How Accurate Is the STOP-Bang Questionnaire for Screening Sleep Apnea?
This is a systematic review and meta-analysis of 47 studies including 26,547 adults who were referred to sleep clinics and tested with polysomnography, not a single trial and not a general-population screening study. Across those sleep clinic populations, a STOP-Bang score of at least 3 had excellent sensitivity, above 90%, for moderate to severe and for severe obstructive sleep apnea. Diagnostic accuracy for detecting moderate to severe OSA was high, above 0.80, in every geographic region except East Asia, where it was 0.52 (95% CI, 0.48 to 0.56). So the headline is adequate sensitivity and diagnostic accuracy for triage in sleep clinics, with one clear regional exception.
Dr. Kumar’s Take
I read this as a solid endorsement of STOP-Bang for the job it was tested on: sorting sleep clinic referrals by risk before polysomnography. A score of at least 3 caught more than 90% of moderate to severe and severe cases, which is what I want from a first-pass screen.
Two things temper my enthusiasm. First, the East Asian data. Diagnostic accuracy there was 0.52, essentially no better than chance for moderate to severe OSA, while every other region came in above 0.80. That is not a rounding difference, and I would not assume the same cut point travels everywhere. Second, the negative predictive values differ by severity: 91% for severe OSA versus 77% for moderate to severe. A low score is reasonable reassurance against severe disease. It is much weaker reassurance against moderate disease, and I would not use it to close the door on a symptomatic patient.
Also worth remembering who was studied. These were people already referred to sleep clinics, where the prevalence of any OSA was 80%. That is a high pretest probability population, and predictive values do not transfer cleanly to a general primary care panel.
Key Findings
The meta-analysis included 47 studies with 26,547 participants (mean [SD] age, 50 [5] years; mean [SD] body mass index, 32 [3]; 16,780 [65%] men). Studies were grouped into geographic regions: North America, South America, Europe, Middle East, East Asia, and South or Southeast Asia.
Prevalence in these sleep clinic populations was high: 80% (95% CI, 80% to 81%) for all OSA, 58% (95% CI, 58% to 59%) for moderate to severe OSA, and 39% (95% CI, 38% to 39%) for severe OSA.
A STOP-Bang score of at least 3 had sensitivity above 90% and high discriminative power to exclude moderate to severe and severe OSA. Negative predictive values were 77% (95% CI, 75% to 78%) for moderate to severe OSA and 91% (95% CI, 90% to 92%) for severe OSA.
Performance was not uniform across regions. Diagnostic accuracy of a score of at least 3 for moderate to severe OSA was high, above 0.80, in all regions except East Asia, where it was 0.52 (95% CI, 0.48 to 0.56).
Brief Summary
Investigators searched MEDLINE, MEDLINE In-Process, Embase, Emcare Nursing, the Cochrane Central Register of Controlled Trials, the Cochrane Database of Systematic Reviews, PsycINFO, Journals@Ovid, Web of Science, Scopus, and CINAHL from January 2008 to March 2020 for studies that used the STOP-Bang questionnaire alongside polysomnography in adults referred to sleep clinics. Results were pooled overall and by geographic region.
Study Design
Clinical and demographic data were extracted from each article independently by 2 reviewers. Combined test characteristics were calculated using 2 x 2 contingency tables, and random-effects meta-analyses and metaregression with sensitivity analyses were performed. The area under the summary receiver operating characteristic curve was used to compare STOP-Bang accuracy against polysomnography. The PRISMA guideline was followed.
Results You Can Use
A STOP-Bang score of at least 3 identified more than 90% of people with moderate to severe and with severe OSA among sleep clinic referrals, which is what makes it useful as a triage step ahead of a sleep study.
A low score is stronger reassurance against severe OSA than against moderate disease: the negative predictive value was 91% (95% CI, 90% to 92%) for severe OSA and 77% (95% CI, 75% to 78%) for moderate to severe OSA.
Geography matters. Accuracy for moderate to severe OSA exceeded 0.80 in North America, South America, Europe, the Middle East, and South or Southeast Asia, but was 0.52 (95% CI, 0.48 to 0.56) in East Asia.
Why This Matters For Health And Performance
Obstructive sleep apnea is a highly prevalent global health concern associated with many adverse outcomes. A short questionnaire that reliably flags people who need polysomnography helps clinics prioritise the patients most likely to have significant disease.
The authors conclude that STOP-Bang can be used as a screening tool to assist in triaging patients with suspected OSA referred to sleep clinics in different global regions. Earlier identification means earlier treatment, and treatment is where the benefit actually lives.
How to Apply These Findings in Daily Life
- Know the STOP-Bang items: Snoring, Tiredness, Observed apneas, high blood Pressure, Body mass index, Age, Neck circumference, and Gender
- Interpret scores appropriately: A score of at least 3 flags possible OSA and warrants further evaluation
- Seek evaluation for higher scores: If you screen positive, discuss a sleep study referral with your clinician
- Do not ignore symptoms: A low score lowers the odds of severe OSA more than it lowers the odds of moderate disease, so persistent symptoms still deserve evaluation
- Use as screening, not diagnosis: STOP-Bang identifies risk, it does not confirm or exclude sleep apnea
- Ask about context: Accuracy varied by region in this analysis, so local validation matters
Limitations To Keep In Mind
STOP-Bang is a screening tool, not a diagnostic test, and cannot replace polysomnography for a definitive diagnosis. This analysis covers adults referred to sleep clinics, where the prevalence of OSA was 80%, so the predictive values do not automatically apply to lower-risk settings. Diagnostic accuracy for moderate to severe OSA was poor in East Asia (0.52; 95% CI, 0.48 to 0.56), which argues against assuming a single cut point performs identically everywhere. One author reported a patent pending for the STOP-Bang questionnaire, which is proprietary to University Health Network.
Related Studies And Internal Links
- Upper Airway Stimulation for Sleep Apnea Treatment
- CPAP vs Oxygen for Sleep Apnea: Cardiovascular Outcomes
- CPAP Reduces Blood Pressure in Resistant Hypertension
- Tirzepatide for Sleep Apnea and Obesity: Breakthrough Results
- How to Sleep Better: Science Daily Playbook
FAQs
What does each letter in STOP-Bang stand for?
S for Snoring, T for Tiredness, O for Observed apnea, P for high blood Pressure, B for Body mass index, A for Age, N for Neck size, and G for Gender.
What STOP-Bang score indicates I need a sleep study?
In this analysis, a score of at least 3 was the threshold with sensitivity above 90% for moderate to severe and severe OSA. A score at or above that level is worth raising with a clinician to discuss a sleep study.
Can STOP-Bang definitively diagnose or rule out sleep apnea?
No. It is a screening tool. Diagnosis requires polysomnography, which was the reference standard in every study included here.
Conclusion
Across 47 studies and 26,547 sleep clinic patients, a STOP-Bang score of at least 3 showed sensitivity above 90% for moderate to severe and severe obstructive sleep apnea, with negative predictive values of 77% and 91% respectively. Diagnostic accuracy for moderate to severe OSA was high in most regions but only 0.52 in East Asia. The questionnaire is a reasonable triage tool for patients referred to sleep clinics, with that regional caveat attached.
