A team of sleep neurologists at the Cleveland Clinic Sleep Disorders Center analyzed 1,474 video polysomnograms performed between September 2018 and October 2023, drawing on the center’s STARLIT registry. Of these patients, 624—about 42 percent—were taking antidepressants, while 850 were not. The researchers manually scored RSWA using chin (submentalis) and bilateral flexor digitorum superficialis (FDS) electromyography, following the American Academy of Sleep Medicine’s current scoring criteria. Unlike many earlier studies that relied on leg muscles prone to artifact, this approach captured the upper limb signals now considered the most accurate diagnostic markers for REM sleep behavior disorder.

The technical rigor of the scoring matters. RSWA was flagged when any of three criteria were met: tonic activity, in which chin EMG amplitude runs at least twice the atonia level for at least half of a 30-second epoch; phasic activity, in which at least five of ten consecutive three-second mini-epochs contain brief bursts of doubled-amplitude muscle activity in the chin or forearm muscles; or any sustained EMG activity meeting amplitude thresholds regardless of duration. To avoid confounding by sleep apnea, REM epochs during or within 30 seconds of a respiratory event were excluded from analysis.

The headline result is a clear class effect. After adjusting for age, body mass index, gender, and race, patients on selective serotonin reuptake inhibitor (SSRI) monotherapy showed a 3.39 percent increase in the proportion of REM epochs containing RSWA compared with medication-free patients, while those on serotonin norepinephrine reuptake inhibitors (SNRIs) showed a substantially larger 6.66 percent increase. Both findings were statistically significant. In striking contrast, tricyclic antidepressant (TCA) monotherapy was not associated with any increase in RSWA—in fact, the point estimate trended downward, though it did not reach significance.

Combination therapies produced the highest estimates of all. Patients taking both an SSRI and an SNRI showed a 10.44 percent increase in RSWA, and those combining an SNRI with a TCA showed a 12.81 percent increase, although the authors caution that these combination subgroups were small—14 and 9 patients respectively—and the results did not reach conventional statistical significance. Overall, mean RSWA was significantly higher among antidepressant users than non-users, at 15.0 percent versus 9.7 percent of REM epochs, and 18.9 percent of medicated patients crossed the 27 percent RSWA threshold used to diagnose REM sleep behavior disorder on polysomnography, compared with 11.4 percent of non-users.

Why would serotonergic and noradrenergic drugs loosen the grip of sleep paralysis? The answer lies in the brainstem circuitry that governs REM atonia. Glutamate-releasing neurons in the pontine sublaterodorsal nucleus activate inhibitory premotor neurons in the ventral medulla, which in turn silence spinal motor neurons during REM sleep. SSRIs and SNRIs boost serotonin and norepinephrine signaling, and the study’s authors suggest that activation of the dorsal raphe serotonergic and locus coeruleus noradrenergic nuclei during REM may interfere with this normal inhibitory cascade, allowing muscle tone to leak back into dreaming sleep.

The clinical stakes are high because RSWA is the polysomnographic biomarker of REM sleep behavior disorder (RBD), a parasomnia in which patients physically enact their dreams—punching, kicking, and shouting—sometimes injuring themselves or their bed partners. Isolated RBD in older adults is one of the strongest known harbingers of neurodegenerative disease, frequently preceding Parkinson’s disease, dementia with Lewy bodies, or multiple system atrophy by years or even decades. Yet increased REM muscle tone is not specific to neurodegeneration. It also appears with antidepressant exposure, psychiatric illness, and narcolepsy, which creates a genuine diagnostic dilemma for sleep physicians evaluating a depressed patient on an SSRI who also shows excessive REM EMG activity.

One emerging hypothesis, supported by the new data, is that antidepressants do not create RBD from scratch but rather unmask a pre-existing, subclinical neurodegenerative process. Patients with antidepressant-associated RBD have been found to harbor markers of neurodegeneration—olfactory impairment, color vision deficits, and autonomic dysfunction—that cannot be explained by the medications alone. If serotonergic drugs simply reveal an underlying tendency toward RSWA, then a patient whose sleep study shows drug-associated muscle activity may still warrant closer neurological follow-up rather than reassurance that the finding is merely a side effect.

The Cleveland Clinic study is not without caveats, and the authors are candid about them. The cohort was clinically selected: RSWA scoring was performed only for studies with suspected parasomnia or observed dream enactment, enriching the sample for muscle activity and inflating absolute percentages relative to an unselected sleep laboratory population. The cross-sectional design precludes causal inference, potential confounders such as melatonin use could not be assessed, and the software could not distinguish tonic from phasic RSWA. The cohort was also predominantly White, and the medicated group skewed female and slightly younger, limiting generalizability. Formal inter-reader reliability testing was not performed, though average RSWA percentages did not differ significantly across the four scorers.

Even with those limitations, the study stands as the largest to date to quantify RSWA with modern AASM criteria and the SINBAR-style montage incorporating upper extremity EMG, and its message is unambiguous: SSRIs and SNRIs, but not TCAs, are associated with measurably more REM sleep without atonia in patients evaluated for parasomnia. For clinicians, the practical takeaway is that medication history must be weighed carefully when interpreting a sleep study suggestive of RBD, particularly since the proportion of patients meeting electrodiagnostic criteria for RBD was higher among antidepressant users in this cohort. For the millions of people taking these widely prescribed drugs, the findings are not a reason to stop treatment—but they are a reason for sleep medicine to look more closely at what antidepressants do to the sleeping brain, and at what those changes might reveal about the decades to come.

Subject of Research: Differential effects of antidepressant classes on REM sleep without atonia measured by chin and upper extremity EMG

Article Title: Antidepressant medications have differential effects on REM sleep without atonia quantified by chin and upper extremity EMG

Article References: Ahdab, J. E., Rodriguez, C. L., Grigg-Damberger, M., Araujo, M. L. D., Andrews, N. D., Thanaviratananich, S., & Foldvary-Schaefer, N. (2026). Antidepressant medications have differential effects on REM sleep without atonia quantified by chin and upper extremity EMG. Journal of Clinical Sleep Medicine, 22(1), Article 109. https://doi.org/10.1007/s44470-026-00127-2

Image Credits: AI Generated

DOI: 10.1007/s44470-026-00127-2

Keywords: REM sleep without atonia, REM sleep behavior disorder, antidepressants, SSRIs, SNRIs, tricyclic antidepressants, polysomnography, electromyography, sleep medicine, neurodegeneration, Parkinson's disease, parasomnia

The research, led by Gustavo Garrido and colleagues at the Bahiana School of Medicine and Public Health in Salvador, Brazil, followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and Cochrane guidelines, and was prospectively registered in the PROSPERO database under registration number CRD42024579831. The team searched three major biomedical databases—PubMed, Embase, and the Cochrane Library—to identify studies that had examined the relationship between sleep paralysis and anxiety. After screening, eleven eligible studies encompassing a total of 5,568 participants made it into the final analysis. Study quality was appraised with the Newcastle-Ottawa Scale, a standard instrument for judging the methodological rigor of nonrandomized research, and study selection was managed using the Rayyan screening platform.

The statistical approach was deliberately conservative. Rather than pooling data with a single fixed estimate that assumes all studies measure the same underlying effect, the investigators used random-effects models estimated with restricted maximum likelihood (REML), a method that accommodates genuine variability between studies. Two complementary outcomes were examined. The first compared continuous anxiety scores between people who experience sleep paralysis and those who do not, expressed as a standardized mean difference (SMD). The second compared the prevalence of formally diagnosed anxiety disorders between the two groups, expressed as a risk ratio (RR). Heterogeneity among studies was quantified with the I-squared statistic, publication bias was assessed with funnel plots and Egger’s test, and the robustness of the pooled estimates was probed with subgroup and leave-one-out sensitivity analyses. All computations were performed in R version 4.4.1 using the metafor package, a widely respected open-source tool for meta-analytic work.

The headline result concerns anxiety symptoms. Across the pooled studies, individuals who experience sleep paralysis scored significantly higher on measures of anxiety than controls, with a standardized mean difference of 0.35 (95 percent confidence interval 0.27 to 0.44, p < 0.01). In the language of meta-analysis, an SMD of this magnitude represents a small-to-moderate but reliable effect. What makes the finding especially striking is the heterogeneity figure: I-squared was 0.0 percent, meaning the studies were in remarkable agreement despite differences in populations, instruments, and settings. In behavioral science, where pooled effects often scatter wildly, a zero heterogeneity value is rare and lends unusual confidence to the conclusion. Publication bias, the tendency for journals to favor positive findings, was judged to be low, further strengthening the result.

The picture was more nuanced for diagnosed anxiety disorders. When the prevalence of formal anxiety conditions was compared between people with and without sleep paralysis, the pooled risk ratio of 1.28 (95 percent confidence interval 0.59 to 2.77, p = 0.53) did not reach statistical significance. However, the sensitivity analysis told a different story. When a single outlying study was excluded, the estimate shifted to a risk ratio of 1.80 (95 percent confidence interval 1.08 to 2.99), indicating that people with sleep paralysis face nearly double the risk of having an anxiety disorder. The discrepancy between the primary and sensitivity analyses underscores how a single influential dataset can obscure an underlying pattern, and the authors concluded that sleep paralysis appears to occur more frequently among individuals with anxiety-related conditions even though the evidence for diagnosis-level prevalence is less consistent than the evidence for symptom levels.

Understanding why this association exists requires a brief tour of sleep neurobiology. During healthy REM sleep, the brainstem actively inhibits motor neurons, producing the near-total paralysis that prevents us from acting out our dreams. Sleep paralysis occurs when this atonia persists into a state of partial or full wakefulness during the transition between sleep and waking—either while falling asleep (hypnagogic) or upon awakening (hypnopompic). Crucially, the fear circuitry of the brain, including the amygdala, is highly active during REM sleep, and Dreaming during this stage is disproportionately negative in emotional tone. When a person regains awareness while still paralyzed, that raw, unfiltered fear state can merge with waking consciousness, producing the intense terror, chest pressure, and hallucinatory intrusions—shadowy figures, sensed presences, incubi—that define the classic episode. An anxious brain, primed to detect threat, may both experience these episodes more frequently and find them more traumatizing when they occur.

The causal arrow, however, almost certainly points in both directions. Prior research has established that anxiety disorders, post-traumatic stress disorder, panic attacks, and pathological worry are all more common among people who report sleep paralysis, and experimental work has documented objective REM-sleep abnormalities in patients with recurrent isolated sleep paralysis. Chronic anxiety disrupts sleep architecture, fragments sleep, and increases nighttime awakenings—conditions that raise the probability of a REM intrusion into wakefulness. Conversely, the episodes themselves are aversive and frightening enough to generate anticipatory fear of sleep, perpetuating insomnia and reinforcing the anxiety loop. Earlier epidemiological work has also tied sleep paralysis to depression and to poor sleep quality more broadly, and studies of high-stress occupational groups such as firefighters have reported elevated rates of the phenomenon in the context of trauma exposure.

The clinical implications of the new synthesis are tangible. Lifetime prevalence of sleep paralysis in the general population has been estimated at roughly 8 percent in earlier systematic reviews, with substantially higher rates among students and psychiatric populations, meaning the phenomenon is far from rare. Sleep medicine specialists emphasize that isolated sleep paralysis is benign in itself, but the new findings suggest that clinicians encountering patients with recurrent episodes should screen for anxiety symptoms rather than treating the parasomnia in isolation. Because the effect on anxiety scores was consistent across all included studies, even modest elevations may accumulate into meaningful distress, particularly given that anxiety disorders impose substantial burdens on quality of life worldwide and their global prevalence has been rising. Behavioral interventions—improving sleep hygiene, regularizing sleep schedules, addressing sleep deprivation, and treating underlying anxiety—may simultaneously reduce both the frequency of episodes and the emotional suffering surrounding them.

The researchers are careful to note the limitations inherent to observational synthesis. The included studies were largely cross-sectional, so the data cannot determine whether anxiety causes sleep paralysis, sleep paralysis fuels anxiety, or shared vulnerabilities—such as genetic predisposition, trauma history, or disrupted REM regulation—drive both. Self-reported sleep paralysis also depends on accurate recall and honest disclosure of an experience many people find embarrassing or frightening to describe. Still, with eleven studies, nearly 5,600 participants, zero heterogeneity in the symptom analysis, and low publication bias, the association itself now rests on a firmer statistical foundation than ever before. For the millions who wake in the small hours, frozen and afraid, the message of this research is oddly comforting: the night-mare is not a supernatural visitation but a measurable intersection of REM physiology and emotional health—and one that science is steadily learning to name, explain, and treat.

Beyond the pooled statistics, the composition of the underlying evidence base offers useful perspective on how the association was detected. Several of the included studies focused on university students, a population in which sleep paralysis is consistently reported at elevated rates, while others examined clinical groups such as outpatients with panic attacks or anxiety disorders. Work among Egyptian college students, for example, linked sleep paralysis to trait anxiety, pathological worry, and post-traumatic stress symptoms, and a survey of Polish students identified comparable psychological risk factors. Studies of African American samples with panic disorder contributed some of the earliest systematic documentation of isolated sleep paralysis in psychiatric settings, and a large international survey has since catalogued the clinical features and coping strategies people use to disrupt episodes, such as attempting small movements or focusing on breathing.

The measurement instruments themselves matter for interpretation. Because sleep paralysis is typically ascertained through retrospective self-report questionnaires rather than overnight laboratory recording, reported prevalence can vary with how questions are framed and how vividly episodes are remembered. One case-control study that did use polysomnography found objective differences in REM sleep characteristics among people with recurrent isolated sleep paralysis, lending physiological credibility to accounts that might otherwise be dismissed as exaggerated. Standardized tools such as the Unusual Sleep Experiences Questionnaire have been developed specifically to capture the phenomenology of these episodes, separating paralysis from the hallucinatory and fear components that often dominate the experience.

For future research, the authors’ findings point toward longitudinal designs that can disentangle temporal ordering, as well as interventions that target both sleep continuity and anxiety. If disrupted or irregular sleep increases the likelihood of REM intrusions, then trials testing whether anxiety treatment reduces episode frequency would provide a direct test of the mechanistic pathway suggested by this meta-analysis.

Subject of Research: The association between sleep paralysis and anxiety examined through a systematic review and meta-analysis

Article Title: Association between sleep paralysis and anxiety: a systematic review and meta-analysis

Article References: Garrido, G., Donato, K., Gonzalez, J. V., Brito, G. N. E., Romeo, B., Machado, V., Guimarães Lopes, L., & Salles, C. (2026). Association between sleep paralysis and anxiety: a systematic review and meta-analysis. Journal of Clinical Sleep Medicine, 22(1), Article 165. https://doi.org/10.1007/s44470-026-00190-9

Image Credits: AI Generated

DOI: 10.1007/s44470-026-00190-9

Keywords: sleep paralysis, anxiety, anxiety disorders, meta-analysis, systematic review, REM sleep, parasomnia, sleep disorders, mental health, sleep medicine, sleep-wake transition, anxiety symptoms