REMStar/Encore Users: Accurate Apnea/Hypopnea Data Capture?

[littlefeat]

Sorry if I missed prior posting on this issue, but.........

Is everyone confident that the REMStar/Encore setup is accurately identifying and recording apneas and hypopneas? In other words, if we are using daily downloads of info from our machines to manage our therapy, but the machine is (even unintentionally) misidentifying or misrepresenting our apnea and hypopnea events, we have a "garbage in, garbage out" problem, don't we? As I have modeled in my own trend analysis, my numbers per the machine bounce between AHI of 35 and AHI of 3 from night to night, with an average in the mid-teens, and a very gradual overall downward trend over the last 90 days but amazing variety night-to-night.

By way of background, I just had a six-month follow-up split-night study and got all kinds of confusing results from the titration--many RERAs (which were new to me), and no clear trends in how my other respiratory events responded to the titration, in any given position. All we know from the titration is that we didn't successfully control the events during the night, which titrated from 10-15 (original diagnosis was AHI of 70, and titration from 5-12 also was inconclusive, but the doc's best guess from the original titration was 10, so that's where we started this titration). Thus, the recent study didn't really confirm or deny that the nightly numbers from my machine are accurate, or give me pressure guidance other than the doc's current best guess that 15 will be about right. I asked the physiologist at my sleep lab (a prominent PhD who worked with Bill Dement for years) about this inconclusiveness, and he did not have any additional words of wisdom or give me much confidence that 15 would somehow be a magic number. Maybe there is no magic number.

You would think that a titration would show that events decrease as pressure increases, until they are eliminated, but it appears to be much less clear than that. I am aware of the fact that, at higher pressures, centrals can be triggered but that fact doesn't seem to have muddied my waters.

Anyway, back to the base question: d you all think that we get an accurate identification of apneas and hypopneas from our REMStar/Encore daily readouts?

[derek]

I have wondered the same question. It seems that we are forced to accept the veracity of Respironics' hardware/software algorithms with no chance of verifying them. The only way I can see of answering your question would be to do a full clinical study comparing the Respironics data with that scored by a clinician. I wonderr if the manufacturers have to submit performance data for FDA approval?

On rthe issue of titration, my own data shows a minimum in AHI around 10 cm H2O with an increase in apneas at higher pressures:
Dependence of AHI on APAP pressue
This is based on almost 600 hours of data. I think that this is a very useful graph.

derek

[littlefeat]

Derek--

For what it's worth, my physiologist thinks that the machine probably does capture basic apnea/hypopnea info pretty accurately, given that it only has pressure and flow sensor info to go on (although he didn't say why he felt that way), but does not yet feel very confident about the autopap adjustment algorithms. Presently, I am not even using the auto-pap function, but am using my auto in cflex mode and doing the titration manually week-by-week (now up to 13) under the review of my doc.

[MarcG]

Derek,

Nice chart. I notice as well that I sometimes see a high AHI at higher pressure, but I'm not so sure that the high pressure results in the high AHI rather than the other way around. How do we differentiate between the cause (independent variable) and the effect (dependent variable)?

I see the high pressures on high AHI nights. It appears to me that the REMstar is cranking up the pressure in response to events, some of which don't respond to the increased pressure. The absolute number of events is not great at those pressures, but the time spent there is relatively small as well, leading to a high AHI.

Marc