Introduction
Obstructive sleep apnea (OSA) is an important comorbidity that is seen in up to 70% of patients who have sustained a stroke (either ischemic or hemorrhagic stroke) or transient ischemic attack. Reference Seiler, Camilo and Korostovtseva1 Despite the fact that untreated post-stroke OSA, which is defined as OSA that is diagnosed after one’s stroke, is linked to increased length of hospitalization, recurrent vascular events, poorer functional recovery, impaired cognition, and decreased mood, Reference Boulos, Dharmakulaseelan, Brown and Swartz2,Reference Swartz, Bayley and Lanctot3 there is a paucity of literature that explores sex differences in this patient population. One study found that male sex was a predictor of post-stroke sleep disordered breathing, Reference Chen and Chen4 and another study found that males with history of stroke had a greater severity of sleep disordered breathing compared to females with post-stroke sleep disordered breathing. Reference McDermott, Brown and Li5 Otherwise, sex differences in post-stroke OSA remain underexplored.
Outside of the stroke literature, it has been demonstrated that females have less severe OSA compared to males. Reference Jordan, McSharry and Malhotra6 In addition, females with OSA are less likely to experience classic OSA symptoms such as snoring and excessive daytime sleepiness compared to males but are more likely to report depression, difficulty falling asleep, and headache. Reference Wimms, Woehrle, Ketheeswaran, Ramanan and Armitstead7,Reference Nigro, Dibur and Borsini8 Previous studies have suggested that differences in upper airway anatomy, distribution of adiposity, control of ventilation, and hormonal status may account for the higher risk of OSA in males in the general population. Reference Ye, Pien and Weaver9 However, patients of both sexes with post-stroke OSA tend to differ in clinical presentation compared to patients with OSA without stroke. For example, stroke patients with OSA tend to have lower body mass indices (BMIs) and do not necessarily snore or experience significant daytime sleepiness compared to those with OSA but no stroke. Reference Bassetti and Aldrich10,Reference Arzt, Young and Peppard11
There is a paucity of literature that explores sex differences in OSA after stroke. Accordingly, the purpose of this study was to explore sex differences in this patient population. Our primary objective was to examine sex differences in functional outcomes in patients with post-stroke OSA. We secondarily assessed sex differences in stroke severity, OSA severity, cognition, and clinical manifestations of post-stroke OSA.
Methods
Standard Protocol Approvals, Registrations, and Patient Consents
We retrospectively evaluated data from three previously conducted studies, Reference Boulos, Kamra and Colelli12–Reference Boulos, Murray and Muir14 all of which were approved by the local research ethics board. Written informed consent was obtained from all patients.
Study Participants and Variables
Inclusion criteria for this study were: (1) stroke confirmed by a stroke physician on neuroimaging, and (2) a diagnostic in-laboratory polysomnography (PSG) or home sleep apnea test (HSAT) within 1 year of stroke demonstrating an apnea–hypopnea index (AHI) ≥ 5. All participants from the three previously conducted studies underwent either level I, technologist-monitored in-laboratory PSG (Compumedics Neuroscan, Australia; scored according to the 2007 American Academy of Sleep Medicine criteria), or via HSAT using the ApneaLink Air, which is a level III portable sleep monitor. As per the 2007 American Academy of Sleep Medicine, recordings were manually scored by a sleep physician. Reference Berry, Brooks and Gamaldo15 We excluded patients who declined undergoing either PSG or HSAT, or those with incomplete sleep data (defined as HSAT that captured less than 4 hours of data).
Sex, age, BMI, neck circumference, and presence of vascular risk factors were obtained. Post-stroke functional outcome was measured using the modified Rankin Scale (mRS). Reference Bonita and Beaglehole16 Stroke severity was measured using the National Institutes of Health Stroke Scale score (NIHSS). Reference Brott, Adams and Olinger17 OSA severity, as assessed by the AHI, was obtained from either in-laboratory PSG or HSAT. The following questionnaires were administered to understand clinical manifestations of OSA: the Epworth Sleepiness Scale, Reference Johns18 which assesses daytime sleepiness; the STOP-Bang questionnaire, which assesses the presence of OSA symptoms such as snoring, daytime fatigue, and observed apneas; Reference Chung, Abdullah and Liao19 the Center for Epidemiologic Studies Depression Scale (CESD), Reference Parikh, Eden, Price and Robinson20 which measures self-reported symptoms associated with depression; and the Montreal Cognitive Assessment (MoCA), Reference Nasreddine, Phillips and Bedirian21 which is a cognitive screening tool.
Statistical Analysis
Descriptive variables were reported for the total sample and the male and female groups. Categorical variables were reported as frequency counts, and male and female groups were compared using chi-square tests. Normally distributed continuous variables were reported as means and standard deviations, and male and female groups were compared using independent sample t tests. Non-normally distributed continuous variables were reported as medians and interquartile ranges (IQRs), and male and female groups were compared using the Mann–Whitney U test.
Linear regression models were constructed to examine our primary hypothesis that sex would be independently associated with post-stroke functional status, as assessed by the mRS score. The linear regression model included the following covariates: sex, age, time between stroke and sleep study, AHI, and stroke severity. We also constructed linear regression models to evaluate our secondary hypothesis that sex would be independently associated with stroke severity (as measured by the NIHSS at the time of stroke). The covariates selected for this model were sex, age, AHI, and time between stroke and sleep study. We further constructed linear regression models to evaluate OSA severity (as assessed by the AHI derived from PSG or HSAT), cognition (as assessed by the MoCA), and OSA-related clinical features (e.g. daytime sleepiness [assessed by the ESS], depressive symptoms [assessed by the CESD], and items on the STOP-Bang questionnaire such as snoring, daytime fatigue, and observed apneas). These linear regression models used sex, age, time between stroke and sleep study, and stroke severity as covariates. Prior to modeling, all variables were assessed for multicollinearity (tolerance statistic value<0.4); if multicollinearity was found, only one variable of a correlated set was retained in the model. The final model was assessed for any potential violations to linear regression modeling using residual plots.
Statistics analyses were conducted using P.A.S.W Statistics 25.0 (SPSS Inc., Chicago, IL). Statistical significance was set at p < 0.05. As our secondary objectives were exploratory in nature, they were not corrected for multiple comparisons.
Data Availability
Anonymized data may be available through a data transfer agreement in discussion with the corresponding author.
Results
Characteristics of the Study Population
A total of 171 participants were included in this study (Fig. 1). Of the total study population, 117 were male (68.4%) and 54 were female (31.6%). Table 1 summarizes demographics, comorbidities, OSA severity, and mRS scores of participants. Sleep study was conducted at a mean of 30 ± 50 days (median 5 days, IQR 31 days) post-stroke. Fifty-nine of the 171 enrolled participants completed PSG, and the remainder completed HSAT. Of the 59 participants who completed PSG, 15 (25.4%) were females.
Figure 1: Participant breakdown.
Table 1: Characteristics of study population
AHI = apnea–hypopnea index; IQR = interquartile range; NIH = National Institutes of Health; OSA = obstructive sleep apnea; STOP-Bang = snoring, tired, observed apneas, pressure, BMI, age, neck circumference, gender.
Linear Regression Analyses
For our primary objective, our linear regression model demonstrated that female sex (β = 0.37, p = 0.03), greater stroke severity (β = 0.29, p < 0.01), and increased time between stroke and sleep study (β = 0.003, p = 0.03) were significant independent predictors for a greater post-stroke mRS score (Table 2). When we used a linear regression model for mRS score using an interaction term between stroke severity and time between stroke and sleep study, we found that the time between stroke and sleep study did not influence the relationship between stroke and post-stroke mRS (β = 0.009, p = 0.92). Female sex (β = 0.13, p = 0.038) and stroke severity (β = 0.62, p < 0.001) were significant independent predictors of mRS in this model.
Table 2: Linear regression multivariable models examining the relationship of sex with outcomes post-stroke in patients with OSA
OSA = obstructive sleep apnea; NIH = National Institutes of Health.
Female sex (β = 1.01, p = 0.04) was the only identified independent predictor for greater stroke severity as measured by the NIHSS. Female sex (β = 3.73, p = 0.04), younger age (β = –0.15, p = 0.01), increased stroke severity (β = 0.60, p = 0.04), and increased time between stroke and sleep study (β = 0.04, p = 0.03) were significant predictors of increased depressive symptoms, as measured by the CESD. Increased stroke severity (β = –0.69, p < 0.01) was a significant independent predictor of greater cognitive impairment, as measured by the MoCA. AHI, a marker of OSA severity, was not a significant predictor of neurological outcome, as measured by the mRS and NIHSS, depressive symptoms (CESD), cognition (MoCA), or daytime sleepiness (ESS) in our study population of stroke patients. There was no interaction between AHI and sex.
Table 3 includes linear regression model results for sleep variables. Male sex was the only significant independent predictor of increased OSA severity, as measured by the AHI (β = 5.93, p = 0.03). Sex was not a significant independent predictor for daytime sleepiness, as measured by the Epworth Sleepiness Scale (β = –1.03, p = 0.22). Finally, sex was not an independent predictor for individual components of the STOP-Bang questionnaire such as snoring, tiredness, and observed apneas (Table 4).
Table 3: Linear regression multivariable models examining the relationship of sex with sleep variables in post-stroke OSA
OSA = obstructive sleep apnea; NIH = National Institutes of Health.
Table 4: Linear regression multivariable models examining the relationship of sex with items from the STOP-Bang questionnaire in post-stroke OSA
NIH = National Institutes of Health; OSA = obstructive sleep apnea; STOP-Bang = snoring, tired, observed apneas, pressure, BMI, age, neck circumference, gender.
Discussion
In summary, our study found that females with OSA diagnosed after stroke had significantly greater functional impairment and stroke severity compared to males despite having less severe OSA. Females with post-stroke OSA also exhibited more depressive symptoms compared to males. There were no significant sex differences in presenting symptoms of OSA such as daytime sleepiness, as well as on individual items of the STOP-Bang questionnaire (i.e. snoring, tiredness, and observed apneas).
The literature suggests that stroke patients who have a diagnosis of OSA, regardless of their OSA severity, tend to have poorer functional recovery compared to stroke patients without OSA. Reference Menon, Sukumaran, Varma and Radhakrishnan22 Our study builds upon the current literature by looking at sex differences in post-stroke OSA. We demonstrate that females with post-stroke OSA show poorer functional recovery compared to males with post-stroke OSA, despite males having more severe OSA compared to females. This relationship was found even after controlling for age, OSA severity, stroke severity, and timing of sleep test post-stroke. Besides female sex, other factors that were associated with increased functional impairment post-stroke included increased time of sleep study from stroke and increased stroke severity. Time between stroke and sleep study was evaluated in this study, given that literature suggests that OSA severity tends to decrease with more time from stroke. Reference Liu, Lam, Chan, Chan, Ip and Lau23 However, timing of the sleep study was not found to be a significant contributor of sleep apnea severity in this population. This could be because participants were included if their sleep study was within 1 year of stroke. Interestingly, there was also no significant sex group differences in age, BMI, and presence of vascular risk factors that could potentially explain why females with post-stroke OSA had milder OSA but more functional impairment than males. This is unique from the general population, where differences have been documented in age, vascular risk factors, and BMI between males and females with OSA. Reference Bonsignore, Saaresranta and Riha24 The lack of difference in a stroke population could suggest a unique relationship between stroke and OSA. This is also consistent with previous literature which suggested that traditional risk factors for OSA in a general population are not necessarily the same in stroke patients. Reference McDermott, Brown and Li5
The current literature also suggests that OSA is often misdiagnosed in females, and there is low prevalence of OSA among females compared to males. Reference Redline, Kump, Tishler, Browner and Ferrette25 This may, in part, be due to a proportionate lack of sleep study referrals for females. Reference Auer, Frauscher, Hochleitner and Hogl26 OSA has generally been depicted as a disease of obese males. However, in the stroke population, it is well accepted that OSA manifests atypically. Reference Arzt, Young and Peppard11 In our study, females with post-stroke OSA exhibited significantly greater depressive symptoms, and there was a trend toward females being more cognitively impaired compared to males. There is a risk that treating physicians may consider other possibilities such as post-stroke depression or cognitive impairment, rather than OSA, for nonspecific symptoms such as fatigue, and thereby potentially underdiagnose and undertreat women with OSA. Furthermore, previous studies in a non-stroke population have found that females are more likely to report symptoms of insomnia, morning headaches, and mood swings, which could result in underdiagnosis of OSA. Reference Morris, Mazzotti, Gottlieb and Hall27,Reference Votteler, Knaack, Janicki, Fink and Burghaus28 Insomnia symptoms may be endorsed, since the airway is most vulnerable to collapse during sleep–wake transition and can present as more awakenings at this time. Reference Bonsignore, Saaresranta and Riha24,Reference Morris, Mazzotti, Gottlieb and Hall27 Therefore, this warrants more research in clinical tools to predict OSA in stroke patients, since typical symptoms of OSA such as those that are used on the STOP-Bang may not fully capture symptoms experienced by female stroke patients.
This study had a few limitations. One of the limitations is potentially the sample size. As previously noted, even though the HSAT used in this study has been validated against PSG, PSG remains the gold standard. Reference Dharmakulaseelan, Chan-Smyth, Black, Swartz, Murray and Boulos29 Also, we did not include participants with significant physical impairment, since they were generally unable to complete the study requirements. Therefore, our results may not be generalizable to patients with more severe strokes. Another limitation of this study was the lack of a control group with participants without OSA. Finally, a control group of patients who had not sustained a cerebrovascular event was not available for comparison. It would be difficult to make conclusions in a non-stroke population since our primary outcome, which was functional outcome measured by the mRS, and stroke severity, are not validated in non-stroke patients.
Overall, this study found that there were significant differences between males and females with post-stroke OSA in terms of functional outcomes, neurological deficits, and OSA severity. We postulate that this may be due to structural, hormonal, or other physiological differences between males and females, which may contribute to or be an explanation for why OSA manifests differently between the sexes. Current measures for sleep disordered breathing and respiratory physiology may also explain the sex differences we found in OSA severity. For example, females may have more frequent upper airway resistance and be less likely to have characteristic oxygen desaturation events, which are currently key to the diagnosis of sleep disordered breathing. Reference Lin, Davidson and Ancoli-Israel30 Hormonal differences may also drive such anatomical and physiological differences between males and females. It has been postulated that in a general population, incidence of OSA may also be associated with decreased sex hormones in postmenopausal females and increased prevalence of hypothyroidism in females; however, more research is warranted in a stroke population. Reference Bonsignore, Saaresranta and Riha24,Reference Redline, Kump, Tishler, Browner and Ferrette25 Given that majority of females enrolled in this study with post-stroke OSA were of postmenopausal age range, this could suggest that closer screening for OSA in postmenopausal women with stroke may be beneficial in limiting underdiagnosis in female stroke patients.
Furthermore, given these known anatomical and physiological sex differences, there may need to be adjustments to the current diagnostic criteria used in OSA. For example, relies greatly on pulse oximetry, and since females are less likely to have large oxygen desaturations compared to males, the HSAT may underestimate overall AHI in females. Currently, OSA is typically diagnosed if an individual has an AHI greater than or equal to 5, and this is the same criteria used among both sexes. However, published normative PSG parameters in healthy adults have incorporated sex-related differences, such as an increase in the AHI by 0.2 for every 10% increase in the percentage of male participants. Reference Boulos, Jairam, Kendzerska, Im, Mekhael and Murray31 These reference values suggest that clinicians should consider using different AHI cutoffs for females versus males when diagnosing OSA. The importance of sex differences when developing diagnostic criteria and cutoffs can be seen with acute coronary syndrome (ACS). Just like in OSA, women experiencing an ACS tend to present atypically compared to men. Based on the current literature, there is clinical value for the use of sex-specific cutoffs for the troponin, in that females should have a lower troponin cutoff compared to males due to physiological differences that result in females having a lower baseline troponin than males. Reference Bhatia and Daniels32 This same practice could be implemented when diagnosing OSA. More work is required in this area both in the general population and in the post-stroke OSA setting specifically.
Conclusion
In a stroke population, females tend to have less severe OSA but greater neurological deficits and poorer functional outcomes compared to males. Understanding sex differences in patients with post-stroke OSA will likely facilitate better recognition of OSA and potentially improve clinical outcomes.
Acknowledgments
None.
Funding
Heart and Stroke Foundation of Canada, Canadian Institutes of Health Research, Canadian Stroke Network, Innovation Fund of the Alternative Funding Plan of the Academic Health Science Centres of Ontario, and Branch Out Neurological Foundation.
Competing interests
ResMed provided the home sleep apnea tests used in this study as in-kind support to Dr Boulos’ research program. ResMed was not involved in the design of this study. Outside of the submitted work, Dr Mark Boulos has received personal compensation for serving on a scientific advisory committee for Paladin Labs, as well as speaker fees from Jazz Pharmaceuticals and Paladin Labs. Outside of the submitted work, Dr Mark Boulos has received in-kind support for his research program from Braebon Medical Corporation and Interaxon. Dr Mark Boulos has received grant funding for his research program from the Ontario Genomics. Dr Mark Boulos’ research program has also received support from the Mahaffy Family Research Fund. The other authors have no disclosures to report.
Statement of authorship
LD: data aquisition, data analysis, and manuscript writing; SB: conception and design of work, manuscript review, and critical analysis; RS: conception and design of work, manuscript review, and critical analysis; BJM: conception and design of work, manuscript review, and critical analysis; MIB: conception and design of work, data analysis, interpretation of data, and manuscript writing.
Open access
