Non-sleepy obstructive sleep apnoea: to treat or not to treat?

Definitions of sleep disordered breathing

OSAHS is the commonest form of sleep disordered breathing within industrialised communities, affecting at least 2–7% of the middle-aged population [7, 8]. OSAHS is defined as the occurrence of disordered breathing during sleep, characterised by objective measures of recurrent apnoeas and hypopnoeas, resulting in symptoms of daytime sleepiness and possible cognitive impairment.

The current “gold standard” for treating moderate-to-severe OSAHS is CPAP, which comprises mechanical splinting of the airways using compressed air delivered through a nasal or full-face mask worn during sleep. [7].

The objective metric of OSA is the apnoea–hypopnoea index (AHI), which has been subject to a variety of definitions over the decades [8]. The third edition of the International Classification of Sleep Disorders (ICSD-3) [9], for instance, encompasses a very broad definition of OSAHS, defining it as: 1) an AHI >5 events·h^–1 of sleep with one or more symptoms (e.g. sleepiness, fatigue, insomnia, snoring) or an associated medical or psychiatric disorder (e.g. hypertension, coronary artery disease, atrial fibrillation); or 2) an AHI >15 events·h^–1 of sleep without symptoms or associated conditions. The implication of these definitions is that OSA should be treated, since the daytime symptoms are being driven entirely and exclusively by the disturbed breathing during sleep. Currently, however, there is little evidence to support this argument, since definitions of OSAHS are derived from two components rated separately and very differently, namely severity of AHI and daytime sleepiness [8]. Indiscriminate application of the ICSD-3 criteria to data from large community-based cohorts results in 50–70% of the population being defined as having OSAHS [10]. This is clearly not the case.

When polysomnography (PSG) is used to diagnose sleep apnoea, OSAHS is defined as mild (AHI 5–15 events·h^–1 ), moderate (AHI 15–30 events·h^–1) or severe (AHI >30 events·h^–1) [11]. This does not account for age- or sex-related changes in sleep disordered breathing and there are few population-specific, normative data available [7, 8].

According to ICSD-3 [9], excessive daytime sleepiness (EDS) is defined as the inability to stay awake and alert during the major waking episodes of the day, resulting in periods of irrepressible need for sleep or unintended lapses into drowsiness or sleep. Pathological sleepiness has been found to occur in 7–13% of the general population in large surveys and in 20–25% of a primary care population [8, 12]. Men and women appear to be equally affected [13]. One review estimated that 42% of all sleepiness was likely to be secondary to a medical or psychiatric disorder rather than a sleep disorder per se [13]. However, up to 25–50% of OSAHS patients do not report subjective tiredness or sleepiness [14].

Quantifying EDS is difficult. The most commonly used scale worldwide is the Epworth sleepiness scale [15], which has numerous limitations and the subjective nature of which has not been shown to correlate with objective measures of sleepiness [16]. An ESS score of ≥11 out of 24 is considered to indicate EDS in populations with and without OSA [8]. Objective measures to quantify sleepiness, though available, are not readily applicable to everyday clinical practice and most are utilised in the assessment of ability to drive [8].

Thus, definitions of OSAHS and OSA are dependent on the technical acquisition of electrophysiological signals, fluctuating scoring definitions and somewhat arbitrary definitions of sleepiness.

Epidemiology of OSA

The prevalence of OSA in adults worldwide is currently estimated to be 938 million people, nearly 10 times greater than previous estimates [17].

The most frequently cited paper on the topic showed that 24% of men (n=325) and 9% of women (n=250) had an AHI of >5 events·h^–1 of sleep [18]. However, when sleepiness was factored in as causally related to the AHI, the prevalence fell to 4% in men and 2% in women. Several population prevalence studies since then have been analysed with the resultant mean (range) prevalence of OSA found to be 27.3% (9–86%) and 22.5% (3.7–63.7%) in men and women, respectively, and the mean prevalence of OSAHS to be 6% (3–18%) and 4% (1–17%) in men and women, respectively [19]. Thus, OSA without self-reported sleepiness is almost three times as common in men and women as OSAHS. The big question remains: is OSA detrimental to health?

Is there a non-sleepy OSA phenotype?

With the prevalence of non-sleepy OSA being so high within the community, numerous studies have been undertaken over the last two decades to determine why some people with OSA develop diurnal symptoms related to nocturnal sleep disordered breathing and others do not. Additionally, sleepiness has been associated with higher levels of cardiovascular and metabolic comorbidity and many trials have attempted to ascertain whether CPAP treatment leads to improved outcomes, apart from improving quality of life primarily by reducing EDS [5, 20, 21]. Attempts at phenotyping OSA/OSAHS patients using presenting complaints have suggested other associations with comorbidities e.g. insomnia was more likely to herald cardiovascular abnormalities in European patients with OSA [22].

Table 1 summarises non-randomised clinical studies undertaken in an attempt to distinguish sleepy from non-sleepy OSA. Overall, the studies are heterogeneous and not directly comparable, due to technological variations and different definitions utilised. Although AHI has been shown to increase with age, the prevalence of OSAHS has not [8]. Despite differences among the studies, patients with OSAHS were more likely to have a higher body mass index, to be older and to have greater metabolic dysfunction than those with OSA without EDS [36, 37]. Whether OSA carries the same prognosis and symptom burden long term as the lack of any sleep disordered breathing and lack of sleepiness has never been formally studied. Natural history studies of sleep disordered breathing, as typified by the first paper to publish such observations in 2005 [38], suggest that in those with “mild” OSA and snoring, there is no significantly increased risk of mortality and morbidity compared to “normal” controls.

Treating OSA

CPAP therapy has a long history of generally poor adherence, which has been attributed to numerous factors, ranging from severity of the AHI, perceived benefit, personality of the patient, interface issues and type of machine [39, 40].

The seminal paper by Weaver et al. [41] examining the minimum duration per night of CPAP usage in 149 patients with severe OSAHS found that 4 h·night⁻¹ led to a reduction in self-reported sleepiness, 6 h·night⁻¹ led to improvement in objective sleepiness and 7.5 h·night⁻¹ led to improvement in functional status. In terms of optimal adherence to CPAP for improving cardiovascular symptoms, the evidence is less clear. A meta-analysis conducted in 2007 of 12 trials (572 patients) suggested that there was a mean reduction in blood pressure for every 1 h of increased use of CPAP per night [42]. However, further studies, including a meta-analysis published in 2012, suggested that sleepy patients benefitted most from CPAP when self-reported sleepiness (using the ESS) fell [43].

Lastly, even with adequate compliance, many patients with a moderate-to-severe AHI will still complain of post-CPAP sleepiness [44, 45]. This can also give rise to a reduction in long-term adherence and has been the subject of numerous studies and troubleshooting guidelines [46]. The previous duration and severity of OSA and its impact on physiological functions may not be entirely reversible [47–49], or the sleepiness may be due to other factors, physical or psychological, which have not been simultaneously addressed.

More than 40 randomised controlled trials investigating the effects of CPAP on blood pressure in OSAHS have demonstrated a modest reduction in blood pressure, especially in those patients with resistant hypertension [50]. Other randomised controlled trials have shown that CPAP use in OSAHS can improve surrogate markers of cardiovascular and metabolic health, e.g. endothelial function, insulin sensitivity and cardiac function [51, 52]. However, the evidence has not been as consistent for OSA without sleepiness.

Results from a meta-analysis of randomised controlled trials published in 2016 [53], examining the impact of CPAP on patients with non-sleepy OSA, found minimal overall benefit on subjective sleepiness, systolic blood pressure or cardiovascular risk. Six published studies were included [54–59], as well as a seventh study [60] which incorporated and re-analysed data from a previously published dataset [56]. In total, data from 1541 patients were reviewed. The studies were published between 2001 and 2016 from sleep medicine centres in Spain, Canada, Great Britain and Sweden. Inclusion criteria were based on an AHI of ≥15 events·h^–1 of sleep, an oxygen desaturation index (ODI) of 4% from baseline >10 events·h^–1 of sleep, or an ODI >7.5 events·h^–1 of sleep. Lack of sleepiness was defined as an ESS score of ≤10 out of 24 in all studies. The percentage of men in these studies ranged from 76% to 91%; the mean±sd age of the patients ranged from 53.1±2.2 years to 66±8.3 years and their body mass indexes ranged from 28.5±3.6 kg·m⁻² to 33.2±5.3 kg·m⁻². CPAP use was recorded in four studies with average overall usage ranging from 2.65±0.73 h·night⁻¹ to 6±1.7 h·night⁻¹ and study duration from 4 weeks to 4.75 years. CPAP did not improve ESS score overall but it did improve the AHI/ODI significantly versus controls (p=0.026). Diastolic blood pressure in those treated with CPAP fell on average by 0.92 mmHg (95% CI −1.39 to −0.46 mmHg; p<0.001).

Of course, there are limitations to every study. This should be borne in mind when interpreting post hoc analyses. Barbe et al. [55] reported cardiovascular risk reduction in patients who used CPAP for ≥4 h·night⁻¹, as did Peker et al. [59]. Choosing 4 h·night⁻¹ usage as a cut-off appears arbitrary in the context of what is known about CPAP use as discussed above.

Two studies have examined whether reasonable adherence can be reached with CPAP in non-sleepy OSA. Campos-Rodriguez et al. [61] reported that after following 357 non-sleepy patients with an AHI ≥20 events·h^–1 in multiple Spanish sleep centres for a median of 4 years, 230 (64.4%) had a compliance of ≥4 h·night⁻¹. They found that a higher AHI at baseline and the presence of hypertension were associated with better adherence. In their group of patients with coronary artery disease and OSA, Luyster et al. [62] demonstrated that the probability of remaining on CPAP at 2 years was 60% in non-sleepy patients and 77% in sleepy patients. A positive experience of using CPAP initially was a strong determinant of long-term use.

These results bring to attention the difficulties in defining EDS in these groups of patients and possibly do not account for issues with bed-partners (who may reinforce CPAP use even in the non-sleepy to control noise levels), the amount and quality of effort and input that a nurse or physician contributes to encouraging a patient to continue on CPAP, and personality. In the context of a trial, the Hawthorne effect can cause bias in both control and treatment groups not present in real life.

Finally, one trial published in 2016 is worth discussing in the context of non-sleepy OSA, although the enrolment criterion of an ESS score of ≤15 out of 24 does not conform to the accepted definition of not being sleepy. This was the “CPAP for prevention of cardiovascular events in obstructive sleep apnoea trial” (SAVE trial) [63], which recruited a very large population (n=2717 total; aged 45–75 years) in order to examine the effects of CPAP on cardiovascular and cerebrovascular outcomes. The trial was designed to be multicentre, parallel-group and open-label, enrolling moderate to severe OSA patients with an ESS score of ≤15 out of 24 with cardiovascular disease or cerebrovascular disease. Follow-up was for a mean of 3.7 years only. CPAP failed to protect patients from a composite death score which included heart failure, myocardial infarction, hospitalisation for unstable angina, stroke or transient ischaemic attack, compared to controls without CPAP, despite significantly reducing daytime sleepiness and improving quality of life, mood and work capacity. However, mean duration of CPAP adherence was 3.3 h·night⁻¹ during follow-up. A post hoc analysis [64] used latent class analysis to identify high-risk OSA clinical phenotypes. Latent class analysis identified four OSA clinical phenotypes: coronary artery disease (CAD) alone, CAD+diabetes mellitus (DM), cardiovascular disease alone and cerebrovascular disease +DM. Composite cardiovascular events were highest for the CAD+DM phenotype. The presence of diabetes in OSA patients with CAD or cardiovascular disease increased the risk of the composite outcome occurring despite treatment. However, adequate CPAP treatment (>4 h·night⁻¹) reduced cardiovascular risk in diabetic patients with OSA, with the strongest effect being seen in patients with cerebrovascular disease. Although disappointing, the results of the SAVE trial may have been subject to a number of important biases and complications. The majority of the subjects enrolled in the study were from two distinctly different ethnic groups (62%); there was a high dropout rate (83%) and high cross-over of control group patients (57 patients) into the CPAP treatment arm. For ethical reasons, patients with EDS and severe hypoxaemia (oxyhaemoglobin saturation <80% for >10% of sleep study time) were excluded, thus limiting the sample to a potentially lower-risk group, and finally, the CPAP usage rates were low. The use of CPAP for only a short part of the night may have resulted in patients missing most of their rapid eye movement (REM) sleep on treatment. REM-linked apnoeas and hypopneas are typically lengthy and are associated with greater oxyhaemoglobin desaturations and higher sympathetic tone. REM-related OSA is specifically associated with incident or recent onset hypertension and cardiovascular disease [65–67].