Mask leakage - should I be concerned

[snoozysue]

I have just updated my version of Sleepyhead as I have not used it in a while. My pressure settings are 8-18.
Last night I had an AHI of zero.
The welcome message said "your mask is leaking way too much"
I had a large leak for 3min 14 secs.
I feel fine but should I be concerned and trying to get less leaks.
Thanks

[Pugsy]

First of all....you are using a Respironics machine...is that correct?
If so you need to make a little adjustment in SleepyHead so that it pulls more accurate leak %.
Go into Preferences...cpap tab and adjust the "Flags over leak threshold" number from 24 (that's the ResMed threshold and way to low for a Respironics machine) to something more in line with what a Respironics threshold is...which is kinda hard because it moves with pressure and the mask used but something around 60 to 70 L/min is probably a good conservative number.
SleepyHead defaults to the ResMed threshold number and it's too low for a Respironics user. Makes it look like there are bad leaks when maybe there aren't any.

Or just remove the check mark so that it doesn't do the leak over threshold % calculation and instead just look over in the events graph...if you have a large leak it will be flagged over on the Events graph....see LL near the top...that's Large Leak.

I mainly use a ResMed machine so I just leave the 24 L/min threshold as is but every time I use my Respironics machine SleepHead will give me that same warning about leak but I go look and no LL flagging at all. Since I know I am not really leaking I just ignore the warning.

Finally....even if someone is really spending 3 % of the night in real large leak...as long as they sleep through it I don't think it's anything to get all worried over.
At 7 hours of sleep..3 % is only 21 minutes and as long as they don't wake me up I wouldn't even give that statistic a second glance.

[snoozysue]

Thank you Pugsy
I changed it to 60 L/min and the message changed.
I did have one large leak event for just over 3mins but that seems minor for 6.5 hours of sleep.

[linuxman]

First of all....you are using a Respironics machine...is that correct?
If so you need to make a little adjustment in SleepyHead so that it pulls more accurate leak %.
Go into Preferences...cpap tab and adjust the "Flags over leak threshold" number from 24 (that's the ResMed threshold and way to low for a Respironics machine) to something more in line with what a Respironics threshold is...which is kinda hard because it moves with pressure and the mask used but something around 60 to 70 L/min is probably a good conservative number.
SleepyHead defaults to the ResMed threshold number and it's too low for a Respironics user. Makes it look like there are bad leaks when maybe there aren't any.

Pugsy, I'm just not seeing this behavior in Sleepyhead. The redline % is based on the SH calculated excess leak when using a Dreamstation, not total leaks. It's clear to see this by manipulating the redline threshold, and looking at the calculated leak line. I have my threshold set at 24L/min, and I have 0% over redline as long as the calculated leaks stays below 24L/min. If over that, the redline % goes up as the time over redline increases, just as one would expect. If it was using total leaks, I would see 100% over redline all the time, because total leaks at my pressure (and just about any pressure) are *always* over 24L/min. Now whether 24L/min of excess leak is excessive or not is a different matter..that's probably too low for both machines. But using that as the redline threshold is going to exhibit the same behavior whether it's a Resmed machine or Respironics. The only difference is where the excess leak estimation occurs. With Resmed, the machine does that internally and SH uses that number. With Respironics, SH has to do that estimation. The Resmed machine and SH (with a Respironics machine) both have the same information to work with (total air flow, a mask leak model, and current pressure). For maximum accuracy, one should set the SH mask curve setting to match their mask, but even without that the stock setting is reasonable for most masks.

Here's an example that shows how the redline % responds using a Dreamstation. I have the over redline % threshold set at 23% for this graph. As you can see, time over redline was .374%, which is due to the few minutes of excess leak just over the redline a little before 6am.

[Pugsy]

Pugsy, I'm just not seeing this behavior in Sleepyhead. The redline % is based on the SH calculated excess leak when using a Dreamstation, not total leaks. It's clear to see this by manipulating the redline threshold, and looking at the calculated leak line. I have my threshold set at 24L/min, and I have 0% over redline as long as the calculated leaks stays below 24L/min. I

It happens for some Respironics users because their excess leaks don't exceed the 24 L/min.
The threshold is just the line between not large leak territory and large leak territory.
So not everyone has it happen that the ResMed threshold creates a big problem for them with the statistics because they don't have much excess leak.
In fact I think that most never notice and rarely get a SH warning about leaks simply because there is just minimal excess leak.

But trust me...when SH says I have 10% of the time in large leak..and I look over at the Events graph for my Respironics machine and I am NOT seeing one large leak flag at all...that threshold number is indeed incorrect for me. Part of it is my pressures used on my Respironics machine which are in the teens and upper teens most of the night and part of it is that Respironics threshold is simply higher than 24 L/min.
Respironics threshold varies by pressure and by mask so we can't pin it down real easily but from past experience seeing lots of Encore and SH reports showing large leak flagging has shown that threshold for Respironics large leak territory is probably up between 60 to 70 to 90 L/min.
Even if you factor out the vent rate to leave excess leak from a total leak number the Respironics large leak threshold is much more forgiving than ResMed's 24 L/min.
A vent rate of say 30 plus the 24 excess number would equal 54 total leak and it's a rare Respironics user that would earn a LL flag with total leak hitting 60 L/min.

So for some people with Respironics machines SH can and will tell them that they are having a problem when there is not real problem.
Doesn't happen to all Respironics users and doesn't even happen to me all the time even with my higher pressures unless I get up real close to where Respironics would actually flag a large leak.

The cold hard fact is that this particular calculation in SH is not as accurate with Respironics as it is with ResMed.
The machine trumps SleepyHead...so if the machine isn't flagging a large leak over on the Events graph..and SH is saying 3 % of the night in large leak...SH is wrong if it is using the 24 number.

The fact it never happens to you...hey that just means that your excess leaks are very minimal indeed.
Most people it isn't a problem but for some people it can cause unnecessary panic if they don't know to look over at the LL graph and in the case of OP in this thread there was only one short lived real large leak but not enough to account for the % warned about.

Setting the mask curve...you understand what you are doing and what it means..most of the people here do good to understand basic leak and basic AHI. I am NOT going to go down that road with people because I have my doubts as to Mark getting that particular feature totally accurate. SleepyHead is Beta software and from one man at that who admitted to me the problems he had coming up with accurate calculation coding for Respironics leak stuff. It ain't perfect.

As for your graph...I don't see a large leak flag on your events graphs. ....the machine didn't flag that little spike. You might think it qualifies but the machine didn't. I am sorry but the machine trumps SH especially a SH using a feature that is questionable in accuracy.
How would you feel if it had said 10% of the time in large leak and you look over to the Events graph and not see one single LL flag?
That's what is happening to some people and those people get all in a panic when there is simply no need.

[linuxman]

Pugsy, I'm just not seeing this behavior in Sleepyhead. The redline % is based on the SH calculated excess leak when using a Dreamstation, not total leaks. It's clear to see this by manipulating the redline threshold, and looking at the calculated leak line. I have my threshold set at 24L/min, and I have 0% over redline as long as the calculated leaks stays below 24L/min. I

As for your graph...I don't see a large leak flag on your events graphs. ....the machine didn't flag that little spike. You might think it qualifies but the machine didn't. I am sorry but the machine trumps SH especially a SH using a feature that is questionable in accuracy.
How would you feel if it had said 10% of the time in large leak and you look over to the Events graph and not see one single LL flag?
That's what is happening to some people and those people get all in a panic when there is simply no need.

Who knows about the LL flag...I don't care about the LL flag...that's irrelevant to me. I might have it turned off. That's like a dummy light on a car that says "you're overheated" vs a analog gauge that tells you how hot the car actually is. What's important (at least to me) is whether my mask is leaking, or not, and roughly how much, so I can strive to keep that reasonably low. The percent over redline is nice quick indicator of how much time spent over that readline, and it works just fine, but I don't use it for anything special other than noting quickly that I'm getting close the edge.

On machine vs Sleepyhead, I don't see how the machine has any particular advantage in calculating leaks at the mask. You can't actually measure excess leak directly (well, with a standard CPAP setup). It always has to be calculated, whether by the machine or by SH. Of course I don't know what exact method Resmed uses to do that calculation, but it's going to be similar to what SH does..which is calculate mask leak rate at the current pressure, and subtract from the total air flow. I don't know what Resmed knows about one's mask, but I've seen that it does allow for mask type selection. It could conceivably use that to pick a different mask curve for different mask types, then use that for the leak lookup. I don't know if it uses that info or not. Or it just uses some sort of aggregate, average, curve that's "close enough" for most masks. It could also use some slightly more advanced mathematics to try determine what the constant offset of the leak signal is. That's tough to do if pressure changes, though, so I doubt they're doing that. In any case, it's always a calculation based on assumptions about the mask leak rate at current pressure. SH does the same thing, but in that case, it actually has a user settable mask curve (albeit, linear fit) that should get you very close to the actual mask expected leak rate. Calculated excess leak should be comparably accurate or even better than what Resmed is providing.

[Pugsy]

Who knows about the LL flag...I don't care about the LL flag...that's irrelevant to me.

Suit yourself.
I have beat this dead horse to death.
You believe you way and I believe my way and let's leave it at that okay?

For most people the dummy light is quite sufficient for them...myself included.

[palerider]

Pugsy, I'm just not seeing this behavior in Sleepyhead. The redline % is based on the SH calculated excess leak when using a Dreamstation, not total leaks. It's clear to see this by manipulating the redline threshold, and looking at the calculated leak line. I have my threshold set at 24L/min, and I have 0% over redline as long as the calculated leaks stays below 24L/min.

the only word I can come up with to describe taking *RESMED* parameters and trying to cram them onto RESPIRONICS machine data is "stupid", though to be charitable "ignorant" or "misguided" might be more polite.

sleepyhead is a mishmash of default settings, some things are appropriate for resmed, some things are appropriate for respironics, it's unfortunate, but the defaults, for the most part, don't change based on the type of machine, and that causes people to get confused, like it's done for you.

[linuxman]

Pugsy, I'm just not seeing this behavior in Sleepyhead. The redline % is based on the SH calculated excess leak when using a Dreamstation, not total leaks. It's clear to see this by manipulating the redline threshold, and looking at the calculated leak line. I have my threshold set at 24L/min, and I have 0% over redline as long as the calculated leaks stays below 24L/min.

the only word I can come up with to describe taking *RESMED* parameters and trying to cram them onto RESPIRONICS machine data is "stupid", though to be charitable "ignorant" or "misguided" might be more polite.

sleepyhead is a mishmash of default settings, some things are appropriate for resmed, some things are appropriate for respironics, it's unfortunate, but the defaults, for the most part, don't change based on the type of machine, and that causes people to get confused, like it's done for you.

So, you're saying that 24L/min of excess leak as a redline threshold is appropriate for Resmed, but not for Respironics? Sure, if the Respironics machine is just so much better at handling excess that 60L/min is ok..then sure, but I see no evidence of that. Why wouldn't the amount of total flow they can tolerate be at least in the same ballpark? Total flow for the same mask with 24L/min of excess for both machines would be identical (75-ish for mask with native leak at 50L/min).

[palerider]

Who knows about the LL flag...I don't care about the LL flag...that's irrelevant to me. I might have it turned off. That's like a dummy light on a car that says "you're overheated" vs a analog gauge that tells you how hot the car actually is.

the manufacturer of YOUR MACHINE knows about the LL flag, they designed the machine to complain about any leaks that IT calculates to be sufficiently large to exceed YOUR MACHINES threshold.

the 24lpm redline is *totally irrelevant* to YOUR MACHINE. if you want to dabble with the 24lpm redline then you need to TRADE IN YOUR MACHINE FOR A RESMED.

It always has to be calculated, whether by the machine or by SH. Of course I don't know what exact method Resmed uses to do that calculation, but it's going to be similar to what SH does..which is calculate mask leak rate at the current pressure, and subtract from the total air flow.

no, that's not how resmed does it. resmed calculates a baseline leak by monitoring the variations in flow, you can see this happen on vpap, theres a lot of math involved, sleepyhead used to try to do that for respironics, ,but the math was insufficient and would occasionally result in artificially low 'excess' leak lines, thus, at my insistence (and showing several examples of this problem) the math was thrown out and the simple subtraction method was used instead.

That's tough to do if pressure changes, though, so I doubt they're doing that. [

you'd be wrong. there's a lot more calculation going on in the computer in there than you realize, and a lot more data available than is presented on the sd card.

you'll do whatever you bloody well want to, and people will keep saying that it is wrong for the type of machine.

sleepyhead should turn off the 24lpm redline for any machine that isn't resmed, simply because the 24lpm redline is irrelevant for any machine that isn't resmed.

other machines have other tolerances for how much leak they can work with, and trying to apply the resmed rules is like ... I don't know, trying to put redhat config files on debian.

[palerider]

So, you're saying that 24L/min of excess leak as a redline threshold is appropriate for Resmed, but not for Respironics? Sure, if the Respironics machine is just so much better at handling excess that 60L/min is ok..then sure, but I see no evidence of that. Why wouldn't the amount of total flow they can tolerate be at least in the same ballpark? Total flow for the same mask with 24L/min of excess for both machines would be identical (75-ish for mask with native leak at 50L/min).

because they are not the same machine, they don't use the same technology, they don't have the same hardware, and they don't have the same programming.

resmed can 'tolerate' a whale of a lot more leak without compromising therapy pressure, download a resmed clinical manual and look at the chart in the back that shows how much leak they can handle and still maintain pressure. then pick your jaw up off the floor.

the 24lpm on RESMED is a comfortable, safe level where the machine can still accurately differentiate between central and obstructive apnea. though in practice, it will still do that up into the mid 30s of leak, after that the machine starts flagging unknown apneas because it can't reliably tell the difference.

since your machine uses a near totally different method to differentiate, ASSuming that the parameters of operation are the same is.... welll, you supply a word.

this is more complicated than you understand.

[linuxman]

So, you're saying that 24L/min of excess leak as a redline threshold is appropriate for Resmed, but not for Respironics? Sure, if the Respironics machine is just so much better at handling excess that 60L/min is ok..then sure, but I see no evidence of that. Why wouldn't the amount of total flow they can tolerate be at least in the same ballpark? Total flow for the same mask with 24L/min of excess for both machines would be identical (75-ish for mask with native leak at 50L/min).

because they are not the same machine, they don't use the same technology, they don't have the same hardware, and they don't have the same programming.

resmed can 'tolerate' a whale of a lot more leak without compromising therapy pressure, download a resmed clinical manual and look at the chart in the back that shows how much leak they can handle and still maintain pressure. then pick your jaw up off the floor.

the 24lpm on RESMED is a comfortable, safe level where the machine can still accurately differentiate between central and obstructive apnea. though in practice, it will still do that up into the mid 30s of leak, after that the machine starts flagging unknown apneas because it can't reliably tell the difference.

since your machine uses a near totally different method to differentiate, ASSuming that the parameters of operation are the same is.... welll, you supply a word.

this is more complicated than you understand.

Ok, great... I'll take your word that Respironics machines suck at maintaining pressure at higher flow rates, and start to drop off pressure at point that's some amount lower than Resmed. If that's the case, why would one increase the redline threshold when using the Respironics? Apparently it can't handle as much excess, so you should lower the redline to warn you at a lower excess pressure.

[Pugsy]

Ok, great... I'll take your word that Respironics machines suck at maintaining pressure at higher flow rates, and start to drop off pressure at point that's some amount lower than Resmed. If that's the case, why would one increase the redline threshold when using the Respironics? Apparently it can't handle as much excess, so you should lower the redline to warn you at a lower excess pressure.

Huh?

Respironics machines seem to handle "excess" leaks better than ResMed not worse. Respironics machines don't get clueless until excess leaks get up around 90 to 100 L/min or 120 L/min total leak.
I have seen it documented on software reports when the machine in auto mode throws up it's little hands and behaves like a fixed pressure mode when leaks are up around 120 L/min....seen it many times on Encore reports.
ResMed machines will do it at 40 to 50 L/min excess leak and sometimes it will even shut itself off because it doesn't know a human is on the other end the leaks are so bad.

Respironics machines take a lot large excess leak before they become compromised.

[linuxman]

Ok, great... I'll take your word that Respironics machines suck at maintaining pressure at higher flow rates, and start to drop off pressure at point that's some amount lower than Resmed. If that's the case, why would one increase the redline threshold when using the Respironics? Apparently it can't handle as much excess, so you should lower the redline to warn you at a lower excess pressure.

Huh?

Respironics machines seem to handle "excess" leaks better than ResMed not worse. Respironics machines don't get clueless until excess leaks get up around 90 to 100 L/min or 120 L/min total leak.
I have seen it documented on software reports when the machine in auto mode throws up it's little hands and behaves like a fixed pressure mode when leaks are up around 120 L/min....seen it many times on Encore reports.
ResMed machines will do it at 40 to 50 L/min excess leak and sometimes it will even shut itself off because it doesn't know a human is on the other end the leaks are so bad.

Respironics machines take a lot large excess leak before they become compromised.

I have no idea which is better at handling pressure losses... I was just going with PR's assertion that Resmed is awesome at it, and Respironics less so. If Repsironics's can really handle 35L/min or so more of total flow than Resmed without falling over..awesome..then I understand why you'd increase the redline..but that's the only reason. It shouldn't be done because Respironics reports total flow.

[linuxman]

no, that's not how resmed does it. resmed calculates a baseline leak by monitoring the variations in flow, you can see this happen on vpap, theres a lot of math involved, sleepyhead used to try to do that for respironics, ,but the math was insufficient and would occasionally result in artificially low 'excess' leak lines, thus, at my insistence (and showing several examples of this problem) the math was thrown out and the simple subtraction method was used instead.

And that was a good recommendation, because if you're able to use the actual mask curve (or something very close to it), you're going to have an accurate excess leak number (to the extent that total leaks and pressure from the machine are accurate). Fundamentally it's a very simple calculation. The machine doesn't have that knowledge of the mask, so yes it has to do more complicated things..but the simple calculation trumps it, if you have knowledge of the actual mask curve, and SH does.