[Equipment] Two Machines - Two AHI Readings

[johnalus]

I'm sorry if this has been covered already, but I did a quick search and didn't find a recent thread covering the difference in AHI readings between two machine manufacturers. My appologies.

My primary CPAP is a Resmed Elite S9 and my AHI reading is usually a low 0.3 plus-or-minus 0.2. I have a travel CPAP, a Trascend EZEX, and with the same mask it shows my overnight AHI in the whole number range of 3.0, plus or minus, with most of the events on the Transcend being hypopnea interrupts.

I have OSA. Both machines are set for 12.0ml and I use the Resmed Mirage FX standard mask for both as well.

Should I be concerned about the diffence between the two different AHI readings? Is the difference significant?

=J=

[trish6hundred]

Hi johnalus,
WELCOME! to the forum.!
Though it's a curious thing, I wouldn't sweat it too much, as long as you are below 5, you're ok.
Hang in there for more answers to your question.
Best of luck to you with your CPAP therapy.

[robysue]

My primary CPAP is a Resmed Elite S9 and my AHI reading is usually a low 0.3 plus-or-minus 0.2. I have a travel CPAP, a Trascend EZEX, and with the same mask it shows my overnight AHI in the whole number range of 3.0, plus or minus, with most of the events on the Transcend being hypopnea interrupts.

I have OSA. Both machines are set for 12.0ml and I use the Resmed Mirage FX standard mask for both as well.

Should I be concerned about the diffence between the two different AHI readings? Is the difference significant?

Each manufacturer has their own algorithm(s) for scoring events. And, yes, those algorithms can make a (big) difference in the reported AHI. This is part of why the "official" medical care people tend to say that full data machines are good for trending data, but should not be thought of as being 100% accurate every single night.

My guess is that the definition for scoring a hypopnea on the Transcend is different than that on the S9.

I'm working from memory here, so I may get this WRONG, but let me climb out on the limb anyway:

All full data machines use some kind of a "threshold" for how low the air flow into the lungs must drop before a hypopnea can be scored. Many machines, including Resmed if I recall correctly, require the airflow to drop by 50% for at least 10 seconds for a hypopnea to be scored. But that threshold is more than just a fixed number: The drop has to be 50% of something, and the something is usually referred to as the baseline.

The thing is, different machines use different ways of computing the baseline airflow into and out of the lungs. Most use some version of a "moving average" where the machine is focusing on just the most recent data. Some machines use 5-minute windows, replacing the data one minute at a time. Some machines use 5-minute windows with continuous updating of the window. Some machines use a certain number of breaths, regardless of how long or short those breaths were.

So whether or not a particular stretch of breathing will be flagged as a hypopnea by a particular brand of CPAP depends strongly on how the machine determines the baseline breathing pattern AND the threshold amount the airflow must drop before a hypopnea can be scored.

My guess is that the Transcend's algorithm for scoring hypopneas is more "sensitive" than the S9's. And my guess is that some of the Transcend hypopneas show up as peaks in the Flow Limitation graph on the S9 because the S9 considers them to be "not quite bad enough" to be scored as a hypopnea.

And how much you should worry about the fact that the Transcend's AHI is roughly 3.0 and the S9's is 0.3? I'd say that as long as you are feeling decently enough regardless of which machine you are using, you shouldn't lose any sleep worrying about which machine is more accurate in recording the hypopneas.

[zonk]

Different machines uses different algorithm to score events and different machines from the same manufacturer uses different algorithm

For example, my S8 AutoSet scores heaps of hypopneas while S9 AutoSet hardly score any hypopnea at all
We can go on the details but that was not your question

Other factors might affect the score such as how many hours using the machine, humidity setting and leak rate
Again, I have no idea how Trascend humidifier works (if any) and how leak rate is calculated
The S9 compensate for leak and display/report unintentional leak

[PaytonA]

Zonk,

I would be interested in the details of why your S8 scores more hypopneas than your S9. I tend to get a lot of hypopneas on my S8 VPAP Auto.

Best Regards,

PaytonA

[robysue]

Zonk,

I would be interested in the details of why your S8 scores more hypopneas than your S9. I tend to get a lot of hypopneas on my S8 VPAP Auto.

Best Regards,

PaytonA

Short answer is: Resmed made some pretty serious changes in the algorithms before releasing the S9.

I don't remember all the details, but basically they reprogrammed the hypopnea detection algorithm to require a bit more of a reduction in airflow for an H to be scored and they may also have changed how the baseline is calculated.

[PaytonA]

Now I've got it. I can reduce my AHI by purchasing an S9 machine!! What a marketing ploy. (Just kidding)

[archangle]

Here's what I think is happening.

Some apnea events are in the gray area in terms of how long they last or how completely you stop breathing. They aren't as harmful as "Big Bad Apneas," (BBA). All machines record the BBA's the same. They will differ in whether they count a particular minor apnea event. Even human respiratory technicians will differ on whether to count a particular event.

There's not a 100% precise method for measuring how long an apnea lasts or how much your breathing decreases. Even two well trained human respiratory technicians will sometimes disagree. Different CPAP machines may use different algorithms.

This isn't as bad as it sounds. You get accurate numbers for the bad events, but for minor events, there's a gray area.

BTW, this is yet another reason why you need a fully data capable machine and someone needs to look at the actual waveform data, not just the AHI numbers.

There also might be some differences in your breathing between machines. Features like EPR do affect some people's apnea numbers. Even without EPR, there is a difference between machines in how the pressure in the mask varies as you inhale and exhale. This is probably mostly in the borderline apnea events as well.

[Sleepster]

Reading this thread gives me even more evidence for my belief that the precision of the AHI is not very good. In other words, anything below 5 is considered "good" so clinically there is no difference between a 0.3 and a 3.0. In other words, from a statistical point of view they're equivalent.

On the other hand, for a specific patient, an AHI of 0.3 might feel significantly different from an AHI of 3.0. And for another patient they might feel the same.

And as we see from this thread, they may actually BE the same.

In other words, if the machines rounded off every AHI to the nearest 5 we might not lose any information of any value!

[zonk]

In other words, anything below 5 is considered "good" so clinically there is no difference between a 0.3 and a 3.0. In other words, from a statistical point of view they're equivalent.

0.3 = 3, so if I give 0.3 dollar (33 cents), you exchange for me for 1 dollar (100 cents)
Nice doing business with you

Not only the number of apnea events per hour counts, also the duration of each apnea count too
10 seconds apnea is not exactly the same as 50 seconds or more apnea
'How you feel' is fine but I can tell, I feel a lot better with AHI 0.3 than AHI 3