Low AHI and Poor Sleep

[musculus]

For a lot of people with SDB, the AHI cannot measure the sleep quality. Below I have 2 figures:

1. From one of my sleep studies. You can see how an airway resistance (see the waveform and corresponding chest movement) almost instantly caused a microarousal.



2. From my biPAP (11/8). You can see the airway blocked (rough waveforms, delayed inhalation) and opened (smooth & round waveforms) repeatedly without an "event" detected. Nonetheless, the sleep quality was compromised. I suspect many will see similar things in your cpap data.

[Sclark08]

My numbers are great but my data shows like yours. And Im still tired

[musculus]

My numbers are great but my data shows like yours. And Im still tired

It's arousal that interupts your normal sleep structure. Apnea or hypopnea are just those arousals that happen a little bit late after airway blockage. There are potentially many more microarousals that are caused by airway resistance but will never be noticed if you only look at AHI.

Sleepyhead is not very good at picking up flow limitations. Or did I do something wrong with the settings?

[SleepWrangler]

Sleepyhead is not very good at picking up flow limitations. Or did I do something wrong with the settings?

I believe SleepyHead reports on the data as recorded by the PAP machine. It doesn't have algorithms or do any analysis to flag anomalies of its own.

[robysue]

Sleepyhead is not very good at picking up flow limitations. Or did I do something wrong with the settings?

SH doesn't "pick up" anything about SDB. SH simply shows you the data your machine records.

The PR System One BiPAP that you are using may or may not even try to flag flow limitations of the sort you are after: The PR S1 BiPAP PRO simply doesn't try to flag FL at all and the PR S1 BiPAP Auto only flags them when run in Auto mode. That's how PR designed the machines and SH can't change that.

That said: When the PR S1 BiPAP Auto is running in Auto, it is not clear what criteria PR decided to use to flag FLs. Certainly they're looking at a certain amount of raggedness in the inhalation part of the breath, but just how ragged it has to be and whether certain kinds of raggedness (chairs vs wiggles vs overall flatness vs angularity etc) are more heavily weighted than other kinds is simply not known to us the mere users of this equipment. It's also not clear how many breaths in a row need to exhibit raggedness before a FL will be flagged by the PR S1 BiPAP Auto or APAP when running in auto mode. For all I know the decision to flag a FL may depend on both the severity of the misshapen inhalations and the number of misshapen inhalations. Perhaps fewer badly misshapen inhalations are needed to score a FL than less badly misshapen inhalations.

You can (as your example shows) use SH to zoom in on the wave flow data and manually look what you believe to be flow limitations followed by what may be "recovery" breaths. The PR RERA algorithm attempts to score respiratory patterns that display increasing flow limitations followed by clear recovery breaths, but quite frankly it's not clear just how reliable the RERA algorithm is.

Finally: In the snippet of data you posted, those "apnea-like" events at the beginning of what you've manually flagged don't get scored because they're less than 10 seconds in length. The other places that you've manually flagged only involve one or two breaths at a time. And that may simply be too few breaths to trigger an FL flag if your machine is running in Auto. As stated above, if your machine is not running in Auto, then it won't even try to flag something as a FL.

Please note: I'm not disputing what you are saying about flow limitations and microarousals. As near as I can tell you have a very plausible interpretation of the data that you've posted. And it is known that microarousals can indeed be very disruptive of some people's sleep.

All I'm trying to do is to explain why this stuff was not automatically flagged as something by your machine.

[djhall]

I know I am probably just stating the obvious, but our machines don't actually look for arousal other than guesses at Respiratory Effort Related Arousals which aren't counted in the AHI. The algorithms look for an arbitrary cutoff of >50% reduction in airflow for > 10 seconds and flag that. The arousal and sleep disruption is assumed to be present if that condition is met. Of course, strings of 49% reductions for 15 seconds, or 99% reductions for 9.5 seconds, very likely cause terrible sleep quality and arousal as well, but that won't get flagged and added to AHI. Restless legs, PLMD, etc can also all cause arousal, but don't get flagged by the machine.

These machines are a vast improvement over no sleep data at all, but they are rather blunt instruments looking for mostly for severe problems. The finer stuff still requires the added data that a full sleep lab brings to the table.

[SleepWrangler]

These machines are a vast improvement over no sleep data at all, but they are rather blunt instruments looking for mostly for severe problems. The finer stuff still requires the added data that a full sleep lab brings to the table.

I assume the issue for a patient is knowing that these arousals are there can the doctor specify better Rx settings or has the patient hit the limitations of the technology. Even if a sleep lab has better tools the PAP device is still just the PAP device and it can only make corrections based on the limitations of its sensors and algorithms.

The finer stuff still requires ...

Is that what we're discussing or is there still major stuff left untreated because of device limitations? Also are the current limitations because we've exhausted research or are we just waiting for monitization?

[musculus]

Thanks for the clarification. The data was recorded with PR S1 BiPAP PRO, however, I do have an PR S1 BiPAP Auto and will give it a try. The auto mode is not useful for me since the pressure increase is just not fast enough. I guess I can set min=9 and max=12 and see if it picks up any flow limitations.

I am not 100% sure about the microarousals marked in the second plot (no one can without a simulatenous EEG recording). I just assumed there must have been nervous system stimulation/arousal to make the airway more open. Below is the summary plot of the sleep study. My N3 sleep was frequently distrupted, much more often by microarousal than by apnea/hypopnea. AHI from that study was only 2.5. I checked the full sleep study data and found many microarousals classified as 'spontaneous' were actually related to airway resistance.

Sleepyhead is not very good at picking up flow limitations. Or did I do something wrong with the settings?

SH doesn't "pick up" anything about SDB. SH simply shows you the data your machine records.

The PR System One BiPAP that you are using may or may not even try to flag flow limitations of the sort you are after: The PR S1 BiPAP PRO simply doesn't try to flag FL at all and the PR S1 BiPAP Auto only flags them when run in Auto mode. That's how PR designed the machines and SH can't change that.

That said: When the PR S1 BiPAP Auto is running in Auto, it is not clear what criteria PR decided to use to flag FLs. Certainly they're looking at a certain amount of raggedness in the inhalation part of the breath, but just how ragged it has to be and whether certain kinds of raggedness (chairs vs wiggles vs overall flatness vs angularity etc) are more heavily weighted than other kinds is simply not known to us the mere users of this equipment. It's also not clear how many breaths in a row need to exhibit raggedness before a FL will be flagged by the PR S1 BiPAP Auto or APAP when running in auto mode. For all I know the decision to flag a FL may depend on both the severity of the misshapen inhalations and the number of misshapen inhalations. Perhaps fewer badly misshapen inhalations are needed to score a FL than less badly misshapen inhalations.

You can (as your example shows) use SH to zoom in on the wave flow data and manually look what you believe to be flow limitations followed by what may be "recovery" breaths. The PR RERA algorithm attempts to score respiratory patterns that display increasing flow limitations followed by clear recovery breaths, but quite frankly it's not clear just how reliable the RERA algorithm is.

Finally: In the snippet of data you posted, those "apnea-like" events at the beginning of what you've manually flagged don't get scored because they're less than 10 seconds in length. The other places that you've manually flagged only involve one or two breaths at a time. And that may simply be too few breaths to trigger an FL flag if your machine is running in Auto. As stated above, if your machine is not running in Auto, then it won't even try to flag something as a FL.

Please note: I'm not disputing what you are saying about flow limitations and microarousals. As near as I can tell you have a very plausible interpretation of the data that you've posted. And it is known that microarousals can indeed be very disruptive of some people's sleep.

All I'm trying to do is to explain why this stuff was not automatically flagged as something by your machine.

[musculus]

I think the issue is currently there is no EEG device available that is at the same time portable, convenient, reliable/accurate, and economical for in-home sleep monitoring. If such device exists and be integrated with PAP machine (a mobile sleep lab), things will be much easier.

These machines are a vast improvement over no sleep data at all, but they are rather blunt instruments looking for mostly for severe problems. The finer stuff still requires the added data that a full sleep lab brings to the table.

I assume the issue for a patient is knowing that these arousals are there can the doctor specify better Rx settings or has the patient hit the limitations of the technology. Even if a sleep lab has better tools the PAP device is still just the PAP device and it can only make corrections based on the limitations of its sensors and algorithms.

The finer stuff still requires ...

Is that what we're discussing or is there still major stuff left untreated because of device limitations? Also are the current limitations because we've exhausted research or are we just waiting for monitization?

[djhall]

These machines are a vast improvement over no sleep data at all, but they are rather blunt instruments looking for mostly for severe problems. The finer stuff still requires the added data that a full sleep lab brings to the table.

I assume the issue for a patient is knowing that these arousals are there can the doctor specify better Rx settings or has the patient hit the limitations of the technology.

I guess my point was that we don't ever actually know any arousals are there from a PAP device. We just look at breathing patterns, guess at what that means for sleep quality, and assume arousals must accompany breathing disruptions of sufficient severity and length. The smaller the breathing irregularities observed, the less confidence we can place in the assumption that an arousal or disruption is happening, and it also becomes less likely that we can assume breathing issues were the cause of the arousal if it was there, and not some other issue like PLMD, or even some other perfectly normal event like a restless bed partner or loud noises from the street.

At some point it is quite possible the only thing we know is that the breathing pattern changed somewhat, and that may or may not have accompanied an arousal, which may or may not have been the result of a medical issue or just something normal in our sleep cycle or a normal reaction to our sleep environment (loud noise, restless bed partner, etc). That puts us way out on a chain of speculation when we start guessing what the smaller breathing changes mean or don't mean. That doesn't mean it is worthless, as it may well indicate the presence of something that should be looked at using more sensor data than a PAP device has, but I don't expect we will be able to improve much outside of RERA detection using breathing only other than perhaps a statistic that reflects the likelyhood that some unknown sleep issue is causing other disruptions and a sleep lab evaluation using more sensors may be warranted.

[SleepWrangler]

I don't expect we will be able to improve much outside of RERA detection using breathing only other than perhaps a statistic that reflects the likelyhood that some unknown sleep issue is causing other disruptions and a sleep lab evaluation using more sensors may be warranted.

Your detailed answer wrt the algorithm suggested there are some useful but arbitrary limits and thresholds. Makes me hope that there is room for improvement of existing CPAP platforms using existing technology through more advanced algorithms, configuration of same during titration study, and additional flagging and corrections using on-the-fly heuristics.

Only six weeks into therapy, AHI as measured by my CPAP device, doesn't seem to correlate very well to my state of sleep health. Peak hourly AHI is lot more indicative. But you know, peak AHI is also a better approximation of the time my breathing flow spends in a crescendo-diminuendo pattern with hypopneas, clear airway, and obstructive events so I think OP is onto something. Hopefully after six months of therapy I won't give a crap because I'll be too darn healthy to care.

[robysue]

Thanks for the clarification. The data was recorded with PR S1 BiPAP PRO, however, I do have an PR S1 BiPAP Auto and will give it a try. The auto mode is not useful for me since the pressure increase is just not fast enough. I guess I can set min=9 and max=12 and see if it picks up any flow limitations.

Is your BiPAP Auto a Series 60 or Series 50 machine?

If your PR S1 BiPAP Auto is a Series 60 you can fool it into acting just like your BiPAP Pro does by setting:

• Min EPAP = EPAP setting from the BiPAP Pro
• Max IPAP = IPAP setting from the BiPAP Pro
• Min PS = Max IPAP - Min EPAP
• Max PS = Max IPAP - Min EPAP

These settings will induce:

• Max EPAP = Max IPAP - Min PS = Max IPAP - (Max IPAP - Min EPAP) = Min EPAP = EPAP from your BiPAP Pro
• Min IPAP = Min EPAP + Min PS = Min EPAP + (Max IPAP - Min EPAP) = Max IPAP = IPAP from your BiPAP Pro

And hence the pressures will NOT change during the night even though the machine is in BiPAP Auto mode, and so it will record FL without trying to respond to them. NOTE: This trick won't work on a PR S1 BiPAP Auto Series 50 machine because the default min PS cannot be changed from 2cm.

[musculus]

Thanks for the clarification. The data was recorded with PR S1 BiPAP PRO, however, I do have an PR S1 BiPAP Auto and will give it a try. The auto mode is not useful for me since the pressure increase is just not fast enough. I guess I can set min=9 and max=12 and see if it picks up any flow limitations.

Is your BiPAP Auto a Series 60 or Series 50 machine?

If your PR S1 BiPAP Auto is a Series 60 you can fool it into acting just like your BiPAP Pro does by setting:

• Min EPAP = EPAP setting from the BiPAP Pro
• Max IPAP = IPAP setting from the BiPAP Pro
• Min PS = Max IPAP - Min EPAP
• Max PS = Max IPAP - Min EPAP

These settings will induce:

• Max EPAP = Max IPAP - Min PS = Max IPAP - (Max IPAP - Min EPAP) = Min EPAP = EPAP from your BiPAP Pro
• Min IPAP = Min EPAP + Min PS = Min EPAP + (Max IPAP - Min EPAP) = Max IPAP = IPAP from your BiPAP Pro

And hence the pressures will NOT change during the night even though the machine is in BiPAP Auto mode, and so it will record FL without trying to respond to them. NOTE: This trick won't work on a PR S1 BiPAP Auto Series 50 machine because the default min PS cannot be changed from 2cm.

Thanks a lot for the instruction and formula.

It 's a PR SYSTEM ONE BIPAP AUTO 750, not a REMSTAR. It has a min PS at least 2 cm and max PS at least 3 cm. Now I set max IPAP = 11 cm, min EPAP=8 cm, min PS = 2 cm, and max PS =3 cm. The EPAP will be between 8 and 9, IPAP will be between 10 and 11, both within my comfortable range.

[SleepWrangler]

It 's a PR SYSTEM ONE BIPAP AUTO 750, not a REMSTAR. It has a min PS at least 2 cm and max PS at least 3 cm. Now I set max IPAP = 11 cm, min EPAP=8 cm, min PS = 2 cm, and max PS =3 cm. The EPAP will be between 8 and 9, IPAP will be between 10 and 11, both within my comfortable range.

Good work. If things go as planned these troubling flow issues you noted will show up in the device sleep metrics somehow? Hopefully you will have a reliable method to flag flow problems. Could be more reliable and accepted by sleep doctor than manually spot checking SH.

[ironhands]

AHI hasn't gone above 1.8 in 3 months, and I'm still just as exhausted in the morning. Lethargic, and I'd do just about anything for even an extra 5 minutes in the morning. still showing around 30-50 spontaneous arousals on my PSG's