AHI and 95 Percentile

I may have asked this before, so please bear with me. I understand that the 95 percentile is the average pressure reading 95% of the night. Last night, I recorded 0 AHI, 0 AI incidents and 0 centals incidents. This was a first for me. My 95 percentile last nite was a 14. My autoset pressures are set at 11-18 cm H2O.

Why would my machine up the the pressure (95 percentile) when I had no incidences? Why would it not stay at 11 until it detected an incident?

Sheriff

You had no incidents BECAUSE the pressure was allowed to increse, as needed. If you had it set at straight 11, I'll bet you would have had incidents.

Percentile pressure number just means that you were AT OR BELOW that number during the night. It does not mean that you were at that number all night.

If leak did not drive the pressure up then the only thing left to drive the pressure up was event triggers and the fact that your AHI was quite low just means it did its job in preventing full blown obstructive events. It doesn't wait for full blown event before it responds.

Sheriff Buford wrote:I may have asked this before, so please bear with me. I understand that the 95 percentile is the average pressure reading 95% of the night.

No. Mathematically speaking "average pressure reading for 95% of the night" doesn't make much sense since an average has to be taken over multiple data points---i.e. over time.

As pugsy explains, the 95% pressure reading is the pressure value for which you were AT or BELOW for 95% of the night. And that also means that your pressure was AT or ABOVE that level for 5% of the night.

To make better of sense of this, let's look at what you posted about last night:

Last night, I recorded 0 AHI, 0 AI incidents and 0 centals incidents. This was a first for me. My 95 percentile last nite was a 14. My autoset pressures are set at 11-18 cm H2O.

First the S9 Autoset increases pressure for several reasons: Among them are flow limitations and snoring. So if the S9 sees flow limitations in the wave form data or detects snoring, it will increase the pressure until the flow limitations or snoring disappears. At that point, the S9 will slowly start to decrease the pressure back down to the min PAP pressure setting.

Second, the meaning of 95% pressure = 14cm means that of the total time the machine was running last night, its therapeutic pressure level was between 11 and 14. In other words:

• your pressure was (AT or ABOVE 11cm) and (AT or BELOW 14cm) for 95% of the night, and
• your pressure was (AT or ABOVE 14cm) and (AT or BELOW 18cm) for 5% of the night.

Since 5% of one hour is three minutes, you can easily determine how long your pressure was ABOVE 14cm:

• In minutes, your pressure was ABOVE 14cm for no more than (three minutes)*(number of hours machine ran)

So if your machine was on for seven hours, your pressure was at or ABOVE 14cm for no more than (3 minutes)*(7 hours) = 21minutes.

To continue with the numerical example. It is possible that the 95% pressure level is far, far greater than the pressure you were at for most of the night. It is also possible for the 95% pressure level to be very, very close to the pressure you were at for most of the night. Consider the following two scenarios:

Night 1: You sleep 7 hours with the S9 running. Your pressure is at 11cm for 6 hours. But in one particular hour you have a lot of flow limitations and the S9 increases pressure to respond to them. During a 25 minute period of this hour with flow limitations, the pressure is AT or (slightly) ABOVE 14cm, before it starts to decline back down towards 11cm. Because the 21 minutes with the HIGHEST pressure readings all have pressure AT or ABOVE 14cm, your 95% pressure reading will be 14cm for this night.

Night 2: You sleep 7 hours with the S9 running. During the first hour the machine detects flow limitations and snoring and starts to increase the pressure. By around 14cm the flow limitations and snoring are eliminated, but during the remaining 6 hours of sleep, every time the machine starts to lower the pressure, the flow limitations reappear. Hence for six hours your pressure is fluctuating right around 14cm---sometimes it's a bit more, sometimes it's a bit less. But the pressure is STRICTLY GREATER THAN 14 cm for only about 15--20 minutes for the whole night. Because the 21 minutes with the HIGHEST pressure readings all have pressure AT or ABOVE 14cm, your 95% pressure reading will be 14cm for this night too.

Note that the 95% pressure reading has little or nothing to do with the "median" pressure level (which is reported in ResScan) or the "average" pressure level in the way we normally think of "average." Both Night 1 and Night 2 have the same 95% pressure level---14cm. But for Night 1, your median pressure level is 11cm and you average pressure reading for Night 1 is pretty close to 11cm. For Night 2, your median pressure level is about 14cm and your average pressure level is apt to be a bit below 14cm because of the time during the first hour when the pressure level was still close to 11cm. From the data provided on the S9's LCD, you cannot tell Night 1 and Night 2 apart. In order tell them apart, you have to look at the data in ResScan.

As to the night in question: As shown in my above numerical examples, the S9 may very well have been running at 11cm for as much as 80% of the night, but when it detected the flow limitations or snoring, it started raising the pressure. And the pressure stayed AT or ABOVE 14cm for at least 5% of the total night. If your S9 was running for 7 hours, the pressure would only need to be AT or ABOVE 14cm for a mere 21 minutes for the 95% pressure level to be reported as 14cm.

Hope that helps,
robysue, the friendly math prof

Thanks Robysue... you were a great help!

Now say that backwards in Chinese!

Robysue, thanks for the math lesson. I'm still trying to figure out my machine, with no info from Dr. until appt. on 6/30. The machine shows a number for 90%, but at the bottom on the screen, it says cmH2O. So, I'm not sure if that is registering humidity or pressure. My pressure is set 4-10, and is an auto machine. No printouts available, just a screen with some numbers, that I don't think the Dr. expects me to look at. I'm at 18 nights, with #s on that screen ranging from 23.6 to 0.3. Also, there is another screen, that I would think is leaks, for it gives a number, and printed in small letters on the screen is says l/min, with a pressure looking icon. Those #s have been 67 to 123. Anybody have any idea what they all mean?
Out of curiosity, I did a "control night" last night. In other words, I didn't sleep with the CPAP, but had it on while reading, dong computer, etc. for 6 hrs. The numbers showed I had 3.6 apneas, and 14 "Sensawakes" (fancy thing on my machine that senses if you are waking up and adjusts pressure a bit). That's hard to understand since I never was asleep and just breathing normally, as when awake. I thought the control night might give me some insight into the data, but didn't prove a thing, other than my BP was higher than it has been on awakening.

I know I went off onto my own thing, instead of staying on Sherrif's question.

busylady wrote:Robysue, thanks for the math lesson. I'm still trying to figure out my machine, with no info from Dr. until appt. on 6/30. The machine shows a number for 90%, but at the bottom on the screen, it says cmH2O. So, I'm not sure if that is registering humidity or pressure.

That 90% number with units labeled as cmH2O is your 90% pressure reading. It's similar to the 95% pressure reading explained above. CPAP pressure is measured in units called cmH2O. One cmH2O is the amount of (additional) air pressure needed to raise a column of water by one centimeter. The cmH2O unit is basically a metric version of "inches of mercury" that we use in the US to measure the barometric pressure for weather forcasting. And just how much additional pressure does 1 cmH2O represent in terms that are meaningful to an average American? Well, if we go to http://www.unitconverters.net/pressure-converter.html and scroll way far down the list, we can convert 1cm H2O into (about) 0.029 inches of mercury. Since standard CPAP pressures range from 4 cmH20 to 20 cmH2O, that means they range from 0.116 to 0.58 inches of mercury.

Standard atmospheric pressure varies from day to day, but typical values are in the 28 to 30 inches of mercury range. Right now (June 13, 2011 at 4:37pm), wunderground is reporting the air pressure in Buffalo, NY as 29.92 inches. When a typical storm front comes through, the air pressure drops anywhere from 0.25 to 1.5 inches of mercury. Some people are really sensitive to these changes in atmospheric pressure and get "pressure" or "weather" headaches or lots of joint pain. But most people don't notice them very much. At the maximum CPAP pressure of 20cmH2O, the machine is ADDING about 6/10 of an "inch" of additional pressure on top of the current atmospheric pressure---about as great of a pressure difference between a typical raining day and a typical sunny day.

My pressure is set 4-10, and is an auto machine.

That's a pretty wide range. What's that 90% figure look like? If it's close to 10 and you are not comfortable at the beginning of the night, you may want a higher starting pressure for comfort. Many folks (but not me---I am an outlier after all) find that they feel as though they are not getting enough air when the CPAP pressure is super low.

No printouts available, just a screen with some numbers, that I don't think the Dr. expects me to look at. I'm at 18 nights, with #s on that screen ranging from 23.6 to 0.3.

I don't have an Icon. The labels for these numbers has to be somewhere in your owner's manual. So I'd suggest starting by rereading that. And is there ANY printing on the screen to indicate what these numbers mean? Since you later on talk about your AHI and the SenseAwake figures, I'll assume these numbers are NOT on that screen. That correct?

Also, there is another screen, that I would think is leaks, for it gives a number, and printed in small letters on the screen is says l/min, with a pressure looking icon. Those #s have been 67 to 123. Anybody have any idea what they all mean?

Yep, that's your leak data. The units are Liters (of air) per minute (of run time). It represents how much air (in terms of VOLUME measured in liters) is being "lost" from the system per minute. Now every mask has a built-in intentional leak rate---an amount of leaking that is built-into the mask to make sure the CO2 you exhale into the mask is properly vented OUTSIDE the mask and into your bedroom. This is to prevent you from re-breathing your own CO2.

In order for you to understand more about your leak rates you need to determine the answers to the following questions:

1) Does the Icon report leak rate figures in terms of TOTAL leak, which includes both INTENTIONAL leaks and UNINTENTIONAL leaks, or does it report leak rates in terms of UNINTENTIONAL leaks only? Read the owner's manual that came with your machine. The answer to this question is likely buried somewhere in there.

2) Does the Icon's LCD report the average leak rate for the night? Or is this the 90% leak rate for the night? Or something else? You need to know that to know how long the leaks are AT or ABOVE this figure each night.

3) If the Icon reports UNINTENTIONAL leaks, then what is the intentional leak rate for your mask at the pressures you are using? You can find this information in the owner's manual for the MASK that you are using. There will likely be a table somewhere near the back of the booklet.

Even though I don't know the answers to these two questions, I'll go out on a limb and say that I think you've got a leak problem and it might be a BAD leak problem. If the UNINTENTIONAL leak numbers are ranging from 67 L/m to 123 L/min for any length of time every night, that's definitely high enough to cause problems. If these figures include your mask's INTENTIONAL leak rate, then to find your UNINTENTIONAL leak rate, you need to subtract off your mask's INTENTIONAL leak rate. The mask I use (the Swift FX) has leak rates that range from 20 L/min to about 35 L/min at 4--10 cm of pressure (if I recall correctly). If your mask has similar intentional leak rate figures and the Icon is reporting TOTAL leak, that would mean your Unintentional leaks could be anywhere from roughly 30 L/min to 100 L/min. And leaks that large could easily affect the quality of your therapy.

Out of curiosity, I did a "control night" last night. In other words, I didn't sleep with the CPAP, but had it on while reading, dong computer, etc. for 6 hrs. The numbers showed I had 3.6 apneas, and 14 "Sensawakes" (fancy thing on my machine that senses if you are waking up and adjusts pressure a bit). That's hard to understand since I never was asleep and just breathing normally, as when awake.

Even with Sensawake, the Icon can't really tell when you are genuinely awake---particularly for LONG periods of time. And the funny thing is, when we are WIDE awake and doing things, our breathing turns out to be a lot less regular than when we're asleep. So during that 6 hour experiment with the machine on when you were reading, going stuff on the computer, and so on, the machine was ASSUMING that the breathing it was detecting was when you were (sound) asleep. And the AHI = 3.6 figure means that it thought it saw about 21 or 22 times during the entire 6 hour period when you stopped breathing for 10 seconds or longer (apnea) or were breathing with very little air going into your lungs for 10 seconds or longer (hypopnea). But the dirty little secret is that the machine didn't know you were awake and all humans will occasionally hold their breath some when they're awake: When you're concentrating on something particularly hard it's not uncommon to hold your breath unconsciously. When you take a deep sigh, it's not uncommon for there to be a 10 second pause before the next inhale. And so on. Well, all these little inconsistencies and irregularities in our wakeful breathing can and sometimes do get mis-scored as apneas or hypopneas by a CPAP/APAP machine.

As for the 14 Sensawakes? Well, the Icon was doing its best to determine when it was seeing a pattern that was indicating you MIGHT have aroused: Arousals are often (but not always) reflected in the wave form by a change from a very regular, very periodic graph to a much more ragged one: The breaths during a brief arousal at night are often far "bigger" than the sleeping breaths, they may be closer together, and they may have a different "shape" to the inspriatory part of the breath. And while our WAKEFUL breathing is less regular than our SLEEP breathing, it's still pretty regular. So every time your WAKEFUL breathing was good and regular and then suddenly changed---you unconsciously sighed, you took a deep breath because you had been unconsciously holding your breath for a few seconds while in deep concentration, you swallowed, you coughed, etc.---there was a good chance that the Icon ran that set of breaths through its Sensawake algorithm to determine whether the change in your breathing pattern was sufficiently great to flag it as a possible arousal. And 14 times during the 6 hour experiment, the Sensawake algorithm said, "Yes, the change is great enough to maybe be an arousal."

I thought the control night might give me some insight into the data, but didn't prove a thing, other than my BP was higher than it has been on awakening.

Any of the above help provide you with that insight you were hoping for?

I know I went off onto my own thing, instead of staying on Sherrif's question.

Happens a lot and at least your post is piggy-backing with some related questions

Thanks robysue!! If I hadn't logged onto this chat site I'd have really felt lost in this whole process. I've read a lot of comments and only posted a couple. Some of you folks are on daily and giving great advice. You'd commented about reading the manual. That's the problem, I have 11 photocopied pages that tell me how to set the clock, an alarm, and other basic stuff, but only a couple lines on the section called "Sleep Data." When I called the Dr. ofc. with a couple questions, I could tell the nurse had a common trite answer that "the Dr. will talk to you about those things when you come in. Are you using the machine..." Sounds like all she wanted to know was compliance for Medicare. The Dr. seemed quite easy to talk to and respected in his field, so hopefully he'll be open to sharing info and I'll feel more in the know after my appt. 6/30. In the meantime the DME gave me a S pillow for the Swift FX instead of XS. We met in a parking lot on Sat., but he didn't have the right one to give me. He said the XS had been mailed to me from 50 mi. away 8 days ago. Yeah, sure... someone never followed through. I left another message today. We're headed out of town for a few days and it would be nice to have the right gear to see if that makes a difference with leaks. Sometimes I can move my head 1" and the pillows can need adjusting. I hear a breathing, mainly exhale noise sometimes; other times, no noise at all. The machine makes no noise at all. Of course, anytime I have to make adjustments, there is a period of leakage. I've only slept in 1 hour increments many nights. I have dreamed, so I do go into REM quite quickly. I understand that as we get older we skip to REM quicker. Don't know if I've answered any of your questions, but you did help clarify mine..... as much as I can learn until I see the Dr. Software for my machine is apparently not available to end users, just to medical staff. Bummer. I'm going to plead though! I monitor BP and glucose and can then make decisions about eating or life style patterns to avoid. It would be good to know what is working and what isn't with the CPAP. I feel like I'm taking a trip without a map and don't know if I should turn right or left, or if I'm just going in circles.

busylady wrote:Thanks robysue!! If I hadn't logged onto this chat site I'd have really felt lost in this whole process. I've read a lot of comments and only posted a couple. Some of you folks are on daily and giving great advice. You'd commented about reading the manual. That's the problem, I have 11 photocopied pages that tell me how to set the clock, an alarm, and other basic stuff, but only a couple lines on the section called "Sleep Data."

Post a NEW thread here with the title: Need the clinician's guide to the Icon Auto and someone will clue you in on where to find it and if you post your confusion about what the user's guide does have to say about the Sleep Data, somebody is bound to help too.

When I called the Dr. ofc. with a couple questions, I could tell the nurse had a common trite answer that "the Dr. will talk to you about those things when you come in. Are you using the machine..." Sounds like all she wanted to know was compliance for Medicare. The Dr. seemed quite easy to talk to and respected in his field, so hopefully he'll be open to sharing info and I'll feel more in the know after my appt. 6/30.

Some unsolicited advice for that appointment: Take some time to figure out what you want to get out of the appointment. Be sure to write your questions down on paper and bring the paper with you so that you don't forget to ask about something that's important to you.

In the meantime the DME gave me a S pillow for the Swift FX instead of XS. We met in a parking lot on Sat., but he didn't have the right one to give me. He said the XS had been mailed to me from 50 mi. away 8 days ago. Yeah, sure... someone never followed through.

I've got an XS lying around in its original package from the last time I got a new mask. Did the DME only give you ONE pillow instead of the set of three sizes to try? If you'd like my XS, PM me with your address and I'll be happy to send it to you since I don't need it.

I left another message today. We're headed out of town for a few days and it would be nice to have the right gear to see if that makes a difference with leaks. Sometimes I can move my head 1" and the pillows can need adjusting.

Post here on a new thread about how to stablize the Swift FX and you'll get lots of tips. For me, I find that if I've got a really good seal, it takes a fair amount of moving to shove the pillows around to do much damage. But if the seal is NOT good, then, yeah, every little movement seems to create a problem. In time this may get better for you.

I hear a breathing, mainly exhale noise sometimes; other times, no noise at all. The machine makes no noise at all. Of course, anytime I have to make adjustments, there is a period of leakage.

The sound of your own breathing may be conducted through the hose if the hose is touching your pillow or your arms or head. Hanging the hose so it doesn't touch your pillow may help. Or simply running a fan on low for some white noise or getting a nature sounds machine may help mask it. I use my iPod: It's parked in an iHome and I have it cycle through a very long playlist of Gregorian chant music all night long at very, very low volume. The point is NOT to try to drown the conducted sound of your breathing out with a louder noise; rather the point is to give you mind a more pleasant, competing sound to focus on. Earplugs will likely make the tendency to hear your own exhales WORSE because they block out all ambient noise and NO conducted noise. Hence the conducted noise winds up sounding LOUDER when you use earplugs.

I've only slept in 1 hour increments many nights. I have dreamed, so I do go into REM quite quickly. I understand that as we get older we skip to REM quicker. Don't know if I've answered any of your questions, but you did help clarify mine..... as much as I can learn until I see the Dr. Software for my machine is apparently not available to end users, just to medical staff. Bummer. I'm going to plead though! I monitor BP and glucose and can then make decisions about eating or life style patterns to avoid. It would be good to know what is working and what isn't with the CPAP. I feel like I'm taking a trip without a map and don't know if I should turn right or left, or if I'm just going in circles.

You've just written a fantastic explanation of why so many of us here at cpaptalk regard the software as so critical. Best of luck in fighting that battle. And do let us know if you manage to win it!