Obstructive sleep apnea (OSA) affects roughly 1 billion people worldwide, according to a report by the American Academy of Sleep Medicine. Severe OSA has been associated with an elevated risk of all-cause and cardiovascular-specific mortality. Studies support an association between OSA and a host of comorbidities, including hypertension, stroke, atrial fibrillation, mood disorders, and neurocognitive outcomes. Undiagnosed and untreated OSA also has major economic and societal costs, reducing workplace productivity and increasing one’s risk of accidents both on the job and while driving.
Positive airway pressure (PAP) is widely considered the most effective treatment for OSA. The majority of patients tolerate CPAP: real-world estimates using international big data show good adherence in over 70% of patients. Robust evidence shows that PAP reduces snoring, decreases daytime sleepiness, and improves quality of life in a dose-dependent manner. Economic analyses have also found CPAP to be cost-effective (Streatfeild, et al. Sleep. 2019;42:zsz181).
But what do we know about the impact of PAP on health outcomes? Perhaps the best studied outcome is cardiovascular disease. Results of observational trials have suggested that CPAP adherence was associated with survival (Pepin JL et al. Chest.). However, it has been speculated that these findings may have been driven, at least in part, by the “healthy user effect.” This phenomenon refers to the tendency for people who engage in one health-promoting behavior (eg, CPAP adherence) to engage in another as well (eg, eating well, exercising, taking prescribed medications). When we observe that patients who use CPAP live longer, we must ask ourselves whether perhaps their better outcomes resulted from healthy habits in general, as opposed to their CPAP usage per se.
Randomization eliminates the potential for the healthy user effect, by assigning patients to a certain intervention as opposed to simply observing whether they choose to use it. And herein lies one of the great disappointments for our field over the past decade: multiple large-scale randomized controlled trials have failed to demonstrate that CPAP reduces cardiovascular mortality, even in patients with pre-existing CAD. The first two of these were the SAVE (Sleep Apnea Cardiovascular Endpoints) (McEvoy R, et al. N Engl J Med.) and RICCADSA (Randomized Intervention with Continuous Positive Airway Pressure in CAD and OSA) (Peker Y, et al. Am J Respir Crit Care Med. ) trials evaluating the effects of PAP on a composite endpoint that included cardiovascular death and nonfatal cardiovascular events. Both trials found no difference between PAP and control groups, leading to a conclusion that PAP did not prevent cardiovascular events in patients with moderate-to-severe OSA and established cardiovascular disease. The ISAAC study (Impact of Sleep Apnea syndrome in the evolution of Acute Coronary syndrome) also failed to show a benefit of CPAP for secondary prevention of cardiovascular events in patients with moderate to severe OSA.
These negative findings were echoed in a recent report by the Agency for Healthcare Research and Quality evaluating a variety of long-term health outcomes in obstructive sleep apnea. The authors stated that “RCTs do not provide evidence that CPAP prescription affects long-term, clinically important outcomes. Specifically, with low strength of evidence, RCTs do not demonstrate that CPAP affects all-cause mortality, various CV outcomes, clinically important changes in psychosocial measures, or other clinical events” (AHRQ, Project ID: SLPT0919, 12/1/2022).
What plausible explanations have been offered for these negative results? Perhaps trials were underpowered. Perhaps patients did not use PAP for a sufficient duration to achieve benefit (usage was under 3 hours in most studies). Perhaps the patients selected for these trials were at such low-risk of adverse outcomes in the first place that treating their OSA didn’t have much impact. Many trials have excluded sleepy patients due to ethical concerns about withholding treatment from this population. But this may have effectively excluded the patients most likely to benefit; in other studies, sleepy patients seem to experience the greatest cardiovascular risk reduction with CPAP. For example, a meta-analysis showed that CPAP is most strongly associated with blood pressure reduction in patients who are sleepy, compared with those with minimally symptomatic OSA (Bratton D, et al. Thorax.). And, recent work suggests that even among non-sleepy patients, it might be possible to identify a subset who could benefit from CPAP. A recent analysis suggested that non-sleepy patients who exhibit a higher change in heart rate following a respiratory event may derive greater cardiovascular benefit from CPAP therapy (Azarbarzin, et al. Am J Respir Crit Care Med. ).
Another, distinct reason for these negative results is that the AHI – our main metric for quantifying OSA severity for several decades – fails to capture the disorder’s heterogeneity. Identifying different phenotypes of OSA may enable more personalized approaches to prognostication as well as treatment. For example, one study identified four symptom clusters of OSA – patients with disturbed sleep, minimally symptomatic, excessively sleepy, and moderately sleepy – who may exhibit different responses to CPAP treatment. Further work is needed to discern whether these clusters reliably predict outcomes in a manner that can be useful clinically (Zinchuk A, et al. Sleep Med Rev.).
So, what is the verdict for CPAP? Sleepy patients with even mild OSA warrant treatment, as is common practice, and these patients are more likely to adhere to therapy. Patients with other symptoms potentially related to untreated OSA should be offered treatment as well. But in asymptomatic patients, it is difficult to make a compelling case to start CPAP on the basis of the AHI alone. It is our hope that novel ways of classifying OSA severity and phenotype will allow better prediction of which patients will experience a protective effect from CPAP. For example, certain subsets of patients may realize greater benefits from CPAP, including those with a high hypoxic burden (Trzepizur W, et al. Am J Respir Crit Care Med.).
For now though, we can allow the evidence that has accumulated in recent years to guide our collaborative decision-making with patients about whether to try CPAP. Depending on how exuberantly we sang CPAP’s praises, we may need to temper our song – at least with regards to cardiovascular risk reduction. In the sleep world, patients are educated not only by sleep providers but also by respiratory therapists who help patients with initial CPAP setups. Consistent, evidence-based messaging by the entire health care team is key. We cannot say that “using CPAP prevents heart attacks” but rather “we’re still not quite sure.”
As in other areas of medicine, sleep medicine may see a shift in focus toward symptoms and patient-oriented outcomes as opposed to the presence of comorbidities. In fact, the recently revised International Classification of Sleep Disorders (ICSD-3-TR) released this year eliminated comorbidity criteria from the definition of Obstructive Sleep Apnea in adults. If adopted by Centers for Medicare & Medicaid Services and other insurers, patients with mild OSA by sleep testing (AHI≥5 but <15) who lack symptoms will no longer qualify for CPAP on the basis of having hypertension, a mood disorder, cognitive dysfunction, coronary artery disease, stroke, congestive heart failure, atrial fibrillation, or type 2 diabetes mellitus. How will this major revision impact the sleep medicine world? Practically speaking, it is likely that fewer patients who present without symptoms and are found to have only mild OSA will end up on PAP.
There will undoubtedly be frustration related to these greater restrictions on who qualifies for PAP. On the other hand, perhaps our energy is better focused on procuring PAP not for asymptomatic patients but rather promoting access and adherence in those who are symptomatic. Differential access to CPAP remains a major problem that very likely contributes to health disparities. In fact, a recent international committee acknowledged that the current CMS criteria for PAP coverage create disproportionate difficulties for non-white patients and those of low socioeconomic background to meet adherence criteria. Their specific recommendations to reduce this disparity in PAP access included eradication of requirements for repeat polysomnography and eliminating the 4-hour rule.
We are moving toward a more personalized approach to characterizing OSA, which eventually may allow for more nuanced, individualized counseling rather than a “one-size -called-CPAP-fits-all” approach. Until we are there, a patient-centered approach that elicits the presence of sleep-related symptoms and daytime impairment, as opposed to isolated comorbidities, provides the most compelling justification for CPAP.