Reliability of Objective Data Download Information: Part I


In a prior post, we discussed “when the numbers don’t add up,” noting several discrepancies in the ways in which patients report their subjective outcomes on various scales measuring symptoms of sleepiness or insomnia. The key observation from that post demonstrated how a sleep physician may have a difficult time sorting out how to move forward with patient care when the measurement scales provide inaccurate or misleading information due to patients’ confusion or otherwise poor skill at accurately gauging their own symptoms. That post dealt with the subjective side of patient sleep problems, and this post will delve into the objective numbers, particularly those related to the more frequently used data downloads from PAP devices. 

A major problem in this area of clinical care is all data downloads are not equal in reliability, accuracy, and utility. Different PAP manufacturers obviously attempt to come up with the best design and algorithms to collect information about breathing events, degree of mask leak, hours of use and level of compliance, the latter often based on some insurance-related metric. We will examine each of these items as well as other more intricate or subtle aspects of the data that may arise.

As we use ResMed devices in the majority of our patients, the information provided in this post will be highly relevant to users of ResMed devices, but general overlap with other devices can still be assumed with one caveat. Each device manufacturer clearly uses different algorithms to gauge things, and the most striking difference would likely be found in the realm of measuring flow limitation events (FLE), the other term used to designate respiratory effort-related arousal or RERAs. Both FLEs and RERAs are presumed to be equivalent, and both serve as indicators or measures of discrete breathing events either for UARS if the events meet diagnostic criteria for that condition or in OSA patients who show additional FLEs on either their diagnostic or titration sleep studies in addition to their apneas and hypopneas. 

To clarify the above terms, as a reminder when the airway is shutdown (Apnea), it only takes a little positive airway pressurize delivered by a PAP machine to begin the process of opening the airway. At this point in the treatment of the Apnea, we are usually converting the complete cessation of breathing into a wider opening that is only 50% closed (that is, also 50% open). We label this breathing event Hypopnea. As we continue to apply PAP, the airway opens further but does not reach normal yet, so let’s use the percentage 75% open (or 25% shut). At this stage, we are looking at an FLE or RERA status. Notice with each of these progressive stages of opening of the airway, more pressure was required. Thus, when we reach the last treatment step, that is, adding still more pressure to convert the FLE to a fully opened airway, we have also reached the highest pressure level needed by the patient. Keep these points in mind because they reflect on how we would choose to alter someone’s pressure settings if we only have the objective data download (ODD) to make a decision.

Also, to further confound the issue, we do not know the specific measurements or algorithms used by each of the PAP manufacturers when they designate these three main breathing events, which clearly means there are circumstances where the same individual could try three different brands of PAP devices on the exact same settings and yet the ODD from each one would not provide identical data. There’s no getting around this issue, but as you will see one of the main themes of this post points to the reliability or lack of reliability of this information.

With this backdrop, let us start with the breathing event indexes from the downloads, because they often prove the most revealing of the patient’s response to PAP therapy, and in some cases these indexes also prove the most misleading. In the ResMed system, we are provided with an Apnea Index (AI), Hypopnea Index (HI), Apnea-Hypopnea Index (AHI). Next, it also gives a breakdown of whether or not the breathing events are Obstructive or Central or of an Unknown type. Theoretically, if someone is receiving an excellent to outstanding response to PAP, then we would expect zeroes across the board for all these indexes. However, more commonly we will see small numbers, measured in fractions, for the majority of patients, something like AI = 0.2 and HI = 0.8, so AHI = 1.0. With such numbers, most likely they would all be labeled obstructive, which would then read 1.0.

In the above situation, we would expect the patient to report excellent outcomes, but such numbers would be very misleading if the data download also showed a marked degree of persistent flow limitation. On the ResMed device, the term flattening is used, and we can gauge from the graph on the download record whether or not the the degree of flattening appears clinically relevant. Sometimes, we see flattening occurring the whole night long and at an intense severity, which strongly suggests the patient will not be reporting a good response. In fact, this finding is a very common occurrence at our sleep center, because despite our very aggressive approach in treating UARS, we routinely observe that a subset of patients cannot immediately tolerate higher pressure settings needed to eliminate the flow limitation. However, once these patients have adapted both psychologically and physiologically to pressurized airflow, they can usually tolerate higher settings. In the current example, we would explain to the patient the need for raising the settings to treat the UARS, and most patients are eager to attempt this trial at the new, higher settings.

What if the flattening does not occur all night? Then, the chances are high that some other factor is involved in causing the problem. The two most obvious examples would be a change in mask seal or a change in body position or sleep stage. And, making matters more simple and complex simultaneously, it could be all three things occurring at once: the person turns on his back, enters REM sleep, drops his jaw, loosens the seal of the mask, develops mask leak and mouth breathing, and as a result of all these changes, the flattening graph on the download now spikes to higher levels of UARS for anywhere from 5 to 30 minutes. Then, seemingly for no apparent reason, the flattening disappears, but in reality, the patient may have changed position again, exited REM sleep, and now the mask fits well again so the leak or mouth breathing ceases. 

What if the flattening is all night long but of low intensity, is it still an indication to raise pressures? Again, we want to know how the patient is responding to PAP therapy. This scenario is more subtle because the patient might say she is doing really well, but upon further probing we note she still wakes up at least once per night to urinate (nocturia), or she notices residual and seemingly unexplained fatigue. Overall, it is exceedingly difficult for most sleep apnea patients to sort out with great certainty their most accurate description of what they understand to be normal sleep. Because they have seemingly reversed problems with insomnia, nocturia, fatigue or sleepiness or all of the above, they do not necessarily expect more benefits and may suffer from the problem of the “ceiling effect”. Gentle probing to stimulate discussion will often reveal the patient has no good reason for waking up at night to urinate, and therefore tweaking the pressure settings or returning the patient to the sleep lab for a retitration usually turns up something useful. If the patient shows persisting flow limitation in the sleep lab, usually the pressure tweaks can normalize the airflow signal, and the patient will report benefit. Guesstimating the pressure tweak based solely on the ODD is something to consider, but its value tends to be more general. That is, if the patient reports improvement, we know we moved the settings in the right direction, but we do not know if we have determined optimal settings.

Recently, a mildly obese young man returned to clinic for a routine follow-up after having initiated an ABPAP device last fall following a split night diagnostic/titration study. At the appointment, things appeared to be going reasonably well. His subjective symptom reports and his insomnia and sleepiness scores were all on the very low to slightly borderline symptomatic levels, and the patient was very pleased with his results. Leak was low, averaging only 5 lpm, his respiratory event indexes were nearly perfect with Apnea Index: 0.1 Hypopnea Index: 0.1 AHI: 0.2, all deemed obstructive. He was compliant with his device for insurance criteria, and he averaged more than 5.5 hours per night for a 60 day period. He consistently used his device 6 nights per week and therefore was actually on PAP for 6 to 8 hours on any given night. His pressure settings were ABPAP 14/8, PS 3, but he offered a mild complaint they felt too low, so pressures were raised to ABPAP 15/9, PS 3. Last, the ODD graphs showed breakthrough snoring and flow limitation, but the degree of these residual breathing events was notably on the low side, nearly at the level of just registering on the scale provided by the download. We might have expected the raised pressures to solve the problem. Yet, the patient had never undergone a full night titration since having started with ABPAP 8 months ago after the split night.

On the titration, the patient responded very well objectively to noticeably higher pressures up to 20/14, PS 4 and was subsequently prescribed new settings of 18/12 PS 4. Most remarkably, the patient reported sleeping “much better,” which in our clinical judgment equates to the patient having done well previously but in reality he had been experiencing the ceiling effect. To raise the pressures by 3 cm for both inspiratory and expiratory settings indicated the flow limitation and breakthrough snoring were actually much large factors as residual breathing events than could be surmised by looking at the graphs. Nonetheless, their presence on the ODD, the complaint of too low pressures, and the absence of a full night of titration data since his original sleep study raised our suspicions for a less than optimal response. As an additional bonus, we observed a mild amount of leg jerks on this study (a repeat finding from the first study) that appeared to be causing independent arousal activity, something that cannot currently be tested or measured on the ODD. This finding motivated the patient to continue his Vitamin D supplements for recently noted Vitamin D deficiency.

In the second case, an elderly male with mild obesity had not seen us in 4 years since having started ABPAP therapy. He returned to clinic and reported substantial benefits from the device and missed only 1 night of use in the past 60 days while averaging more than 6 hours per night. However, he did report residual sleepiness, which he attributed to a very heavy work schedule and some residual nocturia episodes for which he had no explanation. Leak was low, and the residual breathing events showed Apnea Index: 0.8 Hypopnea Index: 0.2 AHI: 1.0, still on the very low side, but there was the additional finding of Obstructive: 0.5 and Central: 0.3. As in the first case, he also suffered negligible but nonetheless breakthrough snoring and flow limitation. His current pressure settings were ABPAP 19/10.2 PS 5.2, yet at the retitration study they were increased to 23/13, PS 6, a substantial raise from what he had been using for the past 4 years. With these new pressure settings he experienced only 1 central apnea on the titration and reported sleeping “much better,” another case where the patient had likely been experiencing a ceiling effect for some time for which further evidence was provided by his decision not to return for follow-up until 4 years after having started PAP.

In both examples, we were seeing subtle findings on the ODD. We would like to have seen more demonstrable snoring and flow limitation to believe the retitrations were warranted. However, the ODD was most useful in hinting at these problems, which in fact turned out to be substantial in light of the large pressure swings required to normalize the airflow signal. As a result, the patients experienced large subjective gains in sleep quality on just one night of the new raised settings. In the next post, we will continue our analysis of ODDs and how to maximize their use in the clinic or the sleep lab.

Barry Krakow MD


Dr Krakow’s 27 years of sleep research have focused on the complex relationship between physiological and psychological sleep disorders. Dr Krakow currently operates private sleep medical center, Maimonides Sleep Arts & Sciences, Ltd., and serves as Classic SleepCare’s paid Medical Director.

Leave a comment

Comments have to be approved before showing up.