- Design
- prospective sleep-clinic cohort linked to the French national health database, Cox models with interaction testing
- Population
- 3,370 PAP-treated patients with moderate-to-severe obstructive sleep apnoea, median follow-up 9 years
- Primary outcome
- first major adverse cardiovascular event
- Effect
- PAP adherence adjusted HR 0.53 (95% CI 0.46-0.62); interaction with high-risk status HR 0.61 (0.43-0.87; p = 0.006); 740 events
A sleep-clinic cohort of 3,370 patients on positive airway pressure for moderate-to-severe obstructive sleep apnoea, linked to the French national health database and followed for a median of nine years, recorded 740 major adverse cardiovascular events. Adherence, defined as at least four hours a night, was associated with an adjusted hazard ratio of 0.53 for those events.
The finding of interest is the interaction. Patients were stratified by two physiological markers taken from the sleep study — sleep apnoea-specific hypoxic burden and the heart-rate response to events — and the association with PAP adherence was substantially stronger in the high-risk group, with an interaction hazard ratio of 0.61.
This speaks to the central puzzle of the field. Randomised trials of PAP have repeatedly failed to show cardiovascular benefit, while observational data consistently suggest it. One explanation is that trials enrolled unselected apnoea populations in which most patients had little modifiable cardiovascular risk from their apnoea, diluting any effect. These markers offer a way to identify the group in whom the effect might be real — and the authors note the association was stronger in non-sleepy patients, precisely the group current practice is least confident about treating.
That the same stratification worked using a simplified version derived from the oximetry signal alone is the practically important detail, because it does not require full polysomnography scoring.
- Record hypoxic burden, not only the apnoea-hypopnoea index, when a sleep study is reported
- Push adherence hardest in patients with high hypoxic burden, including those who are not sleepy
- Do not use daytime sleepiness as the sole gatekeeper for treating cardiovascular risk in apnoea
- Ask whether your sleep service can derive these measures from oximetry, which requires no extra test
- Keep the four-hour nightly threshold in mind when counselling; it is the adherence definition this evidence rests on
The statistics, in plain English
An adjusted hazard ratio of 0.53 (95% CI 0.46-0.62) for adherent versus non-adherent patients is a large association, and that size is itself a warning: adherent patients differ systematically from non-adherent ones in ways no adjustment fully captures — they take other medications more reliably, attend more appointments, and are generally healthier. This is observational and the headline figure almost certainly overstates any causal effect. The interaction, however, is less vulnerable to that criticism: healthy-adherer bias should apply similarly in both risk strata, so a difference between them is harder to explain away, and an interaction hazard ratio of 0.61 (0.43-0.87) with p = 0.006 is a reasonably robust signal.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for pulmonology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free