Obstructive sleep apnea (OSA) is a disorder characterized by snoring and repetitive upper airway collapse during sleep that can be complete (apnea) or partial (hypopnea). This leads to repetitive acute episodic oxygen desaturation, fragmentation of sleep, and marked negative intrathoracic pressures that negatively impact on cardiovascular and metabolic health through adverse effects on endothelial function, increased blood pressure and sympathetic nervous system activation, and inflammation.
The importance of OSA in women has received increased recognition in the past two decades and contemporary population-based studies report a prevalence of objectively determined moderate to severe OSA in women of 20–23% [1
]. Population-based studies consistently show objectively determined OSA is twice as common in men compared to women [2
], which may be accounted for by sex differences in airway pathophysiology [3
]. However, women with OSA symptoms are less likely than men to be diagnosed and treated for OSA [7
], and women are under-represented in studies of clinic samples [6
]. Population-based [8
] but not all clinic-based samples [9
] suggest that typical OSA symptoms (snoring and daytime sleepiness) are similar in both genders. However, clinic-based samples have identified a significantly increased propensity for women with OSA to report depression [13
], insomnia, restless legs, and morning headaches, to be older and more obese compared to men [9
], and demonstrate a higher rapid eye movement (REM) sleep apnea hypopnea index [14
]. Women may also have greater impairments in quality of life and healthcare utilisation related to OSA than men [16
Awareness of the importance of OSA in Australia has been high for several decades with the first report of an effective treatment in 1981 by Sullivan et al. [18
] and the relatively open access to sleep studies. Universal health insurance and reimbursement for sleep studies has led to a 240% increase in the provision of government-funded sleep apnea diagnostic services in Australia since 2005. However, the dramatic increase in provision of sleep studies does not guarantee that the right men or women are being screened and identified for treatment. We have previously identified a high prevalence of objectively measured undiagnosed OSA in community-dwelling men aged over 40 years [19
]. We hypothesize that there is also a large burden of women with symptoms of sleep-disordered breathing without a diagnosis of OSA in the community.
The aim of this study was to obtain a contemporary picture of the gender differences in polysomnography-diagnosed OSA and symptoms suggestive of undiagnosed OSA in a population-based study of community-dwelling adults. We sought to determine the characteristics of men and women with diagnosed OSA and possible undiagnosed OSA and their associations with chronic conditions and quality of life in The North West Adelaide Health Study cohort.
In a population-based sample of adults aged at least 30 years, men were over 3 times more likely to report having been diagnosed with OSA than women. This differential is larger than the twofold difference found in population studies where the diagnosis was made in unselected participants with overnight sleep studies, raising the possibility of under-diagnosis in women in our sample. Loud snoring, obesity, and sleepiness were associated with diagnosed OSA in both men and women; however, women had a higher level of obesity. Sex-specific differences in the associations of diagnosed OSA with chronic conditions were seen.
Consistent with recent results in a large Turkish clinical cohort [11
], our findings suggest that a diagnosis of OSA may rely upon presentation features that differ depending on sex. Women who report being diagnosed with OSA are more likely than diagnosed men to have the classical clinical features of the syndrome (obesity, loud snoring, daytime sleepiness) as well as have independent associations with insomnia, respiratory disease, high cholesterol, and poor overall health. In contrast, men may be referred for investigation and diagnosed with OSA as part of CVD secondary prevention. These findings may reflect different elements at work in the investigative and diagnostic process. Prior population-based studies indicate that women with OSA are more likely to suffer insomnia and depression and here our results are consistent [9
]. However, population studies also show in unselected populations that OSA symptoms (snoring, sleepiness) are similar in men and women [8
]. Our findings and those from sleep clinic samples of patients referred for investigation report significantly higher mean BMI in women compared to men [25
]. We can speculate that the cause of the relative under-diagnosis of OSA in women in our sample is that men with symptoms of OSA or with a CVD history are more likely to be referred for investigation for a condition that is regarded as male-predominant. For women to be referred, the clinical picture may need to be overtly related to OSA. Further work is needed to determine the prevalence and clinical characteristics of OSA in women.
It is difficult to make prevalence comparisons with national data as the prevalence of OSA in Australian women in unclear. A 2013 study of 412 community-dwelling Australian women estimated a similar prevalence of moderate to severe OSA (apnea hypopnea index > 15/h) of 6%. However, this study used a single-channel sleep monitor that underestimates OSA severity [27
], may underestimate prevalence, and does not match current standards for assessing OSA [28
]. In the Busselton health study from 1990, of 102 women, 5 (5%) recorded a respiratory disturbance index ≥ 15/h. Given the increasing burden of obesity that has occurred since this study, these OSA estimates are likely underestimated.
Regardless, there is also a health burden in those without a diagnosis of OSA but who report OSA symptoms. In both genders, those with loud snoring and daytime sleepiness frequently reported witnessed apneas and also showed independent associations with depression, insomnia, and fair to poor general health. In men, OSA symptoms were also significantly associated with hypertension and nocturia. The prevalence of symptoms increases the potential OSA population over that diagnosed by 100% in men and 260% in women. Participants identified with OSA symptoms without a diagnosis experience very poor health status as their impairments in physical and mental health quality of life scores approximate those seen in arthritis and heart failure populations [29
]. Our data suggests that sleep should be included when managing health-related lifestyle risks (smoking, exercise, hypertension), and that investigations for sleep apnea also need to be considered. Our findings support those of Lindberg et al., who have recently reported that women with OSA symptoms are less likely than men to be diagnosed and treated for OSA [7
]. We therefore propose that a higher index of clinical suspicion of sleep apnea is required particularly in women who may not be inclined to report snoring and daytime sleepiness but who have other potential sleep symptoms.
A strength of this study is the sample that was derived from a random sample and that weighting of that sample may overcome the non-response in this follow-up assessment of the cohort [24
]. Respiratory disease was based on self-report combined with spirometric measures obtained in a previous wave of data collection. Self-report of chronic conditions is a limitation of this study; however, misclassification of participants as “no condition” results in our findings being biased towards the null. Self-reported height is likely to be overestimated in older adults [30
] leading to underestimates in BMI. Diabetes and hypertension were ascertained through self-report and medication use, which reduces recall bias, and self-reports of macrovascular events have been shown to be valid [31
]. Notwithstanding effects of lifestyle changes, 77% of people self-reporting diabetes and 78% reporting hypertension or their treatment were determined at the previous clinic stage in 2008–2010 to have these disorders. We did not assess excessive daytime sleepiness (EDS) with the Epworth Sleepiness Scale. However, we have previously shown that objectively assessed OSA is not associated with Epworth scores but rather with a more broad description of sleepiness (at least two or three problems) as used in this study [33
Menopause status and hormone replacement therapy (HRT) may influence the relationship of OSA with comorbidities. We did not assess menopause status, which prevents us from directly examining any relationship of menopause and OSA [34
]. However, of note, 85% of women with diagnosed OSA were at least 55 years, and very likely to be post-menopausal, compared to only 41% of women in the OSA symptoms group. This suggests that pre-menopausal women may also be a target group for investigation for OSA in contrast to the commonly held assumption that OSA is a post-menopausal concern. HRT was used in only 5% of women in our study, of whom 60% were aged under 60 years, which is the age group showing a more favorable HRT-related risk profile for CVD and mortality [35
]. Furthermore, given that over 70% of HRT use occurred in women with no diagnosis of OSA/OSA symptoms regardless of age group, HRT is unlikely to have been responsible for residual confounding in our analyses adjusted for age.
In conclusion, women were disproportionately less likely than men to receive a diagnosis of OSA based on the prevalence of possible undiagnosed OSA/symptoms in the population. When women were diagnosed, overt OSA characteristics were present suggesting different diagnostic processes for a condition generally regarded as affecting men. Nevertheless, there is a sizable population of men and women with possible undiagnosed OSA in the community who have major health impacts and warrant investigation. International studies in populations with relatively low prevalence of obesity (mean BMI ~ 25 kg/m2
) suggests that moderate to severe OSA occurs in approximately 20% of women [1
]. There are major challenges to better clinical care and cost-effective policy to address OSA in Australia. This first is to objectively determine the prevalence of OSA in women in Australia where nearly 30% of women are obese [37
]. The second challenge is to correctly identify from within this potentially large community pool of OSA which women are at risk of OSA-related sequelae and therefore require treatment with a possible focus on presentation of insomnia symptoms. Women with symptoms of sleep problems are at risk of accidents, reduced work productivity and social functioning as well as the reductions in quality of life we have identified. A societal focus on identifying and ameliorating the drivers of poor quality sleep is appropriate given the huge direct and indirect costs of sleep problems to the community [38
]. In addition, limited system resources will require novel screening strategies to identify, prevent, and reduce OSA-related disparities in health in Australian women.