POLL: Typical AHI for ResMed Users

Velbor wrote:

-SWS wrote: So where exactly is the above logic failing?

No failure of logic. Just a sense of incompleteness. And defensiveness on my part, with overreaction to a non-existent "accusation" that I might be "unhygenic". Since my own data was being used as an example for discussion, I would much prefer that my own "excessive wakefulness" not be attributed to any presumed violation of the standard "sleep hygiene" rules (which, again being defensive, I would insist that I follow rigorously - being a rigorous - read obsessive - sort of person by nature).

By the way, I regularly cycle through a careful regimen of Ambien or Klonopin, not to get to sleep, but with the goal of lowering my threshold for waking through the night. Codeine works best, but it's not something a "hygenic" person would want to take regularly!

-SWS wrote: I'm still hoping to hear more about those Respironics summary definitions from my query a mere twelve posts up. And thank you for that, BTW! For instance, I'm wondering if the Respironics hypopnea definition(s) were compiled from various patent descriptions, clinician set-up manuals, training or marketing literature, etc. There seems to be a certain amount of discrepancy among the hypopnea definitions contained in the various Respironics text media.

Sorry, didn't realize that this was a puzzlement. The source of information for my now infamous chart, the source for all of the Respironics and ResMed definitions (the PB data was contributed by ozij) is direct quotation from manufacturer clinical manuals. Here is an image of the Respironics defintion page from the M-Series Auto manual:



Thanks!! Velbor

Thanks, Velbor! So those two recovery breaths are clearly a hard hypopnea-scoring requirement for the M-Series auto. The 50% amplitude reduction is clearly at odds with the clinician's training simulation claiming 40% amplitude reduction instead.

Also a comment that manufacturers probably don't release every little implementation detail about their algorithms. So while Respironics clearly requires those two recovery breaths for hyopnea scoring, I don't think we can comfortably assume that Resmed doesn't also bother to look---in one way or another--for that same post-hypopneic survival reflex (the large-volume recovery breaths).

A part of that Respironics decision-weighting criteria includes spotting FL in the hypopnea's valley. That would probably imply an obstructive component is primarily associated with that hypopnea's etiology, versus central hypoventilation for instance.

I'm still thinking that some of what Resmed scores as hypopneas may be considered hypoventilation events by Respironics instead.

ozij wrote:

-SWS wrote: So my hunch is that if it's sleep/wake transitional hypopneas being recorded, that perhaps the underlying hypopneic etiology has more to do with central homeostasis rather than obstruction.

Dunno about that last part, though, Muffy. Just thinking out loud as I attempt to sort of muddle through what factors might be involved...


Anyone interested in how Muffy's VB scoring hypothesis took roots may want to read this thread:
viewtopic.php?f=1&t=44022&st=0&sk=t&sd= ... 15#p394855

My apologies for responding without having read the root -- and even most of the beginning parts of this thread -- however, that last part in -SWS's post reminded me that the ResMed algorithm is based on the assumption that all hypopneas are central in origin, and it does not attempt to treat them. (See that often quoted Berthon-Jones interview

Why doesn’t ResMed's AutoSet respond to hypopnoea?
When you are lying quietly awake, or when you first go to sleep, or when you are dreaming, you can have hypopneas (reductions in the depth of breathing) which are nothing to do with the state of the airway. For example if you sigh, which you do every few minutes, you usually have a hypopnea immediately afterwards. This can also happen if you have just rolled over and are getting settled, or if you are dreaming. And the annoying thing is that when you are on PAP, this tendency to have what are called central hypopneas - hypopneas that are nothing to do with the state of the airway - is increased. If you make an automatic CPAP device that responds to hypopneas, you will put the pressure up to the maximum while the patient is awake.
Do you think there is a misconception clinically that all hypopneas should be treated ?
For simple obstructive sleep apnea, central hypopneas should not be treated. They are not a disease. Everyone has them. And they don’t go away with CPAP.

ResMed assumes it treats (prevents) osbtructive hypopneas by the way its algorithm responds to what it identifies as flow limitations. That fast response by the S8 seen in the bench test McCoy, Eiken and Diesem:

is a direct result of a planned ResMed escalation when encountering intractable flow limitations.

Here is the S8 responding to a series of intractable flow limitations -- the pressure gets even higher than in response to hypopneas -- and does so very quickly. OTO, the Respironics' response to those is much more careful.




The breathing pattern thrown at the machines can be seen here "Created Unequal" McCoy and Eiken 2006 - a previous version of the benchmarks, but the breathing patterns were not changed:



O.

And thanks for that really great reminder, Ozij!

I know at first Berthon-Jones claimed that Resmed's A10 algorithm did not respond to any hypopneas, since Resmed considered them central. When I now look back, I suspect Berthon-Jones just may have considered what most people had been calling "obstructive hypopneas" to instead be wave-flattened "flow limitations" by Resmed's own proprietary definition. After all, Resmed had always been adept at addressing what they called wave flattening (flow limitations) as preemptive.

I vaguely recall reading in recent years, however, that Resmed was increasing pressure in response to hyponeas---but specifically the ones that happened to have wave flattening (hence obstructive). So I'm not sure if their hypopnea pressure response really changed, or if instead they finally decided to embrace that same obstructive hypopnea terminology/concept everyone else had been using for quite some time.

-SWS wrote:I vaguely recall reading in recent years, however, that Resmed was increasing pressure in response to hyponeas---but specifically the ones that happened to have wave flattening (hence obstructive). So I'm not sure if their hypopnea pressure response really changed, or if instead they finally decided to embrace that same obstructive hypopnea terminology/concept everyone else had been using for quite some time.

In theory (...) we could have an empirical answer to that from people with ReSlink and various versions of ResScan. That would show response (or lack of it) to hypopneas -- and IIRC flow limitation too?

O.

-SWS wrote:Unfortunately my hunch is that wakefulness tends to ameliorate obstructive SDB during wake.

What means that phrase? If you're awake, you don't have obstructive SDB. And even you did, they'd call it obstructive WDB.

-SWS wrote:Dunno about that last part, though, Muffy. Just thinking out loud as I attempt to sort of muddle through what factors might be involved...

Which last part? The whole last part...

Muffenstein wrote:In summary:

Variable Breathing may be a reliable marker of poor sleep quality.

The exclusionary capability of the Variable Breathing Mode may make the Respironics algorithm more accurate in determining SDB events over the entire sleep period in that wake events may be ignored.

Intractable HI in ResMed A10 results may, in some patients, be a result of wake phenomena. In these cases, while an arbitrary "correction factor" may be applied, the first thing that should be examined is Sleep Efficiency.

Extraordinary high ResMed HI, however, should not be summarily dismissed under any circumstances.

...or the last last part...

Muffburger wrote:Extraordinary high ResMed HI, however, should not be summarily dismissed under any circumstances.

The last last part seems obvious to me. While the first thing that one would think if using ResMed would be that the residual events are undertreated obstructive events, I would move that to #4 on the list, given the aforementioned aggressive nature of the A10 HI algorithm (and assuming you're using relatively free-range parameters). If it works, then there won't be any residual events.

#2 would be that the events have a central flavor, and if one is fortunate to have access to Respironics HI, and it is significantly lower, then the Respironics algorithm has discarded them under it's central criteria;

#3 would be that there is fixed obstruction, a subject that has been effectively covered in the past, most notably in the GK420E IFL1 threads with mountainwoman, rested gal, and somebody else who seems to escape me at the moment.

Muffy

Velbor wrote:... (I) might be "unhygenic".

How about dyshygenic?

At any rate, "sleep hygiene" is kind of a catch-all to consider everything surrounding sleep.

To preface, I believe that the comment tossed out by a great number of people that sleep architecture results gathered in the sleep laboratory setting should be arbitrarily tossed out is, to a very great degree, untrue. You know, the ol' "how can you sleep in such a strange place", "all those wires", "the mattress is lumpy", "I can't go to sleep unless I [insert disgusting ritual here]". However, there is a great deal of pre-test information gathered during sleep evaluation, and, through existing comparative studies and a seasoned eye, "lab effect" can be easily be identified and accounted for. Abnormal results may be tempered somewhat, but certainly not summarily dismissed. For instance, if someone has a Sleep Efficiency of 40% in the sleep laboratory, I think I can pretty much guarantee that they don't somehow miraculously sleep like a log when they get home (unless they chew up 6 Ambien before they go to bed).

However, what will throw a monkey wrench into the works is time. Velbor's studies are from years back, rendering the data highly questionable in trying to interpret events today:



However, that said, what I see to be highly encouraging in the CPAP titration is the normalization of Sleep Stage Percentages gained through turning Stage 1 sleep into anything else (Stage 1 sleep is inherently worthless. It's only purpose is to get you to Stage 2) as well as near-zero arousals.

What is bad is that residual Wake totalling nearly 2 hours.

Muffy

ozij wrote:

-SWS wrote:I vaguely recall reading in recent years, however, that Resmed was increasing pressure in response to hyponeas---but specifically the ones that happened to have wave flattening (hence obstructive). So I'm not sure if their hypopnea pressure response really changed, or if instead they finally decided to embrace that same obstructive hypopnea terminology/concept everyone else had been using for quite some time.

In theory (...) we could have an empirical answer to that from people with ReSlink and various versions of ResScan. That would show response (or lack of it) to hypopneas -- and IIRC flow limitation too?

Great suggestion, Ozij.

Velbor and other Resmed data gurus- Does your data set ever show pressure increases in response to any hypopneas? Thanks in advance if you can help.

Muffy wrote:

-SWS wrote:Unfortunately my hunch is that wakefulness tends to ameliorate obstructive SDB during wake.

What means that phrase? If you're awake, you don't have obstructive SDB. And even you did, they'd call it obstructive WDB.

I agree. As obstruction-inclined sleep is repeatedly disrupted by lighter stages of sleep, and especially wake, those airway occlusions are ameliorated. Thus, obstructive
SDB is very common, whereas wake obstruction is not.

Muffy wrote:

-SWS wrote:Dunno about that last part, though, Muffy. Just thinking out loud as I attempt to sort of muddle through what factors might be involved...

Which last part? The whole last part...

Sorry about the confusion!

I wasn't referring to the last part of what you said. Rather, I was doubting the last part of what I said. I can definitely see obstructive apneas being associated with excessive sleep-to-wake transitions. But I was intuitively thinking that if hypopneas accompanied excessive sleep-to-wake transitions, that the underlying etiology is probably going to be central or homeostasis related instead. I still have serious doubts about that intuitive assumption for lack of empirical knowledge.

But on closer look, your patient didn't appear to have those hypopneas---rather it was really erratic breathing that Resmed scored as hypopneas:

Muffy wrote:I have recently had the opportunity to observe a patient with highly erratic breathing noted on NPSG, that when placed on a ResMed unit, exhibited a relatively high HI (and AI, for that matter). Arguably, there were few "scorable events" in NSPG, as well as confounding wake and sleep transition. Close examination of the defining window(s) for the determination of the calculation of "baseline" (key to hypopnea identification) is pending.

However, I believe that "erratic breathing" is the key as to why some ResMed users exhibit higher relatively high(er) HI when compared to Respironics. I am hypothesizing that it is the Variable Breathing Controller in the Respironics algorithm that acts as a filter for erratic breathing. As I understand the operation of the VB Layer, event scoring is suspended.

So that leaves us with one reasonable assumption that the Resmed algorithm falsely scored hypopneas because it couldn't establish a reliable baseline amidst highly erratic breathing. And it also leaves us with a most interesting Muffpothesis to explore.

The Muffpothesis thinks that the Respironcs VB control layer may have discarded those hypopneas. Essentially the Muffpothesis stitches together the facts that: 1) erratic breathing generally presents very tough times for algorithmic hypopnea-scoring, 2) Respironics has a control layer dedicated to erratic breathing, 3) Resmed AutoSet patients tend to report higher HI scores than Respironics Auto patients.

Then the Muffpothesis goes on to speculate that the Respironics VB control layer is primarily associated with wake breathing---that the other erratic breathing conditions mentioned by Respironics are rare, and can be discarded unless proven otherwise. And thus the VB control layer functions for the purpose of detecting moments of wake to discard what would have otherwise been falsely-scored SDB events.

So there are clearly a lot of dots being connected with that hypothesis. For the sake of maintaining balance as we forge ahead and explore the Muffpothesis, I'll point out that while we know the Resmed algorithm overscored hypopneas for that patient, we don't know exactly what the Respironics hypopnea-scoring algorithm would have done with that particular patient.

And if the Respironics algorithm had refrained from scoring false hypopneas during that same sleep session, then we wouldn't know if that was because: 1) the VB control layer recognized the wake states and discarded scored hypopneas according to the Muffpothesis, or 2) because the baseline established amidst that erratic breathing was somehow more workable for that sequence. But for all we know, Respironics may have also had a difficult time correctly scoring that particular patient as well---with the general manufacturer HI discrepancy being attributable to yet other algorithmic scoring differences (besides Respironics successfully discarding wake hypopneas).

While the Muffpothesis is very interesting and reasonable IMHO, proving all those dots are connected the right way remains an outstanding challenge---with plenty of learning for us along the way.

BTW there, -SWS, it looks like your Sleep Hygiene Quiz is spreading faster than H1N1.

Muffy

-SWS wrote: Velbor and other Resmed data gurus- Does your data set ever show pressure increases in response to any hypopneas? Thanks in advance if you can help.

I offer these examples strictly for learning purposes.
I can only hope to, someday, be included in any Resmed/cpap data guru group.

Here are some typical "bad" nights, since I have been using the S8.

7/24


8/25


9/5


Thoughts/comments/suggestions appreciated

-SWS wrote: Velbor and other Resmed data gurus- Does your data set ever show pressure increases in response to any hypopneas? Thanks in advance if you can help. :)

This is difficult. Even if it appears that a hypopnea is temporally related to a pressure increase, we don't know whether that hypopnea caused the pressure response. There may have been a contemporaneous snore or flattening, which we don't know about since ResMed doesn't report these out in the absence of a ResLink module.

ozij wrote: In theory (...) we could have an empirical answer to that from people with ReSlink and various versions of ResScan. That would show response (or lack of it) to hypopneas -- and IIRC flow limitation too?

And even with ResLink, there are difficulties. The data is available only graphically, which makes it difficult to establish cause/effect relationships. And worse, we don't know the "trigger levels" at which the algorithm responds to snore or flattening. Of course, if there were a pressure rise immediately following a hypopnea, with NO apnea and NO snore and NO flattening (which is a difficult interpretation from the graphic data), that might tell us something. But such a creature would be very hard to find.

WOW! I didn't realize ResScan, even with ResLink was that bad. Puritan Bennet's SilverLining for the 420E is so much better!

However, all you need is a few hypopneas without pressure changes to show the machine doesn not respond to them -- in whichever version of machine / software we are discussing.

Resmed's Australian site still shows its Three lines of defense, and the term is still there in American education seminars too: http://www.resmed.com/us/dealers/clinic ... nc=dealers

ResMesd USA wrote:Auto-Adjusting Technologies: How to Optimize Therapy

Program Overview: Discusses the technology of automatic positive airway pressure (APAP) devices and its role in the treatment of SDB.

Functional differences between auto-adjusting devices and standard CPAP
Three lines of defense that differentiate AutoSet™ algorithm
Explanation of disease processes and clinical settings requiring APAP therapy
Clinical data review to assist in determining the effectiveness of APAP treatment
Target Audience: Clinicians, case managers and other allied health professionals striving to achieve optimal adherence to PAP therapy.

Contact / Credit Hours: 1

Elsewhere on the site ResMed USA also wrote:The AutoSet Advantage

Only ResMed devices feature AutoSet technology, which:

Treats automatically - only the pressure you need, and only when you need it (that means less pressure-related side effects)
Treats comfortably - AutoSet delivers a gentle minimum pressure while constantly monitoring the airway
Responds effectively - AutoSet automatically increases the pressure if it detects flow limitation, snore, or apnea – giving you three lines of defense
Acts early, works pre-emptively - to keep your airway open. This can mean lower overall pressures, increased comfort and less pressure-related side effects
Provides assurance confidently - with the clinically proven, safe, trusted and effective AutoSet algorithm

And also

How do AutoSet devices know to adjust pressure?
After you fall asleep and your pressure needs begin to vary, your AutoSet device responds to three separate parameters: inspiratory flow limitation, snore, and apnea. AutoSet devices act pre-emptively by increasing pressure in response to inspiratory flow limitation and snore, both of which typically precede obstructive apneas. The AutoSet device calculates the pressure you need based on the severity of the event. This early intervention prevents obstructive apneas and reduces respiratory arousals.
How do the AutoSet devices handle hypopnea events?
AutoSet devices respond to obstructive hypopnea events when they are associated with flow limitation or snoring. Hypopneas that are central in origin (related to your central nervous system, not physical obstruction) should not be treated with increased pressure.

O.

ozij wrote: However, all you need is a few hypopneas without pressure changes to show the machine doesn not respond to them -- in whichever version of machine / software we are discussing.

Even this needs to be approached with care, if the goal, as I understand -SWS, is to determine whether ResMed Autos ever respond to hypopneas (even if ResMed documentation does not claim to do so). Respironics, for example, states that "When 2 apneas/hypopneas are determined to have occurred within 3 minutes, pressure is increased ..." so there may be additional constraints IF ResMed ever does respond to hypopneas.

PROVING empirically that ResMed DOES respond to hypopneas, as I have described, would be difficult; PROVING empirically that ResMed DOES NOT respond to hypopneas is also not a simple task, since if it DOES EVER respond, we don't know under what circumstances it might or might not do so.

My impression from ResMed's wording about their approach is that ResMed assumes any hypopnea, or apnea, that happens out of the blue with no tell-tale flattening and no snoring associated with it would most likely be central anyway; whereas an obstructive event is generally preceded, in their judgment anyway, by the kind of snore or flattening that the machine would already have responded to.

ResMed's approach of responding significantly and quickly to the precursors of looming obstuctive events instead of waiting for them to happen is something that is greatly misunderstood, in my inexperienced opinion. ResMed warns that its machines score "hypopneas" that a PSG would not score as such. So I would not expect the machine to respond to any so-called "hypopnea" that it saw no reason to attempt to preempt.

jnk wrote:My impression from ResMed's wording about their approach is that ResMed assumes any hypopnea, or apnea, that happens out of the blue with no tell-tale flattening and no snoring associated with it would most likely be central

I don't believe that's quite right as far as apneas go. When breathing ceases (a-pnea), the flow is non-existent, totally flat, and below 10 cm/h2o, ResMed responds to apnea, whenever they happen, as far as I know. ResMed assumes that apneas above 10 are central, and will not respond to them.

ResMed's approach of responding significantly and quickly to the precursors of looming obstuctive events instead of waiting for them to happen is something that is greatly misunderstood, in my inexperienced opinion. ResMed warns that its machines score "hypopneas" that a PSG would not score as such. So I would not expect the machine to respond to any so-called "hypopnea" that it saw no reason to attempt to preempt.

Agreed.

Velbor wrote:PROVING empirically that ResMed DOES respond to hypopneas, as I have described, would be difficult; PROVING empirically that ResMed DOES NOT respond to hypopneas is also not a simple task, since if it DOES EVER respond, we don't know under what circumstances it might or might not do so

I don't think the issue needs us to know under what circumstances the algorithm does not respond to hypopneas. People have managed to see the machines not responding to apneas above 10, I assume we can manage to see the same for hypopneas at any pressure.

As I noted in my previous post::

In response to a Hypopnea FAQ Resmed wrote: How do the AutoSet devices handle hypopnea events?
AutoSet devices respond to obstructive hypopnea events when they are associated with flow limitation or snoring. Hypopneas that are central in origin (related to your central nervous system, not physical obstruction) should not be treated with increased pressure.

However, Resmed in describing its algorithm is very consistent in not describing a (ResMed defined) hypopnea on its own as a reason for raising pressure.

Thanks for the data, carbonman!

Figuring out whether a machine is responding to an "apnea" or to a "hypopnea" can, in itself, be a tricky thing, I think.

I believe that PSG scorers may use a 90% drop in sensor signal for 90% of an event to define apneas (if that is still the definition with modern sensors) but that some home machines may score an 80% reduction in flow or a 75% decrease in ventilation as the home-machine-estimated so-called "apnea." So it seems to me that the line between "apnea" and "hypopnea" can be a cloudy area when discussing home machines, in view of the differences among brands in sensing and defining them. I am unclear when a machine, using it's own internal definitions, chooses to label an event and decide on its reaction. Does it react one way when it estimates a 74% drop and an entirely different way when it estimates a 76% drop? I guess so. But I don't know. That's OK, though, because I'm unclear in my head on a LOT of stuff!