The Baby, the Bathwater, and the Polysomnogram

Donovan and Patel make some important points regarding the utility of the polysomnogram (PSG) (1). We are also strong believers in technological innovation and that optimizing patient care in obstructive sleep apnea (OSA) will require approaches that are portable, scalable, and cost effective. However, we offer several counter arguments to our colleagues for consideration:

• 1.Although the PSG currently serves primarily to yield an apnea–hypopnea index (AHI), the rich data available from these recordings are largely ignored (2). We agree that if the only goal is to obtain an AHI, this goal could probably be achieved with home sleep testing or wearable technology. By analogy, if the only information gleaned from an ECG were a heart rate, the ECG would disappear as an antiquated instrument. However, we and others have done extensive processing of PSG signals and have developed robust methods to assess endotypes (mechanisms) underlying OSA (3) and to define better the heterogeneity underlying disease (4, 5). Furthermore, research into more advanced PSG metric parameters to characterize better individual patients (6, 7) and to prognosticate long-term outcomes (e.g., hypoxic burden [8]) is rapidly progressing. These approaches will likely help us move past the current “one size fits all” strategy, which is typically employed, and toward a more precision medicine approach (9). The abandonment of PSGs would yield a situation in which the recent progress and eventual clinical uptake may well be compromised.
• 2.The authors emphasize the importance of patient-reported outcomes, with which we concur (10). However, we would offer several situations in which the patient voice must be heard but objective data are crucial. For example, we commonly see patients who have a vested interest in obtaining an OSA diagnosis (e.g., government employees or military) because they may then qualify for service-related disability (11). In this context, we frequently observe very high prevalence rates of snoring and sleepiness ostensibly because patients are motivated to obtain an OSA diagnosis. In contrast, in some occupational medicine settings, based on human nature, patients may try to avoid an OSA diagnosis and minimize self-reported sleepiness and snoring (12). We estimate currently at University of California San Diego that roughly 30–40% of our patients fall into a category in which self-report may be unreliable. As such, objective data are critical in many cases to complement the patient voice (13).

Another example of when patient-reported outcomes may be complicated is in the follow-up of patients on continuous positive airway pressure therapy. In many cases, we see patients who report feeling well with treatment but have a high residual AHI. In such cases, we feel strongly that the underlying apnea needs to be addressed rather than just declaring victory based on improved symptoms. Clinically, we have seen the effectiveness of in-laboratory titration to assess these patients with residual apnea on positive airway pressure therapy. An extreme analogy might be giving a patient stimulants for untreated OSA; this action may be to the patient’s short-term satisfaction but likely to their long-term detriment. Thus, again, we view patient-reported symptoms as helpful but not definitive data.

• 3.The authors appropriately emphasize the importance of OSA given the global prevalence of disease, estimated to be up to 1 billion people (14). However, we would argue that many other sleep conditions, including periodic limb movements, narcolepsy, insomnia, hypoventilation, and parasomnias, exist and need to be addressed. In many cases, home testing is not adequate to make these diagnoses, and PSG is quite helpful. Even in patients with OSA, comorbid conditions can frequently complicate therapy and interpretation of diagnostic information. For example, Gooneratne and colleagues (15) have reported high percentages of OSA with comorbid insomnia in the elderly, a situation in which home sleep testing (without any sleep assessment) may be complicated if not misleading.

We welcome a discussion regarding the optimal strategy to diagnose the major burden of sleep disorders in general and agree that implementation of technological solutions may be accelerated by the coronavirus disease (COVID-19) pandemic. However, we are reluctant to discard an important test until alternative strategies have been rigorously tested and proven effective.