In response to this article, I was very impressed with the respirology theories put forward to address different approaches to the treatment of OSA in general and the specific breathing events of UARS (usually dubbed RERAs or Flow Limitation Events). As I am not trained in respirology or pulmonology, my intent in writing this response is to clarify several clinical aspects about how patients with UARS need to be managed when using these advanced PAP devices, including pointers on what occurs when using these devices in the sleep lab during titration sleep studies. Some of the comments will indirectly speak to the machine algorithms for treating UARS.

To be perfect honestly, the field of sleep medicine remains far behind in its clinical applications of advanced PAP treatments for sleep-disordered breathing, whether OSA or UARS.  Ironically, it has achieved more progress in treating CSA.

It may also be of value to the reader to understand the miserable state of affairs in clinical sleep medicine in which arguably less than 10% of all sleep doctors have demonstrated any interest in advanced PAP devices.  Indeed, an advanced PAP device for most sleep doctors would be CFLEX or APAP or APAP with EPR.   To be as precise as possible, thousands of board-certified sleep physicians either have no interest in testing any of their patients on bilevel modes or worse are thoroughly persuaded bilevel modes are unnecessary, costly and inappropriate, because…wait for it…all patients can learn to use CPAP successfully.  Ergo, no other PAP devices are ever to be considered.

Some readers might find this harsh, especially since there are actual guidelines from the AASM indicating appropriate uses for bilevel modes, particularly as it relates to expiratory pressure intolerance or other discomfort.  Notwithstanding, most sleep doctors don’t lose any sleep over AASM recommendations and instead have developed highly entrenched and intransigent ideas about how PAP works.  Namely, all PAP devices are perceived as the same and all brands are the same.  As this conventional wisdom has taken hold dating back at least three decades, it is a point of fact the average patient who is failing CPAP is informed that no other device mode is available or even worth trying.

This sad state of affairs is not unusual in modern (some would say regressive) medicine, where protocols and algorithms have supplanted the “arts and sciences” of medical treatment. Sleep medicine suffers the same fate, only worse in that those of us who know bilevel machines are superior to CPAP cannot get very far discussing these ideas with our colleagues, even though by superior we mean greater improvement in patient symptoms. The biggest losers are the patients who would hope their doctors look around to find suitable alternatives when things are failing of flailing.

Most sleep doctors do not look around for alternatives to CPAP, because most are unwilling to explore “off-label” PAP strategies.   And, when bilevel modes are brought up for discussion, the conversation ends very quickly. This background information is not offered to trash the field of sleep medicine, which performs well in many other realms. Paradoxically, its weakest area is in those areas related to PAP and titrations for the treatment of sleep-disordered breathing, the very disorder that birthed the field.

Rather, it is offered to help patients recognize it is often an uphill battle to attempt new modes of PAP when failing CPAP.  We stopped prescribing CPAP in 2005 and used bilevel modes thereafter.  By 2010, all our patients were prescribed ABPAP or ASV, with about one-third of patients needing the ASV mode.  To most sleep doctors, our work is promptly dismissed, because they do not envision circumstances indicating a need for advanced PAP prescriptions.

Which brings up the big question of why are bilevel modes so clinically relevant to a sizeable majority of OSA/UARS patients?  The answer is quite simple as far as we have observed in clinical sleep medicine:

To fully normalize sleep inspiration as well as sleep expiration, the inspiratory and expiratory airflow curves must be fully rounded on the pressure transducer tracing of airflow. To be fully rounded, each limb of the airflow curve needs its own distinct pressure. That is, higher pressure when you breathe in to round the inspiratory curve and lower pressure when you breathe out to round the expiratory curve. Hence, the use of bilevel.

As CPAP only has one fixed pressure, it can rarely round both curves, either because the sleep technologist is timid about raising the pressure to eliminate flow limitation on inspiration, or the reverse, the tech is aggressively (and correctly) rounding the inspiratory curve with higher pressures, which then are intolerable to the patient on expiration, what we call expiratory pressure intolerance (EPI).

Most individuals, be they sleep apnea patients or sleep healthcare professionals, do not seem to understand auto-CPAP (APAP) only adjusts pressures between breaths so to speak. At any point in time and space, a CPAP or APAP is delivering the same fixed pressure on the unique inspiration and expiration of each unique breath.  Although an APAP pressure may rise or fall based on its algorithm, it doesn’t rise and fall during each breath in and out, like bilevel.  APAP with EPR provides a mini-bilevel, but the gap or as some would call it pressure support is small ranging from 1 to 3 units.

So, in a nutshell, smarter bilevel modes like ABPAP and ASV work better than the other devices for rounding the curve because they can deliver higher pressure as needed for inspiration and then lower pressure as needed on expiration. Other features of these devices including their algorithms kick in throughout sleep to more aggressively or less aggressively improve the airflow signal.

But, here’s the big caveat that most sleep clinicians do not appreciate and which was not mentioned in the article:  There is virtually no evidence, as I understand things clinically, to indicate the algorithms inside ABPAP or ASV can normalize the airflow curve if left solely to the influence of their auto-adjusting features.

Once we realized this lack of perfectibility of the devices in 2010, we knew the only compensatory solution was to manually titrate patients in the sleep lab to override the auto-adjusting settings, based on their failure to aggressively or correctly address flow limitation events.  Moreover, notwithstanding the nuanced theories put forward in the description of the two brands of ASV, I am remain unpersuaded these devices can attain an optimal response to all flow limitation events.

Ironically, the terminology “auto-adjusting” has taken on a life of its own, and sleep techs and clinicians have mistakenly assumed a patient can be started on either default settings or other arbitrary settings and expect good results. Although such effects are not impossible, I have not seen this approach work consistently to yield optimal results.

And, this point brings us back to the need for manual titrations in the sleep lab to find more optimal pressure settings that will function in tandem with the auto-adjusting algorithm.

A simple example might see a default setting of Minimum Pressure Support of 3 on an ResMed ASVAuto device.  A data download shows persistent flattening indicative of flow limitation.  The same flow limitation should also be visible on this setting when the patient undergoes testing in the sleep lab, during which the sleep tech would raise the Min PS by units of 0.2 up to a high of Min PS 6 (ResMed device).  In reality, 6 is rarely attained.  Instead, such a patient’s Min PS might top out at 4.6, assuming it rounds the inspiratory airflow curve.  Thus, a default setting of Min PS 3 would not have corrected the residual flow limitations, and the patient would not have received optimal therapy.

The above example is quite common among patients who have acquired an ASV device yet received minimal assistance in adjusting initial or follow-up pressure settings from the prescribing physician or sleep center support staff. And, contrary to the statements in the article, we have found consistent benefits in using Min PS as opposed to EPAP in treating flow limitation with a ResMed device.

These patients often seek educational appointments with me through my coaching services ( And, in these circumstances, I explain how the auto-adjusting algorithm was insufficient to resolve flow limitations, again either on a data download they may possesses or hypothetically if they were to return to a sleep lab.  One of the first steps to take upon encountering such individuals is to gradually raise Min PS and monitor for improvements.

If you acknowledge the theory of imperfect algorithms, it still doesn’t answer the question on whether the ResMed or the Respironics ASV devices are equivalent or is one superior to the other? We’ve stuck with ResMed devices for a solid decade, and in the occasional instances where patients sought second opinions from us while using the Respironics mode, many of these patients reported a more satisfying experience when they switched to the ResMed ASVAuto mode.  Nonetheless, we have also worked with a few patients that did not perform well with ResMed and they reported greater satisfaction with Respironics.

One other piece of encouraging input comes from the most senior sleep technologist at my old sleep center in Albuquerque (we moved to Savannah this year and the new owners have renamed it, The Sleep Spot,  He reports similar efficiency in conducting manual titrations with either the ResMed or Respironics brands of ASV.  Notwithstanding, the question remains as to whether one brand yields a better response.

To answer such a question beyond the data extracted from a device or its impact on breathing events in the sleep lab, we have spent considerable time looking for corroborating evidence that bilevel modes are superior to CPAP modes.  What I will describe below could also be tested to clarify distinctions between ResMed and Respironics devices.

To start, what’s the greatest corroborating evidence to assess in comparing different modes? Clearly, it should be something to do with sleep outcomes.  In our clinical experience, the single most reliable variable is what the patient reports the morning after using an advanced PAP mode.  In fact, just the single question, “Did you sleep better?” provides a great deal of prognostic information as most such patients answering affirmatively almost invariably adapt well to the advanced PAP mode and declare its superiority to previously used modes, usually CPAP or APAP…and all this after just one night of use and in a sleep lab no less!

A variant of the above question for some people will be “What is your daytime energy like in the afternoon, after you’ve gotten a good night of sleep with your PAP?”  Many CPAP users may report sleeping better in the morning and then suffer flagging energy in the afternoon, a likely indicator for insufficient sleep quality depth on the previous night.

Still greater corroboration is required or desired beyond these subjective reports, albeit these self-reports are incredibly reliable.   Here is where we delved into typical sleep metrics like percent of Delta (Stage 3 NREM sleep), percent of REM, percent sleep efficiency, and REM consolidation index (to be explained).  Obviously we want higher numbers for these 4 variables.  And, we want lower numbers for percent Stage 1 NREM, sleep stage transitions, arousals, and awakenings.

In years of clinical practice, we repeatedly saw these metrics moving in the right direction when switching patients from CPAP to BPAP and then saw further improvements when patients switched from fixed BPAP (standard practice 2005 to 2009) to ABPAP or ASV (2010 onward).  The abstract of our most explanatory paper on these themes can be seen at this link

During this period, we gave much attention to the continuity of REM sleep, because you probably know the worst breathing events and worst oxygenation problems often occur in this stage of sleep.  Most sleep labs score REM in what we call non-specific ways, that is, suppose you see a 15 minute segment of REM interspersed with a few arousals or awakenings of brief duration, say 2 awakenings each of one minute.  Most sleep labs follow rules to count the full 15 minutes as one block of REM despite these brief interruptions.  We don’t see the logic of this approach. Why not measure the continuity of REM since the expectation is that REM will consolidate best when breathing is most normalized.

In the above example if the individual’s REM goes 5 minutes, then 1 min awakening, then 5 min of REM and another 1 minute awakening, and finally 5 min of REM, the total sleep period is 15 minutes, but no REM segment was longer than 5 minutes, so the REM consolidation would be 5 units or 15 divided by 3 distinct segments of REM.

When we observed patients on ABPAP and ASV going 15 to 30 consecutive minutes without any EEG disruption, it occurred to us this might be a signal of higher quality sleep.  So, we calculated the REM consolidation index on every patient for the past decade, because we noticed that a large proportion of patients with a greater REM consolidation index (i.e. more continuous REM) fared better on that actual night of testing and reported faster adaptation to PAP and a better overall response regarding daytime benefits.

Some of this information is experimental or speculative, but I would be remiss if I did not mention we’re talking about thousands of patients that we observed from 2010 to 2019 to guide us in developing this clinical perspective.

Summing up, I believe the technology has a ways to go before we can regularly rely on auto-adjusting algorithms to replace the sleep lab experience.  Whenever that technology arrives, it would be welcome given how much patients still complain about the sleep lab experience.  Until then, looking for corroborating evidence is one very important pathway to use in attempting to determine the superiority of one PAP device over another.   Regrettably, a lot of the corroborating evidence requires overnight testing in a sleep lab, but surely sleep researchers will raise their curiosity levels to a point where this type of comparative data will be studied extensively over the coming years.

Nevertheless, each OSA/UARS patient owes it to himself or herself to dig deep and ask whether or not the quality of sleep feels consistently like it is as good as it gets. If it doesn’t feel optimal, then it’s probably not, which means something else is going on that needs to be addressed.

Many are confused and for good reason in trying to gauge what exactly constitutes a great response to PAP.  Though highly personalized to say the least, some clear-cut markers for the majority of OSA/UARS patient would include:  zero trips to the bathroom at night (see nocturia); minimal daytime sleepiness in the afternoon; minimal or sporadic caffeine use; noticeably more active dream life and at least some recall of dream content, sharper wits, potential changes in reflexes like catching your keys before they drop to the floor, and of course fewer awakenings at night, although you could sleep all through the night and still report some of these daytime symptoms.

Depending upon the age range (younger is more likely here), we would expect clear cut improvements in executive function. Finally, we would expect some degree of enhanced creativity or alertness that leads the individual to engage in some new activity. The one I love to hear about is the patient telling us, “you know I started back at the gym,” and when we ask if they think better sleep is the cause, a surprisingly large proportion don’t see the connection.  The other inspiring one is the person who reports going back to school to finish a program or a degree. We presume such patients have achieved a near normal or normal state of sleep or as some of my colleagues insist…Super Sleep.

For now, if you are struggling with CPAP and want a more advanced PAP device, consider printing up and showing your doctors our randomized controlled study published in The Lancet, demonstrating the superiority of ASV over CPAP in treating chronic insomnia:

For those who require more hands-on expertise, I welcome your consideration of my coaching services at

Best Wishes,

Barry Krakow, MD