Article Text

Download PDFPDF
S27 Predictive performance of STOPBANG questionnaire for diagnosis of sleep apnoea in a cardiac surgical cohort
  1. M Mason,
  2. J Hernández-Sánchez,
  3. D Horton,
  4. A Clutterbuck-James,
  5. I Smith
  1. Papworth Hospital NHS Foundation Trust, Cambridge, UK


Introduction and objectives Questionnaires to assess the risk of obstructive sleep apnoea (OSA) prior to surgery could reduce the need for screening sleep studies. STOPBANG questionnaire is user friendly and was previously validated in a general surgical population. A high risk of OSA has been defined as a score of ≥3 and low risk as a score 0–2. We aimed to validate the STOPBANG against nocturnal oximetry in a population undergoing major cardiac surgery and assessed its prognostic value for postoperative outcomes.

Methods Patients were screened for high risk of OSA with the STOPBANG questionnaire. The presence of sleep apnoea (SA), prior to surgery, was assessed with overnight oximetry. SA was defined as mild with a 4% oxygen desaturation index (ODI) of 5–14/hr, moderate with ODI of 15–29/hr and severe ODI ≥30/hr. Predictive performance of STOPBANG against nocturnal oximetry was assessed for diagnosis of mild and moderate SA by assessing the area under curve receiver operating characteristic (AUC-ROC) and sensitivity and specificity were calculated. A multiple-logistic regression model was used to assess association of STOPBANG and post-operative outcomes.

Results The AUC-ROC for mild SA was low 0.57 (95% CI = 0.47–0.67). Good performance was observed for moderate SA with AUC-ROC 0.82 (95% CI = 0.69–0.95) (Figure 1) but specificity of STOPBANG at the conventional cut of value of ≥3 for moderate SA was very low at 5% whilst sensitivity was 100%. The best predictive STOPBANG cut-off value for moderate SA was ≥6 with sensitivity and specificity of 75% and 77% respectively. Assessing predictive value for severe SA was not possible due to the lack of severe SA cases in our cohort. STOPBANG was not found to be an independent predictor of worse post-operative outcomes.

Abstract S27 Figure 1

ROC curves for STOPBANG to predict ODI ≥5 and ODI ≥15

Conclusion Predictive performance of STOPBANG in our patient cohort at the conventional cut off value was poor. The probable explanation is that the cardiac surgical population is preselected as male, older and most suffer with hypertension. Thus the majority will score as high risk for OSA. STOPBANG had no prognostic value on worse postoperative outcomes in our study, which again contrasts with the findings in general surgical cohorts.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.