Select Page

Sourced from Classic Sleep Care- Increasing Compliance with Advanced PAP Therapy in PTSD Patients

Our co-authored (Krakow, Obando, Ulibarri, McIver) paper on PTSD and PAP compliance which was recently published, provides me with the opportunity to delve more deeply into the nuances of the research paper itself. To my knowledge, this paper represents a milestone in demonstrating (for the first time) that trauma survivors appear to adapt easier and use more consistently advanced PAP devices (e.g. ASV and ABPAP).

There are many aspects of the paper that also relate to OSA/UARS patients in general and not just PTSD patients. As we delve into it, keep in mind the material may be relevant to many mental health patients using PAP therapy as well as those with more classic presentations of sleep apnea. The most difficult aspect in organizing a retrospective chart review of this type (which was the research design for this study) is the need to avoid cherry picking the data that might make something appear more effective than it really is. Moreover, since our study was not with a sophisticated control group, the results only represent what are called “associations,” which is a lower level of evidence in research. Just because lots of people do well with ABPAP or ASV doesn’t necessarily clarify whether ASV or ABPAP are superior modes of treatment when they were not matched against CPAP devices.

One factor in the research, however, that provides a highly relevant degree of credibility to the work is that every patient in the study had tried CPAP in some way, but failed it, found it intolerable, or outright rejected it. Such patients then are said to serve as “historical controls,” because they tried one treatment in the past that clearly did not work for them and then tried a different mode (ASV, ABPAP) that did work. Of course, you can see if you don’t know all the circumstances surrounding their failure with CPAP, then we can’t be certain the comparisons are valid. For example, suppose before visiting our center, all the patients had attempted CPAP at a sleep center that provided very little support for its patients. Or, suppose the CPAP settings were frequently too low or too high, because the sleep laboratory they used did not take the time to fine-tune their pressure/pressures if the patient had been prescribed an Auto-CPAP (APAP) device, an increasingly common practice. In other words, there are so many variables that could alter our perspective about the nature of CPAP failure, we just cannot know reliably if we should assume CPAP could never work in these individuals.

Another major concern about the flow of patients in this sample is the problem of inherent chart reviews where individuals will have missing data or drop out along the way. Which means how could you declare that your findings are fully representative of the points you are trying to establish? In fact, you cannot, and you must report this problem as a major limitation of the study. Let’s look at our published flow chart to understand the problem:

We start out with 229 PTSD patients who marked their PTSD as moderate or worse and who had been tested and shown to suffer from OSA or UARS. The PTSD scale was a well-known, validated survey known as the PSS-SR published by Foa and colleagues a few decades back. Nonetheless, we did not conduct specific PTSD interviews with the patients to confirm the diagnosis. Instead, we selected the higher severity level (moderate or greater) as a more likely indication PTSD was probably a valid diagnosis from the patient’s perspective.

Next, to examine whether or not these patients would do well with a certain type of PAP therapy, they would also need to undergo a titration study and of course be prescribed a PAP device to use at home. This step brings us to a major design flaw in our work. Because 47 people refused to return for a titration study despite receiving clear information of their diagnosed sleep disordered breathing, and another 35 patients who despite undergoing a titration never filled a prescription to attempt PAP at home. You see that we are already down to a new subtotal in our chart review of 147 who completed the additional steps noted above. But that means we had nothing to look at in 82 (47 + 35) patients or nearly one-third of our original sample.

At this point, we don’t throw in the towel because we know that in medical research in particular, it remains a worthy exercise to make observations and analyze data on active treatment-seeking patients. Although we recognize our data would be more generalizable if all 229 patients had followed through, and in reality patients drop out of care all the time and for numerous reasons. They may even still be treatment-seeking for their sleep problems, but they simply were unwilling or unable to proceed with the PAP approach to treatment.

Of the remaining 147 patients who attempted PAP, some early good news was that only 17 patients provided no further data, because 8 got the device but let us know they stopped using it. Another 9 were lost to follow-up, so we have no info on them. That leaves us with 130 patients who were diagnosed with sleep-disordered breathing, filled a prescription for PAP therapy. All these patients appeared to be making some reasonable effort to use the device at home. There were four criteria we applied to make this observation, any one of which was sufficient to declare the patients a current user:

  • They had recently received a prescription to renew PAP supplies
  • They had a recent clinic appointment where they confirmed they were using PAP
  • They returned to the sleep lab for retitrations and confirmed they were using PAP
  • They had been in contact with the sleep center staff and confirmed their use

With these categorizations, we would have hoped to provide insights on the 130 PAP users, but here again we have the dilemma of capturing downloads off their machines either through our efforts at the sleep center or from data provided by their DME company. In fact, 28 patients had no such data available in their charts, and even after we searched other resources such as contacting their DME company, we were left with a final sample of 102 PTSD patients who met all criteria. This included the essential data acquired from the machine. Now, you see that we are down to a sample that remains sizable and is considered large as it involves 100 patients or more, but it is less than half of the original working group. Again, these adjustments may bias the research because we do not know what the outcomes were for the other folks in the original sample who did not meet all the requirements. As you can see, these are the major weaknesses of a chart review.

Forging ahead, we can still learn a lot from the 102 PTSD patients, but we must respect the fact that these individuals may be more highly motivated to pursue PAP than others who dropped out of care. Thus, when we come to the point of making our observations about their use of advanced PAP devices, we need to be realistic about how our approach may apply only to PTSD patients who are not only treatment-seeking but perhaps with a higher level of motivation than other PTSD patients.

And this flow of patients now brings us to the main data point of the study, which was 59 of these patients (or 58% of the working sample) were not only using the device, but using at a high level. They were all compliant or adherent, using nearly 7 hours per night on 94% of possible nights. Another 19 patients were called “sub-threshold” compliant because they did not meet the insurance standard of 4 hours per night for 5 nights/week or about 20 hours per week. Nonetheless, many of them were nearly compliant because they average more than 25 hours per week.  However, since they might have lower hours on some nights or not used on some nights, they did not meet the insurance convention.

This concept of sub-threshold compliance was a major reason we wanted to conduct the chart review, because we knew that many PTSD patients require a longer interval of time to adapt to a device, even if the device should be an advanced one (more on this idea later). The last group used minimally, averaging a few hours for a couple nights per week. However, as we described later in the discussion section of the paper, the question must be asked, what would make a minimal user become a sub-threshold user, and what would make a sub-threshold user become a compliant user? Clinically, these questions are extremely important to answer, because they directly address some of the difficulties so many sleep centers experience in trying to help patients use their PAP machines.

Just how difficult are these patients to treat? The concept of “difficult to treat” is really hard to define. But we can at least describe some aspects of our sample so you get the big picture. All these patients suffered something approaching severe insomnia. On average, they took more than an hour and a half to fall asleep at bedtime. Once asleep, they would later spend more than two hours awake in bed. The insomnia issues alone are very problematic, since several studies have described how insomnia often interferes with efforts to use PAP therapy.

In addition to the PTSD problem in all the patients, 76% of them suffered from one or more additional anxiety disorders (generalized anxiety, panic attacks, obsessive compulsive disorder), and 82% suffered from one or more depressive disorders (depression or manic-depression). Most of the patients suffered moderately severe sleep apnea with an average AHI of 26 and an RDI of 57.  With this information, we know the difference between these two metrics that our patients (on average) suffered many RERAs. The exact mean was 31 events per hour, so many patients suffered from more RERAs than apneas and hypopneas. And, patients with more RERAs may prove more difficult to treat, because when trying to eliminate the RERAs with higher pressures, expiratory pressure intolerance (EPI) is often triggered.

Indeed, this finding of the high RERA index explains in part the finding that 64 of the 102 patients developed the problem of complex sleep apnea, the condition where central apneas emerge as a side-effect to the use of pressurized air. The other 38 patients were on auto-bilevel (ABPAP).

Now that you know some of the major findings in the research, let’s back up to the beginning of the story to see why we elected to conduct this chart review in the first place. Then, we’ll spring forward again to further elaborate on key clinical points in our patients experiences with advanced PAP. Finally, we’ll end with our discussion section where we talk about how patients and sleep professionals might make use of this research and what new research is needed in this area.

The beginning of the story is not exactly what you might expect. Of course, we delved into how difficult it seems to be for PTSD patients to use CPAP therapy. And, we report on one of the most recent studies from the Walter Reed group in Maryland that shows low rates of compliance—around 30%. But, we also cited a much more interesting and motivating reason to explore PAP in PTSD patients – which is the early findings that treatment of OSA/UARS might actually be associated with a decrease in post traumatic stress symptoms. We cited two studies, the work of Youakim and colleagues in 1998 (1) and our own study in 2000 (2). Although we did not go into the details of these two works, it is worth mentioning here that both of them reported on small samples of PTSD patients who appeared to improve their symptoms by using PAP therapy.

The clinical relevance of our paper could not have been timelier, as several other researchers have published on the same or similar ideas in just the past two years. Just inserting the words PTSD and CPAP into a PubMed search will bring a list of 10 recent papers during 2016 and 2017, covering either the difficulties in PTSD patients using PAP therapy or the effects of PAP therapy on relevant PTSD symptoms. A search of “PTSD and OSA” brings up additional relevant articles. This new input is a radical change from the past where the comorbidity of these two disorders was virtually ignored, except by a very small number of research groups. Now, the advances are coming so fast that there have even been articles looking at the relationships between OSA and their impact on PTSD patients responding to specific PTSD treatments. For example, exposure therapy or cognitive processing technique, which are two of the most advanced and consistently validated therapies for trauma survivors.

We could not be more delighted to see this sudden change in interest. In fact, we co-authored a commentary on this topic, (3) pointing out the fallacy in one study that had attempted to argue that sleep disorders assessments did not required prior proceeding with the treatment of PTSD. (4) In our commentary, we cited the excellent work of my co-authors who demonstrated how the presence of untreated OSA seemed to be associated with lower therapeutic effects in PTSD patients attempting exposure therapy. (5)

In our lengthy introduction in the paper, we also describe two specific areas of interest that relate to the general population of OSA patients, as well as to psychiatric patients including PTSD cases. The first concept is that of “sub-threshold compliance,” a woefully under-researched area in PAP studies. Sub-threshold compliance is also termed “partial use.” Though research has previously shown that using PAP just 2 or 3 hours per night (and in some cases even fewer hours) leads to clear-cut clinical benefits. We often find no support for this clinical pearl in the general sleep medicine community or in dealing with insurers who seem to be caught up in the decidedly non-clinical construct of adherence. Specifically, the Medicare criteria for compliance is 4 hours per night on 70% of nights. This amount of use translates into 4 hours x 4.9 nights per week or rounded up to 4 x 5, which is 20 hours in the week. Yet, no insurance carrier (to our knowledge) will sign off on a patient using PAP 20 hours per week unless the 20 was achieved through the arbitrary and to repeat decidedly non-clinical construct of 4 hours on 4.9 nights/week over a one month period. We’ll delver further into this idea later in the post.

The second key concept is how our approach is different than most others in managing the problem of PAP compliance. Most centers use behavioral change models, psychoeducation, or other coaching techniques. Although our sleep center in New Mexico uses some of these tools as well, our primary focus targets the patients “experiential and emotional reactions to pressurized air.” Summing up the approach in the paper:

“…psychiatric patients exposed to CPAP are susceptible to the problems of anxiety sensitivity and somatosensory amplification. In our clinical experience, both of these transdiagnostic vulnerabilities are triggered when attempting to use fixed pressurized air. As CPAP delivers the same pressure setting on inspiration and expiration for any given breath, the patient is always exhaling against a pressure greater than required to maintain a patent airway, an objectively proven observation described in the sleep literature since 1990. This fixed pressure on exhalation often produces an uncomfortable sensation sufficient to trigger PAP intolerance, and in vulnerable patients, claustrophobic or panic reactions flare up. Thus, at our sleep medical center, which specializes in the treatment of psychiatric patients with sleep disorders, we observe CPAP technology itself as a major cause of low adherence or outright rejection.”

The way some people might experience headaches provides the most practical way to explain these concepts. Some people develop a headache, come up with an effective plan to treat it, and then continue on with their daily activities. Other people experience a headache, and some part of their personality makes them continually reflect back on or monitor what’s happening with the headache. This extra attention given to the headache is called attention amplification or attention bias. Bottom line is that when you pay too much attention to the headache, the problem not only fails to resolve, it often worsens in intensity. The nearly identical process occurs with unpleasant sensations from CPAP. The patient is unable to stop thinking about the mask on her face or the pressurized air being delivered, which makes either of these sensations more uncomfortable than originally experienced. Eventually, the patient notices these sensations so much that falling asleep becomes impossible.

One might think that teaching a patient to distract the mind to pay attention to something else would solve the problem. In fact, we use an imagery distraction technique. Which is self-guided daydreaming in the mind’s eye, to help people stop thinking about the mask and the pressure. Yet, despite great success with this approach in many patients, it does not bring the patient all the way into a place of comfort to be able to tolerate the fixed pressures of CPAP. This is because imagery distraction cannot always resolve the problem of expiratory pressure intolerance. Instead, by switching patients to bilevel, auto-bilevel (ABPAP) or ASV devices, we find the patients can gain comfort when breathing out against the lower pressurized air. This change in technology is what helps them to avoid paying too much attention to the “unnatural” flow of air into the nose and throat.

These approaches appear to us to be the most relevant ways in which we can help someone overcome their adaptation difficulties with CPAP. First, teach them how to distract themselves from the sensations of CPAP so that the experience is not emotionally overwhelming. Second, provide the patient with advance technology, which in and of itself leads to greater comfort and therefore acceptability from the patients’ perspective.

In the next post we’ll delve deeper into other fine points of the results from the research and finish up with our discussion points on how to use the information in clinic and what research is needed in the future.



  1. Youakim JM1, Doghramji K, Schutte SL. Posttraumatic stress disorder and obstructive sleep apnea syndrome. 1998 Mar-Apr;39(2):168-71.
  2. Krakow B, Lowry C, Germain A, Gaddy L, Hollifield M, Koss M, Tandberg D, Johnston L, Melendrez D. A retrospective study on improvements in nightmares and post-traumatic stress disorder following treatment for co-morbid sleep-disordered breathing. J Psychosom Res. 2000 Nov;49(5):291-8.
  3. Reist C, Gory A, Hollifield M, Krakow B. Potential benefit of sleep assessment prior to PTSD treatment. Psychiatry Res. 2017 Aug 30.
  4. Sexton MB, Avallone KM, Smith ER, Porter KE, Ashrafioun L, Todd Arnedt J, Rauch SAM. Sleep disturbances as predictors of prolonged exposure therapy effectiveness among veterans with PTSD. Psychiatry Res. 2017 Oct;256:118-123. doi: 10.1016/j.psychres.2017.06.044. Epub 2017 Jun 14.
  5. Reist C, Gory A, Hollifield M. sleep-disordered breathing impact on efficacy of prolonged exposure therapy for posttraumatic stress disorder. J Trauma Stress. 2017 Apr;30(2):186-189. doi: 10.1002/jts.22168. Epub 2017 Mar 8.