What is going on???

[derek]

As most of you know, I have written software to keep long-term performance data derived from the Encore Pro database. One of the more useful set of plots shows how the occurence of sleep events depends on APAP pressure. In particular I have noticed a surprising trend in my AHI vs. pressure plot as the pressure increases:



What bothers me, of course, is the increase in both hypopneas (pink) and apneas (red) at the higher pressures. Does anybody have a suggestion as to what is going on? Why do apneas appear at 10 cm? I don't think these are centrals, because those should be registered as non-responsive.

The above plot is based on just under 500 hours of APAP (REMStar Auto) use. The minimum pressure settings have been in the range 7 - 9 cm H2O (a very few nights at 6 cm). The maximum pressure has been 14 cm, but it has rarely hit that. The other indices (snore and FLI) show a continuing decrease with pressure. There are fewer hours spent at the higher pressures, with resulting lower statistical reliability, nevertheless the trend is obvious.

When I look at the Encore Pro daily events plots it does not seem that the apneas/hypopneas are driving the pressure increases, they simply occur while the pressures are high (from snores or whatever).

Beats me

derek

[FL andy]

Derek,

You are trying to explain changes in a *very, very narrow range* of increases in AHI (2.7) which occur at the extreme *low end* of the AHI scale.

Personally, I would be thrilled to have your AHI results.

To answer you question though, the changes in that small a range could be due to outside sources causing a *needed* rise in pressure, such as: degree of cleanliness of air filters, degree of stuffiness of the nares, etc. (I wonder if barometric pressure has any affect on required xpap pressure - especially a range that narrow?)

However, since I have never used an AutoPap, my comments are not worth much. But I will repeat: I would be thrilled to have your problem.

Hope others have more sound advice than this.

Andy

[Mikesus]

Thinking about it, I am wondering if you are looking at the data correctly.

I.E. One night you have a bad night, and your pressure stays fairly high the entire night, and your AHI is high. When you take that with a night that you did very well, you might have the opposite. And when you look at the data as a whole, you would see a spike on the end as you are seeing...

Another possibility is the AHI at a pressure is not weighted correctly. IOW, you might have only spent a few minutes at the higher pressures with a higher AHI, but the rest of the time was normal.

[wading thru the muck!]

Derek,

One thing to keep in mind is that the Encore Pro is only reporting the "events" that were not resolved by the machine. If you could compare the total events that were occuring you may find that the machine is at a higher pressure because more events are occuring, thus the need for the higher pressure. If you were able to graph the unrelved events as a percentage of the total events at a given pressure it may give you a horizontal line. eg; 10 total events at 7cm, 1 unresolved (10% unresolved), 27 total events at 13cm, 2.7 unresolved (10% unresolved).

In the big picture, I agree with FLandy, you are talking, all in all, within a very narrow and acceptable range.

[FL andy]

One more thought, my friend.

Would it require more pressure to move humid air than dry air? Do you use a heated humidifier? Do you change settings? Does the same setting on a humidifier send the same amount of moisture thru the hose on both low and high amounts of humidity in your bedroom (dry day outside vs. rain)?

Andy

[derek]

Mike:
I compute the indices as:
(total events at pressure)/(total hours at pressure)
over all the days in the database. One of the problems is that waking hours (since Encore Pro cannot tell sleep time) have to be included, and these will generally be at the minimum pressure - thus lowering the apparent AHI at that pressure.

My nightly pressure fluctuations are relatively constant, hovering around the minimum pressure then taking a ride up to 12- 13 cm for about 30 minutes with no apparent cause and then back to close to the minimum again. The apneas all seem to occur while the pressure is up. Perhaps the APAP is recognizing breathing patterns that are pre-apneic, is raising the pressure and then the events take place (and perhaps the frequency would have been much higher if the pressure had not been raised). So possibly the APAP is doing its job.

Wader,
I don't know how I can get a handle on things that didn't happen
Sure all of the recorded events were not resolved by the pressure. I really don't understand what Respironics mean by "Non-Resolved Apneas and Hypopneas (NRAH)" since all of the recorded events were unresolved. I have no NRAH's according to the database.

Andy,
Haven't touched the humidifier - it's set on 2.

[Mikesus]

It is possible that you are having clusters of events that the program of the Remstar can't address, therefore the Apnea's and Hypopneas are slipping thru. As the machine is thinking that there is some resolution, i.e. noticed a fl then a hypopnea, raised the pressure, now it is gone, it would not score as a non responsive. From what I understand, NRA is respironics way of scoring centrals.

[chrisp]

Hi,
Humid air is less dense than dry air. That is why clouds are in the sky. They are lighter than the air they are in. Atmospheric pressure isn't a factor as the units have altltude compensation. I myself would say that spring has arrived as your results are reflecting increased HI. Take a sudafed.

Thats my story and I'm sticking to it.

Chris

[derek]

Chris,
Maybe it's spring in Texas, but not in New England - still have snow on the ground
BTW - I do use Sudafed, I find it keeps me awake unless I take the nighttime stuff (with antihistamine).

[chrisp]

We had a freeze last night . The misquette trees haven't bloomed. Can't fool them. I'm going flying. See yall later.

[littlefeat]

Derek--

My theory on your results would be that, in fact, after a certain point, higher pressures may increase apnea and hypopnea occurrence. One way that this might occur is that, at higher machine pressures, increased leak actually causes lowered pressure in your airway (kinda like blow-by). This is why I am trying to factor leak into my analysis of my own titration over time.

Best--

[-SWS]

Pure speculation for the sake of entertainment only : There is a chance that some of those are pressure-induced central events, Derek. Recall that central events will only be logged as "NR" events if they happen to remain outstanding after three unsuccessful pressure increment attempts. However, there is a pressure threshold for most patients at which central events will begin to induce because of xPAP pressure. Above that threshold the central events will increase in numbers for any given patients. However, at that pressure threshold the central events will be fewer in number.

One possibility is that at that pressure threshold the respiratory drive itself will quickly correct or normalize those central events. That the central events are both pressure induced and incipient (or not much beyond incipient)--- but that the fault tolerant human body is able to efficiently bring those pressure induced central events in check. At higher-than-threshold pressures, the respiratory drive might take much longer to bring at least some of those pressure-induced central events in check. But at lower pressures the scant few central events may disappear almost as quickly as they cropped up---- with no credit due the RemStar Auto's algorithm and all the credit due a fault tolerant respiratory drive that tries its best to efficiently normalize pressure-induced central events.

Quite a bit of conjecture there! But if these are central events that the human body can normalize, they would not meet the Remstar's criteria for being logged as NR events.

[-SWS]

I wish I had really worded my above post as two rhetorical albeit simple questions for anyone who finds this topic as fascinating as I do:

1) Do you think the human body is ever capable of resolving central apneas or hypopneas before the Remstar Auto has a chance to administer three pressure increments (criteria for logging an event as NR)?

2) If so, how do you think those self-resolved central events of such short duration would be scored on the Encore Pro reports?

We know that during a PSG titration pressure-induced central events begin to manifest for most patients at a certain threshold. We also know the human body resolves those pressure-induced central events even if that fixed pressure is maintained. The human respiratory drive doesn't indefinitely cease to breath, rather it normalizes and eventually initiates a breath. So the big burning question with respect to Encore Pro is where/how those self-resolved central events get scored if they happen to go away before three pressure increments occur. My hunch is that they get scored as ordinary apneas and hypopneas for lack of algorithmic differentiation.

[derek]

-SWS
Great thoughts! I naturally wondered about centrals, but thought that they occurred at much higher pressures. I just dug into the database to look at the duration of the apneas. All I can find is the total seconds in apnea per session, but dividing that by the number of apneas gives 12 seconds average (except for one session that had 29 seconds!) So these are all relatively short - probably much too short for the REMStar to correct with its three bumps and a drop in pressure. We saw clearly that on loonlvr's data...

Funny thing you should mention centrals, maybe they run in the family. I have a twin sister, and as young kids we shared the same room. I have always clearly remembered that she would simply stop breathing during sleep and then resume many seconds later. She has had chronic fatigue for about 12 years now, and I have been urging her to get a sleep test. Her apneas were extremely gentle, no snoring or spluttering, so I surmise they might have been central.

[-SWS]

Derek, I just thought of something that corroborates our suspicions that Encore Pro does not register all central events as NR. When I used my ResmStar Auto (without C-Flex) for a couple years I never registered any NR events whatsoever. Yet on the 420e I do register occasional central apneas. The 420e algorithm deduces with a specificity of 100% and sensitivity of 62% that central apneas occurred if cardiac oscillations have been detected. When cardiac oscillations are detected by the 420e the patient's throat is open and acting as a cardiac oscillation wave guide. The event is then detected and scored as a central apnea. If no cardiac oscillation is detected the 420e assumes a closed throat condition is present and the event was therefore obstructive. My scant few central apneas apparently register on the Remstar Auto as ordinary apnea/hypopnea events for lack of algorithmic differentiation.

I agree about your sister needing a study, Derek! I hope you can talk her into having one.