AutoPAP, Activa, UMFF and Mouth Taping

[Velbor]

Please follow the link to my document at:

http://velbor.home.comcast.net/cpap/2009%2001%2022.pdf

The bottom lines of this data story:

 Narrow AutoPAP ranges are not necessarily better than wider ranges.

 Mouth taping can adversely affect AutoPAP therapeutic performance; while it reduces or eliminates leak, it can also degrade therapeutic effects.

 Masks are not just comfort interfaces; different masks differently impact AutoPAP performance, in terms of pressure profile, and in terms of therapy results.

[mindy]

Velbor,

Nice job on the detail. I would, however, want to see variance estimates before assuming the numbers really mean anything. What is the standard deviation or standard error?

The number of nights may or may not be sufficient to reach your conclusions - depends on the variance!

Mindy

[jnk]

Here are my sincere, if off-base, questions that come to mind: What about the theory that mouth leaks compromise the reliability of the data? Couldn't it be that when the numbers are "better," it is actually because the machine was unable to detect some events because of the mouth leaks? Or could it be that increased pressure from the machine, as it attempted to compensate for mouth leak, may have overshot a bit and resulted in increased experienced pressure to the point of improving efficacy of treatment? Is there no amount of pressure that can get your AI below one when mouth-taping? Is "narrow range vs. wide range" so much the issue to address, or is it, rather, "a lower minimum vs. a higher minimum"?

Very nice job on the info, and very interesting stuff!

jeff

[Velbor]

I would, however, want to see variance estimates before assuming the numbers really mean anything. What is the standard deviation or standard error? The number of nights may or may not be sufficient to reach your conclusions - depends on the variance!

Point well taken. My night-to-night variation is high; that's why I was careful to provide ranges of nightly values as well as averages. I have not done formal statistical analyses, beyond what I have presented. But my ongoing observation, as I have compiled this data, has been that by the time collection of 20 nights is approached, the measures of central tendency (mean, median) stabilize nicely. With this as a guideline, I have settled upon 20 nights as a reasonable data period for analysis. (Sadly, there's far too much description, on this board, of individuals making therapy changes based on far less data.)

I might also point out that while my selection of therapy from night to night was not "randomized", I did carefully interlace different modalities (when there was more than one overlapping therapy routine) to minimize the effect of any extraneous temporal trends.

Finally, I would note here that I found a typo in my .pdf file: where I state at the bottom of page 4 that Activa and UMFF AI's were "0.8 and 0.2 respectively", that is incorrect; the numbers should be reversed to read "0.2 and 0.8, respectively." Doesn't sit right to modify the original document once people have looked at it. Velbor

[mindy]

I agree that you've been very careful and explicit but means particularly are not very useful without standard deviations (or error). If we don't know the variance, we don't know if it's enough data and if it's significant, imho.

Mindy

[Velbor]

.... means particularly are not very useful without standard deviations (or error). If we don't know the variance, we don't know if it's enough data and if it's significant ....

Hard to argue with good logic. I should have done it right from the beginning. That's the value of peer review. Thanks for your prodding. Laziness is no excuse.

Fortunately, there was room in the tables for an additional entry. (If there weren't, laziness WOULD have been my excuse!) And my raw data was set up to easily have Excel do a stdev function. Not wanting to fiddle with the original file once its been posted, I've left it alone, and uploaded a new one. The only differences are an added subtitle, the addition of standard deviation to the tables on pages 1 and 2, and the correction of the error I cited previously.

http://velbor.home.comcast.net/cpap/200 ... 0Stats.pdf

I will leave formal statistical critique to others, other than to opine that, on quick glance, the sd's appear to support rough equivalence of similar data, and non-overlapping values (particularly in AI) where differences were the point. Velbor

[mindy]

Thanks for adding that, Velbor.

Here's my take on the data:

1. Difference in ranges: probably no significant difference in therapy results although there is a difference in the 95% pressure. I'd bet your titrated pressure is around 12.

2. Activa and tape Vs. UMFF: probably no significant difference in pressure values, no significant difference in AHI, possible significant difference in AI, no significant difference in HI. Possibly significant difference in time in apnea.

3. Probably no significant differences in the 95% pressure or median pressure among the three options. Probably no significant difference in AHI, AI and HI more problematical (pressure may have impacted these results in terms of # apneas and # of hypopneas). Time in apnea looks like it may be borderline.

With what I'm seeing, I'd suggest several steps:

1. Collect more data
2. Do a formal analysis - eg. check for normality of data and then do formal significance test based on normality.

Until then, imho, it's not really possible to draw conclusions at this point.

Great idea and execution and I hope you will follow up. If you want any help with the analysis, I'd be happy to do so.

Mindy

[ozij]

Thank you, Velbor for that systematic data.

O.

[Velbor]

Why might the presence or absence of tape, when using the same nasal mask, make such a significant difference in therapy? My speculation is that the taping of my lips may eliminate snoring, which is one of the drivers of the ResMed AutoSet algorithm.

I hereby RETRACT the above proposal. Following an effort to obtain additional data regarding my previously described observation of increased apneas when taping, I am now convinced that impact on snoring is NOT the mechanism of action.

I utilized an older ResMed S7 AutoSet Spirit machine with a ResLink attachment, which allowed me to obtain and review highly detailed data, including snore data, which are not otherwise observable with standard ResMed software (AutoScan or ResScan). My data showed NO difference in snoring levels between the Activa mask with or without mouth taping (and similarly no difference using the Ultra Mirage Full Face mask). The machine showed NO evidence of ANY significant snoring for me, regardless of mask.

Although I promptly discontinued this line of research (these data, while lacking statistical rigor, presented what I view as results clear enough to demand immediate cessation of the study), the few days of data from the S7 / ResLink were consistent with the more copious S8 data I had presented previously.

At AutoPAP settings of 6 to 20cm, the S7 results were:
UMFF: AI = 0.8 (0.2% time in apnea) and AI = 1.0 (0.4% time in apnea)
Activa NO tape: AI = 0.4 (0.1% time in apnea) and AI = 1.4 (0.4% time in apnea)
Activa WITH tape: AI = 5.2 (1.9% time in apnea; 8h 25m run time)

As I previously documented, leak rates due to “lip flutter” for the Activa WITHOUT taping, while perhaps higher than ideal, remain within acceptable parameters for accurate machine sensor function and maintenance of therapeutic airway pressure.

Examination of the detailed ResLink data suggest no mechanism to explain these observations. While I cannot speculate as the CAUSE of decreased AutoPAP efficacy when taping, I remain convinced that my data confirms the existence of this unexpected phenomenon. While arguments against mouth taping are generally unconvincing, these findings suggest (at least for some individuals, and at least for some interface devices) that the exercise of caution would be prudent.

Regards, Velbor

[mindy]

Thanks, Velbor for all the effort and care you put into evaluating your data. It's a challenging conundrum and it would be interesting to see of someone else can replicate your results.

Mindy

[Velbor]

I have previously provided data demonstrating that, in AutoPAP mode, mouth taping results in a significant increase of both rate of apneas and overall time in apnea. These figures show the nightly average, range of values, and standard deviation:

Thus, use of the Activa without tape reduced the apnea index by a factor of 14, and the percent time in apnea by a factor of at least 11. (Information about the UMFF is provided as a reference only. Effect on AHI and HI is of less convincing statistical significance.) I had also previously demonstrated that leak involving the Activa without tape was not sufficient in my situation to cause significant sensor or therapy failure; data shows the percent of total therapy time time:

My assumption was that this phenomenon of mouth taping worsening apnea scores was related to the AutoPAP algorithm. However, newly compiled data using fixed-pressure CPAP demonstrates the same pattern:

Once again, use of the Activa without tape reduced the apnea index by a factor of 14, and the percent time in apnea by a factor of 11. (As before, Information about the UMFF is provided as a reference only, and effect on AHI and HI is of less convincing statistical significance.)

As I had previously demonstrated that leak involving the Activa without tape was not sufficient to cause significant sensor or therapy failure, that continues to be true despite the higher median pressure (9.2 or 9.5 on auto, vs. fixed pressure at 11.0):

Thus, even with my known “lip-flutter” (Activa leak rates when tape is used show an almost perfect mask-face seal), leak which might potentially be problematic is present for only a small fraction of the night.

I can offer no explanation for these observations. I can only assure that the data collection and compilation was carefully and accurately managed. The conclusion demanded by the data is that the deleterious effect of mouth taping on apnea is a physiologic phenomenon rather than an algorithm-related artifact.

Other recommendations which might be seen to flow from these data are:

● If this effect occurs in one person (myself), regardless of how “unique” my situation may be, it is not unlikely that it can occur in other persons.
● There is no known way of predicting who may be affected by this situation.
● Therefore, a “cavalier” attitude of recommending mouth taping for persons with leak “problems” should be avoided.
● Leak “problems” need to be quantified. If 95th centile leak (using the ResMed paradigm of correcting for normal mask venting) is 0.38 l/s or less, potentially “significant” leak is occurring during less than 5% of the night, which is arguably acceptable.
● If mouth taping is used, and if AI is above 1.0, a careful analysis of whether the taping is helpful or harmful should be initiated.

Regards, Velbor

[rested gal]

The conclusion demanded by the data is that the deleterious effect of mouth taping on apnea is a physiologic phenomenon rather than an algorithm-related artifact.

Thinking about the physiologic side of it -- Velbor, do you usually have nasal congestion? Have you had an ENT exam?

What I'm wondering... wonder if there is a fairly constant issue with congestion (allergies? sinus problems?) or enlarged nasal turbinates that might be a factor when you used the Activa with taped mouth and absolutely had to breathe only through your nose all night?

If so, nasal congestion, for whatever the reason, could be an issue for quite a few other people. I can well imagine there could be others who would have similar comparison results to yours.

Thanks for posting your charts, Velbor. Thanks, too, Mindy for your interesting comments. This is all very fascinating.

[ozij]

I utilized an older ResMed S7 AutoSet Spirit machine with a ResLink attachment, which allowed me to obtain and review highly detailed data, including snore data, which are not otherwise observable with standard ResMed software (AutoScan or ResScan). My data showed NO difference in snoring levels between the Activa mask with or without mouth taping (and similarly no difference using the Ultra Mirage Full Face mask). The machine showed NO evidence of ANY significant snoring for me, regardless of mask.

As I previously documented, leak rates due to “lip flutter” for the Activa WITHOUT taping, while perhaps higher than ideal, remain within acceptable parameters for accurate machine sensor function and maintenance of therapeutic airway pressure.

Velbor, I assume that lip flutters, or open mouth breathing, disable the machine's ability to identify an apnea or a hypopnea. As we know, that identification is based on various properties of a series breaths taken previously, (2 minutes on a Resmed) and to my untutored mind, it seems these parameters change dramatically once part of the air blows out of your mouth. Having seen many mouth leak reports in my data (with nary an even occuring during leak time) I agree with you that a mouth leak does not effect therapy by disabling proper pressure levels - we know some machines can, and do automatically compensate up to a certain level. But a mouth leak (unlike a mask leak) does not let the proper pressure build up against the obstruction in your upper airway - the machine supplies the compensating pressure, but the pressure does nothing against the obstruction - the air pours out of your mouth. No events reported deos not mean no events happening.

Examination of the detailed ResLink data suggest no mechanism to explain these observations. While I cannot speculate as the CAUSE of decreased AutoPAP efficacy when taping, I remain convinced that my data confirms the existence of this unexpected phenomenon.

Your interpretation of the results assumes mouth taping effects therapeutic efficacy, mine assumes mouth leaking effects measurement efficacy.
So a possible explanation of your results would be as follows:
When you are congested, have a flow limitation, apnea or hypopnea, you try to open your mouth.
Without tape, your mouth opens, and the machine can no longer do proper analysis of your breathing - it doesn't register the apnea.
With a full face mask, your mouth opens, you're getting pressurized air through your mouth, the machine records properly, and responds properly.
Taped: Your open albeit taped mouth lets some air blow out your cheeks, instead of open your obstruction, or, alternatively, you've opened you mouth due to some congestion, and the machine has to fight your apneas through a congested nose. Both might be reason to prefer a full face mask and not to tape.

while arguments against mouth taping are generally unconvincing, these findings suggest (at least for some individuals, and at least for some interface devices) that the exercise of caution would be prudent.

Agreed.

O.

[Velbor]

But a mouth leak (unlike a mask leak) does not let the proper pressure build up against the obstruction in your upper airway - the machine supplies the compensating pressure, but the pressure does nothing against the obstruction - the air pours out of your mouth. .... Your interpretation of the results assumes mouth taping effects therapeutic efficacy, mine assumes mouth leaking effects measurement efficacy.

Ozij, we've had this conversation before, though not in quite as explicit terms. This is where you lose me. You appear to view a mouth leak as something qualitatively different from a mask/skin leak. I'm not convinced that this is so. I assume that pressure equilibrates very quickly within the system (machine / tubing / mask / airway), and that the machine - and the body - has no way to distinguish between mouth leak and mask/skin leak. The pressure in the mask is the pressure in the airway, and this is so whether or not there is leak, and whether leak is from the mask/skin interface or whether it is from the mouth. If the machine is able to compensate for leak by maintaining pressure in the tubing / mask, then that same pressure is also present it the airway. Leak, from whatever source, either overwhelms the entire system, or it does not. But connected parts of the system will be at the same pressure.

I also do not see that there is any difference in "measurement" accuracy between the two situations. The machine (PB aside) knows only the current pressure (and the airflow) at the machine's port. Pressure is effectively the same as that in the tubing and mask, which is the same as that in the airway, regardless of origin of any leak. Admittedly, airflow in a mask/skin leak does not enter the airway and so is more likely to be constant, while mouth leak airflow may or may not vary between inspiration and expiration. Still, it's precisely the periodic breathing-related airflow variation which the machine does register and analyze. If the mouth leak airflow is very large, overwhelming the inspiration/expiration differences, it would seem MORE rather than less likely that the machine would call an apnea. In contrast to your description, I do regularly see apneas in my own reports, even during "mouth leak" episodes.

I can see no practical difference between the two types of leak, other than that there is more drying of the airway when the leak flow comes through the mouth. Nor have I seen documentation that during an apneic event, persons reflexively attempt to open their mouths.

That a full face mask might be the preferred option for me is a definite possibility. However, I am unconvinced that a nasal mask, without tape, is providing inaccurate data, so long as pressures remain within machine specifications, regardless of leak source. If it decide to use a FFM exclusively, it may well be because I have concluded that the mouth leaks themselves are causing arousals and thereby interfering with my sleep.

By the way, for RG, I have no significant congestion or abnormalities in my nose, turbinates or sinuses. My seasonal allergies are well controlled by systemic antihistamine and nasal corticosteroid spray (and in any case they are not relevant at this time of year).

Thanks, Ozij, for trying to provide an explanation for my "anomalous" data. Regards, Velbor

[freepostg]

Velbor,
Tremendous work..thanks for providing such interesting data...I think I'll not use tape or chinstrap tonight and see what happens. It's been quite a while since I've done this and am hoping that I've somewhat trained myself to keep my mouth shut and limit lip flutters since my last go!