The Hamilton Anxiety Scale, usually called the HAM-A or HARS, is a 14-item clinician-rated questionnaire designed to measure the severity of anxiety symptoms. Developed by Max Hamilton in 1959, it remains one of the most widely used instruments in psychiatric research and clinical trials for anxiety disorders. Its scores divide into well-established severity bands, with scores of 7 or below generally indicating no meaningful anxiety and scores of 24 or higher flagging severe anxiety. But the scale’s long history comes with accumulated criticism, and understanding what those scores actually capture requires knowing both its strengths and its blind spots.
What the Scale Measures
The HAM-A covers 14 items, each rated on a five-point scale from 0 (not present) to 4 (very severe/disabling). A clinician conducts a semi-structured interview and assigns scores based on the patient’s responses and observable behavior. The items span a range of anxiety-related experiences: anxious mood, tension, fears, insomnia, cognitive difficulties, depressed mood, and several categories of physical symptoms like cardiovascular, respiratory, gastrointestinal, and muscular complaints.
Factor analyses across multiple populations consistently find that the 14 items cluster into two groups: psychic anxiety and somatic anxiety. The psychic factor captures the mental and emotional side of anxiety, including worry, irritability, and difficulty concentrating. The somatic factor captures the body’s responses, such as muscle tension, sweating, heart pounding, and stomach upset. A large study of Korean university students during COVID-19 confirmed that Hamilton’s original two-factor model fit the data better than alternative three-factor models.1The Open Psychology Journal. Psychometric Properties and Factor Structure of the Hamilton Anxiety Rating Scale among Korean University Students during COVID-19 Studies in Portuguese and Arabic-speaking populations have found the same two-factor structure, with strong internal consistency.2PubMed Central. Translation, cultural adaptation and evaluation of the psychometric properties of the Hamilton Anxiety Scale among a sample of Portuguese adult patients with mental health disorders3Clinical Epidemiology and Global Health. Validation of the Hamilton Anxiety Rating Scale and State Trait Anxiety Inventory A and B in Arabic among the Lebanese population
When researchers compare the HAM-A against self-report measures like the Beck Anxiety Inventory, the distinction between these two clusters becomes even more apparent. The Beck Anxiety Inventory tends to break into four components reflecting neurophysiological, subjective, panic, and autonomic symptoms as reported by the patient, while the clinician-rated HAM-A consistently resolves into the two broad psychic and somatic factors.4Journal of Anxiety Disorders. Relationship between the beck anxiety inventory and the Hamilton anxiety rating scale with anxious outpatients This difference matters in practice: the HAM-A gives the clinician’s structured impression of severity, not the patient’s subjective account, and those two things do not always agree.
What the Scores Mean
A total HAM-A score can range from 0 to 56. Research aimed at identifying clinically meaningful cutoffs for generalized anxiety disorder established the ranges most commonly used today: a score of 7 or below suggests no or minimal anxiety, 8 to 14 indicates mild anxiety, 15 to 23 indicates moderate anxiety, and 24 or above indicates severe anxiety.5PubMed Central. Identifying HAM-A cutoffs for mild, moderate, and severe generalized anxiety disorder These thresholds are commonly used as entry criteria in clinical trials for anti-anxiety medications, where participants typically must score above a certain level to qualify.
In practice, clinicians also look at the subscale scores separately. If the somatic anxiety subscale is disproportionately high compared to the psychic subscale, it may signal that physical health complaints or medication side effects are inflating the total score. This is a particularly important distinction in drug trials, where somatic side effects from a new medication could artificially raise the very score the trial is using to measure improvement.
The Problem of Inconsistent Administration
Hamilton’s original publication gave clinicians relatively little guidance on how to ask questions or anchor their ratings. Two different clinicians interviewing the same patient could arrive at meaningfully different scores, not because they disagreed on the patient’s condition but because they were interpreting the rating scale differently. This limitation has been a concern since the scale’s earliest use.
To address it, researchers developed structured interview guides. The Structured Interview Guide for the Hamilton Anxiety Rating Scale, known as the SIGH-A, provides standardized questions and explicit criteria for assigning severity ratings to each item.6PubMed. Reliability and validity of a structured interview guide for the Hamilton Anxiety Rating Scale (SIGH-A) A similar approach, the HARS Interview Guide (HARS-IG), was developed and tested in a group of 30 inpatients, showing improved agreement between raters compared to the original semi-structured format.7PubMed. Hamilton Anxiety Rating Scale Interview guide: joint interview and test-retest methods for interrater reliability Studies of the SIGH-A have found very high inter-rater reliability, with agreement between clinicians reaching an intraclass correlation coefficient of 0.99.8Depression and Anxiety. Reliability and validity of a structured interview guide for the Hamilton Anxiety Rating Scale (SIGH‐A)
Even with structured guides available, scoring errors persist in real-world use. A study examining consistency checks on HAM-A administrations found that about 35% of all administrations had at least one scoring flag, whether a logical inconsistency or a statistical outlier. Roughly 16% had two or more flags.9PubMed Central. Consistency checks to improve measurement with the Hamilton Rating Scale for Anxiety (HAM-A) That is a striking error rate for an instrument used to make decisions about treatment in clinical trials. It suggests that even trained raters benefit from automated checks that catch impossible or unlikely scoring patterns before the data are locked.
Trouble Telling Anxiety from Depression
One of the most persistent criticisms of the HAM-A is that it does not cleanly separate anxiety from depression. Several of its items, including depressed mood, insomnia, and cognitive difficulties, are common to both conditions. In clinical populations where anxiety and depression frequently co-occur, this overlap becomes a real problem. A study examining the psychometric properties of both the Hamilton anxiety and depression scales concluded that they discriminate poorly between the two conditions.10PubMed Central. Psychometric Properties of the Reconstructed Hamilton Depression and Anxiety Scales
This matters most in treatment trials. If you are testing a new drug for anxiety and your primary outcome measure is the HAM-A, but the drug also improves depression, you cannot easily tell from the score change whether the drug actually reduced anxiety or just made the patient less depressed, which then lowered their HAM-A score. An early and influential review of the HAM-A put this plainly: the scale makes it difficult to clearly distinguish anxiolytic effects from antidepressant effects, limiting its usefulness in anxiolytic treatment studies.11PubMed. The Hamilton Anxiety Scale: reliability, validity and sensitivity to change in anxiety and depressive disorders That same review noted the somatic anxiety subscale overlaps heavily with somatic side effects of medications, creating another source of confusion.
Attempts to solve this have included reconstructing the items into purer subscales and developing newer instruments altogether. But the HAM-A’s enormous installed base in clinical research, with decades of trials using it as the primary or secondary endpoint, means it continues to be specified in regulatory submissions and treatment guidelines. Replacing it entirely would require re-establishing norms and thresholds from scratch.
Using the HAM-A in Non-Psychiatric Settings
The HAM-A was developed for psychiatric populations, but it has found its way into cardiology, oncology, dermatology, and other medical specialties where anxiety accompanies a primary physical illness. Its use in these settings introduces specific interpretive challenges, particularly around the somatic subscale. When a patient with heart disease reports chest tightness, rapid heartbeat, and shortness of breath, those symptoms may reflect their cardiac condition rather than anxiety. The same applies to a cancer patient experiencing nausea or weight loss.
A study of patients before and after cardiac surgery illustrates the issue. Researchers found that even before surgery, the typical patient in this group scored below the standard clinical cutoffs on the HAM-A. The score variation was driven primarily by the psychic anxiety items, not the somatic ones, suggesting that in cardiac patients the somatic subscale is contaminated by symptoms of the underlying heart disease and adds more noise than signal.12Psychopathology. Generalized Anxiety or Depression Measured by the Hamilton Anxiety Scale and the Melancholia Scale in Patients before and after Cardiac Surgery Clinicians working with medically complex patients sometimes focus on the psychic subscale alone for this reason, though this is an off-label use of the scale that lacks its own validated cutoffs.
Despite these complications, the HAM-A remains the instrument of choice in many medical studies examining anxiety as a secondary outcome. In a randomized trial of exercise, escitalopram, and placebo for anxiety in patients with coronary heart disease, all three groups showed reduced anxiety over 12 weeks, with escitalopram producing the largest drop. Notably, that trial used a different anxiety scale (the HADS-A) rather than the HAM-A as its primary measure, likely reflecting awareness of the somatic-item contamination problem in cardiac populations.13JAMA Psychiatry. Effect of Exercise, Escitalopram, or Placebo on Anxiety in Patients With Coronary Heart Disease
Older Adults and Higher Scores
Age affects how the HAM-A behaves. Older adults tend to express anxiety through physical complaints more than younger adults do, and they may not use the same emotional vocabulary to describe their inner state. A study comparing anxiety in adults over 65 with younger patients found that the older group scored significantly higher on the HAM-A total score, as well as on both the somatic and psychic subscales.14European Journal of Therapeutics. The evaluation of generalized anxiety disorder in older adults This does not necessarily mean older adults are always more anxious. It may partly reflect the fact that age-related physical conditions, such as chronic pain, gastrointestinal changes, and cardiovascular symptoms, inflate the somatic subscale.
At the other end of the age spectrum, testing in adolescents has shown that the HAM-A’s reliability and internal consistency hold up and are comparable to results reported for adults.15PubMed. Reliability and validity of the Hamilton Anxiety Rating Scale in an adolescent sample Still, the HAM-A was not designed with younger populations in mind, and most of the severity cutoffs were derived from adult samples. Clinicians working with adolescents typically interpret scores with this caveat.
Cross-Cultural Adaptations
The HAM-A has been translated and validated in dozens of languages, and the two-factor structure tends to replicate across cultures. The Portuguese adaptation showed high internal consistency, with a Cronbach’s alpha of 0.92, and strong inter-rater reliability with an intraclass correlation coefficient of 0.91.16PubMed Central. Translation, cultural adaptation and evaluation of the psychometric properties of the Hamilton Anxiety Scale among a sample of Portuguese adult patients with mental health disorders An Arabic-language validation in over 1,300 Lebanese participants confirmed the same two-factor structure and similarly strong reliability.17Clinical Epidemiology and Global Health. Validation of the Hamilton Anxiety Rating Scale and State Trait Anxiety Inventory A and B in Arabic among the Lebanese population
That said, cultural differences in how people express anxiety can still affect scores. In some cultures, somatic complaints are the primary idiom for psychological distress: a patient may report stomach pain, dizziness, or headaches without describing worry or fear. In others, cognitive symptoms like rumination are more prominently reported. The HAM-A’s split into psychic and somatic subscales can partly accommodate this, but the total score may mean somewhat different things in different cultural contexts. Research looking at somatic symptoms across ethnic groups in the United States, for example, has examined how Hispanic and non-Hispanic patients differ in their endorsement of somatic versus psychic HAM-A items, a reminder that even within a single country the scale’s items are not culturally neutral.
Administration by Video
The shift toward telehealth raised an obvious question: can you administer the HAM-A through a screen instead of in person? The answer, based on the available evidence, is yes. A study comparing telepsychiatry interviews with face-to-face assessments found good agreement on the HAM-A, along with several other clinician-rated scales including the Hamilton Depression Rating Scale and the Brief Psychiatric Rating Scale.18PubMed Central. The reliability of symptom assessment by telepsychiatry compared with face to face psychiatric interviews This is encouraging for clinical trials that increasingly allow remote assessments, though the nuances of observing physical signs of anxiety, like fidgeting, muscle tension, or autonomic flushing, may be harder to catch through a webcam.
Clinician-Rated Versus Self-Report Scales
One reason the HAM-A has endured is that clinician-rated scales and patient-rated scales capture different things. A clinician rating your anxiety integrates what you say, how you say it, and what they observe. A self-report questionnaire captures your own perception. Neither is inherently better, but they can disagree.
Research looking at how well different scale types classify patients with anxiety disorders found that combining clinician-rated scales like the HAM-A with patient-rated instruments improved classification accuracy beyond what either type achieved alone. In distinguishing anxiety disorder patients from healthy controls, the combination of clinician- and patient-rated measures along with cognitive tests reached an accuracy of 98%, with the HAM-A emerging as the strongest individual contributor to the model.19BMC Psychiatry. Patient-rated scales improve the classification accuracy for patients with depression and anxiety disorder: a linear discriminant analysis The implication for clinical practice is straightforward: using the HAM-A alongside a self-report tool like the GAD-7 or Beck Anxiety Inventory gives a fuller picture than relying on either one alone.
Biological Correlates of HAM-A Scores
Because the HAM-A quantifies anxiety severity on a continuous scale, researchers have used it to look for biological markers that track with anxiety levels. Cortisol, the body’s primary stress hormone, is one of the most studied. A study of patients with generalized anxiety disorder found that both blood plasma and salivary cortisol levels correlated positively with HAM-A scores before treatment, with the salivary measure showing an especially strong relationship.20Chinese General Practice. Value of Plasma and Salivary Cortisol Tests in Generalized Anxiety Disorder Similar findings have emerged in patients with oral diseases: people with oral lichen planus who scored higher on the HAM-A also had higher salivary cortisol, and patients with oral submucous fibrosis showed the same pattern with serum cortisol.21PubMed Central. Association of salivary cortisol and anxiety levels in lichen planus patients22PubMed Central. Assessment and correlation of anxiety, depression and serum cortisol levels in patients with oral submucous fibrosis and leukoplakia
These cortisol correlations do not mean the HAM-A is a proxy for a blood test, and cortisol levels are influenced by many things beyond anxiety, including sleep, exercise, time of day, and medications. But the consistency of the association across different patient groups gives some biological grounding to what is fundamentally a behavioral rating. It suggests the scale is picking up something real about the body’s stress response, not just a clinician’s subjective impression. Whether cortisol or other biomarkers could eventually supplement or partially replace clinician-rated scales is an active area of research, though nothing has come close to replacing the interview yet.