The search for realistic evidence on the outcomes of obstructive sleep apnoea

The burden and societal impact of obstructive sleep apnoea (OSA) needs to be more widely understood to guide health policies and improve value-based care. OSA affects one billion people worldwide, representing a major and still under-recognised health problem. Its prevalence is expected to continue to increase, owing to the obesity epidemic and the increase in life expectancy [1]. OSA has been associated with numerous long-term consequences, including cardiovascular and metabolic disorders, and neuropsychiatric diseases [2–4]. Also, OSA impairs quality of life, cognitive function, and productivity in the workplace, and causes road traffic accidents, resulting in injuries and fatalities [5].

The economic burden of OSA is similar to, or greater than, other major chronic diseases: diagnosing and treating appropriately every OSA patient in the USA would allow for saving of USD 100.1 billion annually [6]. To face the challenge of speeding-up equitable access to diagnosis, breakthrough technological advances are developing quickly in the field [7]. While the SARS-CoV-2 pandemic has limited the availability of sleep medical services for several months worldwide, it has strongly advanced various remote and virtual sleep laboratory services, with safe, home-based diagnosis and individualised therapeutic approaches [8].

Continuous positive airway pressure therapy (CPAP) is the first line treatment for symptomatic OSA. Poor CPAP adherence, usually defined as a cut-off of less than 4 h and/or less than 70% of nights, in minimally symptomatic patients remains a challenge for clinicians. However, the objective adherence exceeds use of inhalers in asthma or COPD and medication use in hypertension and diabetes when objectively assessed [9, 10]. In addition, the rapid development of digital medicine in the OSA field in association with a better understanding of users’ experiences is expected to allow for more personalised follow-up pathways and further improvement in CPAP adherence rates [11].

However, the role of CPAP in reducing long-term cardio-metabolic adverse outcomes remains debatable [12, 13]. Early observational data suggested that long-term CPAP usage reduces incident cardiovascular events [14]. In addition, several randomised controlled trials (RCTs) and meta-analyses have consistently demonstrated a clinically meaningful reduction in blood pressure, particularly in patients less than 60 years old, with uncontrolled hypertension, and severe nocturnal oxygen desaturations, as part of a goal seeking secondary prevention [15, 16]. However, RCTs have failed to demonstrate a reduction in incident major cardiovascular outcomes [17–20].

Although RCTs produce the most robust evidence-based information, they are also subject to several important limitations. Most RCTs in OSA have been conducted in minimally symptomatic patients [18–20], included relatively small sample sizes [19, 20], were flawed by a limited number of incident cardiovascular events during follow-up [19], or they weighted heterogeneous composite cardio- and cerebrovascular outcomes equally [18–20]. Most importantly, on average the adherence to CPAP in the RCTs was very low (less than 4 h per night) [18, 20]. Moreover, in some studies, the population was recruited from acute cardiovascular settings, not representative of routine patients evaluated in sleep laboratories.

Ethical concerns have limited the inclusion of sleepy OSA patients for RCTs, which potentially influences outcome in several ways. First, sleepy patients represent a highly responsive group, not only for daytime symptoms but also for cardio-metabolic outcomes [21, 22]. Second, the systematic selection of less symptomatic patients likely impairs mean CPAP adherence of the study population and thus reduces the exposure to the intervention [23].

Other inherent limitations of the RCTs relate to the short follow-up duration compared to previous observational studies [17]. The pathophysiological consequences of breathing disturbances during sleep refer to chronic rather than acute injuries, vascular damage rather than acute decompensation [24]. Moreover, intermittent hypoxia might have specific effects on different vascular beds [25], with data suggesting a higher impact of CPAP on reducing neurovascular risk [26]. Combining stroke with coronary heart disease in one unique composite endpoint may have increased the RCT's power but it might also have masked CPAP effectiveness for some specific outcome entities [23]. Finally, optimal cardiovascular risk management (i.e. medications, surgical and interventional procedures) substantially reduces the room for improvement, which has to be considered thoroughly in calculating sample sizes for demonstrating a significant effect [27].

These methodological issues limit current evidence regarding the impact of CPAP on cardiovascular outcomes. Future RCTs should comprise larger sample sizes (>20 000patients) in line with those conducted in the cardiovascular field to demonstrate clinically relevant reductions in individual cardiovascular event rates. Such sample sizes would permit conclusions from per-protocol (patients with good adherence to CPAP) and intention-to-treat analysis, and allow sensitivity analyses among important subgroups (phenotypes) especially regarding age, gender, body weight and symptom characteristics. Although it is unethical to randomise excessively sleepy OSA into a no treatment arm, new wake-promoting agents may open novel options for comparative studies [23]. Alternatively, a two-stage propensity matching score design may balance CPAP-adherent and non-adherent patients and allow for analysing a sufficiently rich set of covariates, ensuring that outcome differences are attributable to CPAP [8]. Large RCTs pose major feasibility and funding issues. Thus, a first step would be to gather information from the rigorous analysis of large real-world observational studies of CPAP effects. So far, these data strongly suggest a significant reduction in mortality associated with CPAP treatment in adequately sized and unbiased populations [28].

Although the apnoea–hypopnoea index (AHI) remains the most widely used parameter of OSA severity, its role as the main inclusion criterion in studies should be reconsidered. Alternative measures such as the “hypoxic burden”, that captures the depth and duration of respiratory-related desaturation, seem to be better associated with mortality independent of other confounders [29, 30]. Thus, optimal assessment of OSA patients in future studies should incorporate polysomnographic (positional OSA, REM-predominant OSA), pathophysiological (upper airway morphology, arousability, drive, and muscle responsiveness), clinical phenotypes (mildly symptomatic, insomniac, sleepy) and, importantly, pre-existing comorbidities.

Does the primary and often unilateral focus on major outcome parameters reflect patients’ interests sufficiently? What are the most relevant metrics in clinical trials? What are the most valid patient-reported outcomes (PROs) and what is their relevance to respiratory sleep medicine?

At the time of growing recognition of the importance of personalised and patient-centred care, this major impact of PROs must not be ignored. Thus, PROs are now considered as essential for health agencies and policymakers. PROs represent important outcomes in most pulmonary diseases, being more relevant than mortality, for example in asthma (asymptomatic days, missed school or working days, reliever usage), COPD (self-assessment of dyspnoea or activity, exacerbations), pulmonary fibrosis (dyspnoea, exercise performance) and pulmonary hypertension (6-min walking distance). Similarly, sleepiness, cognition and driving performance play a crucial role in the patients’ awareness of how OSA adversely impacts their experience. CPAP is not only highly effective for normalising breathing disturbances but has also a major beneficial impact in the daily life of millions of OSA patients worldwide. Several RCTs have consistently demonstrated improvement in all parameters of quality of life, daytime performance, work productivity and absenteeism [11, 31].

However, taking into account that the best treatment is dictated by the needs of individual patients, one should consider adjuvant (such as medications with potential positive impact) or an alternative to CPAP therapies for OSA, such as oral appliances, surgical approaches and hypoglossal nerve stimulation in selected patients [32, 33]. A clinical fingerprint tool has been proposed, including OSA disease severity, biological activity and impact on the patient that provides a holistic and individualised visualisation of the clinically relevant treatable traits of a patient at any one time and observes the longitudinal effects of treatment [34].

In conclusion, doubts about the cardiovascular impact of CPAP must not outweigh the extraordinary positive effects of CPAP on the daily life of millions of patients. There is urgent need for a more balanced opinion on CPAP efficacy, integrating real-world data in order to provide a more accurate and generalisable picture of the effects of routine clinical usage of CPAP on the multiple outcomes of OSA.

Shareable PDF