ResScan UnderReporting or SmartLink OverReporting or ?

It seems Bill is content with leaving his post without explanation - though it seems (to me, anyway) that he is intimating that my profile lacks the critically important 'patents of nobility' (for those of you who might have seen Heath Ledger in 'A Knight's Tale') [/sarcasm OFF].

So to provide some context, the following (FWIW):

billbolton wrote: As I've said before on a number of occasions......

• I do not have any interest or involvment of any kind in Resmed, other than as a user who purchases their products on the commercial market.
  
  I do work in the Healthcare sector, in a senior public e-Health Informatics role which has only a general connection to SDB, or no connection to any equipment vendor or retailer.
  
  I am a professional grade member, in good standing, of the Institute of Electrical and Electronic Engineers (IEEE), the Australian Computer Society (ACS), and the Health Informatics Society of Australia (HISA)
  
  I do work on a voluntary basis with Australian organisations whose role is to assist SBD sufferers.

Cheers,

Bill

Bill (et al):

As I think I explained early on and quite unlike Bill, my interest in this is nothing more than as a simple user of xPAP equipment. I came into this totally ignorant and being that I recognize I do not know what I do not know - I expect I remain largely ignorant of many facets of this 'journey' (for want of a better term). In fact, that acknowledged ignorance is what led me to post here - in part, so that I might ask questions of those more knowledgeable (such as, presumably, Bill - though his refusal to provide substantive answers is evident throughout this topic, at least), and also to offer thanks and appreciation to those whose posts have been extraordinarily helpful - and there are many, but the one person who comes to mind immediately is 'restedgal.'

In this particular topic, I was (and to a very large extent remain) xPAP vendor agnostic. I have no vested interest in any of the vendors, and am only interested in using the equipment that is most effective for me. OTOH, there seem to be a few ResMed proselytizers who get a bit cranky whenever a viable alternative is discussed or a criticism is directed at ResMed. Oddly enough, I like my S9 and have been, I think, fairly upbeat about it and that is likely to continue. I also like my DeVilbiss IntelliPAP. Neither are perfect, but both (IMO) are excellent. Having said that - I should go back to the acknowledged ignorance I just mentioned in the previous paragraph. My only other point of reference is a Respironics M-Series Plus machine. These are the 3 devices I have direct first-hand experience with, and it is a very small sampling of the broad range of machines.

OK - since I last posted data, here is an update:

Dec 06/07 (IP): AHI 1.8, 95th Percentile Pressure 7.9 cmH20
Dec 07/08 (S9): AHI 0.0, ODI 9.9, 95th Percentile Pressure 7.5 cmH20
Dec 08/09 (S9): AHI 0.5, ODI 21.3, 95th Percentile Pressure 8.6 cmH20
Dec 09/10 (S9): AHI 0.3, ODI 18.6, 95th Percentile Pressure 10.3 cmH20
Dec 10/11 (S9): AHI 0.7, ODI 17.2, 95th Percentile Pressure 8.6 cmH20

APAP ranges for all days were set at a minimum of 6.0 cmH20 and a maximum of 12.0 cmH20.

Here are the aggregate summaries for the two devices:

AHI - S9: 0.39
AHI - IP: 2.63

95% - S9: 7.9
95% - IP: 6.4

ODI - S9: 13.3
ODI - IP: 7.0

When I first logged on to CPAPTALK, if I had seen these summary stats, I likely would have concluded the S9 is the superior device simply because of the much lower AHI. I now know that AHI values from one vendor to another are not directly comparable - and that is quite clearly the case with ResMed and DeVilbiss. Reported AHI values from the ResMed S9 seem to be much lower than reported AHI values from other manufacturer's devices, hence, the ResMed S9 AHI values are useless for comparison to any other AHI values.

As mentioned upthread, I am not very confident of the value of ODI - and plan to open a separate topic asking other knowledgeable members about their experiences with pulse-ox metrics. Who knows, maybe even Bill (and other ResMed proselytizers) will offer something of genuine value to THAT topic. That would be novel.

Still, in terms of ODI and the 95th Percentile Pressure values, the IntelliPAP results are better than the S9 results - and these metrics are, I believe, directly comparable.

In terms of the IntelliPAP, my wife and I both recently noticed that it has gotten just a bit louder. Enough that I am going to make a sound recording of both machines while operating, and then ask DeVilbiss to review to determine if the IntelliPAP needs servicing. The two machines have a 2 dB difference in their specifications with the IntelliPAP rated at 26 dB and the S9 at 24 dB and the S9 is definitely quieter. If it does need service, that will offer an opportunity to report on the customer service aspect of the device in the future.

Comments (especially productive ones) are welcome.

One thing to keep in mind is that the machines react according to how they score the events. If a machine is "over scoring," it will react with a higher pressure because adjusting the pressure is the only option available to the machine.

Someone posted a paper where there was a correlation between S8 AHI and sleep study AHI, but the S8 AHI was over scored by about 4.2, if I remember correctly. I don't know how other machines stack up to sleep study scoring, but the machine is only looking at the data from its perspective and making adjustments accordingly.

It is possible that pressures obtained from an APAP machine may be higher than those from CPAP during a sleep study. If we follow along this path, higher pressures result in more problems with leaks, and that results in compliance issues because people get discouraged.

It would be interesting to see how the different machines stack up to actual sleep study data.

HoseCrusher wrote:One thing to keep in mind is that the machines react according to how they score the events. If a machine is "over scoring," it will react with a higher pressure because adjusting the pressure is the only option available to the machine.

Someone posted a paper where there was a correlation between S8 AHI and sleep study AHI, but the S8 AHI was over scored by about 4.2, if I remember correctly. I don't know how other machines stack up to sleep study scoring, but the machine is only looking at the data from its perspective and making adjustments accordingly.

It is possible that pressures obtained from an APAP machine may be higher than those from CPAP during a sleep study. If we follow along this path, higher pressures result in more problems with leaks, and that results in compliance issues because people get discouraged.

It would be interesting to see how the different machines stack up to actual sleep study data.

Thanks for the comments.

>>Someone posted a paper where there was a correlation between S8 AHI and sleep study AHI, but the S8 AHI was over scored by about 4.2, if I remember correctly.<<

And in the case of the S9, it *seems* ResMed made a change in their algorithm so that reported AHI values are consistently reported much LOWER than other machines - and in stark contrast to ResMed's own S8.

One of the advantages of the IntelliPAP is that the scoring values can be changed by the user - with the intent being to allow the end-user the ability to replicate (as much as possible) the results from their PSG.

>> If a machine is "over scoring," it will react with a higher pressure because adjusting the pressure is the only option available to the machine.<< and;
>>It is possible that pressures obtained from an APAP machine may be higher than those from CPAP during a sleep study. If we follow along this path, higher pressures result in more problems with leaks, and that results in compliance issues because people get discouraged.<<

You raise 2 very good points. In the case of the S9, the reported AHI values are LOWER than other machines, and LOWER than their own S8. To the point their AHI values from the S9 are useless for any comparative purposes, except for trends using that one machine.

Pressures produced by the S9 are consistently higher than the IntelliPAP and the S9 consistently reports a lower AHI. The ResMed apologist might suggest this means the S9 is providing 'superior therapy' - but that is not my experience.

The other VERY important point you make is that higher pressures can (and probably due) lead to higher rates of non-compliance. Again, one of the advantages of the IntelliPAP is their algorithm is explicitly designed to produce gradual pressure changes and to keep pressures at the lowest possible settings. That algorithm priority is evident in several of the graphs I posted upthread, and it is very evident when using the IntelliPAP.

Thanks again.

DHC wrote: . . . AHI values from the S9 are useless for any comparative purposes, except for trends using that one machine. . . .

I personally consider that to be the case with ALL "efficacy" data from ALL home machines. Still I agree, in principle, with your singling out the S9, for two reasons:

First, ResMed's failure to publish specifics of HOW AHI data is scored in the S9, especially how the breathing baseline is now calculated, I personally consider to be a slap in the face to every user and every clinician. Either that, or they are just chicken. (No offense to Ro(o)ster; there's room for all in the barnyard.)

Second, when trending data often bottoms out with a number of patients, it is no longer as useful for trending, in my opinion. Lowering the usefulness of trending data in order to imply some (impossible) correlation with PSG data is a bad choice to make, even if the reason for doing so is understandable--to keep from scaring or discouraging users/clinicians.

But hey, what do I know? I got no business spouting off about this stuff, as if I have any idea what I'm talking about. Silly me.

DHC wrote:
It looks like the S8 AutoSet defines them as:
Apnea - 75 % decrease in flow for a period of 10 seconds (simplifying the chart upthread)
Hypopnea - 50 % decrease in flow for a period of 10 seconds

DeVilbiss defines them thusly (again simplifying):
Apnea - 90 % decrease in flow for a period of 10 seconds
Hypopnea - 50 % decrease in flow for a period of 10 seconds

Have you by any chance looked at the raw data curves? noticed any visible differences between the time histories of a segment scored as an event?
When one gets to a 90% decrease of low pressure data, one wonders about the amplitude resolution of the data.

Message Redo 2 hrs later: so very sorry, I didn't see 8 more pages of additional data and sometime heated discussion.
Having been dxed with OSA in a sleep study at AHI 55, and after 2 weeks ago on an S9 Autoset, oscillating in the AHI 10-25 range, I fail to feel the excitement on discussing results below AHI 5. I don't believe anybody mentioned a figure for measuring and reporting error ranges, or of units, or of reference pressures being the same. So I'll go back and look for more info, hopefully coming back more informed.
But among all you Electrical Engs and Physics PhD, how come nobody has thought of an undergrad. signal processing technique: take the pressure signal segment in the region corresponding to apneas and convert it into the frequency domain (maybe export ResScan to Matlab), then examine the content.