Sign Up
Login
Try for Free
Login
The GAD-7 is a seven-item anxiety severity measure scored 0 to 21, developed by Spitzer, Kroenke, Williams, and Löwe in 2006 and free to reproduce under the official PHQ Screeners terms. Primary care and behavioral health teams use it for anxiety screening and symptom monitoring. This page covers how to document and interpret GAD-7 results in the chart, with a fictional sample note.
Patient self-report; any clinician or trained staff can administer and score, no license or qualification level required
Treating clinicians and care teams, collaborative care registries, accreditation surveyors, payers and auditors
2 to 5 chart lines, often alongside a paired PHQ-9 · patient completion under 2 minutes
Self-report anxiety severity measure (7 items, scored 0 to 21)
Anxiety screening, symptom monitoring during treatment, collaborative care registries, paired depression and anxiety measurement
Free to reproduce without permission under the PHQ Screeners terms; no US, Canadian, or Australian authority mandates it by name
The Generalized Anxiety Disorder 7-item scale (GAD-7) is a seven-item anxiety severity measure developed by Robert Spitzer, Kurt Kroenke, Janet Williams, and Bernd Löwe, published in 2006 from a validation study across 15 US primary-care clinics, with development supported by a Pfizer educational grant. Patients rate seven symptoms over the past two weeks from 0 to 3, for a total of 0 to 21, with four published bands: 0 to 4 minimal, 5 to 9 mild, 10 to 14 moderate, and 15 to 21 severe. The first two items double as the GAD-2 gateway screen (0 to 6, positive at 3 or above). The official PHQ Screeners terms permit reproduction, translation, display, and distribution without permission or fee, and the developers describe the instrument family as public domain, which covers EHR builds, portals, apps, and printed forms; roughly 90 translations are distributed the same way, though translation availability is not the same thing as language-specific validation.
The load-bearing distinction for documentation is that the GAD-7 measures two weeks of symptoms while the GAD diagnosis requires about six months of excessive, hard-to-control worry: in the original validation, only 67% of patients scoring 10 or above reported symptoms lasting six months, so roughly a third of positive screens fail the duration element alone. The scale is also deliberately transdiagnostic in practice: at the cutoff of 10 it flagged panic disorder, social anxiety disorder, and PTSD with lower sensitivity than GAD itself, which makes a high score a useful anxiety-distress signal and a weak classifier. And it contains no self-harm question, so it can never stand in for a risk screen. Its natural partner is the PHQ-9, documented once each with a single integrated interpretation.
Primary care teams document GAD-7 results inside anxiety screening workflows driven by USPSTF recommendations; therapists and psychiatric prescribers trend it for measurement-based care; care managers track it in collaborative care registries, where the billing codes require a validated scale without naming one; and accredited behavioral health organizations use it to satisfy standardized outcome-monitoring expectations. It anchors the anxiety column of an outcome measure note and usually travels with the PHQ-9. Neighbors matter at the boundaries: the Beck Anxiety Inventory is a licensed Pearson instrument with a heavier somatic emphasis, PROMIS anxiety measures carry their own reproduction terms, and Australian Better Access workflows name the K10 and DASS-21 as their example outcome tools, so a GAD-7 entry there rests on the clinician-discretion clause and should say why an anxiety-specific, two-week measure fits.
No statute prescribes a GAD-7 note format. The elements below are the convention that survives review because each maps to something a future clinician, an accreditation surveyor, or an auditor will need. Most results fit in a few lines beside the paired PHQ-9. Each element carries the pitfall that most often undermines it.
Instrument and administration context. Name the form (GAD-7 or the two-item GAD-2), the mode (paper, portal, tablet, telephone, interview), the language, and any assistance. Pitfall: treating administration detail as noise. A randomized 2026 study shifted scores 2.63 points, enough to cross a band, by changing instructional emphasis alone, and a paper-versus-smartphone comparison found only moderate agreement between modes.
Total over 21 with the published band. Write the fraction and the band: 0 to 4 minimal, 5 to 9 mild, 10 to 14 moderate, 15 to 21 severe. Pitfall: importing the PHQ-9's five bands onto this four-band scale, or calling 15 moderate; 15 begins the severe band, an error live on at least one currently ranking documentation page.
The functional-difficulty item, separately. Record the unscored response about work, home, and relationships beside the total, never inside it. Pitfall: an eighth scored item. The validated range is 0 to 21, and the functional answer is the score's impairment context, not its arithmetic.
Trajectory, in points and percent. Name the baseline score and date, then both numbers. Pitfall: one universal change threshold. The published minimal clinically important difference is about 4 points from a chronic-depression sample, reliable change has been estimated near 6, the 2026 MBHR2 quality measure defines response as a 25% reduction at 4 to 8 months, and research conventions use 50%; name the construct you are claiming.
An interpretation sentence. State severity, concordance with the interview and function, and, for a prospective GAD diagnosis, the duration bridge: the scale asks about two weeks, the diagnosis requires about six months. Pitfall: "GAD-7 = 12, generalized anxiety disorder" charted from the score. In the original validation, a third of patients above the screening threshold did not report six months of symptoms.
What the score did not assess. When the picture includes panic attacks, trauma symptoms, social fear, or safety concerns, say how they were assessed, because the GAD-7 under-detects those disorders and contains no self-harm item; risk gets its own documentation in a risk screening note. Pitfall: a low total charted as anxiety ruled out. Pooled meta-analytic sensitivity at the cutoff of 10 is about 0.74, and lower for panic, social anxiety, and PTSD.
Plan linkage and the next measurement. State what continues, changes, or is deferred because of the result, and when the next administration is due. Pitfall: citing an every-session rule as the instrument's requirement. Every-session administration is a program convention (Ontario Structured Psychotherapy runs one), not a steward rule; accreditation standards leave the interval to the organization.
GAD-7 DOCUMENTATION BLOCK Form: [GAD-7 / GAD-2] Language: [ ] Mode: [paper / portal / tablet / interview] Date: [ ] Total: [ ]/21 Band: [minimal 0-4 / mild 5-9 / moderate 10-14 / severe 15-21] Functional difficulty (unscored): [not difficult ... extremely] Prior score / date: [ ] Change: [points] ([percent]) Construct: [MCID ~4 / reliable change ~6 / response per named convention / remission <5 (local convention)] Paired measures this visit: [PHQ-9 __/27 or n/a] Interpretation: [severity + concordance with interview and function; for new GAD dx: duration bridge, two weeks measured vs six months required] Not assessed by this scale: [panic / trauma / social anxiety / safety, and how each was addressed] Plan linkage: [continued / changed / deferred, and why] Next measurement: [interval + reason] Clinician signature / credentials: Date:
Free to use and share, no signup. The PDF includes a one-page cheat sheet with element-by-element pitfalls and a pre-sign checklist; the DOCX is the blank documentation block, ready to adapt.
Scenario: an intake where the GAD-7 is elevated, the paired PHQ-9 is not, and the clinician needs the note to carry the duration bridge from a two-week screen toward a six-month diagnosis. This is the score documentation only, not the full intake. All details are fictional.
Patient: R.M., 29 · Visit: Intake assessment · Clinician: D. Whitfield, PhD, Licensed Psychologist · Note date: 08/13/2026
Measures: GAD-7 (English, patient portal, self-completed 08/13/2026): 14/21, moderate anxiety symptoms. Functional difficulty reported as very difficult. PHQ-9 same visit: 5/27, mild range, item 9 = 0. No prior administrations.
Duration and context (interview): R.M. describes excessive, difficult-to-control worry across work performance, finances, and family health on most days for approximately nine months, with initial insomnia and concentration interference. The two-week score and the interview history are concordant, and the duration element the questionnaire cannot measure is met by history.
Interpretation: The GAD-7 result is a positive severity screen consistent with generalized anxiety disorder, which was diagnosed today by clinical interview covering duration, impairment, and panic, trauma, social-anxiety, substance, medical, and mood alternatives. The score did not establish the diagnosis. Discrete panic episodes denied; no trauma-related symptoms reported. Safety screening was completed and documented separately in the intake risk section.
Plan linkage: Today's 14/21 is the measurement baseline. Begin weekly CBT targeting worry and intolerance of uncertainty with behavioral sleep strategies; re-administer the GAD-7 in four weeks with the PHQ-9 alongside, and interpret change against the baseline in points and percent with the convention named. Result reviewed with R.M., who found the severity band consistent with her experience.
This sample is fictional and for educational purposes. It does not describe a real patient or record, and the scores are invented for illustration and correspond to no real person.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsNo law in the US, Canada, or Australia mandates the GAD-7 by name; the operative expectations are PAYER POLICY and CONVENTION, and for anxiety they are thinner than clinicians assume. The USPSTF recommends anxiety screening for adults 19 through 64 (Grade B, 2023) with an insufficient-evidence statement for 65 and older, and for ages 8 to 18 (Grade B, 2022), naming no instrument and no interval. From there the US quality infrastructure is asymmetric: as of 2026 there is no national HEDIS or MIPS anxiety-screening measure comparable to the depression set, and the anxiety-specific reporting that does exist lives in a registry: MBHR2, Anxiety Response at 6 Months, takes adults with an anxiety diagnosis and an index GAD-7 of at least 8 and defines response as a 25% reduction at 4 to 8 months, a definition that matches neither the 50% research convention nor any severity band, which is exactly why the note should name the convention it is using. Collaborative care billing (99492 through 99494) requires validated rating scales and a registry without naming an instrument. In accredited behavioral health organizations, Joint Commission standard CTS.03.01.09 (guidance updated May 2026) expects a standardized instrument administered at multiple intervals with results feeding goals and treatment decisions, and leaves the instrument and cadence to the organization. For 96127 there is no national unit rule: state Medicaid programs and commercial plans publish materially different limits, so verify the current plan-specific policy and date the claim.
Cross-border, the labels matter. In Canada, Ontario Health's anxiety-disorders quality standard calls for validated severity measures and regular monitoring without mandating a tool, and Ontario Structured Psychotherapy directs every-session GAD-7 administration inside that program, a workflow rule that should never be cited as a national requirement. In Australia, Better Access requires an outcome tool in a mental health treatment plan unless clinically inappropriate and names the K10 and DASS-21 as examples; the GAD-7 fits through the clinician-discretion clause, and the plan-review cadence rules (generally not more often than every three months) govern plan reviews, not how often a brief measure may be administered. On measurement integrity: keep the standardized wording and instructions, because a 2026 randomized study moved scores 2.63 points, enough to cross a severity band, by changing instructional emphasis alone; record the mode, because paper and smartphone administrations agreed only moderately in a randomized crossover; and record the language, because the roughly 90 distributed translations are not all independently validated.
The GAD-7 may be used, reproduced, translated, displayed, and distributed without permission or fee under the official PHQ Screeners terms, and its developers describe the PHQ and GAD instruments as public domain; one instrument registry currently labels it copyrighted by Pfizer, so this page states the operative permission rather than a formal dedication. BastionGPT is not affiliated with, or endorsed by, Pfizer or the instrument authors. This page reproduces no proprietary test items, norms, or scoring materials from any licensed instrument.
The evidence base is specific about where this instrument gets over-read. Pooled meta-analytic sensitivity at the standard cutoff of 10 is about 0.74, not the original study's 0.89; in the original validation itself, only 67% of patients at 10 or above reported the six months of symptoms the GAD diagnosis requires; and a 2026 randomized study shifted scores 2.63 points with an instruction-emphasis change and no treatment at all. Even the severity bands get mangled: one currently ranking documentation page labels a score of 15 moderate when 15 begins the severe band. The BastionGPT Clinical Advisory Board sees the same errors most often in GAD-7 documentation reviews:
BastionGPT is specifically trained, tuned, and clinically tested on behavioral health progress notes and screening documentation.
See how clinicians use it day to day on the AI therapy notes page.
Many BastionGPT users report saving more than 90 minutes per day on documentation.
HIPAA-compliant with a signed BAA on every plan. Your data is never used to train models. BastionGPT drafts, you review and sign.
The total runs 0 to 21 across seven items, and the four published bands are 0 to 4 minimal, 5 to 9 mild, 10 to 14 moderate, and 15 to 21 severe. Those describe reported symptom frequency over the past two weeks, not a diagnosis. For scale: in a large general-population sample, only about 5% of adults scored 10 or above and about 1% scored 15 or above, so a moderate or severe band is genuinely uncommon background noise. A score of 10 or higher is the conventional threshold for further evaluation, and what follows it should be an interview, not a diagnosis line.
No. The mismatch is structural: the scale asks about two weeks, while the GAD diagnosis requires excessive, difficult-to-control worry on most days for about six months, plus impairment and the differential work a questionnaire cannot do. In the original validation, only 67% of patients scoring 10 or above reported six months of symptoms, so roughly a third of positive screens failed the duration element alone. The defensible chart language is "positive severity screen, consistent with GAD once the interview establishes duration and rules out the alternatives," with the diagnostic evaluation carrying the conclusion.
Name the construct, because several circulate and they disagree. The published minimal clinically important difference is about 4 points, estimated in a chronic-depression sample; reliable change, the threshold for exceeding measurement error, has been estimated near 6 points; the 2026 MBHR2 quality measure defines response as a 25% reduction at 4 to 8 months from an index score of at least 8; and research programs commonly use a 50% reduction, with remission below 5 as a local convention. A note that says "16 to 8, down 8 points, a 50% reduction, meeting this clinic's response convention" survives review; "improved" does not.
No. Pooled meta-analytic sensitivity at the cutoff of 10 is about 0.74, and the follow-on validation found it lower still for the neighbors: 0.74 for panic disorder, 0.72 for social anxiety disorder, and 0.66 for PTSD. A patient with disabling panic attacks or trauma symptoms can sit comfortably under every GAD-7 threshold. When the presentation argues with the score, document both and let the interview decide; and remember the instrument has no self-harm item, so safety is always assessed and documented separately.
No steward or national authority sets a cadence. The USPSTF found no evidence for an optimal screening interval; Joint Commission outcome-monitoring expectations leave the interval to the organization; and every-session administration is a program rule where it exists (Ontario Structured Psychotherapy runs one), not an instrument rule. A defensible pattern ties frequency to treatment phase: more often during active treatment changes, less often in maintenance, always at decision points, written into the plan with the next date. Billing follows its own rules: collaborative care requires systematic measurement, and 96127 unit limits are payer-specific.
No. The official PHQ Screeners terms permit reproduction, translation, display, and distribution without permission or fee, and the developers describe the family as public domain (one instrument registry still carries a "copyrighted by Pfizer" label, which does not change the operative permission). Two cautions: a reworded or restructured version is no longer the validated instrument, and the neighbors are licensed differently. The Beck Anxiety Inventory is a paid Pearson instrument, and PROMIS measures carry their own reproduction terms, so the free-to-copy habit stops at the GAD-7's edges.
Each score once, then one integrated interpretation: "PHQ-9 = 11, previously 15; GAD-7 = 8, previously 13; both domains improving, residual sleep disruption tracks the depressive symptoms; plan below." The two scales correlate around 0.75 while still measuring separable constructs, so duplicated narratives under separate headings add length without information. Where a service wants a single combined trend, the published PHQ-ADS composite sums both instruments into a 0 to 48 measure with cutpoints of 10, 20, and 30, which is a documented option, not a requirement. The PHQ-9 documentation page covers the depression-side specifics, including item 9.
No. None of the seven items asks about self-harm, death, or suicide, so a GAD-7 of any value says nothing about risk in either direction, and a note should never cite it as a risk screen. Safety gets its own assessment and its own documentation: a risk screening note for the screening layer and a suicide risk assessment when concern exists. When the GAD-7 travels with the PHQ-9, that instrument's item 9 provides a safety flag, with its own documented follow-up rules.
Yes. Paste the responses, or just the total and prior score, and it drafts the documentation block: the band, trajectory in points and percent, the paired PHQ-9 integrated once, the duration-bridge sentence when a new GAD diagnosis is in play, and the plan linkage, ready for your review. It can also check a finished note for a band that disagrees with its total, a diagnosis carried by the score alone, or a change claim with no named construct. BastionGPT is HIPAA-compliant with a signed BAA on every plan, and your data is never used to train models.
The instrument facts and compliance claims on this page trace to these sources, last verified August 2026:
Educational content, not legal or billing advice. Sample notes are fictional. Follow your organization's policies and your board, payer, and jurisdiction requirements.