POLL: What is your TYPICAL nightly AHI?

[Jason S.]

I thought it might be interesting. I've been in the 1.0-2.0 range consistently of late. Never below a 1.0 and rarely above a 2.0.

[jnk]

The AHI estimates don't compare well between brands, since it is only trending data. It has been said, for example, by respected equipment experts, that the HI portion of the AHI seems to run about double in ResMeds compared to others. So do you want ResMed users to convert (by taking the HI part out, cutting it in half, and adding it back in to AI to get AHI to make the number more comparable to a Respironics estimate), or do you want them to post actual estimates as reported by their machines?

This may end up more confusing than the gender poll!

[twokatmew]

Or poll could be modified where we also select machine manufacturer?

[builta]

Hi,

Actually you need to two dimensional array. One for the manufacturer of the blower and the other for the AHI. Without that I don't think that the data will separate into a clear pattern. It is even possible that you would need to have a third dimension of mask type.

[Jason S.]

The AHI estimates don't compare well between brands, since it is only trending data. It has been said, for example, by respected equipment experts, that the HI portion of the AHI seems to run about double in ResMeds compared to others. So do you want ResMed users to convert (by taking the HI part out, cutting it in half, and adding it back in to AI to get AHI to make the number more comparable to a Respironics estimate), or do you want them to post actual estimates as reported by their machines?

For most people HI is not much relative to AI. If my AHI is 1.5 on a Respironics machine, my HI might be 0.2. So even if my HI would change to 0.4 on a ResMed machine, it doesn't change the overall AHI that much. The poll should be accurate but could possibly be shifted to a slightly lower (definitely less than 0.5) AHI mean due to the somewhat higher numbers provided by ResMed users.

[fishhead]

For most people HI is not much relative to AI. If my AHI is 1.5 on a Respironics machine, my HI might be 0.2. So even if my HI would change to 0.4 on a ResMed machine, it doesn't change the overall AHI that much. The poll should be accurate but could possibly be shifted to a slightly lower (definitely less than 0.5) AHI mean due to the somewhat higher numbers provided by ResMed users.

well, in my case, my AHI has been less than 2, some nights less than 1, and all hypopneas. according to my resmed, i haven't had an apnea in over a week.

i'm not sure what that means.

[twokatmew]

it means you're a hypop kinda guy, fishhead. I had all hypops at both my diagnostic and titration studies. I have plenty of apneas now according to my ResMed machine, but still a lot more hypops than apneas. Even w/ only hypops, we still get a diagnosis of OSA, though I hear the full name is OSAHS -- obstructive apnea hypopnea syndrome.

[jnk]

The AHI estimates don't compare well between brands, since it is only trending data. It has been said, for example, by respected equipment experts, that the HI portion of the AHI seems to run about double in ResMeds compared to others. So do you want ResMed users to convert (by taking the HI part out, cutting it in half, and adding it back in to AI to get AHI to make the number more comparable to a Respironics estimate), or do you want them to post actual estimates as reported by their machines?

For most people HI is not much relative to AI. If my AHI is 1.5 on a Respironics machine, my HI might be 0.2. So even if my HI would change to 0.4 on a ResMed machine, it doesn't change the overall AHI that much. The poll should be accurate but could possibly be shifted to a slightly lower (definitely less than 0.5) AHI mean due to the somewhat higher numbers provided by ResMed users.

My AI is consistently below 1, but my HI is generally between 5 and 8. That HI would likely be 2.5 to 4 on a Respironics.

[Jason S.]

My AI is consistently below 1, but my HI is generally between 5 and 8. That HI would likely be 2.5 to 4 on a Respironics.

I said most people. Exceptions such as yourself would not be statistically significant in a large sample.

[jnk]

I believe it is the rule, not the exception, since brands of machines use very different definitions for what they are estimating, as often explained by rested gal and proved by Velbor's chart:

[NOTE: colorization as shown in the quote below in rg's post is mine, not hers]

Location of full post: viewtopic.php?f=1&t=43948&p=390817#p390817

Re: AHI question...
by rested gal on Sun Aug 02, 2009 1:09 pm

. . . trying to compare the AHI numbers from ResMed machines with the AHI numbers from other brands of machines is somewhat like trying to compare apples and oranges.

A person can be getting EXACTLY the same good, effective treatment from a ResMed machine that shows them an AHI of 9.5 (when the AHI is made up mostly of a high HI number and very low AI number) as they would be getting from a Respironics machine displaying half that AHI number.

. . . The "why" (regarding the AHI you see) has to do with where each manufacturer chooses to draw that imaginary line in the sand for calling a flow limitation a "hypopnea."

When you read "your AHI should be 5 or below" -- that statement DOESN'T APPLY TO RESMED MACHINES, in my opinion. With a ResMed machine, if the AI (apnea index) is extremely low and it's the HI (hypopnea index) making up the bulk of the AHI number, I'd say "Your AHI should be below 10 with a ResMed machine." Only then would you be really comparing AHI's between ResMed and Respironics machines, imho.

If you're going to compare AHI's between various brands of machines, you really must take into consideration that each manufacturer uses his OWN DEFINITION of "hypopnea" -- and that definition is different from one manufacturers to another. The different definitions are going to yield different identifying and reporting results. And that's perfectly ok.

I think you should either quit looking for "5", or else convince yourself to chop the HI that you see in half, then re-add your AI and HI (the new chopped one) together to see an AHI that can be more realistically compared (imho) to the AHI you'd be seeing from other brands of machines.

That doesn't mean one brand is right about the HI and another wrong.

It doesn't mean that one brand misses hypopneas that another brand sees.

It doesn't mean that one brand "over-reports" hypopneas or that another brand "under-reports" them.

It's purely a matter of definition of "hypopnea." None of the manufactures uses the same definition of "hypopnea" that is used in a full PSG sleep study anyway, so I'd not worry it.

. . . Mentally cut the ResMed HI (hypopnea index) in half and do some re-adding, if you want a more realistic comparison of your AHI to the "ought to be under 5" AHI that people who are using Respironics machines often talk about. . . .

Now for one of Velbor's famous charts:



And now back to me and my personally irritating ways:

The only reason I am harping on this point, Jason S., is that it can be disturbing to users of some brands, particularly newbies, when they see their numbers are so different from the numbers obtained by users of other brands. So I hope you don't mind. I just think it is important that they be comforted in knowing that YOU CANNOT COMPARE AHI EFFECTIVELY AMONG THE DIFFERENT BRANDS. IT IS TRENDING DATA ONLY FOR THAT ONE PERSON USING THAT ONE BRAND.

Fun poll, though.

[Jason S.]

I believe it is the rule, not the exception, since brands of machines use very different definitions for what they are estimating, as often explained by rested gal and proved by Velbor's chart:

You believe what is the rule, that ResMed machines register AHI differently than other machines? I don't dispute that, and I wasn't referring to that when I said "exception". What I am saying is that if you take the subset of users with ResMed machines, and then further take the subset of users with ResMed machines who have high HI indexes relative to their AI indexes, that sample of people is small relative to the total sample, and therefore statistically insignificant.

[Jason S.]

Another telling fact:

Do sleep doctors or physicians consider, inquire, or care what brand of CPAP machine is being used and thus qualify their opinions based on the answer? Certainly they do not!

[carbonman]

My AI is consistently below 1, but my HI is generally between 5 and 8. That HI would likely be 2.5 to 4 on a Respironics.

My AI is consistently below 1, but my HI is generally between 5 and 8. That HI would likely be 2.5 to 4 on a Respironics.

I said most people. Exceptions such as yourself would not be statistically significant in a large sample.

I agree w/JNK.
I don't consider myself an exception nor statistically insignificant.
With 2 complete mths of data from the S8 Auto, I completely
concur w/JNK. My daily AHI, before /2 is above 5, consistantly.
AI is .2 avg and HI is 5avg.
At first I was very disappointed w/this machine.
With my Mseries, w/a year of trend data, my AHI was 1.5, consistantly.

What do these numbers mean to me now.....not much more than trend.

How do I feel??? Much better since I got the Resmed.
Bottom line for me. I will be getting another Resmed backup machine.

What does this mean for others?? YMMV

[Georgio]

Typically less than one.

[jnk]

I believe it is the rule, not the exception, since brands of machines use very different definitions for what they are estimating, as often explained by rested gal and proved by Velbor's chart:

You believe what is the rule, that ResMed machines register AHI differently than other machines? I don't dispute that, and I wasn't referring to that when I said "exception". What I am saying is that if you take the subset of users with ResMed machines, and then further take the subset of users with ResMed machines who have high HI indexes relative to their AI indexes, that sample of people is small relative to the total sample, and therefore statistically insignificant.

The point, Jason, is that statistically speaking, because of the way ResMed counts HI, most ResMed users have high HI indexes relative to their AI indexes. They are the majority in the sample, I believe, and they are therefore particularly significant statistically. That is why the point is made over and over on the board--because it is the majority, not a small subset.

I do not make the point to attack the poll. It is a good poll. I participated. None of the polls here are meant to be scientific. Maybe I shouldn't even have mentioned it. Sorry to be a pain. But just a little sorry.

jeff