Re Centrals (and other events):
Yes, a lot of newbies get worried (even freaked out) because they see a central or two scored overnight. And they shouldn't. Of course, we also see newbiew who get worried (even freaked out) because they see an OA or two scored overnight ...
Fact is: Our machines are not perfect. But if you understand the limitations on the algorithms used for scoring the events, you do come to realize that our PAP machines really are technological wonders, and you do develop a real appreciation for the relative accuracy of most of the data they record.
For most of us, we ARE asleep MOST of the time we're using the PAP; and hence it's not an unreasonable for the manufacturers' algorithms to make the assumption that all breathing
is "sleep breathing" and score apneas and hypopneas based on defined measurable decreases in airflow as compared to the running 5-10 minute baseline airflow. Two readily identifiable things can affect how the reported AHI and real AHI are related. Recall:
- reported AHI = (number of scored events)/(run time)
- real AHI = (number of real events)/(actual sleep time)
If the number of scored events is relatively close to the number of real events, then (lots of) wake time will tend to make the reported AHI look smaller than it should be because (run time) > (sleep time) implies:
reported AHI = (number of scored events)/(run time) < (number of scored events)/(actual sleep time)
But having lots of FALSE events, turns the inequality around:
reported AHI = (number of scored events)/(run time) > (number of real events)/(run time)
And so the reported AHI would be greater than the real AHI if run time and sleep time are roughly the same.
For people with "run time" roughly equal to "actual sleep time", the number of FALSE events caused by wake breathing is also likely to be low, and so the reported AHI and the real AHI will be in the same ballpark.
Likewise, MOST of the time the central/obstructive apnea detection algorithms works decently enough; although it's useful to remember that these algorithms are far more likely to misclassify a real CA as an "OA" than they are to misclassify a real OA as a "CA".
But I still think the PR RERA detection algorithm is not a reliable as the algorithms used for scoring OAs, CAs, and Hs. A real RERA requires an arousal, and while there are some clear earmarks as to what the flow rate trace of a RERA looks like, it's not at all clear if the PR RERA algorithm detects most RERAs (i.e. misses very few real RERAs) or scores lots of things as RERAs that aren't (i.e. scores a lot of non-RERA breathing as RERAs).
That's also my concern about the SensAwake algorithm on the F&P: Without the EEG, the machine is taking a (hopefully pretty good) statistical guess that you are are awake. And it lowers the pressure, perhaps substantially when it thinks you are awake. Now if the SensAwake algorithm is more prone to missing real wakes, all that leads to is a bit of middle of the night "potential" discomfort when the patient is awake. But if the SensAwake algorithm tends to score "false wakes"---i.e. if it thinks you are awake when you are really still asleep---then the fact that the machine lowers the pressure down might just be an issue.