STOP-Bang estimates the risk that a patient has obstructive sleep apnoea. It does not measure sleepiness, sleep quality or apnoea severity - it produces a pre-test probability that guides whether a sleep study is warranted.
Eight yes/no items each score one point, for a total of 0 to 8. Four are symptom questions, captured by the STOP acronym: snoring, tiredness, observed apnoea, and high blood pressure. Four are demographic or physical risk factors, captured by Bang: body mass index above 35 kg/m², age over 50, neck circumference above threshold, and male sex. The design means a patient can score highly on physical risk factors alone, without reporting a single symptom.
Where it comes from.
Chung and colleagues published the instrument in Anesthesiology in 2008, developed and validated in preoperative surgical patients. The clinical problem was specific: undiagnosed sleep apnoea raises perioperative risk, and anaesthetists needed a screen that could be completed in a minute or two during a preoperative assessment. The four STOP items were selected by factor analysis from fourteen candidate questions; the four Bang items were added on the strength of their individual predictive performance.
It has since been applied well beyond the operating theatre - in sleep clinics, primary care, bariatric services, occupational health and commercial driver assessment - and is now among the most widely used sleep apnoea screens. Administration takes one to two minutes and response rates in validation studies were 90 to 100 per cent.
How scoring works.
Each "yes" scores one point. The conventional bands are 0-2 low risk, 3-4 intermediate risk, and 5-8 high risk.
Two refinements are worth knowing. First, the rights holder's current scoring guidance also classifies a patient as high risk when they answer yes to two or more of the four STOP items and additionally are male, or have a body mass index above 35, or have a neck circumference above threshold. These combination rules trade sensitivity for specificity: for moderate-to-severe apnoea in surgical patients, a score of 3 or more has 87% sensitivity but only 31% specificity, whereas two positive STOP items plus a body mass index above 35 has 85% specificity at 21% sensitivity.
Second, the neck circumference threshold is not consistent across published versions. The original and the 2016 Chest review use a single cut-off of 40 cm (16 inches); the updated form distributed by the rights holder uses sex-specific thresholds of 43 cm (17 inches) for men and 41 cm (16 inches) for women. Record which version you used.
Score
Severity
Interpretation
0–2
Low risk
Low probability of obstructive sleep apnoea. Reassess if symptoms change.
3–4
Intermediate risk
Intermediate probability of OSA. Reclassify as high risk if the STOP score is 2 or more plus male sex, BMI above 35 or a large neck circumference; otherwise consider sleep testing based on clinical context.
5–8
High risk
High probability of moderate-to-severe OSA. Refer for polysomnography or home sleep apnoea testing.
How well it performs.
In the original surgical validation, a score of 3 or more detected 84% of patients with any apnoea, 93% with moderate-to-severe apnoea and 100% with severe apnoea, at specificities of 56%, 43% and 37% respectively. Nagappa and colleagues' 2015 meta-analysis reproduced that pattern across settings: in sleep clinic populations, sensitivity at the same cut-off was 90%, 94% and 96% for any, moderate-to-severe and severe apnoea, with specificity falling from 49% to 25% as the severity threshold rose.
Chiu and colleagues (2017) pooled 108 studies covering nearly 48,000 patients and compared screens directly. STOP-Bang had significantly higher sensitivity than the Berlin questionnaire, the STOP alone and the Epworth Sleepiness Scale for mild, moderate and severe apnoea - 88%, 90% and 93% respectively - with correspondingly lower specificity of 42%, 36% and 35%. The Epworth had better specificity but sensitivity around 50%, which is why it is not a substitute.
The score behaves as a graded probability rather than a binary. In surgical patients, the probability of moderate-to-severe apnoea rises from 0.18 at a score of 0-2 to 0.36 at 3, 0.50 at 5 and 0.60 at 7-8.
93%
Sensitivity, mod-severe OSA (≥3)
43%
Specificity, mod-severe OSA (≥3)
0-8
Score range
1-2 min
Administration time
How it compares to the alternatives.
Instrument
Items
Time
When to reach for it
STOP-Bang
8
1-2 min
The default apnoea risk screen: highest sensitivity of the brief questionnaires, and quick enough for preoperative and primary care use. UHN licence required.
Berlin questionnaire
10
~5 min
Alternative primary care apnoea screen; lower sensitivity than STOP-Bang in head-to-head pooling.
ASA checklist
14
~5 min
Perioperative apnoea risk stratification where an anaesthesia-specific framework is preferred.
Sleep quality and insomnia, not apnoea. Use when the complaint is poor sleep rather than breathing.
Polysomnography
-
Overnight
The confirmatory test. No questionnaire replaces it.
Right tool, wrong tool.
Reach for it when
-Preoperative assessment, where undiagnosed apnoea changes anaesthetic and postoperative planning
-Triaging who needs a sleep study in primary care or a psychiatric clinic
-Patients on sedating psychotropics, or with obesity, treatment-resistant depression or refractory hypertension, where undiagnosed apnoea is common and easily missed
-Occupational assessments where daytime sleepiness carries safety consequences
Reach for something else when
-You need a diagnosis. Only polysomnography or home sleep apnoea testing can confirm or exclude apnoea
-The question is daytime sleepiness itself - use the ESS
-The question is sleep quality or insomnia - use the PSQI or the ISI
-The patient is a child or adolescent. No authorised paediatric version of STOP-Bang exists; adaptations in the literature are independent research instruments
-The patient is pregnant. Recent evidence does not support STOP-Bang as a stand-alone screen in pregnancy
Reading the score with care.
STOP-Bang is deliberately tuned towards sensitivity, and its specificity is genuinely poor - between roughly 25% and 45% depending on setting and severity threshold. Read a positive result as "this patient warrants testing", not as "this patient has apnoea". In a low-prevalence population the positive predictive value falls sharply: in the Sleep Heart Health Study general-population sample, a score of 3 or more had a positive predictive value of 16% for moderate-to-severe apnoea while the negative predictive value was 95%.
That asymmetry is the point. The score is most informative when it is low: 0-2 makes moderate-to-severe apnoea unlikely. Where specificity matters more than case-finding, use the combination rules above, or the finding that adding a serum bicarbonate of 28 mmol/L or higher to a score of 3 or more raises specificity for moderate-to-severe apnoea from about 30% to 82%.
What it cannot tell you.
The dominant limitation is low specificity, which generates false positives and unnecessary sleep-study referrals. Validation cohorts drawn from sleep clinics and surgical lists have higher apnoea prevalence than general practice, so published predictive values overstate performance in unselected populations. Performance is weak in some groups: in a veteran cohort specificity was 5%, and it performs poorly in renal failure. Because male sex, age and body mass index score automatically, an older overweight man reaches intermediate risk before answering a single symptom question, while women are systematically scored lower - sex-specific thresholds have been proposed for this reason. Optimal cut-offs also differ by ethnicity, with a threshold of 4 recommended in several Asian and Caucasian cohorts. Neck circumference requires a tape measure and is measured inconsistently. And it is a screen, not a diagnostic test.
How licensing works.
STOP-Bang is the property of University Health Network in Toronto. The questionnaire form itself carries the line "Proprietary to University Health Network", and the rights holder directs all users - academic and commercial alike - to complete a licence request through UHN Commercialization, which offers separate express licences for academic and commercial use.
That is why the eight items are not reproduced here. Describing the structure, scoring and evidence is fine; publishing the item wording on a commercial site would require an executed licence. Clinics wanting to use it in routine care, or to embed it in an electronic record, should apply through UHN before doing so.
If you need a sleep instrument you can score immediately, the ISI and the PSQI are available on this site, though neither screens for apnoea.
Get our case study pack by email, with practical examples from clinical teams.
We will only use your details to send relevant Aisel resources. Unsubscribe at any time.
We value your privacy
We use cookies to analyse site usage and improve your experience. Analytics and embedded media (e.g. YouTube) only load if you accept. Read our cookie policy.