Resmed VPAP Adapt SV - for Central Sleep Apnea

[-SWS]

The patent text clearly says that if the airflow signal is of "poor quality" then a predetermined fixed rate of ventilatory assistance will be provided (i.e. that "permeating 15" back up rate) .

My PFTs indicate inflexibility in my lungs. Just putting aside the Swift question for a moment, might there be an influence from the quality of my respiratory response?

Frequen, I don't know how inflexible lungs might change a patient airflow signal. However, I also suspect that inflexible lungs (COPD, emphysema, etc) do not deteriorate the quality of the transduced signal itself.

Rather, if an etiologic nuance caused a back up rate of 15 to "permeate", then I suspect it would have to do with the algorithm being unable to characterize your breathing pattern. Signal quality has to do with instantaneous sampling of airflow in this case. The ASV algorithm also assigns weighted rules associated with common variants of central dysregulation. That implies temporal analytic methods toward breathing disorder characterization.

Here is pure conjecture on my part about what any association-based algorithm might do when an atypical signal is presented:

1) attempt to associate or "best fit" that patient breathing-pattern regardless of unrecognition, in which case an associated pressure/frequency response will be used. Here a mis-associated pressure/frequency response might cause improvement, no change whatsoever, or even cause greater dysregulation.

-or-

2) make no attempt to associate the uncharacteristic signal pattern, but rely on a predetermined response for such atypical presentations.

Speculative scenario two above might lend an uncharacteristic patient that permeating and "off-the-shelf" backup rate of 15 as well---at least in my own way of analyzing.

[christinequilts]

By contrast Respironics samples, calculates, and targets just the top-most instantaneous data point on the same flow curve: hence "peak flow". SAG had mentioned this earlier, and I agree. If you sample and deliver one "peak flow" data point per inspiratory cycle, then that same patient flow-curve slope and thus "minute volume" (area under the curve) pretty much fall in line. From my perspective these two flow-curve data-point targeting methods can accomplish the same thing regarding how much ventilation occurs on each breath.

But minute volume by definition accounts for RR- volume of air which can be inhaled (inhaled minute volume) or exhaled (exhaled minute volume) from a person's lungs in one minute; calculated by taking the tidal volume and multiplying the respiratory rate.

I can't find nice neat definition for Peak Flow as quickly, but it doesn't take into account the RR, which is sounds like is what you're seeing too:

But another difference between these two machines is that the Respironics Auto SV employs a more traditional time-based back-up implementation rather than attempting to directly increase or decrease respiration rate (that variable "F" which Resmed's ASV algorithm manipulates). ... Working with peak flow versus minute volume, in and of itself, probably results in no discernable functional difference regarding resultant ventilation.

Wider mask options sounded good to start, but I couldn't imagine going back to a more rigid backup rate. If its backup rate is closer to ST then the Adapt, then you run into keeping it high enough to provide enough support without being so high it gets in the way, which can destabilize breathing fairly significantly. It sounds like for now, the backup rate issue may be the key thing to keep an eye on.

[-SWS]

But minute volume by definition accounts for RR- volume of air which can be inhaled (inhaled minute volume) or exhaled (exhaled minute volume) from a person's lungs in one minute; calculated by taking the tidal volume and multiplying the respiratory rate.

Unlike Resmed, the Respironics Auto SV uses its flow-curve data samples only to calculate pressure support and a time-based back up. Respironics makes no direct attempt to speed up or slow down RR. That Respironics machine truly doesn't need multiple data-point samples along that flow curve for its technical objectives. These two machines are instantaneously sampling the very same flow curve.

The Respironics case is a task of patient-flow sampling and calculations being used in a machine-based pressure-delivery feedback loop. And for that task, working with multiple points along the way (minute volume) makes no sense. Rather, using one peak point (peak flow) to determine variable PS is just as efficient as utilizing multiple points on that same curve (minute ventilation).

Algorithmically attempt to directly regulate patient RR by moving machine F back and forth as Resmed does, and you will benefit with the algorithmic technique of multiple data-point sampling of minute ventilation. But Respironics' simpler technical objective of calculating only machine PS (of those two variables) leaves it the viable option of sampling only that top-most flow-related data point. And that peak-only strategy frees up real-time system processing cycles to accomplish whatever else needs to algorithmically occur within that Respironics algorithm. If all you're doing is spitting out PS (and not a machine-based F) it makes absolutely no sense to work with minute volume.

Resmed needs its PS min of 3 to move machine-based F back and forth, in an attempt to directly regulate patient RR. Respironics does not attempt to directly regulate patient RR (by moving F back and forth), and they can thus offer CPAP+PS mode. When the Respironics machine sits in CPAP mode for long periods, it has no machine-based cycles with which to even attempt to directly regulate patient RR. You must have a moveable F to attempt RR changes. And to have an F, you must have an alternating PS.

[dsm]

But minute volume by definition accounts for RR- volume of air which can be inhaled (inhaled minute volume) or exhaled (exhaled minute volume) from a person's lungs in one minute; calculated by taking the tidal volume and multiplying the respiratory rate.

Unlike Resmed, the Respironics Auto SV uses its flow-curve data samples only to calculate pressure support and a time-based back up. Respironics makes no direct attempt to speed up or slow down RR. That Respironics machine truly doen't need multiple data-point samples along that flow curve for its technical objectives. These two machines are instantaneously sampling the very same flow curve.

The Respironics case is a task of patient-flow sampling and calculations being used in a machine-based pressure-delivery feedback loop. And for that task, sampling multiple points along the way (minute volume) makes no sense. Rather, sampling one peak point (peak flow) to determine variable PS is just as efficient as sampling multiple points on that same curve (minute ventilation).

Algorithmically attempt to speed patient RR by moving machine F back and forth as Resmed does, and you will benefit with the algorithmic technique of multiple data-point sampling of minute ventilation. But Respironics' simpler technical objective of calculating only machine PS (of those two variables) leaves it the viable option of sampling only that top-most flow-related data point. And that peak-only sampling strategy frees up real-time system processing cycles to accomplish whatever else needs to algorithmically occur within that Respironics algorithm. If all you're doing is spitting out PS (and not a machine-based F) it makes absolutely no sense to work with minute volume.

[-SWS]

Doug, Resmed needs more instantaneous samples per respiratory cycle to pull off their technical objectives. That's also what makes their algorithm so susceptible to leaks. But since Resmed needs more instantaneous samples, they require that smaller diameter pressure sensor line as well.

Cool experiment: think of drawing fluid half way up a straw. Now make that fluid quickly flutter up and down. You certainly couldn't get such sensitive (instantaneous) fluttering with a wide-diameter CPAP hose. That fluid-in-a-straw exercise is analogous to the sensitivity Resmed's approach actually requires regarding instantaneous change.

The Respironics approach doesn't require that degree of sensitivity regarding instantaneous change. That, in turn, is why the Respironics machine doesn't have a small-diameter hose. However, being much less reliant on instantaneous change is also why the Respironics machine doesn't suffer leak problems as easily as Resmed.

I don't see the Respironics machine doing "guess work at the mask" Doug. I just see it not requiring significant instantaneous sensitivity because the algorithmic approach is so very different.

[StillAnotherGuest]

Overall, using Peak Flow to provide ASV shouldn't make a difference in providing for ventilation, and consequently Minute Ventilation, as long as one provides for the inspiratory time component. If AutoSV is sending in a breath at 30 L/M, and does so for say, 1 second, there's your 500 ml tidal volume. And there is a Tinsp in there.

There looks to be two concepts here of importance. The first concerns how it deals with flow reductions (or loosely, hypopneas) where it will support with a percentage of peak flow, and second, how it deals with apneas, where yes, it will have to come up with a rate. And there's no way you could build a machine that would rely exclusively on the patient rate, cause if you're saying you're gonna fix CSR, but ventilate at say 90% of the patient's flow at that rate, you are pretty much doomed.

If one is free of SBD, it looks like the first timed breath is delivered 8 seconds after the end of the last inspiration, and subsequent timed breaths are delivered after 4 seconds plus the average breath period from the beginning of the last inspiration.

If one is having SBD events, the first timed breath is delivered within 4 seconds of the last inspiration.

LOL! Hey, guess what that respiratory rate will be?

And again, I think a big concern will be how it deals deal with periods of variable breathing, such as might occur in Wake and Wake/Stage 1 transitions. This thing might be just as "pushy" as AdaptSV. A key component will be how long it takes to get out of ASV Mode.

Of interest is the concern Respironics had/has with Variable Breathing, which occurs not only during the above periods, but also during phasic REM, an example of which is back here, and one the reasons they put in the Variable Breathing component.
SAG

[christinequilts]

If one is having SBD events, the first timed breath is delivered within 4 seconds of the last inspiration.

LOL! Hey, guess what that respiratory rate will be?

Hmmm, 4 goes into 60 15 times, so I'll take a wild guess and say the RR will be 15!

15 seems to be the magic number lately...we need to come up with some more so we can buy lotto tickets Maybe we could take AHI, % of time in each sleep stage & what else?

But what type of backup rate is it? Rigidly set, like with BiPAP ST or more of a 'failsafe backup rate', as with the Adapt?

[christinequilts]

Here is the wording from ResMed's sample letter of med necessity for the Adapt:

Significant improvement of the sleep-associated hypoventilation was seen with the use of the ASV mode which automatically adjust inspiratory pressure (IPAP) and the timed backup to effectively support the patient's respiratory efforts during central apnea/hypopnea periods.

ResMed 2007 Reimbursement Handbook pdf

Further supports adjustable backup rate theory...and I knew you would get a kick out of them mentioning central hypopneas specifically after last falls debate.

[-SWS]

But what type of backup rate is it? Rigidly set, like with BiPAP ST or more of a 'failsafe backup rate', as with the Adapt

Three possible configuration modes:

1) Back up Rate Off (spontaneous breath triggering only),

2) Back Up Rate: 4 - 30 (set minimum number of BPM) coupled with
Time Insp: .5 - 3 (set duration of inspiration time)

3) Back Up Rate on Auto (short-term time-based computation of the above two parameters are automatically maintained)

You can see that third choice comes the closest to approaching what Resmed does with recent-average-based F. However, Respironics option three above is neither association-rule driven nor fuzzy-logic based. Respironics decides simply to provide a back up breath based only on functions of time. The Resmed algorithm endeavors to stabilize central regulation by moving machine-based F back and forth (driven as a function of disorder types requiring unique "F adjustments"). When Respironics accomplishes the same end result of central stabilization, it does so as a result of variable PS combined with that indirect by-product of a "time-only" derived back up rate. Resmed does so with that same variable PS. However, Resmed maintains the underlying premise and thus calculates to actively affect machine-based F, with the intended purpose of directly maintaining RR (not just as an incidental and beneficial by-product of back up rate).

Additionally the Resmed algorithm may occasionally opt to place that machine-delivered variable F aside, in favor of simply delivering a fixed back up rate of 15 (should all else fail). Either machine is guaranteed to miss the mark (regarding back up or F) for different real-world reasons. Regardless, there should also be plenty of functional overlapping regarding efficacious back up being provided to the CSA/CSR patient population as a whole. At least that's my take.

Looking for ways in which these two machine algorithms might create different patient efficacy patterns is like shooting fish in a barrel. These two machines are absolutely positively guaranteed to produce some efficacy differences in the real world, based on such radical differences in algorithmic approach as those examples recently discussed.

[-SWS]

...and one the reasons they put in the Variable Breathing component.

SAG, we've reviewed that variable breathing controller in both the RemStar Auto and the BiPAP Auto. However, those machines are both auto-adjusting/auto-titrating platforms targeted for OSA. The variable breathing controller in each of those two machines thus exists to refrain from treating central dysregulation, or at the very least to keep from worsening incidental dysregulation in OSA patients.

By contrast the Auto SV is supposed to actively treat central dysregulation via that per-breath timely variation of PS. That is not at all what the OSA-based BiPAP Auto above does with its gradually-changing PS.

Did you per chance run across an OSA-based and thus equivalent variable breathing controller in the Auto SV algorithm?

[StillAnotherGuest]

SAG, we've reviewed that variable breathing controller in both the RemStar Auto and the BiPAP Auto. However, those machines are both auto-adjusting/auto-titrating platforms targeted for OSA. The variable breathing controller in each of those two machines thus exists to refrain from treating central dysregulation, or at the very least to keep from worsening incidental dysregulation in OSA patients.

I was of the belief that the Variable Breathing Control Layer in the AutoCPAP and BiPAP was concerned with limiting the response of the machine to erratic breathing associated with Wake, REM and periods of distress. And besides, if you were having central events, the machine would already be responding using the Apnea-Hypopnea Control Layer (you'd never make it to the Variable Breathing Layer), and you would hope that the NR part of the algorithm would bail you out.

Did you per chance run across an OSA-based and thus equivalent variable breathing controller in the Auto SV algorithm?

In looking at the patent, the Variable Breathing Control Layer doesn't look equivalent, it looks identical. Except now instead of being OSA-based and limiting activity, they seem to have turned around and made it centrally-based and are using it to identify Periodic Breathing, call that an event and react.

Now that's great, but that's a sleep event, and again, I don't think that's the issue. What I want to see is if this thing is going to respond to the erratic breathing in the Wake and Wake/1 transition that gives people problems settling in with ASV technology. And if they're saying that erratic breathing is now fair game for pressure attack, I'd be keeping a critical eye out for that.

Significant improvement of the sleep-associated hypoventilation was seen with the use of the ASV mode which automatically adjust inspiratory pressure (IPAP) and the timed backup to effectively support the patient's respiratory efforts during central apnea/hypopnea periods.

Further supports adjustable backup rate theory...and I knew you would get a kick out of them mentioning central hypopneas specifically after last falls debate.

Yeah, I know, and even the reimbursement for CompSAS says

Complex sleep apnea (CompSA) is a form of central apnea specifically identified by the persistence or emergence of central apneas or hypopneas upon exposure to CPAP or an E0470 device when obstructive events have disappeared.

People talk about unicorns, too, but you don't actually see any of them, either.
SAG

[Lubman]

Well, to echo SWS - ya don't keep up on this thread daily, you miss a lot.
I think CQ (who has completed all she needs for a PhD from my way of thinking)

Intellectual Property (e.g. patents and trade secrets) are the best assumption of why these 2 products use different triggers, and to quote SAG, regardless if you like one technology versus the other - it's the results for a given patient that is key.

Now let me ask a question.
We have dissected FQ's PSG data -- let me ask a question based on my own split night PSG

My CSDB has no Stage IV and virtually no Stage III or REM sleep stages.
For the initial test, I was baselines without the Adapt and had severe oxygen desaturation down into the 70% range.

If the Adapt algorithm was not quite the answer for me, without EERS.
Why the ability of the ResMed algorithm to cope with my arousals better
by adding CO2 under the MD controlled conditions of EERS? (No taping -- notice I said controlled)

Also, if we assume leaks are more critical to the ResMed algorithm, that helps explain
a) why the machine only recommends certain masks, because the feedback loop from the pressure sensor tubing, has been "corrected" or "compensated" for only a few mask models. Maybe the effort to consider other models, where the volume and exhaust hole placement, couldn't be guaranteed in future models of competitors was a factor. Maybe it simply sells more of one's own mask, if a manufacturer believes they make a superior performing product as compared to the competition ... )I'm not saying they do or do not.

b) And with a well fitted mask, and a patient in the lateral position (where leaks seem more likely to occur with a moderately well fitted mask that was adjusted to a sitting or supline oriented person), would be more critical to proper machine operation.

For me, I think the mask size, fit and headgear design are key. The over the head feature of the NV Mirage FF mask, Series 2 gives me a better fit.

It's as if the ASV solution works when the environment is as close to ideal as possible. E.g., the mask volume is known, the vent holes are placed for the proper washout (or controlled washout) of CO2, and it doesn't leak or the leak rate is constant where the machine can compensate or partially compensate for the leak

There are other specifics to be gleaned from the past few days of posts, but
in general, the key differences are beginning to be understood.

Lubman

[-SWS]

People talk about unicorns, too, but you don't actually see any of them, either.

Well, you can't find unicorns if you're looking for a stampede. Similarly you can't find central hypopneas if you're looking only above the frank threshold of central cascade (a.k.a. "the stampede").

You just need to know where to look for both. Those unicorns are hiding in the metaphysical realm, and the central hypopneas are hiding below the cascade.

[-SWS]

What I want to see is if this thing is going to respond to the erratic breathing in the Wake and Wake/1 transition that gives people problems settling in with ASV technology.

I noticed that Resmed recommends trying a twenty-minute wide-awake acclimation period:

In some subjects, the period of unstable breathing immediately after sleep onset could be eliminated by asking the subject to remain awake for 20 minutes after applying the mask and before going to sleeps in order to minimize overbreathing immediately prior to sleep onset.

Wonder how well or poorly that countering technique works. Any empirical runs on that one?

[StillAnotherGuest]

Well, you can't find unicorns if you're looking for a stampede.

Hmmm. I can't see that one making Bartlett's.

However, it's easy to find a unicorn in a stampede. Look right behind the horse going "OOH! AAH! OUCH! STOP!"

In some subjects, the period of unstable breathing immediately after sleep onset could be eliminated by asking the subject to remain awake for 20 minutes after applying the mask and before going to sleeps in order to minimize overbreathing immediately prior to sleep onset.

Wonder how well or poorly that countering technique works. Any empirical runs on that one?

I think that ends up only applying to a subset of a subset (patient asynchrony (PA) immediately prior to sleep-onset plus the tendency for sleep-onset events). I would look for a patient with an inappropriately high Minute Ventilation (MV) (and hence a high Target Ventilation), indicating that it's in the midst of a runaway (PA) and say, "Stop that!" (to the patient)(probably to the machine, too). Using judicious amounts of PS might help to reduce that. Or in a bind, a settings change will reset Target (not a real practical solution, for the most part).

It would not apply to Wake/1 transition period and/or the patients with poor sleep efficiencies.

Looking at the daily trend values for MV might help to identify those patients who have sustained PA, and consequently, "good nights" vs "bad nights".

If the Adapt algorithm was not quite the answer for me, without EERS.
Why the ability of the ResMed algorithm to cope with my arousals better by adding CO2 under the MD controlled conditions of EERS?

Did you try EERS alone? It strikes me that there is a point where any kind of Pressure Support and CO2 enrichment become counterproductive.

We have dissected FQ's PSG data...

There's still a couple of things in there. One area that I think needs closer scrutiny is the part that says

In the original study report, this what I was referring to when I said "cyclic":

She has periodic breathing during unable NREM on CPAP.

This was added in the addendum, so I'm thinking that was kinda subtle, and we should really toss that under the microscope.

When you're getting more stuff there, FQ, can you get a print-out of that area (60 second epochs, plus 300 second epochs to get a bird's eye view, with accompanying histogram to see the location?

Since this area was termed "periodic breathing", and not "central apnea", "central hypopnea" or even "central unicorns", I would be curious about the presence of desaturations and/or arousals.
Cause a unicorn without a horn is a horse.
SAG