Epidemiology of obstructive sleep apnoea: diagnosis, prevalence, comorbiditis, and risk factors

© 2019 Baddewithana Chamara Visanka SenaratnaObstructive sleep apnoea (OSA) occurs due to repetitive partial or complete upper airway obstruction during sleep and may lead to generalised hypoxia and/or arousals from sleep. OSA has both direct and indirect health and economic costs. However, design of public health interventions to control OSA is constrained by some critical gaps in knowledge. OSA is widely regarded to be highly prevalent but the reported prevalence estimates vary widely. The last population prevalence study in Australia was performed nearly twenty-five years ago. New estimates on population prevalence and the comorbidities associated with OSA would help better understand the burden of disease. Several OSA-screening questionnaires have been recommended for use in primary care settings in Australia, but these questionnaires have not been validated in the Australian primary-care population. Although OSA is seemingly associated with some chronic disorders such as cardiovascular and cerebrovascular disorders, metabolic disorders, respiratory disorders and psychiatric disorders, evidence for some associations are limited. Furthermore, the currently known modifiable risk factors for OSA are limited and rarely based on longitudinal evidence. No childhood risk factors are known for OSA. Given these knowledge gaps, during my doctoral work I aimed (1) to determine the validity of commonly known OSA-screening-questionnaires, (2) to describe the prevalence of OSA and its co-morbidities in the general population, and (3) to describe some potential early-life and lifetime risk factors for OSA. My specific objectives were (a) to systematically summarise the current evidence on the validity of the common OSA questionnaires for which such systematically synthesised evidence was unavailable, (b) to determine the validity of some common OSA-screening-questionnaires to detect OSA among middle-aged Australian adults, (c) to systematically summarise the available evidence on the prevalence of OSA in the adult general population including in age and sex- specific subgroups, (d) to describe the prevalence and associated chronic co-morbidities of OSA in middle-aged Australian adults, (e) to determine the role of lifetime lung function trajectories on OSA in adult life, and (f) to determine the role of childhood respiratory risk factors on OSA in adult life. To achieve the first aim, I systematically synthesised the literature on the validity of the Berlin questionnaire (Chapter 3). The evidence on the validity of the other questionnaires were already synthesised at the time I commenced this work. The included studies reported varied sensitivity and specificity based on the OSA assessment method used. I then used the data from the Tasmanian Longitudinal Health Study (TAHS) to validate the Berlin (BQ), STOP-Bang and OSA-50 questionnaires against type-4 sleep studies (Chapter 6). To determine if the flow-based AHI or ODI is more suitable as the reference standard, I investigated their agreement in detecting and classifying OSA (Chapter 6 [6.1]). I found that ODI identified more clinically significant OSA than flow-based AHI. Unlike in the previous studies that relied only on conventional AHI/ respiratory disturbance index (RDI)/oxygen desaturation index (ODI) cut-off levels for the reference test, I derived and used a new reference standard, namely, ‘clinically-relevant OSA’ that included moderate-severe OSA (ODI≥15) and mild (ODI≥5) OSA with symptoms as a combined group, as this is the group that needs to be captured for clinical management. I found that all three questionnaires were not clinically useful on their own (sub-optimal sensitivity and specificity) in a simulated primary-care sample but could be used to rule-in OSA (high specificity) when combined with Epworth sleepiness scale (Chapter 6 [6.2]). I also found that OSA-50 and STOP-Bang had a moderate sensitivity and low specificity and BQ had low sensitivity and moderate specificity, at the recommended thresholds, in the general population (Chapter 6 [6.3]). The trade-off between sensitivity and specificity remain a limitation of their practical use. To achieve the second aim, I systematically synthesised the literature on the prevalence of OSA (Chapter 4). I found that the prevalence of OSA varied between population subgroups and the OSA assessment methods. I then used the data from the Ten to Men study to determine the prevalence of doctor diagnosed OSA in Australian men and its co-morbidities (Chapter 7 [7.1]). I found that the prevalence increased from 2% in younger adults to 8% in the middle-age and that OSA was associated with several chronic cardiovascular, metabolic, respiratory and psychiatric disorders. Using data from the TAHS, I found the prevalence of medically-confirmed OSA to be 5% and probable OSA as determined by the STOP-Bang questionnaire to be 15% (Chapter 7 [7.2]). Medically-confirmed OSA as well as probable OSA were associated with several co-morbidities as seen in the Ten to Men study. Furthermore, I investigated how nocturnal-asthma-like symptoms and bronchial hyper-reactivity (BHR) contribute to the association between OSA-risk and current-asthma using the data from TAHS. I found no evidence for any role of BHR in this association but found some evidence for some nocturnal asthma symptoms to be suggestive of undiagnosed OSA. To achieve the third aim, I investigated how OSA is related to lifetime lung function trajectories and childhood respiratory risk factors using longitudinal data from the TAHS. I found that three trajectories were associated with both probable OSA (defined using STOP-Bang questionnaire) and medically-confirmed OSA (Chapter 8 [8.1]). Frequent childhood lower respiratory tract infection, asthma and lung function trajectories were associated with probable OSA (Chapter 8 [8.2]). Overall, my findings have significant public health, clinical and research implications and significantly advanced the current knowledge in this area. My assessment between ODI and flow-based AHI is the first such evidence and suggest that ODI should be preferred in clinical practice. Current OSA screening-questionnaires could be useful in epidemiological research but not as clinical tools. This calls for development of better screening methods. My findings challenge the current recommendations for OSA screening at primary-care and I make suggestions as to how this process could be improved. The high prevalence of OSA in general, and in males and elderly, and its association with other priority chronic disorders provide public health foci for interventions. The demonstrated role of nocturnal respiratory symptoms in the OSA-asthma association suggests the need for clinical vigilance for undiagnosed OSA in patients with nocturnal symptoms. Furthermore, the evidence for association of lung function trajectories and childhood risk factors with OSA advances the current knowledge and provides platforms for future research to delineate the seemingly complex pathophysiology and natural history of OSA