Sign Up
Login
Try for Free
Login
The STOP-Bang is an eight-item obstructive sleep apnea risk screen published by Frances Chung and colleagues at Toronto's University Health Network in 2008, scored 0 to 8 with one point per positive domain. Anesthesia, primary care, sleep, and occupational teams use it to stratify OSA risk before surgery and referral. It screens; it never diagnoses. This page covers how to document and interpret STOP-Bang results, with a fictional sample note.
Patient self-report with clinician-measured items (BMI, neck circumference, blood pressure history); scored by nursing, anesthesia, or clinic staff; use is licensed through University Health Network
Anesthesia and preadmission teams, primary care and sleep medicine, occupational and driver medical examiners, bariatric programs, payers and auditors
2 to 6 chart lines plus the decision it drives · completion under 2 minutes
OSA risk-screening questionnaire (eight yes-or-no domains, one point each, total 0 to 8)
Preoperative assessment, primary care and sleep referral, bariatric workup, occupational and driver fitness evaluations, Australia's direct-access sleep-study pathway
Property of University Health Network: clinical, EHR, and web use go through UHN's express-licence process; named in Australia's MBS sleep-study pathway; mandated by no US or Canadian law
The STOP-Bang questionnaire is an eight-domain screen for obstructive sleep apnea risk, published in Anesthesiology in 2008 by Frances Chung and colleagues at Toronto Western Hospital, University Health Network. The four STOP domains are symptom and history items (a clinically significant snoring burden, daytime tiredness, another person's observation of breathing pauses during sleep, and hypertension); the four Bang domains are anthropometric and demographic (a body-mass-index threshold, an age threshold, an enlarged neck circumference, and male sex). Each positive domain scores one point for an unweighted total of 0 to 8. Official versions are not identical: the current UHN interactive tool uses a single neck threshold of about 40 cm while some official 2015 country forms use sex-specific thresholds of about 43 cm for men and 41 cm for women, and the literature holds weighted, continuous, bicarbonate-augmented, and population-adjusted variants, so the chart names the form, locale, or institutional build, not just "STOP-Bang."
Two facts govern the documentation. First, it is a screen and a risk stratifier, never a diagnosis: the American Academy of Sleep Medicine's guideline states that questionnaires and prediction algorithms must not be used to diagnose OSA without polysomnography or technically adequate home testing, and no score establishes OSA severity or treatment eligibility. Second, "positive" is ambiguous unless the algorithm is named. The original validation literature used 3 or more as the positive threshold; the current UHN interpretation runs three bands (0 to 2 low, 3 to 4 intermediate, 5 to 8 high) with a combination rule that moves an intermediate score to high risk when at least two STOP domains are positive together with a BMI above 35, an enlarged neck, or male sex. A chart that imports "3 or more equals high" from an older source will disagree with a chart built on the current algorithm, and both will be wrong about the sleepiness question, which belongs to the Epworth Sleepiness Scale: the two instruments measure different constructs and Australia's Medicare pathway requires both for exactly that reason. Insomnia complaints route to the Insomnia Severity Index, not to this screen.
Anesthesia and preadmission teams are the largest users, because the Society of Anesthesia and Sleep Medicine's 2016 preoperative guideline recognizes validated questionnaires for OSA risk identification, the Society for Ambulatory Anesthesia recommended the STOP-Bang for ambulatory screening in 2012, and the American Society of Anesthesiologists' 2014 guideline asks anesthesiologists and surgeons to decide jointly between presumptive OSA management and further evaluation; none of the three mandates the instrument or a cancellation threshold. Primary care and sleep medicine use it to support referral for home sleep apnea testing or polysomnography, where payer rules key off clinical signs and symptoms rather than a score in the United States and where Australia's Medicare direct-access pathway names it explicitly alongside the Epworth score. Bariatric programs screen because OSA prevalence is high, under a locally chosen protocol rather than a national mandate. Occupational and commercial-driver examiners meet the largest myth in the field: FMCSA's regulations prescribe no OSA screening tool, referral rule, or score, and Canada's and Australia's driver standards treat the questionnaire as an aid to referral, not a licensing criterion. The perioperative reasoning is often documented in a consultation-liaison note, the sleep referral in a referral letter, and the safety-critical-work question in a fitness-for-duty evaluation, each carrying the score with the decision it informed.
No UHN, ASA, SASM, or AASM source prescribes a STOP-Bang note format, and no US or Canadian law names the instrument. What survives review is a record that names the version or build, reports the total with the positive domains and the source of each response, states the band under a named algorithm, keeps the result a screen, documents the decision and its reasoning, and assigns ownership of what happens next. Each element below carries the pitfall that most often undermines it.
Version, build, and licence. Chart "STOP-Bang" plus the form, locale, or institutional build and the scoring algorithm in use (the current UHN three-band interpretation with its combination rule, or an older binary 3-or-more threshold), the completion date, and whether the responses were patient-reported, partner-reported, clinician-observed, imported, or historical. The instrument is UHN property; use it under the licence your organization holds. Pitfall: "STOP-Bang positive" with no version or algorithm, when official versions differ on the neck threshold and older and current interpretations disagree about a 3 or 4.
The total and the positive domains. Report x/8 and which domains were positive (by letter or concept), rather than a bare total; note missing or disputed responses, who supplied them, which response was used for the formal score, and whether the alternate score would change the band. Pitfall: "STOP-Bang 4" with no domains, or a partner's report silently converted into a definitive yes when the patient disputes it.
The band under a named algorithm. State low (0 to 2), intermediate (3 to 4), or high (5 to 8) risk under the UHN three-band algorithm, and whether an intermediate score was reclassified to high by the combination rule (at least two STOP domains plus BMI above 35, an enlarged neck, or male sex; age is not one of the three override variables). Pitfall: Every score of 3 or more charted as high risk from an older source, or a 4/8 that met the override recorded as merely intermediate.
Screen, not diagnosis, with diagnostic status stated. Write "screen-positive, suspected OSA" or "previously confirmed OSA" with any prior AHI or REI, PAP prescription, and adherence, and state that the score establishes neither OSA nor its severity; only polysomnography or adequate home testing diagnoses. Pitfall: "OSA per STOP-Bang" or a severity grade inferred from the number, or a known PAP-treated patient re-screened as if undiagnosed.
Context that changes the meaning. Record the procedure and its invasiveness, anticipated opioids or sedation, airway concerns, comorbid disease, and any evidence of hypoventilation, severe pulmonary hypertension, or unexplained resting hypoxemia; in primary care, the sleepiness picture, driving history, and safety-critical work; in women, thin patients, and East Asian populations, that the conventional threshold performs differently. Pitfall: A high score with no gas-exchange assessment, or a low score in a symptomatic woman charted as "no OSA."
The decision and its reasoning. Document what the score changed: presumptive OSA precautions and proceeding, delay for evaluation, referral for testing, or a documented informed decline, with the reasoning. SASM advises against routine delay or cancellation solely for workup except with additional evidence of disturbed ventilation or gas exchange; ASA permits presumptive management or delay, including when suspicion first arises on the day of surgery. Pitfall: "Score high, surgery proceeded" with no reasoning, or a high score treated as an automatic cancellation or an automatic sleep-study order.
Plan, ownership, and safety advice. Name who will order testing, communicate with anesthesia and the receiving unit, arrange follow-up, or reassess, and record sleepiness-related driving and occupational advice where relevant; a declined workup gets the offered evaluation, material risks, alternatives, capacity, the patient's reason, and the follow-up plan. Pitfall: "Patient refused" alone, or a referral with no owner in a chart where nothing happens next.
STOP-BANG DOCUMENTATION BLOCK Date: [ ] Setting: [ ] Clinician: [ ] Version / build: [UHN three-band build / country form + year / other] Algorithm: [three-band with combination rule / binary 3-or-more] Source of responses: [patient / partner / clinician-observed / imported] Score: [ ]/8 Positive domains: [S T O P B A N G] Missing or disputed: [which; who reported; response used; alternate score [ ]/8; band changed? yes / no] Band: [low 0-2 / intermediate 3-4 / high 5-8]; override applied? [yes: two STOP domains + BMI over 35 / neck / male; no] Diagnostic status: [screen-positive, suspected OSA / confirmed OSA: prior AHI-REI, PAP prescribed, adherence]; screen does not diagnose Context: [procedure, opioids/sedation, airway; hypoventilation, severe pulmonary hypertension, resting hypoxemia; sleepiness, driving, safety-critical work; sex/body habitus/population caveats] Decision + reasoning: [proceed with presumptive OSA precautions / delay for evaluation / refer for testing / informed decline documented] Precautions or referral details: [ ] Owner + follow-up: [who orders / communicates / reassesses; date] Safety advice: [driving, occupational] Record: [full responses and scores retained where required (AU MBS)] 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. Neither reproduces the questionnaire items or forms.
Scenario: a high-risk screen at a preadmission visit before major abdominal surgery, with no prior sleep study, no additional gas-exchange concern, and a documented decision to proceed with presumptive OSA precautions and an owned outpatient referral. All details are fictional.
Patient: J.M., 58 · Setting: Preadmission clinic, elective major abdominal surgery · Clinician: R. Iyer, MD (anesthesiology) · Note date: 08/13/2026
Screening: STOP-Bang completed today in the hospital's licensed EHR module (UHN three-band build, single 40 cm neck threshold), responses patient-reported with BMI, neck circumference, and blood pressure history entered by nursing. Score 6/8. Positive domains: S, O, P, B, A, G; T and N negative. Partner present and confirms the observed breathing pauses; patient agrees. High risk under the three-band algorithm (no override needed at 6). This is a screening result: no prior sleep study, no established OSA diagnosis, no PAP.
Assessment: Denies resting dyspnea and morning headaches; room-air saturation 96% seated; no clinical or laboratory evidence of hypoventilation, no known severe pulmonary hypertension, no unexplained resting hypoxemia on the available record. Procedure is time-sensitive, laparoscopic with possible open conversion, moderate expected opioid requirement. Airway examination documented separately; predictors of difficult mask ventilation present. Screen indicates high probability of undiagnosed OSA; it does not establish severity.
Decision and plan: Reviewed with the surgeon and the patient. Per SASM and ASA guidance, no delay for a preoperative sleep study in the absence of additional ventilation or gas-exchange concerns; proceed with presumptive OSA precautions: difficult-airway equipment immediately available, opioid-sparing multimodal analgesia with regional technique if feasible, no routine continuous basal opioid infusion, minimal sedative premedication, head-up positioning, continuous pulse oximetry in recovery with extended observation, and postoperative disposition decided on oxygenation, ventilation, sedation, and analgesic requirement. Recovery-area and receiving-unit handoff to include suspected OSA. Patient informed that outpatient diagnostic sleep evaluation remains recommended regardless of surgical course; referral responsibility assigned to primary care with a copy of this note, and the patient given sleepiness-related driving advice.
Administrative: Full responses and score retained in the licensed module; score, domains, band, algorithm, and decision entered in the preoperative assessment. Instrument used under the hospital's UHN licence; no questionnaire text reproduced here.
This sample is fictional and for educational purposes. It does not describe a real patient or record; the scores, dates, and details are invented to show documentation structure and are not clinical guidance.
Writing these after every session? BastionGPT drafts complete notes from bullets, dictation, or a transcript.
Generate a note from bulletsUnited States: guideline conventions, payer policies, and one widely misquoted federal standard, none of which mandates the instrument. Ambulatory surgical centers participating in Medicare must keep a pre-surgical assessment policy grounded in age, diagnosis, procedure, comorbidities, anesthesia level, and nationally recognized standards (42 CFR 416.52, LAW), and the regulation prescribes no questionnaire. The ASA's 2014 perioperative OSA guideline (CONVENTION) asks anesthesiologists and surgeons to decide jointly between presumptive management and further testing and permits either when suspicion first arises on the day of surgery; the SASM 2016 preoperative guideline (CONVENTION) recognizes validated questionnaires and recommends against delaying or cancelling surgery solely for OSA workup except in patients with additional evidence of disturbed ventilation or gas exchange (hypoventilation, severe pulmonary hypertension, unexplained resting hypoxemia); the Society for Ambulatory Anesthesia recommended the STOP-Bang for ambulatory screening in 2012; MBSAQIP's 2026 bariatric standards require a committee-approved risk-assessment process without naming the tool (CONVENTION). On coverage (PAYER POLICY): Medicare's sleep-testing national coverage determination (240.4.1, 2009) requires clinical signs and symptoms indicative of OSA and an eligible test and names no questionnaire; among commercial payers, Cigna's delegated eviCore sleep-disordered-breathing guideline (V1 effective 1 May 2026, V2 scheduled 5 November 2026) names a validated questionnaire such as the STOP-Bang or Berlin as one element of the clinical evaluation, while the Aetna, UnitedHealthcare, and Carelon policies reviewed key off signs, symptoms, examination, and test suitability, with Carelon revisions scheduled for September and November 2026; a high score therefore strengthens a referral and guarantees no authorization. For commercial drivers, 49 CFR 391.41(b)(5) (LAW) bars a respiratory dysfunction likely to interfere with safe driving and names no instrument, FMCSA's Medical Examiner's Handbook (2024, CONVENTION without force of law) states that its regulations do not prescribe OSA screening, referral, preferred tests, treatment, compliance thresholds, waiting periods, maximum certification periods, or routine retesting, and the 2016 rulemaking was withdrawn in August 2017; examiners refer on multiple risk factors under the general standard, and "DOT requires the STOP-Bang" is false.
Canada, Australia, and the evidence. Canada's June 2026 CCMTA driver-fitness model standard (CONVENTION, applied through provincial licensing LAW) names no questionnaire and turns on diagnosed severity, sleepiness, crash history, and treatment (untreated disease with an AHI below 30 and no admitted sleepiness is generally compatible with all classes; severe disease generally requires effective treatment or specialist assessment; commercial drivers with OSA are reviewed annually), the Canadian Anesthesiologists' Society 2026 guidelines set no STOP-Bang threshold, and British Columbia's adult OSA guideline names it as a tool that may help assess risk while routing testing through the provincial pathway (PROVINCIAL POLICY). Australia is the exception that names the instrument: for the direct GP-initiated adult sleep-study pathway (items 12203 and 12250, PAYER POLICY, current MBS), the patient must score 3 or more on the STOP-Bang, or 5 or more on the OSA50, or high risk on the Berlin, and 8 or more on the Epworth Sleepiness Scale, and the sleep-study provider must retain the fully administered questionnaires with responses and scores for possible audit; failing the thresholds blocks the direct pathway but not the specialist route, and an older, still-indexed fact sheet with a superseded threshold should not be relied on. Austroads' Assessing Fitness to Drive lists the STOP-Bang, OSA50, and Berlin as aids to deciding who to refer, not as licensing criteria (CONVENTION applied through state LAW), the Heavy Vehicle National Law's unfit-to-drive duty (1 August 2026) is a general fitness obligation, and ANZCA's pre-anaesthesia guidance requires risk assessment and documentation without prescribing a tool. On the evidence: the 2015 meta-analysis pooled 90%, 94%, and 96% sensitivity for any, moderate-to-severe, and severe OSA in sleep clinics at a threshold of 3 (specificity 49%, 34%, 25%) and 84%, 91%, and 96% in surgical populations (specificity 43%, 32%, 29%), with the probability of severe OSA at scores 3, 4, 5, 6, and 7 to 8 running about 25%, 35%, 45%, 55%, and 75% in sleep clinics and 15%, 25%, 35%, 45%, and 65% in surgical patients, and its detailed cut-point table came from only six pooled studies; the 2017 bivariate meta-analysis of 108 studies and 47,989 participants found the STOP-Bang more sensitive than the Berlin, STOP, and Epworth (88%, 90%, 93% by severity) and the Epworth more specific but far less sensitive; a 2022 surgical meta-analysis of 3,247 patients found 85%, 88%, and 90% sensitivity with 47%, 29%, and 27% specificity; the combination rule raised specificity for moderate-to-severe OSA from about 31% for any three positive domains to about 85% (two STOP domains plus BMI over 35), 79% (plus enlarged neck), and 77% (plus male sex) in the 2014 and 2016 analyses, while a 2020 external validation found the refinements added little in that surgical cohort; a midlife-women study found 77% sensitivity and 45% specificity at 3 or more (AUC 0.67), a 2021 meta-analysis of 47 studies and 26,547 sleep-clinic patients found substantially lower accuracy in East Asia, a multiethnic surgical study used a BMI threshold of 27.5 for Chinese and Indian patients, and USPSTF's 2022 I statement found insufficient evidence for screening asymptomatic adults. The chart names the population caveat where it applies.
Rights are stricter than most pages say. The STOP-Bang is the property of University Health Network, Toronto, developed by Frances Chung and colleagues; the official screening page carries the UHN property notice and an all-rights-reserved footer, and the FAQ routes every permission request through UHN Commercialization's express-licence forms rather than an email or an attribution line, with an intake that expressly covers routine patient screening and clinical care, sponsored research, and clinical trials. UHN identifies software integration and redistribution as uses that may require permission or a licence, publishes no blanket clinical-use permission, no public fee schedule, no Creative Commons licence, and no exception for free web calculators, and describes a tiered model (free for non-profit use under agreement, an annual fee for corporate health partners, pay-per-administration for industry); its foundation reported nearly 600 licences and about $3 million in licensing revenue by 2023. The AASM's own patient site states that clinical use requires licensing through the official website. Storing a clinician's calculated score and interpretation is different from reproducing the instrument, but the public pages define no safe harbor for item text or scoring logic in an EHR, so an EHR build, a portal form, or a decision-support rule goes through the licence. Official translations exist in more than twenty languages, and neither translation availability nor free web calculators imply public-domain status. STOP-Bang is the property of University Health Network and its authors. BastionGPT is not affiliated with, or endorsed by, University Health Network or the instrument's authors. This page reproduces no questionnaire items or forms and describes the domains and scoring in original prose.
The numbers behind these errors are specific. At a threshold of 3, the 2015 meta-analysis pooled about 90% sensitivity for any OSA in sleep clinics against 49% specificity, so most screen-positive patients do not have severe disease; the combination rule for intermediate scores raises specificity for moderate-to-severe OSA from about 31% to as high as 85% in the 2016 two-step analysis; in a midlife-women cohort a threshold of 3 caught only 77% of moderate-to-severe cases; the SASM guideline recommends against delay for workup except with disturbed ventilation or gas exchange; and FMCSA's handbook states its regulations prescribe no OSA screening tool. The BastionGPT Clinical Advisory Board sees the same errors most often in STOP-Bang 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.
Eight yes-or-no domains score one point each for a total of 0 to 8: four STOP domains (a clinically significant snoring burden, daytime tiredness, another person's observation of breathing pauses in sleep, hypertension) and four Bang domains (a body-mass-index threshold, an age threshold, an enlarged neck circumference, male sex). Under the current UHN interpretation, 0 to 2 is low risk, 3 to 4 intermediate, and 5 to 8 high, and an intermediate score is reclassified to high when at least two STOP domains are positive together with a BMI above 35, an enlarged neck, or male sex (age is not one of the override variables). The older validation literature simply called 3 or more positive. The bands stratify probability, not disease: in the 2015 meta-analysis the estimated probability of severe OSA at scores of 3, 4, 5, 6, and 7 to 8 was about 25%, 35%, 45%, 55%, and 75% in sleep clinics and 15%, 25%, 35%, 45%, and 65% in surgical patients, which is why a calculator that shows one universal probability is misleading.
It depends on which algorithm your chart names, which is exactly why the algorithm must be named. The original 2008 validation and much of the older literature treated 3 or more as the positive threshold, and many pages still call every 3 "high risk." The current UHN three-band interpretation calls 3 to 4 intermediate unless the combination rule applies: at least two STOP domains plus a BMI above 35, an enlarged neck, or male sex moves the patient to high risk, a refinement that raised specificity for moderate-to-severe OSA from about 31% for any three positive domains to about 85%, 79%, and 77% for those three combinations in the 2014 and 2016 analyses, though a 2020 external surgical validation found the refinements added little there. So write "4/8, intermediate by the UHN three-band algorithm, not reclassified" or "4/8, high risk under the combination rule because two STOP domains and the BMI modifier were present," never just "positive."
No and usually no. The score establishes neither OSA nor its severity: the American Academy of Sleep Medicine's guideline says questionnaires must not be used to diagnose OSA without polysomnography or technically adequate home testing, and even at 7 to 8 roughly a quarter to a third of surgical patients do not have severe disease. On the day of surgery, the SASM 2016 guideline recommends against delaying or cancelling for OSA workup except in patients with additional evidence of disturbed ventilation or gas exchange (hypoventilation, severe pulmonary hypertension, unexplained resting hypoxemia), and the ASA 2014 guideline permits presumptive OSA management or an individualized delay decided jointly by anesthesiologist and surgeon. Document that the score was recognized as a screen, the gas-exchange assessment, the procedure and opioid context, existing PAP or prior testing, the precautions used (airway readiness, opioid-sparing multimodal analgesia, no routine basal opioid infusion, enhanced recovery monitoring, disposition by oxygenation and sedation), and the handoff. "Score high, proceeded" with no reasoning is the version reviewers reject.
It can justify a clinical referral when the whole evaluation raises suspicion; it authorizes nothing by itself in the United States or Canada, and it is required in Australia. Medicare's national coverage determination for sleep testing requires clinical signs and symptoms indicative of OSA and an eligible test type and names no questionnaire; Cigna's delegated eviCore guideline (2026) names a validated questionnaire such as the STOP-Bang or Berlin as one element of the clinical evaluation, while the Aetna, UnitedHealthcare, and Carelon policies reviewed key off symptoms, comorbidities, examination, and test suitability, and all apply plan-specific terms. In Canada, provincial pathways govern testing and none names a score. In Australia, the direct GP-initiated pathway (items 12203 and 12250) requires a STOP-Bang of 3 or more (or an OSA50 of 5 or more, or Berlin high risk) together with an Epworth of 8 or more, and the provider must retain the fully administered questionnaires with responses and scores for audit; failing the thresholds closes the direct pathway, not the specialist route. Document the symptoms, sleepiness, driving and safety context, prior testing, and suitability for home testing alongside the score.
Not on the strength of attribution or the tool being free. The STOP-Bang is the property of University Health Network; the official site's FAQ routes every permission request through UHN Commercialization's express-licence forms, whose intake expressly covers routine patient screening and clinical care, and UHN identifies software integration and redistribution as uses that may require permission or a licence. It publishes no blanket clinical-use permission, no public fee schedule, and no free-calculator exception; its tiered model runs free for non-profit use under agreement, an annual fee for corporate health partners, and pay-per-administration for industry, and by 2023 it reported nearly 600 licences. An EHR questionnaire, portal form, decision-support rule, or auto-scored flowsheet is a software integration even when only your clinicians see it; a commercial host of a free public scorer should use the commercial route. Storing a clinician's calculated score and interpretation is different from reproducing the items, but no published safe harbor defines how much text or logic may be stored without permission. Terms as stated at stopbang.ca and UHN Commercialization in August 2026; the signed agreement controls.
No. The US federal standard (49 CFR 391.41(b)(5)) bars a respiratory dysfunction likely to interfere with safe commercial driving and names no instrument, BMI rule, neck measurement, or screening score; FMCSA's 2024 Medical Examiner's Handbook states that its regulations do not prescribe OSA screening, referral, preferred testing, treatment, compliance thresholds, waiting periods, maximum certification periods, or routine retesting, and the 2016 rulemaking was withdrawn in August 2017. Examiners may refer a driver on multiple risk factors and may withhold or shorten certification while a condition that may interfere with safe driving is unresolved; treated OSA does not automatically preclude certification. Canada's CCMTA model standard turns on diagnosed severity, sleepiness, crash history within five years, and treatment, with provincial licensing law and reporting duties layered on; Australia's Austroads standard lists the STOP-Bang, OSA50, and Berlin as aids to deciding who to refer, with licensing decided on diagnosed OSA, significant sleepiness, crash risk, and effective treatment through state law. Chart the examiner's individualized reasoning, never a mandate that does not exist.
The instrument is binary and no UHN rule or validation study tells you to override the patient, adopt a partner's report automatically, or assign half credit, so document the source and run the sensitivity analysis: who reported the event, that the patient disputes or cannot confirm it, which response was used for the formal score, the resulting score and band, and the alternate score and whether it changes the band or the decision ("formal score 4/8 using the partner's report; 3/8 if that domain is excluded; both intermediate under the stated algorithm, so the referral decision is unchanged"). For a patient with no bed partner, record the domain as not endorsed for want of a witness and say why, rather than silently scoring zero; in women the observed-apnea domain is the most sensitive single item, so its absence is worth flagging. Silently converting disputed information into a definitive yes, or a missing witness into a no, is the less transparent choice either way.
No, and the chart should say so where it matters. Male sex scores a point, so a woman with otherwise identical features sits one point lower and can cross a band boundary; in a midlife-women cohort a threshold of 3 had 77% sensitivity and 45% specificity for moderate-to-severe OSA (AUC 0.67), and observed apneas was the most sensitive single item. In a younger, thinner sleep-clinic sample a threshold of 3 had about 84% sensitivity but only 18% specificity, while in obese surgical cohorts 3 kept high sensitivity and higher cutoffs gained specificity for severe disease. A 2021 meta-analysis of 47 studies found accuracy for moderate-to-severe OSA substantially lower in East Asia, and a multiethnic surgical study used a BMI threshold of 27.5 for Chinese and Indian patients; a modified threshold is a modified instrument, to be named, validated, and licensed as such. The Epworth Sleepiness Scale measures subjective sleep propensity, not OSA risk, and is more specific but far less sensitive for OSA (about 54% sensitivity for any OSA in the 2017 meta-analysis versus 88% for the STOP-Bang); the Berlin is an alternative categorical risk tool with lower pooled sensitivity. They are not interchangeable, and Australia's pathway requires an OSA-risk instrument and the Epworth together for that reason.
Yes. Give it the facts (version or build, score and positive domains, source of each response, diagnostic status, procedure and gas-exchange context, symptoms and comorbidities, decision, and owner) and it drafts the full entry: the band under the named algorithm, the screen-not-diagnosis line, the reasoning for proceeding, delaying, or referring, the precautions or referral details, the owner and follow-up, and the safety advice, ready for your review. It can also check a finished note for a "positive" with no algorithm, a screen written as a diagnosis, a high score with no reasoning either way, a disputed response flattened, and a referral with no owner. 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.