flow limitations expressed as an index?

[cathyf]

TBMx I just read this through again in light of some odd things I've seen in my data this last month, and I'm quite intrigued.

One of the things that I see is that if I look at the flow rate curve, sometimes I will see long steady decays in the tops of the inspiratory peaks, that are happening over such a long period of time that they don't count as any sort of "event".

Here's one example of what I mean. First take a 45-minute view. The inspiratory flow rate starts up about 13-15 or so, and then it just slowly decays. For the last six minutes of the decline it's in the 4-5-6 range. Followed by a clear arousal and we're back to cruising along at 13-15 or so.
[attachment=33215]

Here's those 7 minutes at the long flat bottom:
[attachment=33216]

I've got EPR set to 3 and my max pressure is 15, so those flow limits have pegged the pressures to a 15 IPAP and 12 IPAP. But look at all those breaths where the mask pressure is staying at 12 instead of running up to 15 through multiple breaths in a row -- it's like the machine doesn't know when I'm inhaling!

Consider the AASM Hypopnea rule 1A

AASM Hypopnea rule 1A: A 50 % or greater fall (but less than 90%) in the nasal pressure signal excursion for at least 90% of the event duration from pre-event baseline, the duration of which lasts at least 10 seconds, and is associated with a 3% or greater desaturation or EEG arousal and continued or increased inspiratory effort throughout the entire period

• If you see that chaotic breathing at 3:10:36 as a clear arousal
  
• If you consider the baseline as ~14-15 -- which is the level at the beginning of the long, slow decay, and also it's the level that I returned to after that arousal.
  
• If you compare the SpO2 of 99 that I return to at the end of the arousal to the SpO2 of 95 that I start the arousal with (and you ignore that fact that my SpO2 is bouncing up and down in a totally-normal 96-98 throughout the whole decay  )
  
• If you note that the level of the curve tops drop below 7ish at 3:03:37 and the recovery breath starts at 3:10:36. (and 419 seconds is more than 10 seconds)
  

Then that's a mind-boggling 7-minute-long "hypopnea". (I would say that's not an "event" it's a "lifestyle" LOL). Of course the machine didn't see anything it called a hypopnea...

Ok, one thing that I do notice is that my "hypopnea" is accompanied by a small (6.00) leak. Do I need to add the leak somehow back into the flow rate curve? I thought that if the leak was below redline that the machine compensates, but maybe that compensation doesn't include some needed correction to the flow rate curve? You talk a lot about how ResMed is doing a lot of processing on that Flow Rate curve -- and I'm thinking that maybe they are distorting the data with "corrections" that aren't really correct. (This is what I think you are trying to say, too, right?)

I have incredibly severe positional apnea which I can now produce pretty much on demand, but I can also completely eliminate it with a cheap simple cervical collar. But once I get rid of the chin-tucking problem I've got these very subtle odd things going on with my breathing while asleep, and I'm not really sure if they are problems, serious problems, or just strange.

[kappa]

Cathy, I suspect this is assymetric breathing - where you are exhaling through mouth and inhaling via nose (although your profile suggests FFM so maybe not?). Minute vent is halved, leak increases, but SpO2 is largely maintained. A tighter zoom in on the flow might be informative.

[cathyf]

Ah, right -- big part of my ignorance here is that I wore a FFM for 6-1/2 years, and then discovered the amazing wonders of a cervical collar and have now been exploring a nasal mask and pillows.

Anyway, here's a five-minute view of the slide into that long period of "squashed" flow rate. I was wearing a dreamwear nasal bar mask.
[attachment=33294]
I'm still totally unsure on how to interpret the leak numbers-- the leak is in the 6-7 range which I thought was supposed to mean "acceptable". I have no idea whether the leak represents mouth leak, mask leak, or some combination -- is there some way that you can tell? Also no idea of whether the leak should somehow be subtracted from the flow rate. When people talk about flow limitations they talk about the shape of the curve, while I'm focusing on simply the magnitude of the curve. I see this definite thing in my data overall where leak means squashed-looking flow rate, and I don't know whether that's some simple measurement scaling effect ("well yeah you have to add the leaked air back into the flow rate") or it's some subtle cause and effect thing where the leak is causing the airway collapse or the airway collapse is causing the leak.

[SarcasticDave94]

The leak rate trace you'll see in OSCAR from your ResMed is unintentional leaks. This is all leaks above the ones expected.

[cathyf]

Right -- but the leak during that "squash" is about 6 -- well below leak redline. In my data I have this very strong pattern where I'll see really big flow limits during leaks -- big leaks, little leaks, in-between leaks. And very occasionally I will see massive flow limits with no accompanying leaks. What I don't understand is whether
-- the leaks are causing the flow limits
-- the flow limits are causing the leaks
-- they aren't flow limits at all, because the leaks are causing the machine to badly mis-measure the flow rate

AND I don't know if it's the same answer or different answer depending upon whether the leaks are mask leaks or mouth leaks (and I also can't tell the difference between mask leaks and mouth leaks )

As a long-time FFM user very new to nasal masks, the nasal/pillow masks just adds more layers to my confusion!

I build my OSCAR display with leaks directly below flow limits because whenever I see big FLs I immediately look down one graph and see whether there were leaks going on. But while I'm pretty sure that the leaks matter, I don't understand what the relationship is!

[clownbell]

I am not intelligent enough to participate in this discussion, but would like to point out that it has (in my view) gone too far out into the weeds. To make the point, let's change the subject a bit. I get zero obstructives, a few hypopneas and sometimes several centrals -- almost all of 10-12 seconds. A given number of centrals of a short time would be less of a problem than the same number of centrals of a longer time. The original poster was making the same point with regard to FLs. So I think this discussion is significant even though I don't understand all the technicalities.

[cathyf]

clownbell, in a much broader sense I'm wondering if you aren't on to the crux of the issue. As has been pointed out several times in this thread, sleep doctors don't care at all about flow limitations, while device manufacturers have the devices largely focused on reacting to them. A couple of months back

TBMx Wrote:
Keep in mind: the sao paulo cohort established up to 30% of the night with flowlimitations is healthy and normal. That would mean the 70-percentile needs to be different from 0.00 - If that actually would be the case this could also be seen at the first view: the pressure would be equal to CPAP and the FL-graph would be (nearly completely) black (or colored or whatever).

I've been doing this for 7 years, and I just go back and forth and around and around

-- flow limits are bad -- otherwise why would the machine chase them all night with pressure increases?
   OR
-- flow limits don't matter in and of themselves, they only matter as an indicator of impending apneas and hypopneas
   OR
-- what matters is arousals, so flow limits constantly waking you up are bad, but if you sleep right through them and they don't cause events, who cares?
   OR
-- to really beat back the flow limits I need to get a vauto on craig's list -- I've seen how much great things I can do with an EPR of 3, imagine how wonderful it will be with pressure support that I can set wherever I need it to go (unless of course I end up with a vauto and pressure support set to 2.9, LOL)

I'm just trying to figure out what is reasonable and possible to try, and whether I should do it.

[2SleepBetta]

I just dropped by to review latest posts here and saw cathyf's remark about large FL at leaks. On only that one point for now, couldn't the FL spike stem from loss of pressure due to the leak?  I may be overlooking or not remembering countering points here in this helpful thread.

[cathyf]

2SB -- what has me puzzled comes down to this:

The flow limit discussions are all about these subtle changes in the shapes of different parts of the waveforms -- M shapes, round-to-flat, lots of clever math to calculate what you can eyeball -- and there is a lot of sophisticated stuff going on. But I'm also wondering if there is a very simple factor that the algorithms ALSO count as  "flow limits" and that's just the curve getting squashed down closer to zero. I mean if you think about the plain English meanings of "flow" and "limit" it seems like a narrowing of the curve would qualify, right?

Now IF my guess about "simple narrowing" is true, then I could see that it's possible that nobody talks about it, because the whole shape of the waveform thing is subtle and fascinating, but "oh it's smaller, too" is more "duh!"

AND a simple shift down wouldn't change the shape of the curve so wouldn't factor into that discussion.

AND -- if we are talking a simple shift up/down, then it really could be some simple formula "at X amount of leak, and Y pressure, add Z to the recorded flow rate to get the actual flow rate" -- and after you add that in then maybe the curve isn't actually squashed?

All of this is me stumbling around trying to figure out if there is something interesting going on when I see the leaks and flow limits together, or it's just the way the numbers are coming out the machine. Maybe the machine just isn't making some correction(s) to the numbers that it does know how to do but the engineers decided not to bother?

[2SleepBetta]

Hello to sheepless and to all others probing what the meanings of the  Resmed FL flags are and how they can help us.

Incidental to working on another FL post I needed to use a PLD file's raw Resmed FL data. I used the file accompanying an analysis project another member gave me to work on, which I had belatedly begun  to do. 

Working with the FL in the file, various bits, pieces and approaches to summarization of FL came to mind. Nothing original. Still,  the use of percentile scores seems the best bet. I'm not "up" on the Excel percentile tool so I "wung it" another way for a moment in the tabular area of the graphic.

Note that the maximum value for FL severity (0.63) and the number of "0", zero, FL samples (2170)  are at extremes of the blue trace in the lower graphic. The other trace was truncated to keep the chart small (didnt want to show sleep session time percentage for the 2170 "0" samples).

The upper graphic and the tabular values each reflect moving averages of 10 sample points (=20 seconds-about 5 breath durations). The maximum average FL "severity" level for 10 samples was 0.568

Nevertheless, after looking, readers may have or see far better ways or may point to anything that may appeal to them in the graphic. It is a conglomeration of things--no way do I recommend it all as one package, just some different things to consider separately and one might see as helpful and feasible. 

Our tech people can set the bounds on what is thinkable for the OSCAR development team To Do list some day.

The upper graphic depicts a dramatic subset of the PLD file that is summarized variously below it.