Why all the fuss about Leak values?????

cwied wrote:. . . given a 95th percentile leak value under 24, I had assumed that the leak was within the ability of the machine to compensate. . . .

During the time that leak is within spec, yes the machine can compensate and can measure accurately enough for trending during that time. BUT during the time that leak is outside of spec (such as at times during that 5% that pressure is somewhere above the reported value), many events can occur that are unable to be detected.

That is why ResMed's own words on the matter in the S9 AutoSet clinical manual about "flow derived values" state that "results may be inaccurate in the presence of leaks."

The manual for my ResMed machine, an S8 VPAP Auto, states: "A leak rate of more than 24 L/min (0.4 L/s) affects the accuracy of other measurements and is associated with patient discomfort, disturbed
sleep, and reduced efficacy of treatment." That means, as I understand it, that during the time that leak is above that amount, events may be missed, even if the reported 95th centile is below that value.

Laurie1041 wrote:I'll cross the path of a FFM if my downloaded data is clearly showing significant leakage that will interfere with my therapy. Thank you again for the information.

Unless you are sure that the leak is from your mouth, I wouldn't necessarily go from pillows straight to a FFM. I did that, and I regretted it. People in general, with plenty of exceptions of course, have a harder time getting FFMs not to leak than nasal masks simply because there is more surface area to leak and more anatomy to accommodate (e.g., the mouth seals, but the nose bridge doesn't, etc.). If it's the pillows that are leaking and not the mouth, I'd try a nasal interface first before moving to a FFM.

SleepingUgly wrote:Unless you are sure that the leak is from your mouth, I wouldn't necessarily go from pillows straight to a FFM. I did that, and I regretted it. People in general, with plenty of exceptions of course, have a harder time getting FFMs not to leak than nasal masks simply because there is more surface area to leak and more anatomy to accommodate (e.g., the mouth seals, but the nose bridge doesn't, etc.). If it's the pillows that are leaking and not the mouth, I'd try a nasal interface first before moving to a FFM.

Good Evening SU,

You make an excellent point with respect from jumping from nasal pillows to an FFM and the potential difficulties that this may create. I am getting accustomed to the nasal pillows and am comfortable with them. I am certain that most if not all of my leak is due to my headgear slipping which causes the entire interface to become unstable and the pillows to move around and lose their seal. I finally read the documentation that came with the Swift FX and it clearly shows that the back strap should go under my hair (I have long hair). I will certainly give this a go. Ya never know. Maybe it was just a case of "Read Directions First". Laurie

nanwilson wrote:To keep the headgear from sliding on you hair just add two bits of velcro (the scratchie bit not the fluffy side) to the inside of the bottom back strap. The velcro kinda stops the straps from sliding but doesn't stick in your hair. Make sure its the scratchie side of the velcro not the smooth side.
Nan

Exactly !! ResMed should have made those vertical straps ADJUSTABLE. The only reason I get leaks is the top strap slipping off my head.

Silly design.

Excellent, RS.

Why did ResMed put the 95th % ile on the screen ? Would rather have average adjusted by time, or some figure more useful.

Physician wrote:Excellent, RS.

Why did ResMed put the 95th % ile on the screen ? Would rather have average adjusted by time, or some figure more useful.

I haven't the (CPAP-induced) foggiest idea of why ResMed put the 95% leak rate on the screen. Seems to me the median leak rate is the one they should have put on.

I mean, even if you sleep for 10 hours straight, your leak rate is OVER the 95% percentile rate for a grand total of 30 minutes.

Leak can have wide variation throughout the one night. For that reason, an average or median for that one night could be next to worthless. A 90th or 95th percentile imparts a great deal of information with one number, simply and elegantly. It is just a matter of grasping what information that is. IMO.

I have also been having trouble with the back strap slipping. Going right now to try the velcro. Thanks!

jnk wrote:Leak can have wide variation throughout the one night. For that reason, an average or median for that one night could be next to worthless. A 90th or 95th percentile imparts a great deal of information with one number, simply and elegantly. It is just a matter of grasping what information that is. IMO.

I humbly disagree: The 95% number can be extremely inflated by one largish 3 minute leak/hour and tell you absolutely nothing about what's going on the rest of the time.

For example, lets suppose you slept for 8 hours on Night A. And lets suppose that for a total 7.5 hours on Night A you had a leak rate of 0.0 L/min. And that you had one major leak that lasted about 30 minutes: Lets suppose that it took about 6 minutes that the leak rate to go from 0.1 to 29.4 L/min. And for a grand total of 24 minutes out of the entire 8 hours of time the machine was on, the leak rate was above 29.4 L/min

And now lets suppose you also slept for 8 hours on Night B. And lets suppose that you had a leak rate right around 23.0 L/m for 7 hours on Night B. And lets suppose for 36 minutes over the course of the night, your leak rate was between 23.0 and 29.4 L/min. And for a grand total of 24 miutes out of the entire 8 hours of time the machine was on, the leak rate was above 29.4 L/min.

NOW on BOTH nights, the 95% Leak Rate is reported as 29.4 L/min since on BOTH nights, your leak rate was AT or ABOVE 29.4 L/min for 5% of the total time (5% of 8 hours is 24 minutes)

On Night A, the median leak rate is 0.0 since the leak rate was less than or equal to 0.0 for (at least) 4 hours.
On NIght B, the median leak rate is 25.0 since the leak rate was less than or equal to 25.0 for (at least) 4 hours.

Now, I'll admit that my two examples are a bit extreme, but which night do you think represents a more serious leak problem? I'd say Night B, but there's no way to distinguish between the two of them looking at just the 95% figure. But since your leak rate is LESS THAN or EQUAL to the Median Leak rate number for 50% of the time, I think it gives a better idea of what's going on. To get the best idea of what's going on, though, I think you really need to look at both the median and the 95% percentile numbers.

It is also worth pointing out that on both Night A and Night B, the green smiley face would show up in the short version of the Sleep Quality menu, since the red frowny shows up only when the 70% Leak Rate is 24 L/min or greater. And on Night B, the 70% Leak Rate is 23.0 L/min because the leak was AT or BELOW 23.0 L/min for at least 70% of the night, which is 0.7 x 60 minutes x 8 hours = 5.6 total hours.

Well, I think we all agree that the best thing is to see the graph.

But if I can't see a graph, I would prefer information about what happened most of the night rather than information about half the night. To my mind, that's what makes the 90 or 95th centile more useful info than the 50th. The higher centiles tell me how high leak was but chop off the brief extremes that would skew the number, to my way of thinking.

I've never been very good with mathematical concepts, though. So don't go by me.

robysue wrote:

Physician wrote:Excellent, RS.

Why did ResMed put the 95th % ile on the screen ? Would rather have average adjusted by time, or some figure more useful.

I haven't the (CPAP-induced) foggiest idea of why ResMed put the 95% leak rate on the screen. Seems to me the median leak rate is the one they should have put on.

I mean, even if you sleep for 10 hours straight, your leak rate is OVER the 95% percentile rate for a grand total of 30 minutes.

Agree, robysue:

If the 95th percentile is there, then the AVERAGE should also be there. Imagine a 95th of 24 but an average of 0.5. Now that is meaningful information. And we should not have to use ResScan to look at and understand the leaks.

While ResMed may be making one of the top two CPAP machines, they made some stupid and silly errors in the design and GUI of the S9, and some design flaws in their masks and straps.

By the use of velcro to alter/shorten the vertical straps, my leak rate fell to nearly zero. ResMed should have made the verticals adjustable. That way the top circumferential would not flip off during the night ------> leaks. These design flaws are painfully obvious.

I plan on working with a new company to improve mask and straps.

jnk wrote:Leak can have wide variation throughout the one night. For that reason, an average or median for that one night could be next to worthless. A 90th or 95th percentile imparts a great deal of information with one number, simply and elegantly. It is just a matter of grasping what information that is. IMO.

What ? Maybe I need a second cup of coffee, but what you posted here makes no sense. When my LCD screen reported a leak rate of 7.2, the ResScan graph showed one large spike for a few minutes, and a near ZERO leak the remainder of the entire night. That 7.2 was a useless value, and had the AVERAGE leak rate been on the display it seems it would have been a comforting 0.5 rate.


We should not need to visit a sleep doctor nor have to use software to acquire an accurate assessment of our leak rates and therapy efficacy.

Many months ago I reported to ResMed multiple defects in design and manufacture in the S9 and masks, and gave many suggestions for improvements, but thus far nothing has changed, nor have I received a progress report or intention of change.

But ResMed has not been doing poorly:




ResMed quarterly profit rises 35 percent
BY Reuters
— 4:51 PM ET 10/28/2010
* Q1 EPS $0.37 vs Street view $0.32
* Revenue rose 14 percent to $282 mln
LOS ANGELES, Oct 28 (Reuters) - Respiratory equipment maker ResMed Inc (RMD

Quarterly profit rose 35 percent, beating expectations, on double-digit sales increases both in the Americas and overseas.
The company, which makes masks and devices to assist sleep disorder breathing problems in people who suffer from sleep apnea, posted on Thursday a net profit of $56.7 million, or 36 cents per share, for its fiscal first quarter ended September 30. That compared with a profit of $42.1 million, or 27 cents a share, a year ago.
Adjusting for one-time items, the company earned 37 cents a share, beating the average analyst expectation of 32 cents per share, according to Thomson Reuters I/B/E/S.
Revenue for the quarter rose 14 percent to $282 million.
San Diego-based ResMed (RMD

Sales in the Americas rose 16 percent to $155.6 million, while overseas sales jumped 13 percent to $126.4 million.
"The findings from recent clinical studies continue to demonstrate the importance of diagnosing and treating sleep-disordered breathing (SDB)," Chief Executive Officer Kieran Gallahue said in a statement. "There is abundant evidence supporting the correlation of SDB with heart failure, stroke, atrial fibrillation, diabetes, hypertension and even the dementia associated with Alzheimer's disease." (Reporting by Deena Beasley; editing by Carol Bishopric)

Physician wrote:. . . and had the AVERAGE leak rate been on the display it seems it would have been a comforting 0.5 rate.

Yes, comfortably unaware that you had any leak problems at all the night before, although you did. And you may have had many apneas and hypopneas during those few minutes that were unable to be scored. I would want to know that. But hey, that's just me.

jnk wrote:

Physician wrote:. . . and had the AVERAGE leak rate been on the display it seems it would have been a comforting 0.5 rate.

Yes, comfortably unaware that you had any leak problems at all the night before, although you did. And you may have had many apneas and hypopneas during those few minutes that were unable to be scored. I would want to know that. But hey, that's just me.

But there were no AP or HY events and that duration of that leak was trivial. ResMed needs to add more useful leak info on the LCD. Period.

ResMed seems to accept input, but they need to be more consumer-responsive.

Physician wrote: . . . But there were no AP or HY events . . . .

But how would you know that, if for several minutes the leak was above the amount that allows the machine's sensors to "see" events?