Clinical background
Obstructive sleep apnoea (OSA) is common among surgical patients and is associated with increased perioperative morbidity: more difficult intubation, more postoperative pulmonary complications, more intensive care admissions and a longer length of stay. At the same time, most cases are undiagnosed, and polysomnography (PSG) is resource-intensive and rarely feasible before a planned operation. The STOP-Bang score fills this gap: a rapid, structured screening tool that can identify patients with an increased probability of moderate to severe OSA and thereby guide perioperative monitoring, anaesthetic planning and possible referral to a sleep laboratory.
The strength of the score lies in its high sensitivity: a low score is more useful for ruling out OSA than a high score is for confirming it. It is therefore primarily an exclusion tool, not a diagnostic one.
Calculating the STOP-Bang score
STOP-Bang consists of eight binary questions, each worth 0 or 1 point, giving a total score of 0 to 8:
where:
- = Snoring loudly (louder than talking, or audible through closed doors)
- = Tired, fatigued or sleepy during the daytime
- = Observed apnoea (gasping or choking) during sleep
- = Treated for high blood Pressure
- = BMI >35 kg/m²
- = Age >50 years
- = Neck circumference >40 cm
- = Male Gender
Risk bands: 0 to 2 points corresponds to low risk, 3 to 4 to intermediate risk and 5 or more to a high risk of moderate to severe OSA.
The derivation cohort consisted of 746 patients undergoing elective admission for surgery at Toronto Western Hospital and Mount Sinai Hospital in Toronto, Canada [1]. The median age was 60 years, 49 per cent were men, the median BMI was 30 kg/m² and the median neck circumference 39 cm. Patients underwent either laboratory PSG or portable (level 2) PSG at home. The prevalence of OSA was high: 68.4 per cent had OSA (AHI >5), of whom 20.5 per cent moderate (AHI >15 to 30) and 18.0 per cent severe (AHI >30). The area under the ROC curve was 0.65 for any OSA, 0.67 for moderate to severe OSA and 0.71 for severe OSA [1].
Interpretation in practice
| Risk band | STOP-Bang score | Recommended action |
|---|---|---|
| Low risk | 0 to 2 | The probability of moderate to severe OSA is low. Routine perioperative management without specific OSA-related measures. PSG is not needed to rule out OSA. |
| Intermediate risk | 3 to 4 | Uncertain zone. The probability of moderate to severe OSA is about 36 per cent [1]. Clinical assessment of the type of procedure, opioid requirement and comorbidity determines whether the patient should be managed as high risk or referred for PSG. |
| High risk | 5 to 8 | Increased probability of moderate to severe OSA. Consider perioperative precautions: minimised opioid use, continuous pulse oximetry postoperatively, early mobilisation and consideration of CPAP if the patient already has such equipment. For major surgery or high comorbidity, referral to a sleep laboratory before the operation should be considered. |
A score of 5 or more gave, in the derivation cohort, an odds ratio of 4.8 for moderate to severe OSA and 10.4 for severe OSA [1]. At a score of 7 to 8, the odds ratio for severe OSA rose to 14.9. The predicted probability of moderate to severe OSA increased from 36 per cent at a score of 3 to 60 per cent at a score of 7 to 8 [1].
Validation and performance
A systematic review and meta-analysis of 17 studies and 9,206 patients confirmed the performance of STOP-Bang in both sleep clinic and surgical populations [2]. In the surgical population (pooled data from three studies, about 1,000 patients), sensitivity at a cut-off of ≥3 was 91 per cent (95 per cent CI: 87 to 93) for moderate to severe OSA and 96 per cent (95 per cent CI: 92 to 98) for severe OSA. The corresponding negative predictive values were 84 and 97 per cent. Specificity was, however, low: 32 per cent for moderate to severe OSA and 29 per cent for severe OSA [2]. In the sleep clinic population, sensitivity was 94 per cent for moderate to severe OSA with a negative predictive value of 75 per cent, but specificity of only 34 per cent [2].
A later geographical meta-analysis of 47 studies and 26,547 participants in sleep clinic settings showed similar results: sensitivity above 90 per cent at a cut-off of ≥3, a negative predictive value of 77 per cent for moderate to severe OSA and 91 per cent for severe OSA, but specificity of 28 and 24 per cent respectively [3]. Diagnostic accuracy (AUC) was above 0.80 in all geographical regions except East Asia, where the AUC was 0.52 (95 per cent CI: 0.48 to 0.56) [3]. This indicates that STOP-Bang may perform less well in East Asian populations, possibly because of differences in body habitus and craniofacial anatomy that affect the BMI and neck circumference thresholds.
In the derivation cohort the AUC was moderate (0.67 for moderate to severe OSA), which the meta-analyses confirm: the instrument discriminates better for severe OSA than for milder forms, but overall discrimination is not strong [1, 2].
To improve specificity at intermediate scores, alternative scoring models have been proposed [4]. In a study of 516 surgical patients, specificity for moderate to severe OSA rose from 31 per cent (at a score of ≥3) to 85 per cent when the requirement was at least two positive STOP items combined with a BMI >35 kg/m², and to 77 per cent when combined with male sex. For severe OSA, the combination of STOP ≥2 plus BMI >35 plus male sex gave a specificity of 97 per cent [4]. These alternative models are not implemented in the standard version of STOP-Bang, however, and are not used routinely.
Limitations
STOP-Bang is a screening tool, not a diagnostic test. The low specificity at a cut-off of ≥3 means that a large proportion of patients without OSA are classified as positive, particularly in populations of low prevalence. In a general population (the Sleep Heart Health Study, 4,770 participants), specificity was 30 per cent and the positive predictive value only 16 per cent for moderate to severe OSA [2].
A retrospective cohort study of 26,068 patients undergoing major elective non-cardiac surgery questioned the predictive validity of STOP-Bang for perioperative outcomes [5]. Although the proportion with already diagnosed OSA rose across the risk bands (4 per cent low, 12 per cent intermediate, 44 per cent high), STOP-Bang showed only weak correlation with ASA physical status (Spearman ρ = 0.28), the Revised Cardiac Risk Index (ρ = 0.24) and the Charlson Comorbidity Index (ρ = 0.10). After risk adjustment, the STOP-Bang risk bands were not associated with 30-day mortality, cardiac complications or length of stay [5]. This suggests that STOP-Bang measures OSA risk but not necessarily perioperative risk as such.
The instrument does not apply to patients with already diagnosed OSA, in whom management is guided by the existing investigation and treatment. Nor is it validated for pregnant women, children or patients with central sleep apnoea. The thresholds for BMI and neck circumference derive from a North American surgical population and may need to be adapted for other ethnic groups, particularly in East Asian populations where the instrument has performed less well [3].
References
- Chung F, Subramanyam R, Liao P, et al. High STOP-Bang score indicates a high probability of obstructive sleep apnoea. Br J Anaesth. 2012;108(5):768-775. PMID: 22401881
- Nagappa M, Liao P, Wong J, et al. Validation of the STOP-Bang Questionnaire as a Screening Tool for Obstructive Sleep Apnea among Different Populations: A Systematic Review and Meta-Analysis. PLoS One. 2015;10(12):e0143697. PMID: 26658438
- Pivetta B, Chen L, Nagappa M, et al. Use and Performance of the STOP-Bang Questionnaire for Obstructive Sleep Apnea Screening Across Geographic Regions: A Systematic Review and Meta-Analysis. JAMA Netw Open. 2021;4(3):e211009. PMID: 33683333
- Chung F, Yang Y, Brown R, Liao P. Alternative scoring models of STOP-bang questionnaire improve specificity to detect undiagnosed obstructive sleep apnea. J Clin Sleep Med. 2014;10(9):951-958. PMID: 25142767
- Sankar A, Beattie WS, Tait G, Wijeysundera DN. Evaluation of validity of the STOP-BANG questionnaire in major elective noncardiac surgery. Br J Anaesth. 2019;122(2):255-262. PMID: 30686311
