New Modes of Tracking CPAP Adherence

Vol. 21 • Issue 2 • Page 16

Sleep Apnea Treatment

Continuous positive airway pressure is the standard of choice in treating patients with obstructive sleep apnea.1,2 Furthermore, CPAP adherence, in terms of hours of use per night, has been proven to modify clinical outcomes.3-5 To assess CPAP adherence and treatment efficacy, CPAP manufacturers have implemented tracking systems that monitor CPAP efficacy (residual sleep-disordered breathing), hours of CPAP use, mask leak, and a number of different flow signals. However, it is important to recognize that no universal guidelines exist on how to use these new CPAP tracking systems, nor do we have strong evidence that tracking systems alone can improve OSA outcomes.

Despite the fact that CPAP adherence tracking systems have not yet been rigorously tested to show measurably improved outcomes, their use seems clinically sound. In fact, CPAP adherence tracking now is a requirement for Medicare and other payers to continue reimbursement for CPAP beyond the first three months of treatment. Moreover, we can track CPAP use better than almost any other therapy for a chronic disease and we have the ability to specifically link patterns of use to OSA outcomes.

Archive ImageA

Why do we care about CPAP adherence and hours of use? Because studies have shown that increasing hours of CPAP use results in better outcomes.3,4 In addition, patients routinely overestimate their CPAP usage with self report.6 Thus, objective monitoring of CPAP use has become the standard of care for managing patients with sleep apnea.

CPAP systems can track hours of use, residual event detection, and mask leak (the primary measurements used in clinical practice). The tracking systems are not limited to conventional CPAP alone, but also can be utilized in patients being treated with auto-CPAP, bilevel, auto-bilevel, or adaptive servo-ventilation.

Adherence tracking systems can collect data that measures the date ranges of CPAP usage and the total number of nights the CPAP was utilized (and not utilized); the systems also can manipulate and further sort the data to present the percent of nights CPAP was utilized, percent of nights CPAP was used > 4 hours/night, percent of nights CPAP was used < 4 hours/night, average or median use on the nights that the CPAP was utilized, and average or median use on all nights. In general, the CPAP adherence-tracking systems are accurate in objectively determining CPAP use.

Unfortunately, sleep-disordered breathing event detection and leak data are more problematic to interpret than hours of use. CPAP tracking systems provide averaged data (over many nights, so these data may not reflect the last week or month) for the residual “apnea hypopnea index” (AHI) and residual “apnea index” (AI) while using CPAP. Moreover, the “AHI” reported by CPAP downloads is an entirely different metric than the one recorded during a conventional sleep study. Currently CPAP devices use a reduction in airflow (measured with a pneumotachograph) to estimate the residual AHI or AI.

In contrast, during polysomnography, apnea or hypopnea determination is based on more robust data, including respiratory flow patterns (nasal pressure and a thermistor), EEG arousal, thoracoabdominal effort, and oxyhemoglobin desaturations. Thus, residual AHI or AI measured from a CPAP download is not a true surrogate of the AHI or AI measured during a sleep study.7 Caution therefore must be used in interpreting OSA resolution or persistence from CPAP adherence data reports.

Finally, the residual apnea data appear to be more clinically relevant than the hypopnea data, which is not surprising since hypopnea detection on polysomnography depends on measures that are not obtained with standard CPAP devices (e.g., oxyhemoglobin saturation and EEG arousals).8,9 From a clinical perspective, event detection data should be used in the management of OSA patients if the data are at either end of the spectrum [normal AHI/AI (< 5 events/hour) or very high AHI/AI (> 30 events/hour)]. However, intermediate residual AHI/AI data can be difficult to interpret and should be examined within the clinical context of the patient. A standardized set of guidelines for measuring and reporting residual events needs to be developed, and the efficacy and accuracy across all CPAP manufacturers demonstrated.

CPAP mask leak data also are problematic. Although there is evidence that a reduction in CPAP mask leak can improve compliance and that improved compliance can improve OSA outcomes, there are no large scale studies that have directly examined the relationship between CPAP mask leaks and OSA outcomes.5,10 Mask leaks depend on both the mask (nasal, pillows, or full face) and the pressure being delivered. What is a clinically significant mask leak? There are no data to answer this question but there may be no leak threshold that is “clinically acceptable,” as even a small leak directed into a patient’s eyes can be a problem.

Moreover, mask leak data are averaged measurements and may not reflect recent changes in the CPAP interface. Mask leak may be secondary to leaking through the mouth or around the mask. Moreover, if the CPAP unit is running when a patient goes to the bathroom, this may appear as large leak in the download even though there is not a true mask leak. Leak data, like event detection data, must be examined within the clinical context of a patient; extreme measurements on the spectrum are more likely to be valid than middle of the road numbers. For a patient without a mask leak, no changes in the interface are necessary unless the patient is complaining about mask discomfort. If the patient’s mask leak is significantly greater than the leak threshold specified by the specific CPAP manufacturer, the interface could be changed.

The new CPAP adherence tracking devices measure many other respiratory signals, including periodic breathing (Cheyne-Stokes pattern), vibratory snoring, respiratory effort related arousal (RERA), flow limitation, clear airway apnea (central sleep apnea), and sleep fragmentation. Unfortunately there are essentially no data examining the validity, reliability, reproducibility, or utility of these signals.

There are several different methods to transmit CPAP adherence tracking data. Most systems use cards (smart card-SD cards), memory sticks, or wireless transmission. The accuracy of these wireless systems in measuring CPAP adherence, in comparison to using the SD cards, has been demonstrated in multiple studies.11,12 However, the electronic transfer of the CPAP adherence data can raise privacy and safety issues. There are several other barriers to incorporating CPAP adherence tracking systems routinely in clinical practice. CPAP adherence profiles are not standardized between the different proprietary tracking systems and the reports are not yet easily exportable to electronic medical records. For wireless data transmission, connectivity to server databases can be suboptimal depending on the patient’s geographical location. Most sleep clinical practices are not configured for this type of data management and examining the CPAP tracking reports can slow down patient flow.

In conclusion, OSA is a chronic disease and needs to be managed accordingly, i.e., CPAP adherence must be followed consistently over time. CPAP adherence, in terms of hours of use per night, has been shown to improve clinical outcomes.3-5 CPAP usage can be reliably obtained from CPAP tracking systems and these data are robust.11 However, the residual events (apnea/hypopnea) and leak data from CPAP tracking systems are not as easy to interpret and standards need to be developed to optimally utilize these data.

View a list of references at

Richard J. Schwab, MD, is professor, department of medicine, division of sleep medicine; and pulmonary, allergy and critical care division; co-director, Penn Sleep Center, at University of Pennsylvania Medical Center in Philadelphia. Elizabeth B. Kneeland, BA, is an administrative assistant in the division of sleep medicine at the University of Pennsylvania Medical Center in Philadelphia.