI-18 Score: Sum of all 18 items; scores range from 0 to 54.
- Illness Likelihood / Core Health Anxiety Subscale: Sum of Items 1 through 14; scores range from 0 to 42.
- Negative Consequences of Illness Subscale: Sum of Items 15 through 18; scores range from 0 to 12.
- Clinical Cut-Offs: In the original standardization sample (Salkovskis et al., 2002), a total score of 18 or higher reliably distinguishes clinical hypochondriasis / illness anxiety disorder from normal health anxiety (sensitivity = 89.5%, specificity = 81.2%). A more conservative clinical threshold of 20 or greater is frequently applied in formal clinical trials to minimize false positives.
11. Permissions & Fee and Test Year
- Year of Publication: 2002 (Primary validation study published in Psychological Medicine).
- Copyright & Intellectual Ownership: Copyright © 2002 by Paul M. Salkovskis, Katharine A. Rimes, Hilary M. C. Warwick, and David M. Clark. Published by Cambridge University Press.
- Access and Permissions: The Health Anxiety Inventory is an open-access psychometric instrument for academic research, non-commercial clinical evaluation, and educational purposes. No licensing fee or formal permission is required for non-commercial academic research, provided that appropriate scholarly attribution is cited to Salkovskis et al. (2002). Commercial use, digital inclusion into proprietary software platforms, or commercial pharmaceutical testing may require formal licensing from the primary authors or Cambridge University Press.
12. References
- Abramowitz, J. S., Deacon, B. J., & Valentiner, D. P. (2007). The Short Health Anxiety Inventory: Psychometric properties and construct validity in a non-clinical sample. Cognitive Therapy and Research, 31(6), 871–883. https://doi.org/10.1007/s10608-006-9058-1
- Alberts, N. M., Hadjistavropoulos, H. D., Jones, S. L., & Sharpe, D. (2013). The Short Health Anxiety Inventory: A systematic review and meta-analysis. Journal of Anxiety Disorders, 27(1), 68–78. https://doi.org/10.1016/j.janxdis.2012.10.009
- Bailer, J., Witthöft, M., & Rist, F. (2013). The Short Health Anxiety Inventory (SHAI): Validation of a German version in a clinical and non-clinical sample. Zeitschrift für Klinische Psychologie und Psychotherapie, 42(3), 161–171. https://doi.org/10.1026/1616-3443/a000210
- Clark, D. M., Salkovskis, P. M., Hackmann, A., Wells, A., Fennell, M., Ludgate, J., Ahmad, S., Richards, H. C., & Gelder, M. (1998). Two psychological treatments for hypochondriasis: A randomised controlled trial. The British Journal of Psychiatry, 173(3), 218–225. https://doi.org/10.1192/bjp.173.3.218
- Greeven, A., van Balkom, A. J., Visser, S., Merkelbach, J. W., van Rood, Y. R., van Dyck, R., van der Feltz-Cornelis, C. M., & Zitman, F. G. (2007). Cognitive behavior therapy and paroxetine in the treatment of hypochondriasis: A randomized controlled trial. The American Journal of Psychiatry, 164(1), 91–99. https://doi.org/10.1176/ajp.2007.164.1.91
- Kellner, R. (1986). Somatization and Hypochondriasis. Praeger-Greenwood.
- Pilowsky, I. (1967). Dimensions of hypochondriasis. The British Journal of Psychiatry, 113(494), 89–93. https://doi.org/10.1192/bjp.113.494.89
- Salkovskis, P. M. (1989). Somatic problems. In K. Hawton, P. M. Salkovskis, J. Kirk, & D. M. Clark (Eds.), Cognitive Behaviour Therapy for Psychiatric Problems: A Practical Guide (pp. 235–276). Oxford University Press.
- Salkovskis, P. M., Rimes, K. A., Warwick, H. M. C., & Clark, D. M. (2002). The Health Anxiety Inventory: Development and validation of scales for the measurement of health anxiety and hypochondriasis. Psychological Medicine, 32(5), 843–853. https://doi.org/10.1017/s0033291702005822
- Tyrer, P., Cooper, S., Salkovskis, P., Tyrer, H., Crawford, M., Byford, S., Dupont, S., Smith, G., & Barrett, B. (2014). Clinical and cost-effectiveness of cognitive behaviour therapy for health anxiety in medical patients: A multicentre randomised controlled trial (CHAMP). The Lancet, 383(9913), 219–225. https://doi.org/10.1016/S0140-6736(13)61905-4
- Warwick, H. M., & Salkovskis, P. M. (1990). Hypochondriasis. Behaviour Research and Therapy, 28(2), 105–117. https://doi.org/10.1016/0005-7967(90)90023-c
- Wheaton, M. G., Berman, N. C., Franklin, J. C., & Abramowitz, J. S. (2010). Health anxiety: A comparison of two cognitive behavioral models. Journal of Anxiety Disorders, 24(8), 875–882. https://doi.org/10.1016/j.janxdis.2010.06.010