Response to ‘OSA: innocent bystander or associate in crime?’ (Br J Anaesth 2019; 123: e473–e474)

Abstract

Editor—Corso and colleagues1Corso R. Maitan S. Cattano D. Obstructive sleep apnoea: innocent bystander or associate in crime?.Br J Anaesth. 2019; 123 ([Epub ahead of print]): e473-e474https://doi.org/10.1016/j.bja.2019.04.046Abstract Full Text Full Text PDF Scopus (1) Google Scholar raise several important issues that are relevant to our study,2Sankar A. Beattie W.S. Tait G. Wijeysundera D.N. Evaluation of validity of the STOP-BANG questionnaire in major elective noncardiac surgery.Br J Anaesth. 2019; 122: 255-262Abstract Full Text Full Text PDF Scopus (7) Google Scholar and to research in general, on the STOP-BANG questionnaire (SBQ). They are concerned that our study did not evaluate the validity of this well-known preoperative screening questionnaire. Validity describes the extent to which a survey instrument, such as the SBQ, actually measures what it is intended to measure. The SBQ was developed to screen for undiagnosed obstructive sleep apnoea (OSA) in surgical patients. Its validity for this originally intended purpose is well-established.3Chiu H.Y. Chen P.Y. Chuang L.P. et al.Diagnostic accuracy of the Berlin questionnaire, STOP-BANG, STOP, and Epworth sleepiness scale in detecting obstructive sleep apnea: a bivariate meta-analysis.Sleep Med Rev. 2017; 36: 57-70Crossref PubMed Scopus (293) Google Scholar, 4Nagappa 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; 10e0143697Crossref PubMed Scopus (332) Google Scholar Nonetheless, this evidence should not be automatically extrapolated to supporting validity in roles beyond the originally intended purpose, such as a tool for predicting postoperative complications,5Nagappa M. Patra J. Wong J. et al.Association of STOP-Bang questionnaire as a screening tool for sleep apnea and postoperative complications: a systematic review and Bayesian meta-analysis of prospective and retrospective cohort studies.Anesth Analg. 2017; 125: 1301-1308Crossref PubMed Scopus (60) Google Scholar especially as individual questionnaire items (e.g. hypertension, obesity) often co-exist with known prognostically important comorbidities (e.g. coronary artery disease, diabetes mellitus, heart failure). Our study was designed to help address these important issues. Consistent with our study objectives, SBQ scores were found to correlate weakly with measures of prognostically important preoperative comorbidities, but not to independently predict postoperative 30-day mortality, cardiac complications, or hospital length-of-stay. We agree that composite endpoints introduce important methodological problems, especially when their individual components do not share similar severities, frequencies, or treatment responses.6Tomlinson G. Detsky A.S. Composite end points in randomized trials: there is no free lunch.JAMA. 2010; 303: 267-268Crossref PubMed Scopus (93) Google Scholar As with other similar studies in the perioperative setting, we did not use available, albeit still uncommonly used, formal statistical methods to handle composite endpoints.7Armstrong P.W. Westerhout C.M. Composite end points in clinical research: a time for reappraisal.Circulation. 2017; 135: 2299-2307Crossref PubMed Scopus (53) Google Scholar, 8Mascha E.J. Sessler D.I. Statistical grand rounds: design and analysis of studies with binary- event composite endpoints: guidelines for anesthesia research.Anesth Analg. 2011; 112: 1461-1471Crossref PubMed Scopus (55) Google Scholar Nonetheless, we believe that the limitations of composite endpoints are less relevant to our particular study, especially as it included only one composite endpoint (i.e. cardiac complications), and our overall findings were qualitatively similar across all the three main endpoints (i.e. mortality, cardiac complication, hospital length-of-stay), only one of which was a composite endpoint. We concur with Corso and colleagues1Corso R. Maitan S. Cattano D. Obstructive sleep apnoea: innocent bystander or associate in crime?.Br J Anaesth. 2019; 123 ([Epub ahead of print]): e473-e474https://doi.org/10.1016/j.bja.2019.04.046Abstract Full Text Full Text PDF Scopus (1) Google Scholar that undiagnosed pulmonary hypertension and postoperative atrial fibrillation are important potential concerns in patients with OSA, but would also emphasise that our study focused on the prognostic implications of the SBQ, not the prognostic implications of OSA. When stratifying risks for postoperative complications, a high SBQ score is not synonymous with a formal diagnosis of OSA. For example, in the Postoperative Vascular Complications in Unrecognized OSA (POSA) multicentre prospective cohort study of more than 1300 surgical patients without previously diagnosed OSA,9Chan M.T.V. Wang C.Y. Seet E. et al.Association of unrecognized obstructive sleep apnea with postoperative cardiovascular events in patients undergoing major noncardiac surgery.JAMA. 2019; 321: 1788-1798Crossref PubMed Scopus (134) Google Scholar high-risk SBQ scores had only modest diagnostic accuracy in identifying patients with severe OSA (i.e. positive likelihood ratio 2.5 and negative likelihood ratio 0.51).10Jaeschke R. Guyatt G.H. Sackett D.L. Users’ guides to the medical literature. III. How to use an article about a diagnostic test. B. What are the results and will they help me in caring for my patients? The Evidence-Based Medicine Working Group.JAMA. 1994; 271: 703-707Crossref PubMed Scopus (2150) Google Scholar Such misclassification errors might also explain, in part, why severe OSA was associated with postoperative atrial fibrillation in the POSA study, whereas high-risk SBQ scores were not. As we indicate clearly in our paper,2Sankar A. Beattie W.S. Tait G. Wijeysundera D.N. Evaluation of validity of the STOP-BANG questionnaire in major elective noncardiac surgery.Br J Anaesth. 2019; 122: 255-262Abstract Full Text Full Text PDF Scopus (7) Google Scholar missing neck circumference measurement data were an important study limitation, hence explaining why specific sensitivity analyses were conducted to assess its impact. Notably, these sensitivity analyses showed that incomplete neck circumference data did not account for our findings. These incomplete neck circumference measurements were drawn from hospital-based electronic registries used for clinical care, and highlight another potential issue with the SBQ, namely differing prognostic performances when used in the usual clinical care setting vs a research setting. The protocolised approaches typically mandated in the research setting can plausibly result in improved prognostic performance that might not be evident in usual clinical practice. Similar issues have been seen in other screening tools, such as the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU).11van Eijk M.M. van den Boogaard M. van Marum R.J. et al.Routine use of the confusion assessment method for the intensive care unit: a multicenter study.Am J Respir Crit Care Med. 2011; 184: 340-344Crossref PubMed Scopus (196) Google Scholar The difference in performance between research and clinical settings may also explain, in part, the differences in our findings from those of the POSA study, which found SBQ to be associated with postoperative myocardial injury, but not other individual cardiovascular events such as cardiac death, heart failure, or new-onset atrial fibrillation.9Chan M.T.V. Wang C.Y. Seet E. et al.Association of unrecognized obstructive sleep apnea with postoperative cardiovascular events in patients undergoing major noncardiac surgery.JAMA. 2019; 321: 1788-1798Crossref PubMed Scopus (134) Google Scholar Overall, we believe that the SBQ has a clear and established role in screening surgical patients for undiagnosed OSA. The uncertainty pertains to its use—especially in the usual clinical care setting—as a risk prediction tool for identifying patients at increased risk for mortality, cardiac complications, and prolonged hospital stay. Our study re-emphasises the importance of re-evaluating the validity of screening tools when used outside their original indications and setting. The authors declare that they have no conflicts of interest. Obstructive sleep apnoea: innocent bystander or associate in crime?British Journal of AnaesthesiaVol. 123Issue 4PreviewEditor—The article presented by Sankar and colleagues1 on evaluation of the STOP-BANG questionnaire (SBQ) infers that the study is about an assessment of the sensitivity and predictive value of the SBQ. In contrast, the study actually focuses on the perioperative association of obstructive sleep apnoea (OSA)-SBQ classes and postoperative outcomes (except for those of a respiratory nature). While the authors aim to define the statistical construct, the result was used as premise for justifying perioperative associations. Full-Text PDF Open Archive