Among the mistakes to avoid in any medical field is to assume that the current standards are sufficient for optimal care. In the field of sleep medicine, there may be a general perception that the hourly rate of apneas and hypopneas in a single-night snapshot of sleep is sufficient for diagnosis and risk stratification of sleep apnea. There may be a sense that what is normal and severe is agreed upon, and the gray areas in between are open for discussion. However, there is substantial and growing evidence that the one-night PSG gold standard is not as shiny as it may seem, with potential to both over- and underestimate OSA severity. Understanding the strengths and limitations of the PSG is crucial for accurate patient phenotyping, whether clinically or for research, especially in light of variable and increasingly restrictive insurance coverage for what is perceived by some as expensive and unnecessary testing that can be accomplished with limited channel home devices. In this chapter, some of the key considerations are reviewed in this regard, with a focus on body position as it pertains to sleep apnea phenotyping. This single factor can allow severe sleep apnea to masquerade as normal and vice versa—a reality of which patients and providers alike should be aware.

The gold standard for quantifying the presence and severity of OSA remains the attended laboratory PSG. Like any diagnostic test in medicine, there are trade-offs to consider. Clearly the laboratory PSG provides extensive physiological information, including EEG, muscle tone, leg movements, EKG, and typically a combination of seven respiratory channels (two airflow, two effort, oximetry, intercostal EMG, snoring). Video allows manual scoring of body position. Among those patients who exhibit apparent changes in severity despite no change in body position or sleep stage, the video sometimes shows that the head position has changed, which can also influence breathing in some cases [6].

The rich physiological information provided by this array of sensors can offer important insights into a wide range of sleep pathologies. However, for some patients, the presence of these sensors also compromises sleep due to discomfort or restricted positioning. Sensor discomfort and the foreign environment in general may contribute to what is known as the first-night effect, in which laboratory sleep is more fragmented, with increased N1, decreased REM, and decreased sleep efficiency [7–9]. The fact that sleep occurs away from the habitual home environment may, on the other hand, prove to have a positive impact on sleep consolidation if there are factors in the home that contribute to poor sleep consolidation—this is sometimes called the reverse first-night effect.

There are multiple other factors that might not occur in the laboratory setting but could interfere (or help) with sleep consolidation [10, 11]. For example, late or middle of the night consumption of food or alcohol or smoking is typically not permitted in the laboratory. The lack of typical bed partner presence might lead to improved sleep in the lab, if the partner’s sleep is disruptive, for example, due to snoring [12]. For other individuals, the absence of a routine of sleeping with a bed partner might itself become a source of stress and thus alter sleep in the laboratory. Finally, the single-night snapshot of sleep presents a limitation of “under-sampling” because some of the aforementioned factors can vary from night to night, in addition to the possibility that sleep may have stochastic variability even when such external factors are held constant. In some publications, night to night variability in OSA severity was partially explained by variability in body position [13–16]. The challenge of night to night variability extends to the possibility that individuals may be differentially vulnerable to the potential impact of body position and other factors impacting severity.

The main benefit of home sleep apnea testing is that it can be performed in the habitual sleep environment. The widely claimed cost savings of this method are not supported by cost-benefit models except in very specific populations [17–21]. The limitations listed above for laboratory PSG also apply to home testing, many of which are related to the single-night testing paradigm. In addition, there are other key limitations specific to home testing. For example, home-testing devices do not actually measure sleep, with the exception of the Watch-PAT that extracts surrogate sleep stages using an algorithm based on a combination of actigraphy and autonomic signals [22, 23]. This limitation is important from a practical standpoint, as the reported respiratory event index can underestimate the sleep apnea severity simply because the time over which the events are collected may consist of a combination of wake and sleep, and thus the event rate is per hour of recording rather than per hour of sleep (as it is reported in the laboratory PSG). For patients with high sleep efficiency, the approximation has arguably a negligible effect on respiratory event rate calculation. But for patients with insomnia, or who have for any reason excess waking time on the night of testing, this potential for underestimation can impact clinical decision making, depending on the results. For example, a 20 % difference in total sleep time could move a respiratory index from mild to normal or from moderate to mild, which might impact treatment discussions.

Another limitation of failure to measure sleep stages is that REM versus NREM dependence cannot be determined, which may be important for phenotyping (NREM-dominant disease may be a risk for complex apnea), as well as the potential interaction with body position (supine and REM combination may be the most vulnerable combination for many patients). The limited sensor approach of home testing is clearly focused on sleep apnea, but comorbid sleep disorders are not assessed, such as periodic limb movements, REM without atonia, nocturnal seizures, or parasomnia. The potential degree of misperception is also unknown with home devices, which is of key importance in the evaluation and management of the patient with insomnia or comorbid sleep apnea and insomnia. Finally, technical failures and issues of chain of custody represent important considerations when pursuing home sleep apnea testing.

Regarding body position in particular, only some of the currently available devices include this metric. In the 2007 guideline on the use of home sleep apnea testing devices, 10 of 26 devices had position monitoring [24], and in the follow-up technology review of this field, 11 of 20 devices reported position monitoring [25]. Collop et al. listed body position as a main feature in the systematic evaluation scheme (the “C” in the SCOPER framework) [25]. In the published validation studies of these devices, the focus is usually on detecting apneas and hypopneas rather than optimizing the accuracy of body position detection. Thus, among the many reasons why a negative home-testing night should be followed by confirmatory laboratory, PSG is to ensure that supine sleep (and in particular supine-REM sleep) is observed. Yin et al. showed that patients spent somewhat more time supine in the lab than at home, which explained some of the difference between at-home and in-lab assessments of OSA severity [26]. As Collop et al. point out in their manuscript, body position is not formally part of the diagnostic criteria for OSA [25]. This represents a disconnect between the increasingly appreciated importance of body position in phenotyping OSA patients and the tradition of using the full-night AHI as the single-AHI diagnostic answer. Barriers to implementing guidance regarding position include the lack of standard to define position-dependent OSA (what ratio is clinically relevant? does the absolute rate matter as well?) and the lack of readily available techniques to monitor body position longitudinally in the home.

Consider three patients who undergo laboratory PSG, and each turns out to have the same summary AHI for their night of testing: 20 per hour. Patient A shows marked positional variation in sleep apnea severity and has an AHI of 5 in lateral positions, while the AHI was 35 events per hour in the supine position. He spends half of the night of testing supine, and thus the weighted average of events per hour turns out to be AHI = 20. Patient B also spent half of the night supine, but the AHI was 20 per hour regardless of body position. Patient C spent the entire night supine and has an AHI = 20, with unknown AHI in the lateral positions.

Patient A is reluctant to pursue CPAP despite a supine AHI in the severe range because he maintains that he never sleeps on his back due to back pain. He contends that his lateral AHI, which was borderline, is the more representative value and should guide treatment decisions. This highly positional patient demonstrates two key issues. One is that we do not have reliable and readily available techniques to objectively monitor body position over time in the home, and thus we have little basis for predicting whether any particular patient can successfully and consistently avoid the supine position during sleep. The second issue is that the summary AHI metric is an average of two “extreme” values observed on the single night of testing. This is akin to the satirical comic of a statistician with one hand in ice water and the other in boiling water, saying “the average temperature is perfect,” in the sense that the average is not always the relevant metric. In patients with marked positional variability in severity, it makes little sense to formulate decisions based on an AHI that will be weighted toward high or low values based on what happens to occur regarding position on that night. If the sleep position happened to be mainly lateral on testing, and the full-night AHI dictated treatment, then conservative measures might be emphasized. If the position happened to be mainly supine on testing, then the patient would be characterized as severe, and treatment would be strongly emphasized, including the possibility of surgery if PAP were not tolerated. Alternatives such as dental appliance might not be entertained because the AHI was higher than the mild-to-moderate range generally preferred for this strategy. Without repeated measures in the home setting, it remains unknown how the average over months or years impacts associated medical risks. Does an AHI of 20 per night every night carry the same medical risks as alternating nights or weeks at a time with an AHI of 5 and an AHI of 35 based on fluctuations in body position during sleep? Although the “correct” approach depends on many factors and patients’ preferences, this hypothetical patient illustrates the uncertainties that can surround the single-night AHI—even when it is obtained from an attended laboratory PSG.

Patient B has OSA that is independent of body position, and the AHI severity is in the moderate range. Although body position does not complicate the discussions and planning as occurred with patient A, there may be other factors that influence the nightly severity in patient B, such as alcohol consumption or nasal congestion. However, given that the night in the lab was “typical” according to his report, he elects to pursue treatment with PAP based on the diagnosis of moderate OSA. This patient, compared to patient A, is a reminder of how clinical counseling and decision making can differ substantially despite the “same AHI value” on testing.

Patient C is reluctant to pursue treatment for sleep apnea. Like patient A, he maintains that he does not sleep on his back at home. He is irritable that the technicians told him he had to sleep supine, and he felt constrained by the wires and sensors. His wife reports he only snores on the rare occasion that he rolls supine, and he strongly suspects on this basis that his breathing is normal so long as he is in the preferred lateral position. Like patient A, he thus thinks that the test is not an accurate representation and treatment should not be guided by it. Sleep labs routinely encourage or enforce supine sleep to avoid the potential for a false-negative interpretation, given that the typically vulnerable combination of supine and REM sleep should be observed before ruling out sleep apnea on a lab PSG. However, this cautionary approach has the potential downside of leaving us unable to counsel patients who happened to sleep supine for the whole PSG, who wish to pursue positional therapy, who believe they can avoid supine sleep as a chronic treatment strategy, or who are reluctant to accept a diagnosis based on a perceived unnatural circumstance. Yet without home monitoring devices to track body position, we cannot objectively confirm that positional therapy is effective. Novel devices are in development to answer this challenge, as described elsewhere in this volume. In the case of patient C, his full night of supine sleep did not allow confirmation of the degree of position dependence, which precludes consideration of positional therapy, even if he and his bed partner note decreased snoring in the lateral position.