273 adults starting CPAP, all with 15 or more breathing events per hour on a sleep-lab study. Each had a fixed pressure set during a second night in the lab. All used ResMed machines that sent data to the clinic. The team read the machine report at the 3-month visit.

One hospital and only 19 cases. The machine counts were not routinely confirmed with a sleep test. The plan was built for regular CPAP users and was not tested against another approach. Only fixed-pressure ResMed machines were studied. Other brands count events differently.

Why Events Still Show Up on CPAP

Q: What kind of events were they?

15 were mostly obstructive. The clinic raised the pressure by 1 to 2 cmH2O, and 14 fell to 10 or fewer events per hour. The other one had very variable leaks. 2 were mostly central. One was found to have insomnia too and started cognitive behavioural therapy for it. For the other, the team judged the machine count unreliable. 2 showed a Cheyne-Stokes pattern. A separate breathing test on CPAP did not find it in one. It confirmed it in the other, who was moved to a different machine type called auto servo-ventilation.

A 2026 Brussels study followed 273 people starting CPAP. 19 still had more than 10 machine-reported breathing events per hour, and most were obstructive events that fell after the clinic raised the pressure.

PulseOxTracker Team9 October 2026

Infographic summarising a 2026 clinic study of events still showing on CPAP. 19 of 273, or 7%, had more than 10 machine-reported breathing events per hour in their first 3 months on CPAP. Who was studied: 273 adults starting CPAP at one Brussels hospital, with 15 or more breathing events per hour on their sleep study, using fixed-pressure ResMed machines sending data to the clinic. What kind of events the 19 had: 15 obstructive, 2 central, 2 Cheyne-Stokes pattern. What happened next. 15 mostly obstructive: the clinic raised pressure 1 to 2 cmH2O and 14 fell to 10 or fewer an hour. 2 mostly central: 1 also had insomnia and 1 was judged an unreliable machine count. 2 Cheyne-Stokes pattern: 1 was not seen on a breathing test and 1 was confirmed and moved to ASV. Mouth-and-nose mask: first mask for 13 of 19 with events and 5 of 19 without. Pressure-setting night: oxygen dips per hour were 23.6 with events and 6.9 without. Nightly CPAP use: 5.8 hours with events and 5.7 without, with similar symptoms. What it cannot show: one hospital and only 19 cases, machine counts were not routinely checked by a sleep test, the 7% applies only once mask leaks were sorted out, other brands and auto-adjusting machines were not studied, and the care plan was not tested against another approach. — A summary of the 2026 study: how many people still had events on their CPAP report, what kind they were, what happened next and what the study cannot show.

Your CPAP machine counts the breathing events it still sees each night. Earlier research cited in this paper says 10% to 25% of users have leftover events. A 2026 study from a Brussels hospital looked at how often that happens once mask leaks are dealt with, and why.

Who was studied?

273 adults starting CPAP, all with 15 or more breathing events per hour on a sleep-lab study.
Each had a fixed pressure set during a second night in the lab.
All used ResMed machines that sent data to the clinic.
The team read the machine report at the 3-month visit.

How many still had events?

19 of 273, or 7%, had more than 10 machine-reported events per hour. The team only counted events after leaks were corrected, because leaks can make a machine miscount.

What kind of events were they?

15 were mostly obstructive. The clinic raised the pressure by 1 to 2 cmH2O, and 14 fell to 10 or fewer events per hour. The other one had very variable leaks.
2 were mostly central. One was found to have insomnia too and started cognitive behavioural therapy for it. For the other, the team judged the machine count unreliable.
2 showed a Cheyne-Stokes pattern. A separate breathing test on CPAP did not find it in one. It confirmed it in the other, who was moved to a different machine type called auto servo-ventilation.

The authors say a machine can label obstructive events as central or Cheyne-Stokes by mistake. We covered a case like that.

How did the 19 differ from the rest?

Each was matched with a similar patient who had no leftover events.

13 of the 19 started on a mouth-and-nose mask, against 5 of 19 in the comparison group.
On the pressure-setting night they already had more events and more oxygen dips: an ODI of 23.6 against 6.9.
Nightly use was about the same, 5.8 against 5.7 hours.
Sleepiness, fatigue and insomnia scores were similar.

What can't it show?

One hospital and only 19 cases.
The machine counts were not routinely confirmed with a sleep test.
The plan was built for regular CPAP users and was not tested against another approach.
Only fixed-pressure ResMed machines were studied. Other brands count events differently.

What can you do with this?

If your machine reports events, show the report to your CPAP clinic. Pressure and mask changes are their call.
An oximeter records overnight oxygen separately from the machine. See tracking your oxygen alongside CPAP.
Here's how to share recordings with your doctor.

Home oximetry is not a sleep study.

Not a medical device. Does not diagnose, treat or prevent any condition. Discuss results and changes with a clinician.

References

Hondeghem S, Mettay T, Espana-Dominguez C, Bruyneel AV, Bruyneel M. Prevalence and management of residual respiratory events in OSA patients treated with CPAP: a prospective observational clinical experience. Sleep and Breathing. 2026;30(4):218. doi:10.1007/s11325-026-03767-z