Only 12% of 102,564 Korean sleep apnea patients stayed on PAP therapy

Positive airway pressure is the most effective treatment for obstructive sleep apnea, and the persistent awkward fact about it is that a large share of patients stop using it. A nationwide analysis published in Sleep Medicine on October 19 puts a number on that in a health system designed to make adherence easier to achieve and easier to measure [s1].

The dataset

The study used National Health Insurance Service claims covering 102,564 adults in Korea who were diagnosed with obstructive sleep apnea on the basis of polysomnography and prescribed PAP therapy between July 2018 and December 2021 [s1].

Adherence was not self-reported. It was defined by the insurer's own reimbursement criteria: during the compliance phase, use of at least four hours per night on at least 21 days within any 30-day period; and thereafter, average use of at least two hours per day in each subsequent three-month period [s1].

Kaplan-Meier survival analysis and Cox proportional hazards models were used to describe adherence patterns and identify predictors [s1].

The headline number

Only 12.2% of patients remained adherent over the follow-up period [s1].

The authors describe this as unexpectedly low, and note that it arises despite Korea's universal health coverage and strict adherence monitoring [s1].

That framing is the point of the study. The two explanations usually offered for poor PAP adherence — cost and lack of oversight — are largely removed here. Coverage is universal, and device usage is tracked closely enough to be the basis of reimbursement.

What predicted staying on treatment

Better adherence was associated with younger age, particularly the 20-29 group, with female sex, and with auto-titrating devices [s1].

Poorer adherence was associated with a body-mass index of 30 kg/m² or above, with older age — particularly the 40-49 group — and with the use of continuous rather than auto-titrating PAP [s1].

Prescriptions written by neurologists and psychiatrists were associated with slightly better adherence than those written by otolaryngologists [s1].

Economic status, residential region, disability status and mask type were not significant predictors [s1].

Two of those findings cut against intuition. Higher BMI is a risk factor for more severe apnea, and more severe apnea is usually associated with better adherence because the symptomatic benefit is larger; here it went the other way. And the 40-49 age group, not the oldest patients, showed the poorer adherence.

How it compares with the wider literature

A systematic review covering twenty years of adherence data, published in 2016, found an overall CPAP non-adherence rate of 34.1% based on a seven-hour nightly sleep time, with no significant improvement across the two decades and behavioural intervention improving adherence by roughly one hour per night on average [s2]. Its authors argued the persistently low rate called into question the framing of CPAP as the gold standard of therapy [s2].

The Korean figure is not directly comparable — different definition, different denominator, different follow-up structure — but the direction is the same, and this is a far larger and more complete population than the study-level literature could assemble.

The mechanism the authors propose

The study's conclusion is not that patients are simply non-compliant. It is that the criteria themselves may be counterproductive: the authors call for more flexible maintenance requirements and the introduction of positive incentives to encourage sustained use [s1].

That reading is worth taking seriously. When continued reimbursement depends on meeting a usage threshold, a patient who falls below it loses coverage, and losing coverage is itself a route out of therapy. An adherence metric that functions as a gate can convert a temporary lapse into a permanent discontinuation. A claims database cannot distinguish that mechanism from patients who simply stopped, which is the central interpretive limit of the study.

Other limits

These are administrative claims, not clinical records. They capture what was billed, not why anyone stopped — side effects, mask discomfort, nasal symptoms, partner disturbance, or symptom resolution are all invisible here.

The follow-up window begins in July 2018 and ends in December 2021, which overlaps the COVID-19 pandemic [s1]. Sleep clinic access, follow-up visits and device servicing were disrupted in most health systems during that period, and the study design cannot separate that effect from underlying adherence behaviour.

Predictors identified in a Cox model on registry data are associations, not causes. Auto-titrating devices being associated with better adherence may reflect who receives them as much as what they do.

What to watch

Whether Korea revises its maintenance criteria, and whether adherence rates move if it does; whether similar national analyses in other single-payer systems find comparable figures; and whether the association between auto-titrating devices and adherence holds up when device allocation is randomised rather than observed.

This article describes an analysis of national insurance claims and is informational only. It is not medical advice and does not recommend any device, therapy, or change in treatment.

Sources

• [s1] Analysis of positive airway pressure adherence in 102,564 patients with obstructive sleep apnea in Korea, Sleep Medicine, 2025;136:106859, published online 2025-10-19.
• [s2] Trends in CPAP adherence over twenty years of data collection: a flattened curve, Journal of Otolaryngology - Head & Neck Surgery, 2016;45:43, published 2016-08-19.