flow limitations expressed as an index?

[sheepless]

idk if this is the right forum for this. 

I wonder if it would be possible, and if others' think it worth considering, to express flow limitations as an index in addition to the measure of 'severity' we see now. it strikes me that duration of flow limitations, or the proportion or density of usage time with flagged flow limitations, is at least as important and useful information as the severity of discrete fl events. my uninformed guess is that's where a difficulty might lie, inasmuch as the flagged events are discrete, untimed. but when I look at my fl graph, I can see that apart from the degree of severity there may be many flags or just a few, so maybe some measure of density would capture that aspect for night to night or setting to setting comparisons. I guess med and 95%tile tells us something about the level of severity over time but I don't think that's quite the same as how much of the night is flagged as flow limited. maybe it's as simple as frequency of flow limitations flags divided by usage time. a step further might be to have the ability to select periods of time within which this index is recalculated (which I believe is done in oscar for other metrics).

I'm thinking specifically about resmed machines since that's all I've ever used. idk if the ways other mfgrs report flow limitations would complicate this.

[sheepless]

3 weeks without a single response can only be interpreted as a resounding lack of interest. still, I'm curious why, if anybody cares to say.

[Sleeprider]

I don't know how we would arrive at an index. I often go by what I see in the 95% FL statistic when evaluating a need for pressure support or EPR, but I have seen considerable differences between individual results at the same 95% flow limit. I generally like to see less than 0.1 in that statistic, and it is still disruptive for some people and impossible to get for others. Apnea and hypopnea are expressed as an events per hour index, and I don't know if that's possible for FL. I think your thread starter didn't pick up participants because it is unclear what an index consists of, or how it might be calculated.

[sheepless]

I'm not sure I can say it any differently. I think it's as simple as wanting to know how much of the night my breathing is flow limited, independent of whether it's 0.1 or 1.0 in severity. I too depend strongly on med and 95%tile but they don't tell me if I had 1 minute or 8 hours of flow limited breathing.

I guess I'm thinking that an entire night of low level flow limitations (that don't register high enough in the med and 95%tile columns to be considered problematic) could be as detrimental as a brief bout of of 'severe' flow limitations (with the correspondingly and comparatively higher entries showing up in those columns).

as you know from other threads, I'm not particularly well-versed in calculations, so I can't suggest anything beyond frequency (or duration if that's available) divided by hours of use. frequency data must already be available in order to figure the median severity. I suppose an index of frequency could also be weighted by degree of severity.

I understand that we tend to have much more flow limited breathing than the machines flag. I can imagine trying to capture the unflagged flow limitations based on shape and amplitude would be very difficult. still, I think, simply comparing the 'density' per hour of Flagged flow limitations from night to night, setting to setting, would be useful enough.

how flow limitations are treated would be the same so maybe it doesn't matter that much but I still think useful another means of determining the significance of flow limitations in a given individual at given settings on a given night, in addition to severity (without duration) and how-you-feel reports.

what is 'severity' anyway? some measure of reduction of airflow? now compared to a few minutes ago?

[Crimson Nape]

Does Custom CPAP Event Flagging, listed in File > Preferences > CPAP, not help your quest?

[sheepless]

well, I'm not sure. the machine already flags flow limitations by some definition. I don't want to change the definition, which is what I think custom event flagging does. and we already know severity. rather, I want to know cumulative duration or average duration per hour; i.e., how much of my night is characterized by flow limited breathing and how does that compare to other nights and settings. 1x/hr or 20x/hr? idk if x is minutes/seconds, events or something else. depends on what data the pap machine produces.

[pholynyk]

I think we need a combination of duration and severity. The ResMed doesn't 'flag' flow limitations, it samples a value every 2 seconds.

When we look at the whole night, it looks like we have discrete spikes:
   

But if we zoom in, we see another story:
   

Some events are low valued, but may last 4 to 8 or 10 seconds. Some have an irregular shape, higher, maybe shorter or longer. I suspect the higher value events are more disruptive.

I'm not sure how to calculate a meaningful index, given this sort of data.

[SarcasticDave94]

For my own therapy, I looked at flow limits via mentally converting the number to a percentage. Maybe it's not useful for your goal.

[GuyScharf]

Flow limits also have different values on different machines. For example, on the DeVilbiss DV64, flow limits are reported as events and have values of none, mild, moderate, and severe.

[sheepless]

that seems manageable (said optimistically with no basis in actual knowledge of such things).

event frequency. event frequency/hrs. frequency of mild (moderate, severe) events. frequency of mild (moderate/severe) events/hr. [later maybe distribution of fl events by pressure or time of night].

since severity is already reported in oscar, for resmed at least, maybe only need frequency.

rather than trying to fit all machines and all data into a single chart of rows and columns under the daily/details tab, maybe:

a) oscar layout could be specific to machine type which is listed in the profile or otherwise automatically detected

b) there could be drop down lists in submenus under a heading like flow limitations to click to report different things, like total frequency, frequency/hr, frequency of mild, frequency of mild/hr, this night, this night compared to last night, this week, this month... either preset categories or something like a search menu that allows us to select a date range and the info we want to see displayed.

it could quickly get complicated so if others' think there might be some use for this, frequency/hr by whatever measure is appropriate per machine would be a simple start to test the utility of the metric.