Accuracy of S9 Autoset statistics?

There is no absolute clear line dividing an apnea from normal breathing. Even when you are in a sleep lab with a fancy CPAP machine, EKG, EEG, a chest belt, and O2 sensor, two experts will count some events as apneas, and not count some other events. Some experts don't consider an event to be an apnea unless your SPO2 level drops. Your CPAP machine only has access to airflow and pressure data, not any of these other sensors. Different CPAP machines use different scoring algorithms and can produce different scores with the same data.

Because of this, AHI scores can vary widely.

As someone else already said, your pressure with the CPAP may not be the same as when you did the sleep study. You may have different EPR settings. If you're on auto pressure, pressure varies.

To a large sense, these factors don't really matter that much in most cases. Really bad apneas will get scored by any algorithm. Numbers may vary for "borderline" events. If you're getting really high apnea numbers, there's probably something wrong.

I'll also agree that apnea numbers for periods of time when you were awake are pretty meaningless and will screw up your scores.

Rather than just looking at numbers of events, I think it's necessary to look at actual flow waveforms and see how "real" the apnea event was.

Sometimes an autopressure machine can't really figure things out. Sometimes, they even "take off" and go to a high pressure. If so, pick a manual pressure or set a narrow auto pressure range and stay with it.

I think the OA (Obstructive Apnea) numbers are pretty good. If you score an OA, something happened. The problem is that an 11 second apnea counts as one event the same as a 120 second apnea.

I'm a lot more skeptical about hyponeas, flow limitations, central apneas (CA), and RERAs being scored by a CPAP machine without an SPO2 sensor or other instrumentation. When I look at my own flow data, there are a lot of events that show up that don't look like anything of importance when I look at the waveforms.

My concern is that I hear a lot of people (including doctors) panicking because of someone having a high number of CAs. I'm worried that "minor" or "maybe" CAs from a CPAP machine are being regarded as "gospel" and someone is unnecessarily spending a lot of money or being saddled with an invasive bilevel or ASV machine that they don't really need. Or simply someone suffering unnecessary worry.

Don't panic about numbers, unless they're really high. Use them as a warning that further investigation is necessary.

Pugsy wrote:It is a bit hard to see but it looks like those groupings with the centrals are associated with breaks in the scoring. In other words wakeups. Then you put the mask back on and have some more groups and these could be sleep onset events or even toss and turn arousal. The machine doesn't know if you are awake or not. It just calls them as it sees them.

Thanks Pugsy. You are absolutely right, those are times I woke up during the night. You actually have me realizing all my apparent pauses occur shortly after times I wake up during the night. Hmmm.

Thanks for explaining the other data points on the chart. Thanks also to jy2e71 and Marcman for explaining these points. I thought the CPAP was an overpriced fan but I guess it's an overpriced fan with an onboard computer too apparently. Cheers.

archangle wrote:

Don't panic about numbers, unless they're really high. Use them as a warning that further investigation is necessary.

Thanks archangle this is very good advice. I can see how tracking these #s too closely could give me an unneeded panic attack! If I notice a difference and improvement in how I feel that is the best result I can hope for down the line anyway.

soundersfootballclub wrote: You are absolutely right, those are times I woke up during the night. You actually have me realizing all my apparent pauses occur shortly after times I wake up during the night. Hmmm.

If you don't go back to sleep immediately those events being flagged after the wake ups may not be "real" events in the sense that you need to be concerned about them. So I wouldn't be overly alarmed and wouldn't dwell on the numbers so much right now. Especially those events occurring during possibly sleep onset. With time and experience things should settle down a bit so that you are waking so often and then you will have a better idea what is really going on.

My initial concern was with the central index and that is why I wanted to see the actual reports to see when they were occurring. A few random centrals are normal. To have clusters of them at sleep onset can be normal also.
Just watch things and as time goes by you will have a better handle on all this stuff.

If you can isolate the cause for the awakenings and work on that you would be ahead. Fragmented sleep for any reason will mess with how we feel the next day. If it were me that would be number one on my list right now. That and simply getting used to sleeping with all this "stuff" attached to us.

Hi, I am a clinical specialist for these products and I'd like to help you a little.

The results of the S9 Autoset should be pretty accurate. There is a condition called "Complex Sleep Apnea" which is when people start developing central apneas in response to being on CPAP therapy. It's estimated that as much as 15% of people with sleep apnea actually have complex SA (per physicians at Mayo clinic published studies). The ResMed S9 device has pretty accurate central apnea detection and the results should be considered as reliable. It might get a little better as you get more used to the therapy. But if it doesn't, you might be better served by being switched over to the ResMed S9 ASV machine - which is the treatment for complex and central sleep apnea conditions. ResMed is also the only CPAP company that has central apnea detection approved and cleared by the FDA.

As far as your other questions...
Minute Ventilation: Amount of air you breathe in a minute (size of breath x respiratory rate)
Flow Limitation: a partially obstructed breath (throat is halfway closed)

ResMed S9 Auto devices respond with pressure increases anytime it senses snoring, flow limited breaths and obstructive apneas. And depending on the severity of the event, it will vary the amount of pressure of increase. If a central apnea is detected the device will not increase the pressure (as this could make the problem a little worse). Once the breathing is stabilized the device will slowly start to decrease back to a lower level, but will increase again if you need it. The goal of these devices is to give you the lowest amount of pressure that you need. Pressure needs will vary throughout the night, depending on sleep stage, body position and time of night - all of which will vary the severity of the events.

Please let me know if you have any further questions. Good luck and keep working with your device... your heart will thank you!

One more thing...
As far as your leak question...

A high leak will be recorded anytime the mask either gets dislodged, taken off, etc and left to run. I wouldn't be concerned with the maximum leak. When they do a download of your data the clinicians will be looking at either the "95th percentile" and "median" leak values. The 95th percentile leak means that your leak was at this level 95% percent of the time or lower. The median is the middle number that your leak reach for that night. Most clinicians are really only focused on the 95th percentile leak. Plus as long as you are on a ResMed device, the device will autocorrect/compensate anytime the leak value is less than 24 LPM. You are on a top of the line product!

Hi all,

I just got my new S9 Autoset. I've been using S8 AutoSet before.

With S9, my AHI is around 1.9 for the first couple of nights. The central apnea index is around 1.4. The pressure stat is at 11.0 and my pressure range is set between 9-15. I am wondering if I need to tweak the pressure range to reduce the central apneas... I'll get an USB adapter for the memory card and download my data and post it...

Thanks.

Rachel72 wrote:One more thing...
As far as your leak question...

A high leak will be recorded anytime the mask either gets dislodged, taken off, etc and left to run. I wouldn't be concerned with the maximum leak. When they do a download of your data the clinicians will be looking at either the "95th percentile" and "median" leak values. The 95th percentile leak means that your leak was at this level 95% percent of the time or lower. The median is the middle number that your leak reach for that night. Most clinicians are really only focused on the 95th percentile leak. Plus as long as you are on a ResMed device, the device will autocorrect/compensate anytime the leak value is less than 24 LPM. You are on a top of the line product!

Rachel. Thanks for the responses. So I have been using the S9 for almost a week and my AHI has been between 6 and 11. Has not dropped below 5. I have it set on APAP with a pressure of 5-12 per my doc's orders. The pressure hasn't gone higher than 10 so far. I am wondering why my AHI hasn't gone below 5 yet. Is this normal considering I am still new to treatment?

soundersfootballclub wrote:I thought the CPAP was an overpriced fan but I guess it's an overpriced fan with an onboard computer too apparently. Cheers.

And a damn good one at that. The ability of the xPAP machines to measure pressure fluctuations in the air stream, and the algorithms used to interpret them, astound me. And I'm a mechanical engineer so technology like this should be old hat to me.

It looks like my AHI is still low, mostly around 2 or so. There are mostly clustered CSAs. I'd like to post my graphs. Which graphs (pressure, flow, leak etc..) are the more important ones to interpret what's going on?
Thanks.