Is It Time to Let Go of Sleep Staging in the Clinic?

The opinions expressed in this article are not reflective of AAST.

After a variety of sleep-related jobs, I settled into the role of lecturer at a large state university. Among other courses, I teach a sleep and dreams course. In the first week of class, I talk about defining sleep with a focus on reduced perception of the environment, and in week two, discuss scoring sleep stages. I draw heavily on the lectures I developed for AAST, pointing out K-complexes (and who knows why they are called that) and rapid eye movements (REM). This year I asked myself again, “Why do we do this?”

My Sleep Scoring Background

I had the great fortune to be a student of Al Rechtschaffen at the University of Chicago. The Rechtschaffen and Kales scoring manual for human sleep stages was published in 1968 and remained the standard for nearly half a century. My first research publication documented patterns of slow wave activity in rats that seemed to mirror those of humans.¹ A subsequent study identified low voltage non-rapid eye movement (NREM) sleep in rats, thought to be the rodent equivalent of human N1 sleep.² The zeitgeist in Chicago was that sleep was sleep, the same in all animals.

While working for the American Association of Sleep Medicine (AASM), I collaborated with Steve Van Hout to develop the Inter-Scorer Reliability (Sleep ISR) program. This caught on quickly and we soon had hundreds of sleep technologists providing stage scores for the same two hundred epochs per month. At first, I designated myself as the “expert” of sleep scoring. Thomas Penzel³ described this as “truth by age,” a method that “refers to the scoring performed by one experienced expert scorer, usually a respected and elderly person.” Or, in my case, an elderly person. While most technologists agreed with my scores, Van Hout included a button in bright orange — far too easily accessed on the scoring webpage and instantly emailing the message to my inbox — for users to share feedback (which mostly of consisted of disagreements). In response to this, I recruited a few experienced and respected colleagues to help determine the correct score.

The Sleep ISR provided a huge volume of data, and, at the urging of Con Iber and others, Van Hout and I collected data and wrote a paper on the reliability of sleep stage scoring.⁴ As a retrospective study, we were able to use the majority score as the “truth” rather than the wisdom of a single scorer or a small group. Given that we started with a majority (or at least a plurality) to define the correct score, the agreement on sleep stages was not great. The overall agreement for all stages was 82%. We thought this might be due to a large group of inexperienced individuals, but more than 90% of the users had at least three years of experience as a sleep technologist.⁴

Present Day

I have my students read Jerry Siegel’s article, “Do All Animals Sleep?”⁵ at the beginning of their course and after, ask them a series of questions: What stage do we score for a dolphin with hemispheric dissociation? Is a bullfrog that maintains alertness during quiescence awake or asleep? Niels Rattenborg recently provided a phylogenetic analysis of REM and NREM sleep⁶ suggesting the two stages are more common than we think, but REM may not look the same in birds as it does in humans. If such variability exists in animal sleep, might scoring of human sleep reflect similar variations?

In asking these questions, I’m sensing that the zeitgeist is changing and wonder if it is still important to identify stages in a sleep clinic. I spent much of my career doing this, but now a home sleep test (HST) is enough to diagnose sleep apnea — even in the absence of sleep staging — and a smart watch has been approved to provide apnea data. The current exception to this is the diagnosis of narcolepsy, which currently requires a short sleep onset latency and two or more sleep onset REM episodes for diagnosis. It won’t be long before a smart watch will replace the multiple sleep latency test (MSLT).

Let me be uncharacteristically optimistic here. I think the sleep clinic of the near future will abandon overnight sleep studies and hours of visual scoring, leaving it to artificial intelligence (AI) to interpret data uploads from smart watches. Will the scores perfectly match Penzel’s “truth by age?” Probably not, but I’m confident that AI will provide the clinician with all the necessary information to make a diagnosis. The sleep clinic of the future will focus on treatment for patients with sleep disorders, which are getting more complicated by the day. Patients will have questions about continuous positive airway pressure (CPAP), automatic positive airwave pressure (APAP), adaptive servo-ventilation (ASV), oral appliances, implanted stimulators or pharmacologic treatment for sleep apnea and will benefit from a visit to the sleep clinic. Restless legs, a variety of parasomnias, and, yes, insomnia will be addressed by the clinic — either in person or via telehealth. But just think — no more night shifts, no more electrode paste, no more stage scoring. To the future, and beyond.

References

1. Rosenberg RS, Bergmann BM, Rechtschaffen A. Variations in slow wave activity during sleep in the rat. 1 Dec 1976. Physiology & Behavior, 17(6), 931-938. https://doi.org/10.1016/0031-9384(76)90011-1
2. Bergmann BM, Winter JB, Rosenberg RS, Rechtschaffen A. NREM sleep with low-voltage EEG in the rat. Feb 1987. Sleep, 10(1), 1–11. https://doi.org/10.1093/sleep/10.1.1
3. Penzel T, Zhang X, Fietze I. Inter-scorer reliability between sleep centers can teach us what to improve in the scoring rules. 15 Jan 2013. Journal of Clinical Sleep Medicine, 9(1), 89-91. https://doi.org/10.5664%2Fjcsm.2352
4. Rosenberg RS, Van Hout S. The American Academy of Sleep Medicine inter-scorer reliability program: sleep stage scoring. 15 Jan 2013. Journal of Clinical Sleep Medicine, 9(1), 81–87. https://doi.org/10.5664%2Fjcsm.2350
5. Siegel JM. Do all animals sleep? Apr 2008. Trends in Neurosciences (Regular Ed.), 31(4), 208–213. https://doi.org/10.1016/j.tins.2008.02.001
6. Rattenborg NC, Ungurean G. The evolution and diversification of sleep. 18 Nov 2023. Trends in Ecology & Evolution, 38(2), 156-170. https://doi.org/10.1016/j.tree.2022.10.004

Richard S. Rosenberg, PhD,

Rich Rosenberg was introduced to sleep research by Arnold Leiman while an undergraduate at UC Berkeley. He was a graduate student in the laboratory of Allan Rechtschaffen and received a PhD from the University of Chicago. After a brief postdoctoral fellowship with George Sacher, he worked in sleep disorders centers at the University of Chicago and Evanston Hospital. He was hired by Jerry Barrett at the American Academy of Sleep Medicine and contributed to practice parameters, webinars and a variety of educational programs. He also developed educational materials for AAST. After having his fill of midwestern ice and snow, he returned to California to be a lecturer in the Department of Psychology at California State University, Long Beach.