To Adhere or Not to Adhere—Patients Selectively Decide

THE ABILITY TO MEASURE OBJECTIVE ADHERENCE TO CONTINUOUS POSITIVE AIRWAY PRESSURE (CPAP) THERAPY OFFERS SLEEP PHYSICIANS AND BEHAVIORAL researchers a unique window into the daily activities of patients with obstructive sleep apnea (OSA). Commercially available CPAP machines monitor treatment efficacy by recording the time that the CPAP circuit is pressurized (i.e., mask-on time), as well as the amount of air leaking from the circuit and the presence or absence of snoring, apneas, and hypopneas. This information has proven invaluable in patient management. For example, the finding of a large air leak on the CPAP download may indicate the need to change the patient's mask interface. Downloads with an unacceptably high residual apnea-hypopnea index can alert the provider that the CPAP setting may need to be adjusted.

Using this objective measure of CPAP adherence, Weaver and colleagues were the first to identify that, as early as the first week of therapy, patients segregate themselves into consistent daily users of CPAP or intermittent users who skip days and have low hours of use on days when they do adhere.¹ The individual patterns of usage established within the first week of treatment persist for months or longer. The availability of such reliable, readily obtained data on patient behavior offers an opportunity to develop and evaluate early interventions to improve CPAP adherence before patterns of nonadherence become ingrained. These interventions may be based not only on technical problems relating to the mask interface, pressure setting, or effectiveness of treatment, but also on socioeconomic and psychological factors that may influence the patient's ability and willingness to adhere to CPAP on a daily basis.^2–4

Now that physicians and researchers have a tool to track daily adherence to CPAP and identify patterns of nonadherence, an important question arises: are sleep apnea patients who adhere poorly to CPAP also nonadherent in other medical treatments? In other words, are patients who are nonadherent to CPAP also nonadherent to diet, exercise, and medications? The important article by Villar and her colleagues in this issue of the journal indicates that this may not be the case.⁵ Patients appear to be selective in choosing which medical interventions they will adhere to.

In one of the first studies to compare patient adherence across different medical interventions, the authors compared medication adherence in patients with OSA who were CPAP adherent (> 4 hr/day) versus those who refused or discontinued CPAP therapy. Adherence to medications treating hypertension and hyperlipidemia and reducing platelet aggregation was determined based on medication refills over a two-year period that occurred some time after the patients had accepted or declined CPAP treatment. Villar and colleagues found that patients who declined or discontinued CPAP treatment had similar adherence to all three medication categories as patients who were CPAP adherent.

This intriguing finding rebuts a vexing criticism raised against non-randomized longitudinal studies reporting that untreated OSA is an independent risk factor for incident cardiovascular morbidity and mortality. In 2005, the same group of Spanish investigators, taking advantage of the refusal of some patients with OSA to use CPAP, reported that patients with severe OSA who refused treatment had a greater risk of fatal and non-fatal cardiovascular events over a 10-year follow-up period compared to age- and weight-matched individuals without OSA, even after adjustment for baseline demographic and clinical factors.⁶ In contrast, the risks of fatal and non-fatal cardiovascular events in patients with OSA who had adequate CPAP adherence were similar to those in the control group. Other observational cohort studies have reported similar results.^7–10 However, these nonrandomized studies may exaggerate the effectiveness of CPAP if the CPAP adherent patients were healthier than nonadherent patients in ways not typically measured in cohort studies. Patients who adhere to one treatment may be more likely to adhere to other interventions that are known to influence cardiovascular outcomes. Such a “healthy user” bias has been detected in studies of hormone replacement therapy¹¹ and in landmark cardiovascular trials in which placebo-adherent subjects experienced lower mortality than placebo-nonadherent subjects.^12–15 The observational studies linking OSA with increased cardiovascular risk did not assess whether participants with OSA who did not use CPAP may also have been nonadherent to other, potentially cardioprotective measures such as diet, exercise, and medications for hypertension, hyperlipidemia, and reduction of platelet aggregation. Therefore, differences in cardiovascular outcomes observed between CPAP users vs. non-users may actually have been due to a healthy user bias rather than CPAP treatment for OSA per se.

The finding of Villar and her colleagues that medication adherence patterns are similar in CPAP users and non-users provides important new evidence that the results of their earlier observational study may not be attributed to a healthy user bias, at least as it relates to and can be measured by pharmacy prescription-refill records. Their current study, however, does not completely resolve the issue. One of the study's limitations is the collection of CPAP and medication adherence data over widely different time periods. A second limitation is the failure to track daily CPAP and medication adherence. It is possible that a more refined tracking of daily CPAP usage and daily pill taking for cardiovascular conditions may yet demonstrate some association between nonadherent behaviors in these separate domains of medical care. Moreover, other factors that may have influenced cardiovascular outcomes, such as diet and exercise, were not examined and may have differed between the two groups. Of note, behavior may differ across different patient populations. Therefore, it may not be possible to generalize results from one center. Additional studies addressing the above issues relative to CPAP use and healthy user bias are needed.

Despite these limitations, the study of Villar and colleagues provides an excellent example of how behavioral measures controlling for potential confounding factors can be incorporated into and strengthen future observational studies. In light of the ethical challenge of conducting randomized controlled trials of CPAP therapy that would deny treatment to patients with moderate to severe OSA, well-conducted observational studies such as those from these Spanish investigators are providing mounting evidence of a causal relationship between OSA and cardiovascular outcomes.

DISCLOSURE STATEMENT

Dr. Kuna has received support from Philips Respironics and has received the use of equipment for federally funded studies from Philips Respironics and Embla. Dr. Platt has indicated no financial conflicts of interest.