Sourced from: Classic Sleep Care – Sleep Factors and Suicide: Observations from the 2018 SLEEP Conference
One of the most problematic and regrettable perspectives today in healthcare is the overall neglect or mismanagement of sleep complaints and related sleep disorders among mental health patients suffering suicidal ideation and behaviors. Not only are sleep disorders routinely ignored in these dire circumstances, but worse they are routinely misdiagnosed, under-treated or incorrectly treated. At the sleep conference, several of these points were highlighted, both during presentations of data about suicidal patients and by the failure to provide more penetrating analyses of the same data.
As the best example of the latter phenomenon, several poster presentations as well as lectures given at the symposium consistently described biased recruitment strategies to exclude OSA/UARS patients from protocols examining insomnia or suicidal behavior or both. You might be puzzled on the rationale to exclude sleep-disordered breathing patients, especially if you imagine as I do that SDB is a regularly occurring feature of patients suffering from suicidal ideation and behavior. Why then exclude these patients?
To understand this research approach, we need to delve a bit deeply into the topic of research design wherein researchers in general attempt to narrow the focus on the patients they study, that is, by limiting specific characteristics of the patients under investigation. For example, if you want to understand how insomnia operates in depression, you would pick a sample of patients with insomnia and depression, but you would avoid patients who also reported serious heart disease or chronic pain disorders. These two conditions could have so much influence on insomnia or depression such that you would not know at the end of the study whether or not they affected your results.
As a related example, if you were studying cognitive-behavioral therapy for insomnia (CBT-I) in depressed insomniacs, conceivably heart disease or pain conditions would affect how well or how poorly some patients responded. Of course, you could also choose to study CBT-I in a group of rheumatoid arthritis patients (chronic pain) who reported both insomnia and depression. Such a study would be valuable, but it would narrowly apply to RA patients with depression and how their insomnia responds to CBT-I.
With regards to OSA/UARS, in any of the above examples had you not excluded patients with sleep breathing problems, you would then need to account for this additional sleep disruption factor. Because the patho-physiological effects of sleep-disordered breathing lead to chronic sleep fragmentation, it would then appear to make perfect sense to choose patients without OSA/UARS. Otherwise, you might not know how to interpret the results of your study since either insomnia or SDB could be causing the sleep fragmentation.
Most researchers think in these ways to increase precision and thus keep their eyes on just a few elements. For example, to reiterate, to investigate the precise connections between depression and insomnia, you would prefer to select patients who reported just these two conditions and not much of anything else, assuming you could find such people. You would probably choose to exclude cardiac patients or chronic pain patients or those with poorly treated respiratory conditions like asthma and COPD. And, you would exclude people who had recently suffered the loss of a loved one. You can see how the list of exclusions might be never-ending. Yet, while there are many variables you would try to exclude to maintain your narrow focus, you probably would need to permit depressed patients currently using medications or those previously attempting to treat their depression. These variables are common characteristics among cohorts of depressed patients, so you might lose too many of them for your study if such recruitment criteria were too narrow.
In sum, if you include patients with too many characteristics reflecting too many different disease conditions, your treatment might not work as you expected or if it did work, you might not be so certain as to why. What if a pain patient in your research is using opiate medication for pain? Sometimes opiates improve insomnia. Ultimately, research is designed so the information can be translated, so to speak, into real-world clinical medicine. This process refers to a concept called generalizability, which is itself a bit confusing, so I will explain in a little more detail, which will then circle us back to why some people choose to eliminate OSA/UARS patients from their research.
Since most sleep patients actually suffer from lots of different ailments, in some ways it would seem odd to keep narrowing down the focus of the sample studied in your research. The narrower things get, the more you are losing generalizability. That is, your results would not reflect what happens in everyday life, because it would be peculiar to imagine you could find people who only suffer from depression and insomnia and not be suffering from several other conditions. On the other hand, if you do work with a narrow group of patients who report predominantly problems of depression and insomnia, then your results would show that treating insomnia with CBT-I works in depressed patients. This finding, though narrow, starts the process of generalizability, because you know at minimum CBT-I resolved insomnia in your sample of depressed patients. It’s best then to think of generalizability along a spectrum of usefulness. A narrow sample focus provides a narrower yet still valid generalizability, whereas a broader sample focus provides greater overall generalizability.
In current research studies on sleep complaints in suicidal patients, we are in the earlier phases of research development where the narrow focus paradigm predominates. Therefore, with this point in mind, it would make sense to eliminate OSA/UARS patients from the mix. But, there is a problem here, which some of you might quickly notice when you hear the elimination criteria would be an AHI > 10 or 15.
Think about it: if you wanted to eliminate all OSA/UARS patients from interfering in your study design, then why not pick the standard in the field for the past 30 or more years, namely, AHI > 5? Why pick an AHI that means you are only excluding moderate to severe OSA, but not excluding mild OSA? What would be the rationale for doing so? The obvious argument would likely be the more pronounced the sleep-disordered breathing the more pronounced the underlying sleep disruption. In other words, to examine a treatment for insomnia in suicidal patients, why include the confound of a more noticeable SDB-induced sleep fragmentation. Now, if we accept this principle, then why include any degree of sleep fragmentation caused by SDB? Why not exclude the mild cases of OSA as well? Or, is their sleep fragmentation so mild, it would not affect the study?
The answer to this riddle will not surprise you if you are a regular reader of this blog. The real reason they had to raise the level of AHI for the exclusion criteria is simple—the astronomical prevalence rate of OSA, including mild OSA, among insomnia patients. Thus, had they kept the standard AHI of 5 events/hr, they would not have been able to capture enough patients to conduct the research in the first place. In a way, this change in recruitment strategies is a “hat tip” to the small cadre of research groups around the globe who are actively studying the connections between SDB and chronic insomnia. As other types of insomnia research groups are exposed to the works of this vanguard of scientists, the traditional researchers have begun to realize they must look more closely at the OSA connections. In doing so, they are now confronted with the emerging paradigm that SDB is all too frequent to be ignored among insomniacs.
Overall, this new paradigm leaves many of these research groups in a quandary because they may not have the skill, expertise or motivation to appreciate and treat OSA or UARS in their insomnia patients. They may be “stuck” so to speak in wanting to continue to research through a narrow lens by imagining the prevalence of OSA is not so common. Of course, they can really only do so by claiming that it’s really moderate to severe SDB that’s the big deal, and then they are giving themselves permission to keep mild OSA (AHI < 15) in the mix.
Can we find further problems in these research studies should they choose to discount the role of OSA/UARS? I believe we can, and the way to understand the problem is to consider two of the main perspectives provided by researchers who study sleep and suicide. First, we have the theory of hyperarousal, which in general refers to the increase in EEG brain activity during sleep, which lessens the depth of sleep. In mental health, we also talk about hyperarousal while awake, referring to someone such as a PTSD patient who is hypervigilant, or in lay terms always on edge. The second theory refers to the sequencing or timeline in how we view the longitudinal relationship between specific sleep symptoms or disorders and the final outcomes of suicidal thinking, behavior and ultimately death. Both of these aspects overlap, so our continued discussion will attempt to weave things together.
The ultimate in hyperarousal during sleep would be a disorder that constantly fragments sleep, causing either brief or lengthy arousals from a few seconds to much longer periods of wakefulness. The ironic history on sleep fragmentation research is the out and out failure of so many investigators of insomnia to repeatedly declare that hyperarousal observed regularly in such patients has no obvious cause, other than to declare “insomnia is a hyperarousal disorder.” This statement is much less useful compared to a declaration like “cars driven through red lights spend more time in junkyards.” In both instances, something is being fractured, your sleep or your car, but the first statement (about sleep) provides no cause for the break, while for the car you can at least suspect that driving through red lights must lead to more crashes that total cars, landing them in junkyards.
Insomnia research for decades has attempted to drive home the point that the cause of sleeplessness is the sleep fragmentation due to hyperarousal, yet without providing a clear mechanism through which the brain becomes aroused. Indeed, the most common explanation is “the brain of an insomniac is hyperaroused” and probably genetically predisposed to be so. The irony of course is that OSA/UARS are leading precursors to and causes of hyperarousal. The argument is not that OSA/UARS needs to be understood as the only cause of hyperarousal among chronic insomniacs, because these individuals may also suffer from a predisposition of some sort, but the fact that so much recent research shows high rates of sleep-disordered breathing among insomniacs should be raising a very large red flag about the incomplete state of affairs observed in most insomnia research.
If OSA/UARS are major contributing or causal influences on the excessive arousal activity in the brains of chronic insomniacs, and treatment of these conditions leads to clear-cut decreases in hyperarousal, aren’t we obligated to wonder just how much SDB is operating in these patients? Why continue to espouse the belief that hyperarousal is simply innate to insomnia when you can easily observe and document (with the correct respiratory sensors in the hands of a competent sleep technologist, interpreted by a competent sleep specialist) this connection between SDB and EEG arousal? Again, as described in earlier posts, we are moving closer and closer into the realms of potential medical malpractice. It is not reasonable to keep prescribing hypnotic medications to insomnia patients to tamp down their hyperarousal for more than a decade without ever considering the possibility the hyperarousal has a cause other than genetics. It is a requirement of standard of care paradigms to ask the question: why don’t these drugs work or why don’t they work very well? Moreover, it is an ethical requirement of a physician or a prescribing psychologist to be curious enough to wonder if something is being missed in the differential diagnosis (the list of conditions or diseases we suspect might be causing the problem at hand).
Perhaps the greatest barrier to accurate and timely recognition of the presence of OSA/UARS in a suicidal patient with sleep complaints is the nearly ubiquitous example where these patients present with symptoms of insomnia or nightmares or both. Because insomnia and nightmares are so vexing to the patient and often described with desperation and other emotional distress, it would be surprising for a psychologist and even a sleep physician to not be swayed into thinking that insomnia and nightmares are the primary sleep conditions to be evaluated, diagnosed and treated. Unfortunately, what we have seen for more than two decades is that either or both insomnia or nightmares are more often than not a flare in the sky signaling the presence of a sleep breathing disorder as a potentially much larger threat to mental and physical health. This sleep breathing disorder resides, figuratively, in a place hidden from the eyes of the patient and the provider, thus leading to its relative invisibility and delayed identification unless someone interprets more comprehensively the meaning of the flare above.
Still another confound is the link between insomnia and hyperarousal as well as nightmares and hyperarousal. Both links have been written about for decades and have led researchers and clinicians to reasonably make the assumption nightmares and insomnia are the source of excessive arousal activity. Our point is nightmares or insomnia could also be considered downstream effects. Evidence for this perspective is mounting, because research continues to show OSA/UARS treatment decreases both insomnia and nightmares. In this paradigm, we would then look at the sleep breathing disorder as the source of the hyperarousal activity, which somehow caused insomnia and nightmares or somehow catalyzed a part of the brain’s systems to increase vulnerability to develop the problems of insomnia and nightmares. Either way, the net effect is logical: if SDB treatment decreases insomnia and nightmares along with their related hyperarousal activity, then we would no longer link the hyperarousal directly to these usual suspects despite their being the more obvious sleep disorders in plain sight. Instead, we would need to declare the primary culprit as sleep-disordered breathing, the ultimate upstream activity that all along the way was leading to hyperarousal.
However, we could also trace things further back in the sequence to investigate the brain itself and ask the question of whether or not something in the brain of the suicidal patient adversely alters the way the person breathes night or day and whether such a finding could lead to the eventual development of OSA/UARS. In some ways, this theory is even more appealing, because there is no apparent logic that supports the idea most suicidal patients with insomnia and nightmares should have sleep-disordered breathing. Whereas, if you could discover something “wrong” in the brains of suicidal patients, it would make intuitive sense to believe brain dysfunction was the ultimate starting point of the problem, that is, the ultimate upstream factor. Next, the research would need to confirm whether or not this brain problem/change/condition in fact can be shown to alter nocturnal respiration and lead to sleep-disordered breathing.
In looking back over this post (multiple times) I can see how it feels circuitous. Though not necessarily intended, I wish to point out that a lot of ideas in research come about through circuitous pathways. Some may imagine there is a more logical stream of ideas that neatly package themselves together in linear fashion and then proceed step by step to inevitable outcomes. As far I have observed in research, the pathways are often choppy, misdirected, and lacking any obvious destiny to a final set of definitive observations.
Again, here we are 45 years after Guilleminault and colleagues reported in 1973 the first cases of insomnia linked to sleep apnea. And, yet in the present research environment that investigates conditions such as we have been discussing suicidal ideation and behavior and completed suicide there remains tremendous resistance if not outright ignorance on the rather obvious problem of chronic sleep fragmentation induced by OSA/UARS in these fragile patients. Instead, these deeply suffering individuals are repeatedly informed the problem is in their minds, their neurotransmitters, their genetic predisposition, or perhaps their sensitivity to environmental or psychological stressors, without ever once hearing about the possibility that their rotten, degraded, no-good sleep is the perfect substrate that feeds the fire for all these other contributors to their disheartening and demoralizing anti-life perspective and attitudes.
It cannot go without saying what a dramatic turn of events may arise AFTER A GOOD NIGHT’S SLEEP!
Yet this pearl of wisdom continues to be relegated to a footnote in the patient’s care, if that, due to all the biases described above. Where does that leave us in our efforts to help these patients?
Not in a very good place in my estimation as both clinicians and researchers demonstrate only a fair degree of motivation to want to engage this way with their patients or samples of patients for studies. This silo effect has been repeatedly described and has proven pervasive in numerous fields of medicine, wherein a subspecialty of any type can develop its own brand of tunnel vision, after which new information is nearly impossible to disseminate within the confines of the subspecialty itself.
Imagine right now that every family member with a suicidal patient read this blog and then proceeded to contact a primary healthcare provider, mental health or medical, demanding their loved one undergo sleep testing. Sadly, far greater than 50% of the doctors or therapists would imagine only that the family member was grasping at straws and the other smaller half might imagine the idea was interesting but would have no conceivable way to take action; or, they would believe there was nothing they could do about this new information.
So, again what would happen? The answer is nothing in most cases. This last point highlights the frustration of so many patients and their family members who ultimately must confront these narrow perspectives so routinely expressed by many healthcare professionals. Internet access and unlimited availability to growing knowledge bases will have some impact on these processes and may speed up the translation of scientific evidence into actual clinical practice.
In the meantime, many individuals continue to suffer and continue to engage in suicidal ideation and behavior, and some successfully kill themselves. Yet, there is no way to declare that had the true nature of their sleep disorders been correctly diagnosed, they somehow would have been spared these tragic endings. There is a great deal of complexity to suicidal behaviors and suicide, so the jury is out on the extent to which sleep fragmentation triggered by undetected OSA/UARS aggravates these mental health problems. Nonetheless, we already know that treating insomnia and nightmares, each one independently, appears to decrease suicidal ideation and behavior. Therefore, it certainly seems reasonable to hypothesize that OSA/UARS is an active component of these processes and merits much greater attention in research circles.
With the above in mind, here’s my modest proposal. Let’s collect the top 100 mental health researchers of suicide in the world along with the top 100 government grant administrators or related officials who oversee the distribution of funding for suicide research. Next we place them in various sleep laboratories around the globe, hook them up to EEG monitors and then proceed to keep them awake for days on end. At most, there are permitted to experience two hours of sleep, but overall whenever they sleep it must be highly fragmented through the use of artificial techniques, such subtle sound waves to kick them out of deep sleep. Then, over the course of days we would observe their psychological status and video tape their rapid decline in mood, culminating in some instances in depression and suicidal ideation and possibly self-harming behaviors. Finally, we would ask each of these research subjects when he or she would like to be permitted to sleep normally again, which would be affirmed by all of the individuals sooner or later.
Once they recovered from their sleep fragmentation/deprivation, they would be asked to watch the videos of their behavior. Then, they would be asked whether or not they have a new perspective on the value of healthy restorative slumber? Finally, they would be shown the videotape sections where they began to demonstrate suicidal ideation or behavior, as applicable. Afterwards, they would go through a series of didactics to learn about the role of sleep-disordered breathing and how it causes the sleep fragmentation/deprivation they had been exposed to in extreme fashion to mimic what suicidal patients might be experiencing chronically year after year.
Would some of these professionals get it? Would the light bulb go off in their heads? Again, sadly, probably not a lot, but for sure some what finally understand. Indeed, a lot occurs when any individual actually experiences the problem of SDB first hand. That alone would be enough to “wake up” the individual from the slumber that had prevented him or her from understanding how badly poor sleep had affected mental health and mood.
Though the experiment above might be a game-changer, we’re more likely to see the change through the methodical development of research studies proving these points. On an individual basis we might see some new trends emerge soon. But overall, unfortunately, I must say I won’t be holding my breath regarding rapid changes in policies and guidelines to help suicidal patients.