In this issue of the Journal (pp. 668–672), Flemons and colleagues (1) provide a fascinating account of how sleep medicine services are provided in different countries. With the increased recognition of sleep apnea, systems for delivering diagnosis and treatment are overwhelmed. Physicians are trying to cope but, even with creative approaches, waiting lists for diagnosis and treatment are unacceptably long. There is a need to rethink current strategies.

Sleep apnea is not a new condition, but its clinical recognition is relatively recent (1965) (2). It was described previously, however, in literature. Much attention is paid to the Dickens description of the fat boy in The Pickwick Papers, but there are earlier descriptions. In the book Edda, one of the Icelandic sagas published in the early 13th century, an individual—Skrymir—is described who snored so hard that the forest resounded. He had excessive daytime sleepiness falling asleep despite repeated blows administered by Thor (T. Gislason, personal communication).

The modern era of sleep apnea began in earnest in the 1980s. The seminal discovery by Sullivan and colleagues of an effective treatment—nasal continuous positive airway pressure (3)—changed the ability to treat the disorder. In this decade, the foresight of the leadership of the National Heart, Lung and Blood Institute to support research in this area resulted in substantial and dramatic increases in our knowledge about the pathogenesis, epidemiology, genetics, consequences, and treatment outcomes. This new knowledge has led to rapidly growing demand for services, but most health care delivery systems are not meeting this demand. There is wide variation in the size of the population served by individual sleep centers (see tables in the article by Flemons and colleagues [1]).

Faced with the dilemma of how to treat the “flood” of patients presenting with symptoms suggestive of sleep-disordered breathing, physicians are using non-conventional approaches for diagnosis and treatment—approaches not based on solid evidence. Most surprising in the article by Flemons and colleagues (1) is the widespread use of ambulatory approaches to diagnosis rather than full in-laboratory polysomnography. Several recent reviews and policy documents (4–7), including one by an independent group (not sleep professionals) (5), indicate that this strategy cannot be recommended. Thus, why is the approach being used widely by thoughtful sleep physicians? As previous reports indicate, it is not the technology per se—indeed the technology looks promising (5)–but rather the lack of compelling evidence due to poor study design. The widespread use of this approach, as reported by Flemons and colleagues (1), suggests that a major focus of future research needs to be to properly evaluate these alternative diagnostic strategies.

There are, however, major barriers to overcome. First, many devices are currently lumped together into classes. Reviews of evidence for a particular class of devices evaluate results of studies of all devices within that class (6, 7). Thus, if one particular technology is outstanding, it will not be recommended because the evidence will be weighed down by results from inferior devices.

The multitude of devices presents another problem. The results from a study of one device cannot be extrapolated to another. This situation is unlike research using polysomnography, where standards have been set so that results are more generalizable and evidence can develop quickly (8). Studies of individual proprietary devices are supported by individual manufacturers, have small sample sizes (because of financial constraints), and hence methodologically rigorous reviewers judge them to be of poor quality.

We need to break through these barriers to properly evaluate the utility of these different diagnostic approaches. An important step would be a document developed by the professional societies that clearly states the essential components of a technology that assesses breathing disturbances during sleep without the need for EEG recording—in the same way as was done for in-laboratory polysomnography. Equipment manufacturers might recognize the value of establishing a common technological platform so that we can rapidly obtain the evidence we need.

A second barrier is the nature of the study designs and analyses of data. Many studies involve simultaneous assessment by the simplified device and full polysomnography. This approach has been criticized because it is not the circumstance in which the device is to be used (5). The alternative design involves assessment by the simplified device on a separate night, often in the home. This introduces, however, an additional source of variance based on known night-to-night variability in the degree of sleep-disordered breathing (9, 10). If full polysomnography were assessed in this way, it, too, would likely be found wanting!

Analysis strategies are also problematic. The goal of an evaluation is not to precisely determine the apnea–hypopnea index, but rather to identify individuals who will benefit from treatment. Currently, however, diagnostic sensitivity and specificity are assessed by whether the apnea–hypopnea index is above or below a fixed threshold, e.g., 15 episodes/hour (11). This results in the absurdity that if the apnea–hypopnea index is 16 episodes/hour on the full polysomnography but 14 episodes/hour on the simplified test, this is a diagnostic failure! Thus, new approaches are needed that evaluate the different sources of variance: differences between diagnostic equipment, differences between nights, differences between the home and in-laboratory setting.

Thus, the article by Flemons and coworkers lays down a clear challenge. We need to use all available strategies to increase access to diagnosis and treatment for patients with sleep-disordered breathing. It seems inconceivable that we should tell a patient the following: “You are highly likely to have severe sleep apnea, a disorder associated with an increased risk of car crashes, high blood pressure, and probably heart attack and stroke. We have an effective treatment for this disorder. We will arrange a study for you in 14 months' time to assess this.” If we are still doing this in 5 to 10 years from now, we will have failed our patients and failed the funding sources that have supported the research that makes treatment of this disorder so important. The major issue now in this field is access. Let us commit to solving it.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES