Understanding Leakage Statistics

[energyguyoly]

Just took a first look at my S9 autoset statistics from the SD card. I'm using a nasal mask. I'm trying to understand the leakage numbers which seem quite high. Can someone point me to a guide for interpreting the numbers?

[robysue]

First, it is important to note that Resmed machines and the ResScan software report unintentional leak rates. In other words, an approximation for the intentional leak rate for the mask you are using at the pressure you are using is subtracted off of the total leak data before reporting the leak rate.

Next, in order to understand how the Resmed S9 handles the leak data and how ResScan reports it, we need some numbers. Since you didn't provide me with your acutal numbers, I'll make some up.

So let's suppose that you slept with the S9 running for 7.5 hours and we're looking at the overnight data in ResScan or off the LCD for the numbers it shows. Now lets suppose that ResScan says that your Leak Rate data looks like this:

Median = 1.0 L/min
95% = 2.5 L/min [This is the Leak number shown on the S9's LCD]
Max = 15 L/min

Now let's look at what the rest of those numbers actually mean.

The median leak rate = 1.0 L/min means that for 50% of the time the S9 was on, your leak rate was LESS THAN or EQUAL TO 1.0 L/min. And so for 50% of the night, your leak rate was also GREATER THAN or EQUAL TO 1.0 L/min. So in the 7.5 hours you slept with the machine, for 3.25 hours (3 hours and 15 minutes) you had a leak rate of LESS THAN or EQUAL TO 1.0 L/min.

The 95% leak rate = 2.5 L/min means that for 95% of the time the S9 was on, your leak rate was LESS THAN or EQUAL TO 2.5 L/min. And so for 5% of the night, your leak rate was GREATER THAN or EQUAL TO 2.5 L/min. Now it's important to realize that 5% of one hour is equal to three minutes. Since you slept for 7.5 hours, that means your leak rate was GREATER THAN or EQUAL TO 2.5 L/min for a grand total of (3 minutes)*(7.5 hours)=22.5 minutes. That 22.5 minutes of "high leak rates" might be in one giant half hour leak or it could be in several smaller leaks---the only way to know is that to look at the detailed graphs in ResScan. [You find the detailed graphs under the "Detailed Graphs" tab. If you only download "summary data" the detailed graphs won't be there.]

The max leak rate=15L/min indicates that at some point when the S9 was sampling leak data, it detected a leak that was as large as 15L/min. It may have been a very short lived thing---when you broke the seal to scratch your nose for example. Or it could be that on one or two of the big leaks that you had, the leak pretty quickly went from in the neighborhood of 2.5 L/min to 15 L/min and stayed there. But you know for sure that you could not have leaked at a rate of 15L/min for more than about 22 minutes because of the 95% leak rate number and the time the machine was on.

Interpreting the leak rate data is also important:

First, on the Short Sleep Quality menu, if you see Mr. Red Frowny Face, that indicates your 75% leak rate is AT or ABOVE 24 L/min. And according to Resmed's litertature if you are leaking AT or ABOVE a rate of 24 L/min for 25% of the night, the ResMed engineers believe that your therapy and data can and probably are being compromised.

Many people take the 95% Leak Rate to be the key stat Leak data. But it is worth looking at the median leak rate as well, in my opinion. The reason I look at both is that the median leak rate tells me whether I was leaking off and on all night long or not, whereas the 95% leak rate only tells me if my worst leaks were bad ones or not. To illustrate what I mean, let's look at two examples:

Person A sleeps for 7.5 hours and is leaking off and on all night at a rate of 10.5 L/M and for about 25 minutes during the night, the leak rate actually is between 10.5 and 15 L/min. So A's leak numbers look like this:

median=10.5, 95%=15.0, and max=28.0

Now suppose Person B sleeps for 7.5 hours and has NO leaks at all for most of the night, but has ONE 25 minute leak where the leak rates are from 10.5 to 15 L/min during this leak. B's leak numbers look like this:

median=0.0, 95%=15.0, and max=28.0

Neither person's leak rates are large enough to trigger a visit from Mr. Red Frowny Face in the short Sleep Quality menu on the LCD. But Person A is likely to be having more significant problems with comfort due to leaking all night long.

Obviously, the closer to 0.0 your leak rate data is, the better your leak data is. Information from Resmed seems to indicate that leaks above 24 L/min (the RED LINE in ResScan's detailed graphs) are enough to cause problems with the algorithms the S9 uses in order to detect and respond to apneas, hypopneas, flow limitations, and snores. So in any case you really want your 95% leak rate under 24 L/min to insure that you're getting some kind of meaningful therapy. But comfort is as a much of an issue as therapy is in my opinion: Nasty leaks cause most people real discomfort and they frequently cause people to wake up multiple times during the night to fiddle with the mask even when the leak is well below 24 L/min. For me, I don't worry at all if my 95% unintentional leak rate is below 10 L/min unless I had to fiddle with my mask more than once in the night. And if I wake up multiple times needing to fiddle with my mask because of leaks, then the leak rate IS a problem regardless of how low the 95% leak rate is.

[xenablue]

robysue - your explanations are so awesome! I've printed this out as I just can't seem to get this info clear in my noggin.

Cheers,
xena

[robysue]

xena,

thanks for the compliment. 30 years of teaching college level math courses does come in useful around here when folks start asking about numbers.

[HoseCrusher]

If I may throw out another consideration...

I am not convinced that a 0 leak rate is optimum. In the case of the Swift FX, a loose fitting mask is very comfortable, but with a loose mask you will have some leakage as you turn in your sleep.

You don't want a high leak rate, but it may be that if you have a leak rate of 0 your mask is too tight...

[robysue]

If I may throw out another consideration...

I am not convinced that a 0 leak rate is optimum. In the case of the Swift FX, a loose fitting mask is very comfortable, but with a loose mask you will have some leakage as you turn in your sleep.

I've used the FX since I started xPAP back in September. And it's true---that mask should be worn really loose.

But back when I was using the S9, I was consistently (as in every night except for my first night as a hosehead) getting a median leak rate of 0.0 while wearing the FX rather loosely---and adjusting it a bit looser as time went on.

My 95% leak rates were not 0.0 night after night and I didn't expect them to be. But they were almost always below 10.0 L/min and I'd have many nights in a row with 95% leak rates less than 5 L/min. And a few nights here and there with 95% leak rate = 0.0 L/min.

Max leak rates? I typically ignored them since my max leak rates were literally all over the place: Some nights I'd hit a trifecta of 0.0 leak rates across the board and on others, that max leak rate would be well above the RedLine even though the 95% was below 10.0. After all, whenever the machine samples the leak rate while you are scratching your nose or pulling at the straps to get those pillows into the right orientation, there's a pretty good chance that the maximum leak rate will be really high, but also totally meaningless.

So my basic advice for S9 users continues to be:

Pay attention to both the median and 95% leak rates. If that median rate is in the double digits, you've got some real problems with leaks. If the 95% is above or close to 24 L/min, you've got at least some short term problems and you really need to look at the leak graph to find out how bad they are. But if the median leak rate is in the low single digits and the 95% is in the low double digits? Well, if the leaks are waking you up or if you're waking up with a super dry mouth, the leaks are worth worrying about. But if you're sleeping soundly through the smallish leaks and your mouth isn't particularly dry, there are better things to do than chase a 0.0 leak rate.