Understanding OSCAR statistics

[GusHauck]

I am new to using OSCAR and trying to understand/interpret some of the statistical calculations. I think I understand the conceptual meaning of the terms, but the reported statistics leaves me scratching my head. For example - Flow Limitation. What does a "99.5% Flow Limit: 0.31" or "95% Flow Limit: 0.16" mean? 

Beyond the glossary of terms, which is great for understanding the concept, is there a primer on the statistical measures and interpreting them?

[BoxcarPete]

Those stats are the 95th percentile and the 99.5th percentile of the relevant event or parameter. That means 95% of the time your flow limits were below 0.16 and 99.5% of the night you were below a 0.31 flow limit. The primer for interpreting them is to look at how close the two value are together. The farther apart they are, the more isolated and quickly resolved your FLs are. The closer they are together, the more FLs hang around at those values. Double is pretty typical from the handful of charts I've seen. Mine are spikier than that, my 95% is like a third or a quarter of my 99.5% on most nights.

[GusHauck]

Thanks. That explains a lot.

[paulag1955]

But what's considered far apart as opposed to close together?

[BoxcarPete]

Dunno, I've only seen my own charts in great detail. But when I get FLs, it's almost always either a weird looking couple of breaths, or a series of breaths that get flat-topped, then the flat tops start to slope down, then the machine starts registering FL, then the machine increases pressure for me, then the breaths go round again. This doesn't happen very frequently, but FL gets kind of high at its peak, so I usually see something like 0.07~0.11 95% and 0.24~0.41 99.5% with the two somewhat proportional to each other at a 3.5-4x ratio.

[Gideon]

Think of 99.5% as the max. It's purpose is to eliminate unrealistic spikes. I go off of the 95% value.

95% there is no medical rule. From experience we know that 0.10 is high enough to be bad and acted on, we prefer to see 0.02 - 0.03 as a value.

With a 95% value of 0.16 most likely you should be increasing your EPR to better manage your flow limits.

[paulag1955]

0.10 what? What is the scale?

[dataq1]

Resmed takes a bunch of criteria and then scales the aggregate from 0 to 1.
O represents excellent, very good, nice, beautiful.
1.0 represents terrible, awful, horrible, really really bad.

One of the leading criterion is the shape of breath shape.

[paulag1955]

So are you saying they don't represent any particular value? They're just arbitrary numbers?

[dataq1]

It’s a grading of a bunch of different elements that Resmed has inferred might represent or are potentially indicative of restrictions to free flow. 
For example a “perfect” inhalation curve is a nice rounded trace. But a trace that has a flattened top, regardless of the volume of air, is considered indicative of something that is a possible restriction. 

As far as I know, the actual grading of the individual  elements is a trade secret.

And of course you see that the flow limitations value is the principle trigger to increase pressure (for auto adjusting machines)