CPAP machines routinely under-report residual sleep apnoea, according to a US study that found standard device downloads missed most of the disease burden picked up by parallel physiological monitoring.

Savit Malhotra

The retrospective study, published in the journal Sleep [link here], analysed 24,939 nights of paired CPAP and physiological monitoring in 465 adults receiving PAP therapy through a single US telemedicine sleep practice. Patients underwent longitudinal monitoring using an FDA-approved home sleep testing device as part of routine care, allowing researchers to compare same-night CPAP downloads with concurrent cardiopulmonary coupling (CPC) and overnight oximetry data.

CPAP devices reported a mean residual AHI of 2.4 events per hour across all nights, compared with 12.1 events per hour on CPC. CPAP downloads classified 88.2% of patients as having a normal residual AHI, compared with just 7.7% by CPC analysis.

Among the 22,338 nights where CPAP reported a residual AHI below five events per hour:

• 85% had a CPC-derived AHI of at least five events per hour
• 47.7% had a CPC-derived AHI of at least 10 events per hour
• 28.3% included at least five minutes with oxygen saturation below 90%
• 10.6% included at least five minutes below 88%

Investigators said the systematic misclassification may have limited accurate assessment of residual disease burden despite apparent treatment adherence.

Lead author Mr Savit Raj Malhotra, a researcher at Empower Sleep in California, whose home monitoring device was the comparator used in the study, said the findings should change how clinicians interpreted CPAP downloads rather than diminish their value. He said adherence and treatment efficacy should not be viewed as synonymous, and that residual AHI should be interpreted alongside a patient’s symptoms and, in selected patients, complementary physiological assessment.

Unlike polysomnography, CPAP devices estimated residual AHI from airflow alone and could not directly detect oxygen desaturation, arousals or respiratory events occurring after patients removed their mask, often late in the night when REM sleep, and the most severe obstructive events, became more prominent. The researchers noted the American Thoracic Society already distinguished CPAP-derived AHI from polysomnography-derived AHI as a fundamentally different physiological measure.

Residual nocturnal hypoxaemia was more common in patients with obesity, COPD, hypertension and previous stroke, and COPD was associated with more than twice the odds of spending at least five minutes below 90% oxygen saturation.

Sleep physician Dr Stacey Gunn, an investigator on the study also affiliated with Empower Sleep, told the limbic two groups warranted particular attention: those with significant disease burden at baseline, where clinicians needed to confirm the burden had actually resolved, and those with comorbidities such as heart failure, atrial fibrillation, COPD, stroke or uncontrolled hypertension, where residual apnoea could worsen the trajectory of other conditions and the threshold for retesting should be lower.

Mr Malhotra said clinicians should first rule out other common causes of persistent symptoms, including insufficient sleep, insomnia, medications, depression and restless legs syndrome. Where residual sleep-disordered breathing remained a concern, he said overnight oximetry was a practical first step, being inexpensive, widely available and able to pick up nocturnal hypoxaemia that CPAP algorithms missed.

The researchers acknowledged some apparent residual disease may have reflected patients removing CPAP during REM sleep, but said this could not explain the entire discrepancy. Analyses restricted to nights where CPAP use and physiological monitoring were closely matched still found substantial discordance, suggesting the flow-based monitoring itself had limitations.

Dr Gunn compared the finding to prescribing insulin without checking the dose was right, arguing clinicians should not simply assume CPAP was working.

“I don’t think we need to wait years for [longitudinal outcome] data to be collected before we start implementing what are, frankly, common sense changes. The data from our study show that there is a significant amount of residual disease being missed. Acting on that doesn’t require a leap of faith – it just requires us to look,” she said.