LAUSR

Many thanks for your response. Indeed, the classification of Obstructive Sleep Apnoea (OSA) in our study deviates away from the standard criteria. However, we aimed to look at how well the Stop-Bang Model and Epworth Sleepiness Scale could distinguish cases of OSA based on the level of intervention as opposed to severity. More importantly, we wanted to see whether the screening tools could highlight patients with OSA who required treatment. Thus, the Apnoea-Hypopnoea Index (AHI) thresholds use to define each group was based on those used to guide treatment of OSA within our hospital. It should be noted that the management of OSA is not only based on the severity, but also the effect that the condition has, and may have, on an individual’s quality of life [1–3]. Consequently, not all patients who receive treatment will have severe OSA. Nonetheless, we agree that that given that there is little evidence with similar categorisation to compare with, further study is necessary to substantiate our findings. Secondly, we agree that collinearity between variables should be considered when performing multinomial logistic regression analyses. However, in our model this was not problematic, as indicated by the calculated variable inflation factors. Furthermore, our models do not include the StopBang Model, age, gender and BMI all at once. There is a non-significant model, which includes the Stop-Bang Model, age and gender, and a significant model which includes BMI, age and gender. Lastly, regarding the ability of the Stop-Bang Model to predict the severity of OSA, we agree that further study including meta analysis using the standard classification is necessary. We must also not overlook the predictive potential of modified versions of the Stop-Bang Model [4] or novel models [5] that consider factors directly related to the underlying mechanism of OSA [3, 6–9], as outlined in the discussion of our study [10].