Longest Apnea?

I was reviewing a few days worth of data today and found one that showed a 39 second apnea event. HOLY CRAP. Everything else is as normal as you can get, less than 4.0 AHI for my 3 week average, but this one data point really freaked me out.

Has anyone else had something like this? Possible causes other than an actual apnea?

JC

Every so often I have apnea's in the same range. They've definitely gotten fewer though, since tweaking my pressure.

Respironics does not provide that information... at least on the M series pro. I just got a respironics S8 autoset 2 and I now have limited data on apnea length. My longest in the last week was 21 seconds. I would venture a guess that my median length is about 13 or 14 seconds.

It might possibly be even worse than you think.

Respironics breaks up the night into 30-second reporting windows, evaluates each window separately, and indicates whether or not there was an apnea present during each of those time windows. An apnea present during one window, however, and continuing into the next window (or windows!) would presumably be reported as multiple separate apneas, rather than as a single long apnea. As already pointed out, Respironics software does not provide information on the duration of individual apnea events. (It does provide information on total apnea duration for a night, but that information has been shown not to be reliable.)

ResMed does a much better job, reporting the duration of each apnea event. Well, almost. ResMed breaks up the night into 60-second reporting windows, evaluates each window separately, and indicates the presence and duration of an apnea within each of those time windows. But if there are multiple apneas during a single window period (in theory there could be as many as 5; I have seen 3, and 2 are not uncommon), ResMed's graph shows only ONE apnea for that minute, and I believe that they display only the most recent, rather than necessarily the longest. And, as with Respironics, if an apnea begins during one window and continues (for at least ten seconds) into the next, it would again presumably be reported as multiple separate apneas, rather than as a single long apnea.

For either machine, an apnea of 18 seconds duration, 9 seconds in one window and the other 9 in the next window, would not be reported as an apnea at all!

Still, we should be thankful for whatever information at all they deign to give us.

All of mine, and there aren't very many with the new ResMed are 10-13 seconds so far, except for the one I'm talking about....seems out of place, almost like there is another reason....

JC

I wonder if the strange ways the data is handled is related to data compression due to the limited memory devices used in these machines.

flylow wrote:I wonder if the strange ways the data is handled is related to data compression due to the limited memory devices used in these machines.

Strange perhaps when we focus on the "simple task" of counting the number of events and recording the duration of specific events.

Not so strange when we think of all the other things they are also trying to do simultaneously. At the simplest level, if reporting is based on a "timeline" paradigm, what one unique "time" should be assigned to a "single event" which has an extended duration?

Perhaps easier to visualize is the task of dealing with pressure and leak, which in essence are varying "constantly". Unless you want a theoretically infinite number of datapoints, some "granularity" of time has to be accepted. Every second? While that sounds good, it could result in a very "spiky" tracing, with every insignificant momentary change recorded. Not terribly useful. And as a practical matter, how many datapoints will fit and be visually distinguishable and meaningful on an 8.5x11 paper printout of a line graph? Practical considerations pretty much require that the manufacturers do what they in fact do: select a specific "time window" (30 seconds for Respironics, 60 seconds for ResMed) and present some sort of "summary" for all the data in that window (mean for Respironics, median for ResMed - we think - and we don't know the "granularity" of the individual data measurements for which they present their measure of central tendency).

Of course, they could employ a "separate system" for dealing with pressure/leak data, as opposed to respiratory event data. That would increase complexity, and cost, and would likely lead those of us who like to dip into the "raw data" to complain about THAT decision. Can't please everybody. Hard enough to please anybody. Still, their software does give us a reasonable quantity and quality of data, for most users, to meet most needs, and to make our therapy status understandable.

Data divers like me are always going to find something we wish were done differently. But at the same time, EVERYONE should be aware of the characteristics and limitations of the data they ARE working with.

jrcmlc wrote:... and found one that showed a 39 second apnea event. ... Has anyone else had something like this? ...

Besides possible issues with the way the machine records the event, it is possible you had a long apnea. They happen. The thing is to not have them happen frequently. My apneas happened so frequently and for so long that my machine assumed I had taken off my mask! Time after time it would turn off during the night. Not a good sign! But then my nickname is "Abby, Abby Someone, Abby Normal, I believe!"

JohnBFisher wrote: Besides possible issues with the way the machine records the event, it is possible you had a long apnea.

Just to be clear that I'm not being misunderstood:
My point is that your actual apneas may, under certain conditions, be of LONGER duration than they are reported by ResMed.
But if ResMed says that an apnea (according to its definitions) lasted for 39 seconds, then that is AT LEAST how long it lasted.

Yikes.

JC

Gee, I've learned something new tonight. I've always been uncomfortable with Time in Apnea Per Day being what I feel is a diluted number, but I didn't have a clue about the possibility of lengthy apneas not being reported because of ocurring in different reporting windows. With 16 windows during an 8 hour night, seems reasonable to expect there are crossover events. Now I'm wondering what's out there where PSG data has been "interpreted" by machine standards looking for such things as crossover events and comparing PSG count and duration to what a machine would have reported. Hmmmm.

Velbor wrote:

JohnBFisher wrote: Besides possible issues with the way the machine records the event, it is possible you had a long apnea.

Just to be clear that I'm not being misunderstood:
My point is that your actual apneas may, under certain conditions, be of LONGER duration than they are reported by ResMed.
But if ResMed says that an apnea (according to its definitions) lasted for 39 seconds, then that is AT LEAST how long it lasted.

Perhaps this is why respironics does not report the length of apneas. I understand that design choices need to be made. It is hard to understand the computing abilities of these machines but memory should not be a significant limitation with the memory capabilities of the memory cards available. It shouldn't but it is. I suspect, in time, the computing and memory capabilities of the machines will improve. There are data compression methodoligies that do not require segmenting in time. They look for data exceptions rather than continuous recording of the same number.

Mine.. are generally 10-11 sec, but I have had a 32+ sec one. Since my AHI was under 4, I did not think anything about it. Maybe I was extra tired, extra sinus congestion, stress. There are many things that could had contributed to it. It does not happen often, but it does happen. My usual AHI is 3 +/-, so I cannot complain!

kteague wrote: .... With 16 windows during an 8 hour night, seems reasonable to expect there are crossover events. ....

Hi, Kathy. The time windows are 30 or 60 SECONDS wide, not minutes, so an 8 hour night provides 480 (ResMed) or 960 (Respironics) distinct reporting periods. Neither of these are bad numbers. Too many and your data becomes erratic, reflecting transient occurrences. Too few and your data suffers from imprecision. The manufacturers haven't done a bad job engineering this aspect of their machines.

Respironics could not handle reporting an apnea of 39 seconds, since their data window is only 30 seconds (and in that sense, their window is "too short"). They do not (and I intentionally avoid using the word "therefore") report individual apnea durations at all. Their "total time in apnea" report suffers from at least two issues: (1) they report total apnea time only to the nearest 6 seconds, and (2) it appears that their "day" is not reflective of usage but is defined by the 24-hour period of their internal clock which is set to Greenwich Mean Time. Even if this figure were accurate, there's little way to distinguish betweeen lots of middling-duration apneas with minimal PO2 drops, versus mostly short duration apneas interspersed by a few whoppers with major oxygen desaturations.

The problems regarding apnea duration reporting with ResMed pop up in two situations: (1) if the apnea is near the beginning or near the end of a window, you may lose the continuous time which occurs in an adjacent window - "clusters" of adjacent apneas might raise suspicions (and if less than 10 seconds long, the "adjacent" apnea time is lost completely), and (2) if there was more than one apnea within the 60 second window, only the last of them gets graphed (i.e., if there was an apnea of 39 seconds, a break, and then an apnea of 11 seconds, only the later 11 second apnea is graphed).

In any system, there is some information which you get, and some information which you don't get. Much depends on what you want to do with the data. Encore and ResScan software both do a nice job, as far as they go. Neither can fully satisfy a data nerd like me.

A few nights ago I had an apnea that lasted 48 seconds, and it was amongst a cluster of others of varying times. Can't seem to get rid of those clusters. Probably during REM time, I guess.