Sleep apnoea · CPAP · Evidence

CPAP in mild OSA: when treatment may do more harm than good

Written by Dr Ari Manuel, Consultant in Sleep & Respiratory Medicine · Last reviewed March 2026

CPAP is one of the most effective treatments in respiratory medicine — for the right patient. In moderate-to-severe obstructive sleep apnoea, particularly where symptoms are significant, the evidence for benefit is strong and the clinical rationale is clear. But mild OSA is a different clinical entity, and the reflex to treat it with CPAP deserves considerably more scrutiny than it typically receives.

The reality is that in mild OSA — broadly an AHI of 5–15 per hour — the evidence that CPAP improves hard clinical outcomes is weak, inconsistent, and in some respects negative. Meanwhile, the potential for CPAP to cause harm in this group is real and routinely underestimated.

What the major trials actually show

The SAVE Trial (2016)

The Sleep Apnea Cardiovascular Endpoints (SAVE) trial randomised over 2,700 patients with moderate-to-severe OSA and established cardiovascular disease to CPAP plus usual care versus usual care alone. After a mean follow-up of 3.7 years, CPAP produced no significant reduction in the primary composite cardiovascular endpoint — including heart attack, stroke, heart failure hospitalisation, and cardiovascular death. This was a landmark negative trial in a population with existing cardiovascular disease and moderate-to-severe disease severity. The implications for mild OSA, where the biological rationale for cardiovascular benefit is even weaker, are considerable.

The ISAACC Trial (2022)

The ISAACC trial specifically recruited patients with acute coronary syndrome and moderate-to-severe OSA — again, a population at high cardiovascular risk. After a median follow-up of over three years, CPAP did not reduce the incidence of cardiovascular events compared with control. Symptomatic benefit was observed, but the anticipated protective effect on cardiovascular outcomes did not materialise.

Mild OSA and the evidence vacuum

Notably, neither SAVE nor ISAACC specifically addressed mild OSA — which actually makes the picture worse, not better. These trials tested CPAP in populations with the strongest theoretical cardiovascular rationale for treatment (established disease, moderate-to-severe AHI) and found no benefit. Extrapolating cardiovascular protective rationale to mild OSA — a lower-severity condition, often with minimal symptoms — is not supported by the available evidence. The NICE guidance on OSA (NG202) reflects this, recommending CPAP primarily for symptomatic moderate-to-severe disease and emphasising that treatment decisions in mild OSA should be symptom-driven rather than AHI-driven.

The specific harms of CPAP in mild OSA

Inducing treatment-emergent insomnia

One of the most clinically significant and underappreciated risks of initiating CPAP in patients with mild or minimally symptomatic OSA is the induction of insomnia. Patients who sleep reasonably well before CPAP — perhaps unaware of mild nocturnal breathing irregularities, or aware but not significantly symptomatic — are suddenly required to sleep with a mask and a machine generating continuous positive airway pressure. The conditioned arousal, claustrophobia, pressure discomfort, and disruption to sleep initiation that can result are well-documented. A meaningful proportion of patients who did not have clinically significant insomnia before CPAP develop it after initiation. This is not a trivial outcome: insomnia is associated with increased cardiovascular risk, depression, and metabolic dysregulation — the very outcomes CPAP was supposed to prevent.

Anxiety, medicalisation, and nocebo effects

Receiving a diagnosis of sleep apnoea — even mild — and being prescribed a medical device for indefinite nightly use is not a psychologically neutral event. There is a substantial nocebo literature demonstrating that patients who are told they have a potentially serious condition, and given a device to manage it, frequently experience increased health anxiety, heightened symptom awareness, and reduced sense of wellbeing — even when the treatment is physiologically effective. Patients with mild OSA who were previously asymptomatic or minimally symptomatic may paradoxically report worse quality of life after CPAP initiation than before.

Mask-related side effects

Skin irritation, pressure sores, aerophagia, dry mouth, rhinitis, conjunctivitis, and noise disturbance to bed partners are common side effects of CPAP that affect a significant proportion of users. These are tolerable costs in a patient with severe symptomatic OSA who gains meaningful symptomatic benefit. In mild OSA, where the symptomatic gain may be marginal, the side effect burden can outweigh the benefit.

Treatment-emergent central apnoeas

Initiating CPAP in some patients provokes the emergence of central apnoeas that were not present before treatment — so-called complex sleep apnoea syndrome or treatment-emergent central sleep apnoea. This occurs in a clinically significant minority of CPAP initiations and is more likely in patients with a physiological tendency toward central respiratory instability. The result is that CPAP resolves the obstructive component while generating a new, different form of sleep-disordered breathing — sometimes requiring escalation to adaptive servo-ventilation, and sometimes representing a net harm compared with no treatment.

The AHI threshold problem

Current diagnostic thresholds define OSA as an AHI ≥5 (with symptoms) or ≥15 (without symptoms) per hour. An AHI of 6 is technically diagnostic. An AHI of 14 is technically mild. But these thresholds are derived from epidemiological population data and do not reflect the biological significance of any individual patient's breathing pattern. A patient with an AHI of 12 but minimal arousals, preserved slow-wave sleep, low hypoxic burden, and no symptoms may have no clinically meaningful disease. Treating this patient with CPAP solely on the basis of a number — without contextualising that number within their sleep architecture, symptom burden, and physiological phenotype — is not good medicine.

The hypoxic burden metric — measuring total overnight oxygen desaturation load rather than event frequency — is a more physiologically meaningful measure of the actual harm being done by sleep-disordered breathing. A patient with an AHI of 10 and low hypoxic burden, high cardiopulmonary coupling, and no symptoms may be at less cardiovascular risk from their "OSA" than a patient with an AHI of 8 and significant hypoxic burden and autonomic dysregulation. AHI alone cannot distinguish these patients.

What should happen instead

In mild OSA, treatment decisions should be driven by symptoms, sleep architecture quality, hypoxic burden, and functional impact — not AHI alone. Where treatment is warranted, the choice of treatment should be individualised. Mandibular advancement devices (MADs) have a comparable evidence base to CPAP in mild-to-moderate OSA and are substantially better tolerated. Positional therapy is effective in position-dependent mild OSA. Weight management — particularly in the era of GLP-1 receptor agonists — addresses the underlying physiological mechanism rather than providing a nightly workaround. Nasal optimisation can meaningfully reduce AHI in patients with nasal obstruction-driven upper airway collapse.

In patients with genuinely asymptomatic mild OSA and low hypoxic burden, watchful waiting with lifestyle modification and annual review is a legitimate and evidence-consistent management strategy. The burden of proof for initiating an indefinite nightly treatment with known side effects and uncertain clinical benefit should be high.

A note on second opinions

Many patients are referred to this service having been prescribed CPAP for mild OSA and finding it intolerable, unhelpful, or anxiety-provoking. A formal review of the original sleep study, the clinical context, and the available alternatives is often the most useful first step. The original diagnosis is not always wrong — but the treatment choice may not have been the right one for that individual, and the alternatives may not have been adequately explored.