Sleep apnea, a prevalent chronic disease, is recognized as an independent risk factor for stroke, cardiovascular disease, and all-cause mortality (1, 2). It has been shown to increase the incidence rate of acute ischemic stroke (AIS) and is associated with worse functional outcomes and higher stroke recurrence rates (3, 4). More than half of stroke survivors exhibit sleep apnea during the acute post-stroke phase, with obstructive sleep apnea (OSA) being the most prevalent subtype (5, 6). The pathophysiology linking OSA to stroke remains unclear (3). Given that stroke and OSA are intertwined conditions with shared risk factors and comorbidities (7), OSA poses additional challenges in the acute-phase management of stroke patients. Continuous positive airway pressure (CPAP) is the gold standard for OSA treatment, targeting both anatomical and non-anatomical mechanisms (8). Some studies suggest that OSA therapy is feasible and may improve long-term outcomes in stroke patients (9–12), while others indicate that CPAP treatment could also aid in stroke prevention (13, 14). However, due to limited evidence in the acute phase, there are no clear clinical guidelines for this patient population.

Understanding the epidemiology and outcomes of patients with acute ischemic stroke and comorbid obstructive sleep apnea (AIS-OSA) is crucial for identifying subgroups that may benefit from CPAP and for recognizing unmet needs within this patient population.

This study analyzed the prevalence, comorbidities, complications, and outcomes of hospitalized patients with AIS-OSA using data from the National Inpatient Sample (NIS). Furthermore, the outcomes of AIS-OSA patients who received CPAP during hospitalization were examined.

A large international case–control study demonstrated a statistically significant association between sleep impairments and the risk of acute stroke (17, 18). OSA is a well-established risk factor for AIS (18). A meta-analysis revealed that OSA is present in up to 70% of stroke and transient ischemic attack patients (19). Despite its high prevalence, only a small proportion of stroke patients undergo OSA testing and receive treatment (20). According to a cross-sectional study using NIS data, the documented incidence of OSA in acute cerebral infarction patients was 3.1% (21). In our retrospective analysis, we identified 103,004 hospitalized AIS-OSA patients between 2010 and 2019, of whom 6,476 (6.3%) received CPAP treatment (Figure 2). Among AIS-OSA patients who received CPAP treatment, most of patients (75.4%) were treated for less than 24 h (Figure 3). However, these aggregate treatment durations do not reflect patients’ adherence to CPAP and are therefore uninformative for assessing therapeutic efficacy (22). As administrative codes do not capture why CPAP was withheld, its absence in the non-CPAP group could reflect patient intolerance, refusal, contraindications, or unavailability of equipment during hospitalization, but these reasons cannot be ascertained from the NIS.

While a correlation exists between OSA and stroke, it remains unclear whether varying demographic characteristics or clinical factors influence outcomes in patients with AIS-OSA (23). CPAP is the gold standard for OSA treatment, though its efficacy in AIS-OSA patients remains uncertain (24, 25). Understanding the risk factors affecting CPAP treatment in AIS-OSA patients is crucial for effective disease management.

In this study, patients with AIS-OSA who received CPAP exhibited higher in-hospital mortality, longer LOS, greater TOTCHG, and a different distribution of discharge dispositions (Table 2). These findings were consistent with reports from other national hospital discharge databases (26–29). We also found that AIS-OSA patients treated with CPAP had more than three comorbidities (Table 2). These comorbidities (e.g., sepsis, heart failure, pneumonia) were pre-existing conditions present before the initiation of CPAP treatment, reflecting a greater baseline burden of illness. Compared to non-CPAP-treated patients, those receiving CPAP had higher in-hospital mortality, longer LOS, and greater TOTCHG. However, among surviving CPAP-treated patients, a statistically significantly higher proportion were discharged routinely (Table 2), suggesting that CPAP may be associated with better prognoses and improved neurological recovery at discharge in this subgroup. This paradox also may reflect selection and survivorship biases: patients able to tolerate or be offered CPAP may have been less neurologically impaired at baseline (we lacked NIHSS or GCS data), whereas the most severe cases, who could not tolerate or were never selected for CPAP, died earlier and were thus excluded from the denominator of survivors eligible for discharge. OSA can induce hypercapnia, which paradoxically reduces blood flow to ischemic brain regions while increasing perfusion to unaffected cerebral vessels (30). In AIS patients, inadequate perfusion of the ischemic penumbra leads to neurological damage and functional decline (31, 32). CPAP treatment may help prolong the survival time of the ischemic penumbra, promote neuronal functional recovery, and reduce the risk of neurological deterioration. Furthermore, CPAP improves sleep quality and oxygen saturation by providing pressure support to maintain upper airway patency during sleep (33). It also exerts multifaceted effects on stroke pathophysiology, including modulation of neuronal apoptosis, mitochondrial bioenergetics, oxidative stress, angiogenesis, glucose metabolism, and blood–brain barrier regulation (34, 35).

Our investigation revealed older age (≥ 65 years) and male sex as independent risk factors for CPAP treatment (Figures 4A,B). While OSA can manifest across all ages, its prevalence increases with age, tending to plateau after approximately 65 years (36). A significant predominance of OSA in men has been observed (37), with hormonal influences on breathing control and upper airway muscle activation during sleep, as well as sex-specific fat distribution, appearing to contribute to this disparity (38, 39). The analysis indicated a higher incidence of CPAP treatment among Black, Hispanic, and Asian or Pacific Islander individuals (Figures 4C,D). The African American race is associated with untreated OSA and serves as an independent predictor for stroke and elevated all-cause mortality, even after adjusting for cardiovascular risk factors, compared to White individuals (40). The prevalence of diagnosed OSA is notably elevated among racial/ethnic minorities, and rates of undiagnosed OSA are particularly high within these groups (41). Factors such as a higher likelihood of residing in economically deprived neighborhoods, statistically significantly greater rates of obesity, and increased prevalence of hypertension (42) contribute to the severity of OSA and influence CPAP usage, consistent with other studies. Glaucylara et al. (43) reported increased levels of cardiovascular risk factors associated with OSA, including elevated neutrophil counts, an inflammatory marker, which was strongly associated with men, younger individuals, and African American individuals. One study suggested that OSA patients exhibiting phenotypes associated with age, gender, BMI, and sleep apnea severity demonstrate slightly higher CPAP usage (44).

Our study revealed a higher likelihood of receiving CPAP treatment in northeastern hospitals (Figures 4E,F), a finding consistent with previous literature. Levi Dunietz et al. reported statistically significant state-level and regional disparities in both CPAP treatment and adherence among Medicare beneficiaries with OSA suggest gaps in delivery of OSA care for Americans (45). These gaps may stem from demographic differences, including racial variation, as well as from socioeconomic status, urbanicity, and access to accredited sleep centers (45). Patients receiving CPAP were more likely to be Medicaid beneficiaries (Figures 4G,H), possibly due to the higher healthcare costs associated with CPAP-treated patients (28, 29) and their increased likelihood of belonging to racial/ethnic minority groups residing in economically deprived regions (41, 42). CPAP treatment was also more prevalent in large hospitals compared to small ones (Figures 4I,J), potentially attributable to the complexity of care and the higher patient volume in larger institutions (46).

In this study, acquired immune deficiency syndrome, deficiency anemia, congestive heart failure, chronic pulmonary disease, coagulopathy, fluid and electrolyte disorders, other neurological disorders, paralysis, psychoses, pulmonary circulation disorders, obesity, alcohol abuse, drug abuse, and weight loss were associated with CPAP treatment; however, these associations do not imply causation (Figure 5). Given the interconnected nature of stroke and sleep apnea, with numerous shared risk factors and comorbidities, it remains uncertain at an individual level whether sleep apnea is a potential cause or a consequence of stroke (3, 6). Obesity, pulmonary circulation disorders, hypertension, diabetes, and other comorbidities are consistently reported in patients with stroke diagnosed with obstructive sleep apnea (5, 47). All these factors influence the severity of the disease and, consequently, CPAP treatment. In cases of severe OSA, treatment is essential, and CPAP is the recommended first-line therapy (38).

Patient-related factors associated with CPAP treatment in our study included sepsis, pneumonia, acute myocardial infarction, deep vein thrombosis, pulmonary embolism, cardiac arrest, acute kidney injury, urinary tract infection, intracranial hemorrhage, thrombolysis, and mechanical thrombectomy (Figure 6). Prior research has indicated that prolonged mechanical ventilation is associated with a high incidence of thrombosis in critically ill patients despite prophylaxis (48). While a prospective randomized clinical trial (49) found that CPAP reduced early pulmonary infection rates and endotracheal intubation rates, our study demonstrated that CPAP treatment was associated with an increased incidence of lung infection. This discrepancy may be attributable to the possibility that CPAP-treated patients in our cross-sectional study presented with more complications, more severe underlying conditions, and a higher predisposition to developing lung infections. Our findings suggest that AIS patients receiving CPAP treatment were more likely to undergo mechanical thrombectomy or thrombolysis and experience acute myocardial infarction or cardiac arrest, potentially due to OSA’s capacity to induce pathological cardiac and cerebrovascular effects (50), and the potential for ischemic stroke lesions in critical areas regulating airway patency or breathing effort to exacerbate sleep apnea (3).

The relationship observed between CPAP treatment and complications is associative rather than causal. Establishing a definitive causal link between CPAP and negative outcomes is inherently challenging due to the methodological limitations of observational and database studies. Patients receiving CPAP treatment frequently present with a higher burden of comorbidities or postoperative complications, which may represent the true underlying causes of adverse outcomes, rather than the CPAP treatment itself. While efforts are made to adjust for these confounding factors, a direct causal relationship cannot be definitively established. Notably, a meta-analysis of randomized trials indicated that early CPAP treatment demonstrated no effect, whereas CPAP treatment initiated beyond 7 days showed beneficial effects (11). Given that conventional CPAP treatment may not invariably be beneficial and could potentially have adverse effects in specific patient populations, it is exceedingly important to consider a personalized approach to OSA treatment in AIS patients.

Further research is warranted to delineate the clinical characteristics of patients who would and would not derive benefit from CPAP treatment. Moreover, large-scale prospective trials are necessary to rigorously evaluate the risks and benefits of routine versus more selective CPAP application among patients with OSA in the setting of hospitalized AIS in general. Should these findings be corroborated in randomized prospective studies, a more selective treatment paradigm will be essential to improving mortality rates, reducing length of hospital stay, decreasing complications, and lowering associated healthcare expenditures.

While our findings offer valuable insights, several limitations inherent in our analysis warrant consideration. Firstly, our identification of OSA-AIS patients between 2010 and 2019 relied on ICD-9-CM/ICD-10-CM codes. This approach lacked comprehensive data regarding disease severity, as well as scores from the National Institutes of Health Stroke Scale (NIHSS), Glasgow Coma Scale (GCS), and Modified Rankin Scale (MRS). Without these metrics we were unable to adjust for baseline neurological impairment, which is a strong predictor of mortality and functional outcome. To partially address this limitation, we included the comorbidity “other neurological disorders” as a proxy indicator of pre-existing neurological disease burden; nevertheless, this surrogate cannot fully capture acute stroke severity, and residual confounding may remain. Furthermore, because the database contains no information on the timing of CPAP initiation (early versus late), actual adherence during hospitalization, or any pre-admission CPAP treatment, we cannot separate the true effect of CPAP from these unmeasured factors (10, 51), any observed associations may therefore reflect timing, adherence, or prior exposure rather than the treatment itself. Secondly, our analysis is further constrained by the absence of device-specific information in the NIS. Knowing whether patients received fixed-level or auto-titrating CPAP (the actual pressure settings) is essential for interpreting any outcome differences (52). Moreover, the beneficial effects of CPAP are highly dependent on adequate adherence: prior work has shown that nightly use of ≥4 h was considered adherent (53). Because the NIS does not record mask-on time or device-download data, we cannot determine whether the observed associations were diluted by sub-therapeutic or intermittent in-hospital use. Thirdly, the inherent challenges associated with incomplete or inaccurate database records and the absence of randomization, characteristic of our study’s retrospective design, introduce a higher risk of unidentifiable confounding factors that could influence the observed results. Fourthly, our study data were derived from the NIS, a United States in-hospital database, which may introduce selection bias due to the exclusion of data from outpatient, nursing home, and global settings. While our larger sample size helps to mitigate the impact of these errors, these limitations should be acknowledged. Despite these constraints, the consistency of our results with previous studies lends further credence to our findings.

Our findings indicate that hospitalized AIS-OSA patients treated with CPAP experienced increased in-hospital mortality, LOS, and TOTCHG. However, among the surviving CPAP-treated patients, a statistically significantly higher proportion experienced routine discharge. This observation may suggest that surviving patients who received CPAP treatment had more favorable prognoses. We also identified independent predictors and associated factors for CPAP treatment in hospitalized AIS-OSA patients. Randomized controlled trials specifically focusing on AIS-OSA patients within various high-risk groups are necessary to refine the current consensus regarding the initiation of CPAP in the acute setting.