Inverse I:E or Incorrect Reported Times?

[Geer1]

In my time on here I have seen numerous instances of people commenting on the inspiration and expiration values that are reported in OSCAR. The term inverse I:E ratio is brought up sometimes when the inspiration time is longer than the expiration time and in some cases this has even lead to persons wrongly believing they have inverse I:E and in one instance even resulted in getting technicians and doctors being involved. 

In short these numbers are influenced by anything that causes the flow rate graph to cross the 0 flow line during the inhalation or exhalation periods. The most common cause of this is cardiogenic oscillations (little oscillations in your flow data caused by expansion/contraction of your heart) which are present on lots of peoples data.

Almost every night I have "Inverse I:E" because of these values being wrongly reported due to the presence of my cardiogenic oscillations. Here are some examples showing measured times vs reported times. 

   
   

As you can see in the summary data my average reported values are inspiration time of 1.98 and expiration time of 1.92, which would be considered "inverse I:E", I:E should be around 1:2. You can that there are ~2 cardiogenic oscillations per breath which is normal and consistent in my results. The breaths are fairly uniform but below them you can see how the reported inspiration and expiration times (especially inspiration) are varying substantially. Inspiration is fluctuating from 1.67 to 2.91 and expiration from 1.52 to 2.28. You can also see how the measured inspiration time on one of these breaths is 1.427 and the measured expiration time is 2.854 (giving a perfect 1:2 I:E ratio). These numbers are so significantly different from the reported times that they don't even fall within the ranges being reported hence why my average inspiration and expiration times are so wrong and indicating "inverse I:E". 

At least in my data having no cardiogenic oscillations is rare. I had to look at a few different nights data to find this but here is a decent example showing how the reported numbers can be correct if no cardiogenic oscillations etc are influencing the data. As you can see the numbers match in that situation, 1.72 reported vs 1.726 measured.  

   

I am willing to bet that most people that appear to have "inverse I:E" usually just have more cardiogenic oscillations present. 

I hope this post will make users more aware of how easily and often these numbers are misrepresented. Inspiration times are almost always overstated and expiration times almost always understated. The reason for this is because the expiration times end once the first oscillation crosses 0 flow and the inspiration time appears to start at roughly this same time. 

If these numbers look strange my recommendation is to first check the data before drawing any conclusions. I personally believe they are wrong more often than they are right but others may disagree with that statement.

[Sleeprider]

I wrote a wiki on the I:E ratio, and I agree there is a high degree of error in machine data for inspiration and expiration time.  When the issue arises, I generally ask to see a zoomed view of the flow rates, and more often than not, find a normal I:E ratio. The Philips machines are especially bad about this.  In fact, Sleepyhead used to report I:E ratio, and dicussions among the OSCAR development team resulted in removing the I:E statistic, but the individual I and E times are reported.  There have been discussions that the use of a zero flow crossover is a significant source of error with the Philips machines.  I believe Resmed actually uses a fuzzy logic to determine the phase of respiration and this is AFAIK the basis of the time of inspirations or expiration that OSCAR reports.  This fuzzy logic will be one of the considerations the OSCAR team will use in a future release to better report I:E time on Philips and other machines.  Until that update occurs, users are advised to look carefully at flow rate charts rather than rely on the statistical summary data.




Another interesting and related issue is the actual reporting of I:E ratio.  As you probably know, Resmed normally report inspiration time in relation to expiration time in this ratio as "One to Ratio".  OSCAR used to simply divide Inspiration by Expiration and report that as the ratio.  The correct formula it to divide expiration time by inspiration time, which usually yields a ratio greater than 1.0. 

The inspiration and expiration time is more likely to be improperly reported when a longer null flow occurs and, as in your case, heartbeats or other anomalies are present.  It is interesting to note in your chart that peak IPAP coincides with peak inspiration and that the incorrect reporting of inspiration time does not appear to trigger IPAP.  Since the Airsense 10 pressure follows spontaneous effort, this is not a surprise, but I have seen where individuals with your cardiogenic oscillations have the misfortune to be using a Philps machine, the respiratory sync can be very poor as  result of the way Flex works.

[Gideon]

This issue, which has existed pre-OSCAR, is one of the reasons the loader code has been revamped. Before we make alterations in the code we have to understand what the data actually is. We are looking at a major change to the way data is handled as a result. Ultimately this will result in a change in the data model for OSCAR and a better program for all. This has also led to a goal to validate all results displayed and to display them as presented by the vendor or indicate any modification of the data. This means that a 100% line is just that a 100% line, not a 99% or a 95% to eliminate outliers, something that was done for a good reason but the modification was not indicated.

[Geer1]

It is easy for a trained eye to determine where inspiration begins and ends but trying to program a machine to determine this would be quite difficult especially considering all of the different breathing waveforms people can have. 

Overall OSCAR is an amazing program that gives huge insight into results and I applaud all involved in maintaining and improving it. Little inconsistencies like this are minor as long as we know about them and how to interpret them. The hardest part is often convincing users that they need to manually confirm data like this rather than just relying on the numbers given, something I know I am not the best at. I'm good with data, not with people lol...

Sleeprider I too have noticed how the easybreath waveform doesn't apply IPAP until inspiration is nearly complete. I believe this is because it is meant primarily to act as exhalation pressure relief. I have wondered how much changing the pressure waveform could help people with restricted airways (by applying the pressure earlier). Part of me ever wonders if by the time a person increases PS high enough to overcome flow limitations that it is creating problems with too efficient expiration causing the central apneas in some people. I had wondered about people switching to S mode to play with rise time in these situations but it fairly significantly changes both inspiration and expiration parts of the waveform though so may be of minimal help. Trigger and cycle sensitivity are the best options available and seem to be helpful in some. I wish it was possible to use a similar shaped waveform that applied IPAP roughly 300 ms sooner and had the same expiration form.

[SarcasticDave94]

I think that despite the I:E ratio report on OSCAR, it's still a great reporting tool. All of the IT guys doing the programming have done an excellent job at it's creation. Despite "one instance even resulted in getting technicians and doctors being involved", that's me BTW, I think OSCAR is doing a good service. Unless your specific therapy requires I:E ratio to be accurate, it should not be a problem. That would likely include most here on Apnea Board. OSCAR reports everything else pretty accurate as far as I can tell.

As for "one instance even resulted in getting technicians and doctors being involved", that's my business that I discussed it with a doctor and an RT. As I stated clearly in my own post, the RT saw the inverted ratio with her own eyes, viewing it non-OSCAR. Let me say that again, it was my business in going to the doc and RT, not yours. It was made AB public knowledge as I posted it. I'm stating it plainly for the last time, stay out of my sandbox, Geer1. Don't include me in your OCD engineer perfection crusades in rectifying an inverted breathing ratio issue that is being addressed.

[Geer1]

I was leaving you out of this hence why I did not include links or names...

OSCAR is amazing and I would trust its data over an RT for a lot of things. I just wish you would do the same...

[DeepBreathing]

OK guys, let's keep this civil. We've already had to close or edit threads because of bickering and I don't want to do it again.

[Dormeo]

I have cardioballistic artifacts that always appear between breaths. When I was using a ResMed Airsense 10 Autoset, it always reported I times that were longer than E times. When I switched to the ResMed VAuto, my E times were always reported as longer than my I times. Nice!

[car54]

I have also wondered why my respiratory rate has been so erratic as I brought up in my thread " Questions for Pulmonologist"

I think this conversation might be the answer.
All of you have valid points.
This is my take on it, if I am wrong please correct me.
 Unless zero is zero all the numbers that are being talked about need to be investigated by expanding the graphs
 Improvements are in the works for more accurate Oscar results
 It is hard to even measure the inhale and exhale time in my case because I am close to zero for a length of time.
 I have multiple crossovers on the zero line which increases the respiratory rate and makes the graph look erratic

  My worst Airsense results 2017

My worst Aircurve

The worst of today

So now I have to research cardiogenic oscillations and the meaning of it in my breathing. Any comments?

car54

[Sleeprider]

"Cardiogenic oscillations" may not be descriptive of your respiratory variation. That term is Geer1's and describes his interpretation of his results, which show small bumps in the respiratory flow. We see similar or larger crossovers and "chaos" with PLM as well other causes, and your charts above do not seem to be these heartbeat fluctuations. Dormeo said that the change from Airsense to Vauto resulted in the inspiration/expiration times being correctly reported, and it's true the Vauto has a much more sophisticated bilevel algorithm with trigger and cycle sensitivity and inspiration timing controls.

Your zero-flow crossovers are more complex and of a higher magnitude than the ones described by Geer1 at the beginning of this thread. You might benefit from the capabilities of real bilevel therapy in stabilizing the respiration rate by providing a more assertive timing for pressure transitions to help guide your respiratory phase. The flow forms you provided appear to be mostly pauses in the expiration cycle, and a relatively weak inspiration. Although you have EPR set at 3, it appears the Autoset has difficulty determining your respiratory phase, and provides only a fraction of the pressure support during those periods. Part of the problem might be resolved by decreasing inspiation trigger sensitivity, and increasing cycle sensitivity with bilevel. You also have relatively high flow limitation that would likely respond to higher pressure support to resolve, and to top it off, we could use Ti controls to ensure pressure support remains on long enough for you to complete inspiration once it starts. I think your issues with respiration rate and I/E phase determination is more complex than what this thread was started to address, but some of the concepts are applicable.

Here is a good article covering cardiogenic oscillations https://www.ncbi.nlm.nih.gov/pubmed/19439200