Less Leaks and Higher AHI - what's up with that?

Hi, I've noticed that my Resmed AHI scores were higher than usual the two times I made my mask more snug, and thus had a lower leakage score. This would seem to be counter to expectation, but here's my theory for why it happens. Let's say that I have 16 apnea events per night without the CPAP. If I adjusted the mask to the point where it had minimal or no leaks, then it might catch and correct 12 of those events, and not be able to correct 4 of them. So my AHI would show the next day as "4", as the AHI is a record of events it recognized but is not able to correct.

On the other hand, if I have a leak, then the CPAP would recognize fewer of the CPAP events, right? Therefore it may catch and correct only 6 of them, and not be able to correct 2 of them, assuming the ratio is the same as above. So in this case my AHI is "2", which gives me the false impression that I did better in this scenario that the prior one where I got a "4", whereas in fact twice as many events were caught and corrected when the mask had a lower leakage.

Has anyone else come to this conclusion? I do recall I felt the very best the day after the night before had a very low mask leak score, but my AHI was higher than normal. So that got me thinking that the AHI is not really the operative number to be aware of in my situation, because there are variable amounts of leaking from night to night. It must be the "hidden number", which is the events caught and corrected.

grifho wrote:So my AHI would show the next day as "4", as the AHI is a record of events it recognized but is not able to correct.

that's not how AHI works. wiki/index.php/AHI

grifho wrote:On the other hand, if I have a leak, then the CPAP would recognize fewer of the CPAP events, right? Therefore it may catch and correct only 6 of them, and not be able to correct 2 of them, assuming the ratio is the same as above. So in this case my AHI is "2", which gives me the false impression that I did better in this scenario that the prior one where I got a "4", whereas in fact twice as many events were caught and corrected when the mask had a lower leakage.

only if you had really really really horrible leaks. otherwise the leaks have no affect on the machines ability to recognize and treat events.

download and post data.

grifho wrote:which is the events caught and corrected.

there's no such thing as "event caught and corrected, sorry. an event is an event. there's no correcting it. events that are prevented aren't events... they're just hypothetical.

Hi palerider, thanks for the response. I was ready to be shown where I'm wrong, but I don't see how anything you say invalidates my theory.

First off, I read your link, but I was aware of what AHI was already. So where was I wrong? I merely said that my AHI in my theoretical situation would be 4, which would be the number of events/hour that the CPAP recognized but was not able to correct. How is that 'not how AHI works?' it's only the uncorrected ones that represent the AHI score, right? If it got corrected, it's a non-event.

Secondly, you say 'the leaks have no affect on the machine's ability to recognize and treat events" Wow, that contradicts everything I've heard, including from my sleep doctor. I don't even have a particularly dramatic mask leak, with an average of maybe 12L/minute, yet my sleep doctor and the resmed 'myair' site sends me emails telling me to fix the leak for best results. Besides, it seems only natural that a leak in a closed system like CPAP would compromise the ability to both recognize and treat events.

As for your contention that events that corrected are not events, but are 'hypothetical'. Well, if I generally have 16 events without the CPAP and only 4 with the CPAP, there's nothing hypothetical about that improvement. One can infer that there were therefore 12 caught and corrected.

So, I still stand by my contention that a higher mask leak can lead to a lower recorded AHI from the machine, when in fact a lower mask leak corrects more events. Again i may be wrong, but waiting for a compelling argument to tell me so. Meanwhile, the numbers I'm getting are confirming it. Well, here they are for the last 6 days, with first number being mask leak and second the AHI. See how the numbers have an inverse relationship, at least in regards to the extremes?

05 8.0
14 4.3
12 4.7
14 5.0
10 3.4
04 6.9

grifho wrote:Hi palerider, thanks for the response. I was ready to be shown where I'm wrong, but I don't see how anything you say invalidates my theory.

ya can lead a horse to water...

grifho wrote:First off, I read your link, but I was aware of what AHI was already. So where was I wrong? I merely said that my AHI in my theoretical situation would be 4, which would be the number of events/hour that the CPAP recognized but was not able to correct. How is that 'not how AHI works?' it's only the uncorrected ones that represent the AHI score, right? If it got corrected, it's a non-event.

your 'theory' is only correct if your 'night' is one hour long.
cpaps don't "correct" events, they prevent them. your thinking is muddled.

grifho wrote:Secondly, you say 'the leaks have no affect on the machine's ability to recognize and treat events" Wow, that contradicts everything I've heard, including from my sleep doctor. I don't even have a particularly dramatic mask leak, with an average of maybe 12L/minute, yet my sleep doctor and the resmed 'myair' site sends me emails telling me to fix the leak for best results. Besides, it seems only natural that a leak in a closed system like CPAP would compromise the ability to both recognize and treat events.

well, I guess I'll just bow to your superior knowledge of the whole cpap infrastructure.

however, I'll correct it so that other people aren't mislead.
first, resmed states in the manual that leaks up to 24lpm are acceptable and the machine can continue to function *perfectly*. second, observation by MANY people here shows that even up into the 30lpm range, the machine can still differentiate between central and obstructive apneas, when you start getting leaks much above that, the machine STILL provides appropriate treatment pressure, but it starts to be unable to tell what kind of apnea you may be experiencing. it can still sense your respiration, but the method that it uses to tell why you stop breathing (can you tell me what that is?) becomes to fuzzy and the machine reports Unknown Apnea instead of Central or Obstructive.

grifho wrote:As for your contention that events that corrected are not events, but are 'hypothetical'. Well, if I generally have 16 events without the CPAP and only 4 with the CPAP, there's nothing hypothetical about that improvement. One can infer that there were therefore 12 caught and corrected.

if you have a regular 16 "events" (of one sort or another) every night, then, presuming you sleep for 8 hours, your AHI would be 2, and you shouldn't have been prescribed a CPAP. you'd also be the first person ever reported to have exactly the same number of events (and no doubt at the exact time). however, since in reality, people are rather more variable than that, there is NO WAY TO KNOW that you would, or would not have had an event, ergo, technically any events you did not experience, are hypothetical.

grifho wrote:So, I still stand by my contention that a higher mask leak can lead to a lower recorded AHI from the machine, when in fact a lower mask leak corrects more events. Again i may be wrong, but waiting for a compelling argument to tell me so. Meanwhile, the numbers I'm getting are confirming it. Well, here they are for the last 6 days, with first number being mask leak and second the AHI. See how the numbers have an inverse relationship, at least in regards to the extremes?

you're welcome to believe whatever you want to believe, however, in this case, you are wrong.

how many years have you been at this? how many years of data have you collected?

grifho wrote: low mask leak score, but my AHI was higher than normal

You'll understand what is happening much better if you forget about AHI and mask leak score, and start looking at the daily detail graph timelines.

your 'theory' is only correct if your 'night' is one hour long.
cpaps don't "correct" events, they prevent them. your thinking is muddled.

Ah, I see where I went wrong. In my theoretical example I said 16 events per night. I meant 'per hour'. I knew what AHI was. I just wrote it wrong. If you had just told me that directly, I would have been able to see the error. I guess "prevent" is a more accurate word than "correct", I suppose. But for the purposes of my example, it's a distinction that is not that important. In fact, you said "the leaks have no affect on the machines ability to recognize and treat events". Since 'treat' is basically the same as 'correct', then I suppose you are saying that your thinking is muddled too?

well, I guess I'll just bow to your superior knowledge of the whole cpap infrastructure.

however, I'll correct it so that other people aren't mislead.
first, resmed states in the manual that leaks up to 24lpm are acceptable and the machine can continue to function *perfectly*. second, observation by MANY people here shows that even up into the 30lpm range, the machine can still differentiate between central and obstructive apneas, when you start getting leaks much above that, the machine STILL provides appropriate treatment pressure, but it starts to be unable to tell what kind of apnea you may be experiencing. it can still sense your respiration, but the method that it uses to tell why you stop breathing (can you tell me what that is?) becomes to fuzzy and the machine reports Unknown Apnea instead of Central or Obstructive.

I didn't mean to imply I have any expertise in CPAP. I'm just trying to reconcile what you said with what feedback I'm getting from others, such as in person with the sleep doctor, emails from resmed, speaking with the company I got it from. They all tell me to address the mask leak. But it doesn't matter all that much to me what kind of apnea it is, as long as it prevents the events. So, if you're right, and I hope you are, thanks for that information! Maybe I should worry about the leak less, as I'm at least in the 'good' range. Just as an aside, I do get a lot of conflicting information on various aspects of CPAP, including the documentation on how to clean the mask.

if you have a regular 16 "events" (of one sort or another) every night, then, presuming you sleep for 8 hours, your AHI would be 2, and you shouldn't have been prescribed a CPAP. you'd also be the first person ever reported to have exactly the same number of events (and no doubt at the exact time). however, since in reality, people are rather more variable than that, there is NO WAY TO KNOW that you would, or would not have had an event, ergo, technically any events you did not experience, are hypothetical.

As I said above, I meant 16 per hour, not night, in my example, which is close to the 22/hour they actually told me I had in the sleep study. Sorry for the confusion. You're right, there's no way to know in any individual night whether I would have had 24 or 18 events or whatever, and therefore I don't know how many were prevented. But for the purposes of evaluating improvement, of course one can estimate how much the CPAP has helped. If my average is 22/hour over a long period before the CPAP, and the CPAP results in an average of an AHI of 4/HR over time, then it makes sense to say that my improvement is 18/hour. We can say this with a fairly high degree of certainty, rather than hand-waving it as "hypothetical". It's similar to how one would evaluate the benefit of a stop light, for example. Let's say there were an average of 50 accidents per year at an intersection over a span of 25 years. If a stop light is added, and the accidents consistently turn into 15 per year, no one would take down the stop light and say "that improvement is merely hypothetical" No, the improvement would be very close to 35/year and it would be statistically significant, as would be an improvement in CPAP, if the data consistently shows around the same average. That's all I was implying in my original example, not an exactly accurate representation of improvement on any specific night, but based on an average improvement.

you're welcome to believe whatever you want to believe, however, in this case, you are wrong.
how many years have you been at this? how many years of data have you collected?

Years of data? Well, does one month count? And my theory is probably wrong. I just needed to know why it was wrong, and you gave me some ideas regarding the mask leak that would repudiate my theory. But I'm going to keep an eye on those numbers to see if that reverse correlation continues. It's hard not to notice that when the leak goes up, the AHI goes down.

You'll understand what is happening much better if you forget about AHI and mask leak score, and start looking at the daily detail graph timelines.

chevie, my machine does display some averages of data over time in the display. However the only thing it displays for an individual day is the AHI and time of use. Online, it shows me the same, plus number of times mask was on or off, and the mask seal number. So maybe I'm not getting that nice detailed daily graph you are referring to. But thanks for letting me know about it. Maybe I can ask about getting that kind of thing

grifho wrote:Let's say there were an average of 50 accidents per year at an intersection over a span of 25 years. If a stop light is added, and the accidents consistently turn into 15 per year,

The best research shows the number of accidents increases when a stoplight is added at a well marked intersection.

grifho wrote:So maybe I'm not getting that nice detailed daily graph you are referring to. But thanks for letting me know about it. Maybe I can ask about getting that kind of thing

Don't tarry in getting free SleepyHead software - https://sleep.tnet.com/equipment You will begin to understand better what you are talking about and your posts will ramble less.

Good luck.

The best research shows the number of accidents increases when a stoplight is added at a well marked intersection.

well, ok sure, ChicagoGranny. Then just change my example to "took away a stoplight" instead of adding one. The particulars are irrelevant to the point I was making.

Don't tarry in getting free SleepyHead software - https://sleep.tnet.com/equipment You will begin to understand better what you are talking about and your posts will ramble less.

Hmmm...can I load that software on my Resmed Airsense 10? I thought it just had modem firmware. It would be nice if I could. Looks like some helpful data comes out of it. I see know that someone from another post says that I will be able to...cool

As for "ramble less", ouch...tough crowd

The software is for your pc. The cpap records lots of data onto its SD card.

grifho wrote:As for "ramble less", ouch...tough crowd

Just trying to help you out here.

grifho wrote:

You'll understand what is happening much better if you forget about AHI and mask leak score, and start looking at the daily detail graph timelines.

chevie, my machine does display some averages of data over time in the display. However the only thing it displays for an individual day is the AHI and time of use.

which part of "GET SLEEPYHEAD" are you unclear on?

grifho wrote:

Don't tarry in getting free SleepyHead software - https://sleep.tnet.com/equipment You will begin to understand better what you are talking about and your posts will ramble less.

Hmmm...can I load that software on my Resmed Airsense 10? I thought it just had modem firmware. It would be nice if I could. Looks like some helpful data comes out of it. I see know that someone from another post says that I will be able to...cool

As for "ramble less", ouch...tough crowd

if you'd read more, or perhaps clicked the sleepyhead link in my sig, you'd know ALL about it, and not make statements that get you laughed at, like the one you made above.

grifho wrote: fact, you said "the leaks have no affect on the machines ability to recognize and treat events". Since 'treat' is basically the same as 'correct', then I suppose you are saying that your thinking is muddled too?

ohhh, you're SO right, I did mistype. ""the leaks have no affect on the machines ability to treat your apnea.
ftfy.

grifho wrote:I didn't mean to imply I have any expertise in CPAP. I'm just trying to reconcile what you said with what feedback I'm getting from others, such as in person with the sleep doctor, emails from resmed, speaking with the company I got it from. They all tell me to address the mask leak. But it doesn't matter all that much to me what kind of apnea it is, as long as it prevents the events. So, if you're right, and I hope you are, thanks for that information! Maybe I should worry about the leak less, as I'm at least in the 'good' range. Just as an aside, I do get a lot of conflicting information on various aspects of CPAP, including the documentation on how to clean the mask.

*excessive* mask leaks must be addressed, because when they get to be too large, they affect, as I said, the ability of the machine to accurately determine what kinds of apnea you have, (and thus respond properly to it) and if the leaks are truly massive, they'll get beyond the machine's ability to maintain therapy pressure, and thus will increase the number of apnea events you experience. however, that amount of leakage is truly huge, as you can see from the chart in the back of the clinicians manual. (somewhere between 115lpm@20cm up to 149lpm@4cm). if you don't have a clinicians manual, then get one, instructions on how to are all over this forum, just do a bit of looking. the leaks you've stated you have (seriously, get sleepyhead so you can see what's going on) are NOT *excessive*, unless you're mistaken as to how bad they are... (again, get SleepyHead).

as to the conflicting advice, you'll find that there is a LOT of that in the sleep medicine world, as well as a HUGE amount of CYA from doctors and DMEs and resmed. they'd much rather tell you "wash your mask and hose EVERY DAY" so that on the remote chance that you are a nutjob that decides to sue them, they can say "this patient didn't follow instructions, so we're not liable for his problems" .... ever look at all the warning stickers on a ladder?

the answer is to do what people here have done, and educate yourself take what people here say, (including me) and try to find *PROOF* that it's wrong, not just 'well, someone said'. ... by researching and trying to find proof, you'll educate yourself and you'll learn far more about your therapy, what works, what doesn't, and your health will benefit.

grifho wrote:As I said above, I meant 16 per hour, not night, in my example, which is close to the 22/hour they actually told me I had in the sleep study. Sorry for the confusion. You're right, there's no way to know in any individual night whether I would have had 24 or 18 events or whatever, and therefore I don't know how many were prevented. But for the purposes of evaluating improvement, of course one can estimate how much the CPAP has helped. If my average is 22/hour over a long period before the CPAP, and the CPAP results in an average of an AHI of 4/HR over time, then it makes sense to say that my improvement is 18/hour. We can say this with a fairly high degree of certainty, rather than hand-waving it as "hypothetical".

by your comments, you seem to have lost sight of the fact that AHI is an AVERAGE over the night, it's almost unheard of that people have events with any regularity, they may sleep 7 hours perfectly fine and then have 40 events in the last hour. giving them an AHI of 5. but just looking at the little display on your screen, or the derpy myair phone app, you'll NEVER KNOW. (GET SLEEPYHEAD*cough*)

your stoplight analogy is not germane because you're comparing varying full nights sleep to a brief snapshot of activity (during your sleep study) and saying "well, I've corrected this many things!" but, again, sleep varies minute to minute, and even more day by day..

yes, the machine makes things much better, but there's NO WAY TO KNOW how many events you didn't have, in an hour, or in a night, because there are so many variables. hence, hypothetical. you said "If my average is 22/hour over a long period before the CPAP" and I ask... how do you KNOW? did you wear monitoring equipment for several months before getting a cpap? no? then you're just guessing. yes, your sleep is better, but to try and say "the cpap corrected 6 events an hour last night" is patently wrong.

grifho wrote:Years of data? Well, does one month count? And my theory is probably wrong. I just needed to know why it was wrong, and you gave me some ideas regarding the mask leak that would repudiate my theory. But I'm going to keep an eye on those numbers to see if that reverse correlation continues. It's hard not to notice that when the leak goes up, the AHI goes down.

get sleepyhead and get some REAL data. maybe you sleep better with the mask looser, even though it leaks a bit more. this stuff is *complicated* there are a lot of factors involved.

It is possible that with more leaks ... the leaks are waking you up before you have an event ... therefore less events to record.
Fix the leaks and you sleep better and go into rem sleep and have more events.

now it you get the sortware for your computer .... you will be able to see what is happening