Reduced REM Sleep Linked to Higher Mortality in Heart Disease

THE PROTOHUMAN PERSPECTIVE#

We've spent decades obsessing over sleep duration. Seven hours, eight hours, the endless debate. But the real signal was always hiding inside the architecture — specifically, in the REM stage most of us barely think about.

For the biohacking community, this is a critical pivot. If you're optimizing sleep with trackers, supplements, and temperature protocols, you need to know that total time in bed is not the variable that best predicts whether your cardiovascular system is degrading. The proportion and quality of your REM sleep may be a more precise biomarker for cardiometabolic risk than anything you're currently tracking. For individuals already carrying coronary artery disease or obstructive sleep apnea — conditions affecting hundreds of millions globally — REM disruption isn't a minor inefficiency. It appears to be a measurable mortality signal. The data arriving from multiple independent cohorts in 2025 makes this harder to ignore than it was even two years ago. This isn't speculative longevity science. It's clinical risk stratification happening in your sleep lab results right now.

THE SCIENCE#

REM Sleep: More Than Memory Consolidation#

Rapid eye movement sleep has historically been studied for its role in memory encoding and emotional regulation. But its cardiovascular implications are now demanding equal attention. During REM, the autonomic nervous system undergoes dramatic shifts — heart rate becomes irregular, sympathetic tone increases, and blood pressure fluctuates in ways that stress the vascular endothelium. For a healthy individual, this is physiological. For someone with coronary artery disease or obstructive sleep apnea, it becomes a nightly cardiovascular stress test that the body may be failing.

The question researchers are now asking is not whether REM sleep matters for heart health. It's whether the pattern and proportion of REM sleep can predict who dies.

The Sleep Temporal Entropy Signal#

Leng, Chen, Cavaillès, and colleagues introduced Sleep Temporal Entropy (STE) — a Shannon entropy-derived biomarker that quantifies fragmentation within specific sleep stages, not just total sleep disruption ^[1]. This is a meaningful methodological advance. Traditional metrics like Wake After Sleep Onset (WASO) and Sleep Efficiency treat all awakenings equally. STE doesn't. It captures the informational complexity of transitions within and between sleep stages, including REM.

Using data from the Sleep Heart Health Study (n = 4,862) and the Shanghai Sleep Health Study Cohort (n = 3,219), the team found a U-shaped association between REM-specific STE and mortality. Compared to the middle reference quintile (Q3), the lowest REM STE quintile (Q1) carried a hazard ratio of 1.97 (95% CI: 1.63–2.38) for all-cause mortality. The highest quintile (Q5) was also associated with elevated risk (HR: 1.35, 95% CI: 1.06–1.73) ^[1]. Similar patterns held for cardiovascular-specific death.

That lowest quintile — representing the most monotonous, least variable REM patterns — essentially doubled mortality risk. I want to be careful here: this is observational, not causal. But the effect size is large enough that it survives adjustment and replication across two independent cohorts.

— Actually, I want to rephrase that. It's not just "large." For a sleep metric, a hazard ratio approaching 2.0 is exceptional. Most sleep-mortality associations hover around 1.2–1.4.

The REM-Dyslipidemia Connection#

Wang, Gao, and Gao examined 5,239 participants from the Sleep Heart Health Study and identified REM stage percentage as an independent predictor of dyslipidemia in OSA patients (OR: 0.987, 95% CI: 0.977–0.998, P = 0.022) ^[2]. Each unit increase in REM percentage was associated with a 1.3% reduction in dyslipidemia odds. That sounds small. It's not, when you consider that REM percentage can vary by 10–15 points between individuals.

Sleep latency, REM latency, mean oxygen saturation, and percentage of time below SpO₂ 95% were also independently associated with dyslipidemia ^[2]. BMI mediated the association between REM stage percentage and dyslipidemia, suggesting a pathway through metabolic regulation that REM sleep may directly influence — potentially via autonomic modulation of hepatic lipid processing and overnight cortisol rhythms.

This is the part where the sleep-cardiology interface gets mechanistically dense. Dyslipidemia drives atherosclerosis. Atherosclerosis causes coronary artery disease. If reduced REM independently worsens lipid profiles, then the REM-mortality link has a plausible biochemical intermediary.

Autonomic Dysfunction: The Counterintuitive REM-OSA Finding#

Here's where I'm less convinced. A 2024 study by researchers publishing in Sleep and Breathing compared HRV profiles between REM-predominant OSA patients (n = 137) and stage-independent OSA patients (n = 252) ^[3]. Their hypothesis was straightforward: REM-OSA patients should show higher sympathetic cardiac modulation, given the known cardiovascular risk of REM-specific apneas.

The data said the opposite. REM-OSA patients demonstrated consistently higher cardiac vagal modulation — lower LF:HF ratio, lower LF power, higher HF power — during both N2 and REM sleep ^[3]. The authors attributed this to confounding by severity and sex: REM-OSA patients had milder disease (median AHI 10 vs. 17) and a higher proportion of females (45% vs. 26%).

This doesn't invalidate the mortality data. But it does complicate the mechanistic story. If REM-specific apneas don't uniformly drive sympathetic overdrive, then the pathway from reduced REM to cardiovascular death may not run through autonomic dysfunction alone. Or perhaps it's the absence of adequate REM — rather than what happens during REM apnea — that carries the risk. That distinction matters.

Sleep Disorders and Structural Heart Disease#

El Jamal, Brooks, Skarke, and FitzGerald extended the cardiovascular risk envelope further. Using two large EHR databases (TriNetX and All of Us), they found that having any sleep disorder increased the hazard for future aortic stenosis by 15% (HR: 1.15, 95% CI: 1.13–1.18) ^[4]. Changes in lipid profile mediated a proportion of this association, independent of classical cardiovascular risk factors.

This is a population-level signal, not REM-specific. But it reinforces the broader pattern: disrupted sleep architecture accelerates structural cardiovascular deterioration through metabolic pathways, with dyslipidemia as a central mediator ^[4].

The ProPASS Dimension: Beyond Duration#

The ProPASS Consortium pooled device-measured sleep data from six cohorts across five countries (n = 14,085) and found that sleep regularity and efficiency were more strongly associated with cardiometabolic risk than duration alone ^[5]. Short sleep carried a modest association (β: 0.05), but irregular sleep patterns (β: 0.14) and low sleep efficiency (β: 0.10) were substantially more predictive of a composite cardiometabolic risk z-score ^[5].

This aligns with the STE findings. It's not just how long you sleep. It's whether your sleep maintains consistent, structured architecture night after night.

COMPARISON TABLE#

THE PROTOCOL#

Based on current evidence, here is a structured approach for individuals concerned about REM sleep quality — particularly those with existing coronary artery disease or diagnosed OSA.

Step 1: Establish a baseline. Request a full polysomnography from your sleep physician. Specifically ask for REM percentage, REM latency, and fragmentation indices — not just AHI. If you already have a PSG report, review these values. A healthy REM proportion is typically 20–25% of total sleep time. Below 15% warrants attention.

Step 2: Track sleep architecture longitudinally. Use a validated wearable (Oura Ring Gen 3, WHOOP 4.0, or Apple Watch with sleep staging enabled) to monitor REM trends nightly. No consumer device matches PSG accuracy, but multi-week trends in REM percentage are informative. Look for consistent patterns below your personal baseline, especially after alcohol, late caffeine, or training changes.

Step 3: Eliminate REM suppressants from your routine. Alcohol is the most common REM suppressant in the general population — even moderate intake (two drinks) significantly reduces REM proportion. Cannabis (particularly THC-dominant strains) similarly suppresses REM. Certain medications, including SSRIs and beta-blockers, can also reduce REM; discuss alternatives with your prescriber if cardiovascular risk is a concern.

Step 4: Optimize sleep regularity over duration. The ProPASS data suggests that night-to-night consistency in sleep timing carries greater cardiometabolic weight than total hours ^[5]. Set a fixed wake time seven days per week. Allow bedtime to flex by no more than 30 minutes. This stabilizes circadian entrainment, which in turn supports consistent sleep architecture including REM distribution.

Step 5: Address OSA aggressively if present. CPAP compliance directly improves REM sleep quantity and quality by preventing the airway collapse that fragments REM stages. If CPAP is intolerable, oral appliance therapy or positional therapy (for supine-predominant OSA) should be trialed. The data from Wang et al. connecting REM percentage to dyslipidemia risk makes untreated OSA a lipid metabolism problem, not just a snoring problem ^[2].

Step 6: Consider targeted supplementation — cautiously. Magnesium glycinate (200–400 mg before bed) may support sleep architecture stability through GABA receptor modulation, though direct REM-specific evidence in humans is limited. Tart cherry extract (providing natural melatonin) has shown modest sleep quality improvements in small trials. I'd avoid exogenous melatonin above 0.5 mg — supraphysiological doses can paradoxically disrupt sleep architecture.

Step 7: Monitor lipid panels in conjunction with sleep data. If you're tracking REM percentage and it's consistently low, request fasting lipid panels every 6 months. The association between reduced REM and dyslipidemia means your lipid profile is a downstream validation metric for your sleep interventions.

What is Sleep Temporal Entropy and why does it matter for heart health?#

Sleep Temporal Entropy is a Shannon entropy-based metric that measures the complexity and fragmentation of transitions within and between sleep stages. Unlike older metrics that just count awakenings, STE captures the pattern of disruption — and the REM-specific version has shown a near-doubling of mortality risk at its lowest values. It matters because it may eventually replace cruder sleep quality scores in cardiovascular risk models, though it's still in the research validation phase.

How does reduced REM sleep contribute to dyslipidemia?#

The mechanism isn't fully mapped, but the association is clear: each percentage-point increase in REM sleep proportion was linked to a 1.3% reduction in dyslipidemia odds in a large cohort study ^[2]. The likely pathway involves autonomic regulation of hepatic lipid metabolism during REM, and BMI appears to mediate part of the effect. Honestly, we don't have the full causal chain yet — but the statistical signal is consistent enough across datasets that ignoring it seems unwise.

Who should be most concerned about REM sleep loss?#

Adults with diagnosed coronary artery disease, obstructive sleep apnea, or both. These populations already carry elevated cardiovascular risk, and the data suggests that REM disruption compounds that risk independently. Anyone on REM-suppressing medications (SSRIs, certain beta-blockers, benzodiazepines) should also discuss this with their physician — particularly if they have concurrent cardiovascular diagnoses.

Why did REM-related OSA patients show better autonomic function in one study?#

This counterintuitive finding from the 2024 Sleep and Breathing study ^[3] is likely explained by confounding: REM-OSA patients had milder overall disease and more women in the group, both of which independently improve HRV profiles. It doesn't mean REM-specific apneas are benign — it means you can't interpret autonomic data without controlling for severity and sex. The study's authors themselves called for more nuanced research.

When will Sleep Temporal Entropy be available in consumer wearables?#

Not soon. STE currently requires polysomnographic-grade data or, at minimum, research-grade EEG. Consumer wearables can approximate sleep stages but lack the resolution to compute entropy-based fragmentation metrics reliably. I'd estimate 3–5 years before we see validated STE algorithms in devices like Oura or WHOOP — and that's optimistic. For now, PSG remains the only reliable source.

VERDICT#

Score: 7.5/10

The convergence of evidence from multiple independent cohorts — STE mortality data, REM-dyslipidemia associations, ProPASS cardiometabolic findings — builds a convincing case that REM sleep quality is an underappreciated cardiovascular risk factor. The STE hazard ratio of 1.97 is striking. But I'm holding back from a higher score because the data is almost entirely observational, the STE metric hasn't been validated in interventional trials, and the counterintuitive autonomic findings from the HRV study remind me that we don't fully understand the mechanism. The clinical utility is real but premature for protocol-level confidence. If interventional data confirms that improving REM proportion reduces hard cardiovascular endpoints, this becomes an 9. Until then, it's a strong signal worth acting on cautiously.