STOP-Bang Questionnaire for Sleep Apnea Screening

The STOP-Bang questionnaire is an eight-item screening tool designed to identify people at risk for obstructive sleep apnea (OSA), the condition in which your airway repeatedly collapses during sleep. Developed and first validated in surgical patients at preoperative clinics, it has since become one of the most widely used OSA screening instruments in the world, appearing in sleep clinics, primary care offices, and even commercial driver evaluations. Its strength is catching nearly everyone who has moderate or severe OSA; its weakness is flagging many people who do not.

What the Letters Stand For

Each letter in STOP-Bang represents one yes-or-no question. The first four, forming the “STOP” portion, are things you report about yourself:

S – Snoring: Do you snore loudly, enough to be heard through a closed door?
T – Tired: Do you often feel tired, fatigued, or sleepy during the day?
O – Observed: Has anyone observed you stop breathing or choking during sleep?
P – Pressure: Are you being treated for high blood pressure?

The second four, the “Bang” items, are demographic and physical measurements:

B – BMI: Is your body mass index above 35?
A – Age: Are you older than 50?
N – Neck: Is your neck circumference greater than 40 cm (roughly 16 inches)?
G – Gender: Are you male?

Each “yes” earns one point for a total possible score of 0 to 8. The original STOP-only version used just the first four questions. Adding BMI, age, neck size, and gender pushed sensitivities substantially higher, reaching 100% for detecting severe OSA in the original validation study.

How Scores Are Interpreted

A score of 3 or higher is the standard cutoff for being considered “at risk” for OSA. Patients scoring 0 to 2 are generally classified as low risk for moderate-to-severe disease, while those scoring 5 to 8 fall into a high-risk category. This is not a diagnosis — it tells a clinician whether a formal sleep study is warranted. The probability of having moderate-to-severe OSA climbs steadily with each additional point. In sleep clinic populations, the probability of severe OSA rose from about 25% at a score of 3 to roughly 75% at a score of 7 or 8.² In surgical populations, those numbers were somewhat lower — around 15% at a score of 3 and 65% at 7 or 8 — reflecting the fact that patients referred to sleep clinics tend to have more severe disease than the broader surgical population.

Higher scores also carry steeper odds. Patients scoring 5 had roughly ten times the odds of severe OSA compared to those scoring below 3, and patients scoring 7 or 8 had about fifteen times the odds.³ These are large effect sizes that make higher STOP-Bang scores genuinely informative for clinical decision-making.

Sensitivity Versus Specificity

Understanding how well the questionnaire performs requires looking at two numbers that always exist in tension. Sensitivity tells you how often the tool correctly flags someone who actually has OSA. Specificity tells you how often it correctly clears someone who does not. STOP-Bang is built to be sensitive — it would rather flag too many people than miss someone with a dangerous breathing disorder.

A large systematic review and meta-analysis of preoperative studies found pooled sensitivity for all OSA of 85%, rising to 88% for moderate-to-severe cases and 90% for severe cases. Specificity went in the other direction: 47% for all OSA, dropping to 29% for moderate-to-severe and 27% for severe.⁴ Another meta-analysis spanning 47 studies and over 26,000 participants reported that a score of 3 or higher had sensitivity above 90% for moderate-to-severe and severe OSA, with negative predictive values of 77% and 91%, respectively.⁵

In practical terms, if the questionnaire says you are low risk, there is a strong chance you genuinely do not have significant OSA. But if it says you are high risk, you might or might not actually have it. The tool is designed to cast a wide net, meaning a positive result is a prompt for further evaluation, not a reason to start treatment.

How It Compares to Other Screening Tools

STOP-Bang is not the only OSA questionnaire. The Epworth Sleepiness Scale (ESS) measures daytime sleepiness through eight situational questions, and the Berlin Questionnaire groups symptoms into three categories. Head-to-head comparisons generally favor STOP-Bang for sensitivity. One study found STOP-Bang’s sensitivity at about 91% versus 70% for the ESS, with both showing similar specificity around 75%.⁶ A separate study in a Persian population found the Berlin Questionnaire had slightly higher sensitivity than STOP-Bang (about 86% versus 81%), but STOP-Bang had substantially higher specificity (82% versus 53%) and a meaningfully better area under the curve.⁷ Results vary across populations, but the pattern is consistent: STOP-Bang tends to pick up more cases of OSA than its competitors, and it takes less than two minutes to complete.

The NoSAS score, a newer instrument that weighs neck circumference, obesity, snoring, age, and sex, has shown comparable performance to STOP-Bang in some studies, with similar areas under the curve for predicting both the standard breathing-event index and blood-oxygen dips during sleep.⁸ Even so, STOP-Bang remains the most widely adopted tool, partly because of its large validation base and partly because it is already embedded in perioperative protocols around the world.

Why It Struggles in Some Populations

One of the most discussed limitations of STOP-Bang is how it handles sex and gender. Because “male gender” is one of the eight items, every woman starts with one fewer possible point. This means women need to score positive on three of the remaining seven items to reach the at-risk threshold, which can make the tool less sensitive for detecting OSA in female patients. A study of midlife women found that using a cutoff of 3 produced a sensitivity of 77% and specificity of only 45% for moderate-to-severe OSA, with a modest area under the curve of 0.67.⁹ Research has increasingly called for gender-specific evaluation of questionnaire results to prevent underdiagnosis in women.¹⁰

The tool also performs differently in people who are very obese. Among obese patients being evaluated for bariatric surgery, overall sensitivity remained high — one study reported 93% for moderate-to-severe OSA — but specificity was only about 34%.¹¹ Another study of bariatric candidates found that the discriminative power of all questionnaires, STOP-Bang included, was “quite modest,” with area-under-the-curve values well below what is typically reported in general surgical populations.¹² The explanation is straightforward: when nearly everyone in the room has OSA risk factors like a high BMI, the questionnaire loses its ability to separate those who have the condition from those who do not.

Geographic variation matters too. The large JAMA Network Open meta-analysis found that STOP-Bang’s diagnostic accuracy for moderate-to-severe OSA was above 0.80 in most world regions but dropped to 0.52 in East Asian populations, essentially no better than a coin flip.¹³ Differences in craniofacial anatomy, BMI distributions, and symptom-reporting patterns all contribute to this gap, and it is an active area of research.

Use Before Surgery

The questionnaire was originally built for the preoperative setting, and that remains its most established use case. Undiagnosed OSA is a real hazard during and after surgery: the sedatives and pain medications given during procedures relax the muscles that keep the airway open, which in someone with an already-collapsible airway can lead to dangerous drops in oxygen. Multiple studies have linked higher STOP-Bang scores to worse perioperative outcomes.

One prospective study of urgent surgery patients found that those who screened positive had respiratory complications during surgery at more than three times the rate of those who screened negative (21% versus 6%) and higher rates of complications in the post-anesthesia care unit as well (57% versus 34%). These patients also stayed in the hospital longer — a median of 6 days compared to 4.¹⁴ Another study found that patients classified as high-risk before surgery had postoperative complications at nearly four times the rate of low-risk patients.¹⁵

In bariatric surgery specifically, a STOP-Bang cutoff of 4 predicted both airway obstruction and mild-to-moderate low oxygen levels during the postoperative period.¹⁶ And a study looking at outcomes in high-risk patients who were not using continuous positive airway pressure (CPAP) machines found significantly higher rates of pulmonary complications, worse oxygen levels, and longer hospital stays, including two cases of sudden death.¹⁷ These findings explain why many anesthesiology departments now require STOP-Bang screening before elective procedures.

Cardiovascular Risk and Beyond the Operating Room

OSA is not just a surgical concern. It is independently associated with high blood pressure, heart rhythm problems, heart failure, and stroke. The STOP-Bang questionnaire has been studied in cardiology populations with mixed results. In patients with atrial fibrillation, one study found STOP-Bang was 89% sensitive but only 36% specific for detecting OSA, meaning it had a high false positive rate in a group where many people already share risk factors with OSA patients.¹⁸ In a broader group of cardiovascular patients, the picture was even more lopsided — sensitivity of 97% but specificity of just 13%.¹⁹ At that level of specificity, the questionnaire is effectively marking nearly everyone as positive, which means it cannot meaningfully separate those with OSA from those without in that population.

Still, STOP-Bang scores carry prognostic weight beyond simple OSA detection. A prospective cohort study of hospitalized patients found that a score of 5 or higher was independently associated with cardiovascular death, with an adjusted hazard ratio above 3 — meaning those high-scoring patients were roughly three times more likely to die of cardiovascular causes after adjusting for other risk factors.²⁰ This finding suggests the questionnaire captures something about overall cardiometabolic risk that goes beyond just sleep-disordered breathing.

Strategies for Improving Specificity

Because the standard cutoff of 3 flags so many people who turn out not to have OSA, researchers have explored ways to sharpen the tool. One approach is simply raising the threshold. Using a cutoff of 4 or 5 instead of 3 drops sensitivity somewhat but pushes specificity much higher. Another approach is two-stage scoring: first screen with a cutoff of 3, then among those who score positive, look at whether at least two of the four “Bang” items (the objective measurements) are positive, or whether a specific Bang item like BMI or neck circumference is present. This alternative scoring has been shown to improve specificity while preserving reasonable sensitivity.²¹

A different strategy adds a blood test. A serum bicarbonate level of 28 mmol/L or higher, which can indicate chronic low-grade carbon dioxide retention from nighttime breathing problems, boosted specificity above 80% for all severity levels of OSA when added to the standard STOP-Bang cutoff of 3.²² This is a meaningful jump from the baseline specificity numbers in the 27-47% range, though it requires a blood draw and is not as simple as answering eight questions.

Overnight pulse oximetry, which tracks oxygen levels while you sleep, has also been studied as a complement. The oxygen desaturation index — how often oxygen drops by a certain amount per hour — correlates strongly with the formal sleep-study breathing-event index, more strongly than STOP-Bang scores alone do.²³ Combining the questionnaire with home oximetry may eventually give clinicians a path between “just a questionnaire” and “full overnight sleep study,” though insurance and logistics vary widely.

Cost-Effectiveness of Screening

Whether preoperative OSA screening actually saves money depends on the time horizon. A cost-effectiveness analysis published in Anesthesiology modeled several strategies: no screening, STOP-Bang alone, STOP-Bang followed by a portable home sleep monitor, and STOP-Bang followed by a full in-lab sleep study. Over just the perioperative period, no screening was cheapest, and adding formal sleep testing after STOP-Bang was not cost-effective at standard payment thresholds. But when the model extended over a lifetime, accounting for the long-term cardiovascular consequences of untreated OSA, STOP-Bang followed by a full sleep study became clearly cost-effective, with an incremental cost of about $2,044 per quality-adjusted life year gained.²⁴ The message is that the real value of catching OSA before surgery is not just avoiding a bad week in the hospital — it is identifying a chronic condition early enough to treat it for years.

Screening Commercial Drivers and Workers in Safety-Sensitive Jobs

Drowsy driving is a well-documented risk, and commercial truck and bus drivers with untreated OSA are at elevated risk for accidents. STOP-Bang has been validated in the general population and in commercial driver cohorts, where a score of 3 or higher showed the same pattern of high sensitivity and strong negative predictive value seen in clinical settings.²⁵ However, some research suggests that objective-measurement-based questionnaires may flag more at-risk drivers than STOP-Bang does, raising questions about whether a tool developed in clinical settings is the best fit for occupational screening where the consequences of a missed case include multi-vehicle crashes.²⁶ For now, STOP-Bang remains widely used in fleet health programs, often as a first-pass filter before drivers are sent for formal sleep testing.

What the Questionnaire Cannot Do

STOP-Bang is a screening tool, not a diagnostic instrument. A positive score does not tell you how many times per hour your airway collapses, how much your oxygen drops, or whether your symptoms will respond to a CPAP machine, an oral appliance, or weight loss. Those answers require a sleep study — either an in-lab polysomnography or a home sleep apnea test — that measures actual breathing events during sleep. The questionnaire’s job is to identify who should get that study, not to replace it.

It also cannot distinguish obstructive sleep apnea from central sleep apnea, a less common form in which the brain intermittently fails to send breathing signals. Central sleep apnea is particularly common in heart failure patients, and the STOP-Bang items do not capture the features that separate it from the obstructive type. This is one more reason the tool performs poorly in cardiology populations — it may flag people who do have disordered breathing during sleep, but for the wrong reason.

Self-reported symptoms are another inherent limitation. Snoring and observed apneas depend on having a bed partner who notices, and many people underestimate their own daytime sleepiness. The questionnaire’s four subjective items are only as good as the information the patient provides, which is why the four objective “Bang” items — measurements that do not depend on anyone’s perception — were added to the original STOP version in the first place. If you live alone and sleep through your own snoring, the tool has less signal to work with.