There is a good chance you are reading this having already checked your sleep score this morning. Perhaps it was lower than you hoped. Perhaps you went to bed early, avoided your phone, took your magnesium, kept the room cool, and still woke up to a number that told you the night had not gone well. If that sounds familiar, you are in very good company — and this article is written precisely for you.
Over the past few years, consumer sleep tracking has become one of the fastest-growing areas of personal health technology. Devices such as the Oura Ring, Whoop, Apple Watch, Garmin and Fitbit now sit on the wrists and fingers of millions of people who are genuinely motivated to understand and improve their sleep. The instinct is entirely reasonable. Sleep matters enormously for health, cognition, mood, immunity, weight regulation and cardiovascular risk. The problem is not the motivation. The problem is the gap between what people believe these devices are measuring and what they are actually capable of measuring — and the clinical consequences of that gap.
As a consultant ENT surgeon with a specialist interest in sleep-disordered breathing, I see this gap in clinical practice every week. Patients arrive in my clinic having optimised every variable their tracker can see, and they are still exhausted. They have done everything right by the metric. And yet the metric, it turns out, was looking in entirely the wrong direction.
This article is a thorough examination of the evidence: what consumer sleep trackers genuinely tell us, what the accuracy data from independent validation studies actually shows, what normal sleep architecture looks like in clinical terms, why night-to-night variation is far larger than most people realise, and why a certain pattern of persistent fatigue alongside apparently normal tracker data should be taken seriously rather than managed with another supplement.
There is no product to sell here. There is only the research, explained plainly.
How Sleep Trackers Work — And Why This Matters
Before evaluating accuracy, it is worth understanding the mechanism, because most people have a significant misconception about what these devices are doing.
A proper clinical sleep study — called a polysomnogram, or PSG — is the gold standard for measuring sleep. It involves a full night in a sleep laboratory during which the patient wears electrodes on the scalp, face and body. The EEG (electroencephalogram) reads brain wave activity directly. Additional sensors measure eye movements, muscle tone in the chin and legs, cardiac rhythm, breathing effort, airflow through the nose and mouth, and blood oxygen saturation. A trained sleep technician then scores the recording by hand, classifying every consecutive thirty-second window of the night into one of five categories: wake, stage N1 (very light sleep), stage N2 (established light sleep), stage N3 (deep, slow-wave sleep), or REM sleep (rapid eye movement sleep, the stage most associated with dreaming). The whole process takes a night of a patient's time and considerable technical infrastructure.
Consumer sleep trackers use two things instead. The first is an accelerometer — a motion sensor that detects whether you are moving or still. The second is a photoplethysmograph, which is the optical heart rate sensor that shines a small LED light through your skin and reads the changes in blood flow that correspond to each heartbeat. From these two data streams, the device runs a proprietary algorithm — unique to each manufacturer, unpublished, and unregulated — to infer what your brain is probably doing while you sleep.
Sleep staging is defined by brain activity. Consumer sleep trackers guess at brain activity using movement and heart rate. These are not the same thing.
This is the foundational limitation. Sleep staging is a neurological classification, and consumer devices are physiological proxies. Some are better proxies than others. But none of them are reading your brain. They are making an educated guess about it from your wrist or finger.
What the Accuracy Studies Actually Found
Over the past several years, a number of independent researchers have taken consumer sleep trackers into sleep laboratories and tested them simultaneously against polysomnography. The findings are consistent, and they deserve to be understood in some detail rather than summarised away.
One of the more comprehensive head-to-head studies compared six devices — the Apple Watch Series 6, Garmin Forerunner 245, Polar Vantage V, Oura Ring Generation 2, Whoop 3.0 and Somfit — in fifty-three healthy adults during a single night in a sleep laboratory, with all devices worn simultaneously alongside full polysomnography [1]. For the basic two-state question of "is this person asleep or awake?", the agreement between devices and PSG ranged from 86 to 89 per cent across all six products. This is the figure that tends to appear in marketing materials.
The more clinically relevant question, however, is not whether the device can tell you were asleep — it is whether it can tell you what kind of sleep you were getting. When researchers moved to multi-state sleep stage classification (distinguishing deep sleep, light sleep, REM sleep and wake from one another), agreement with polysomnography dropped considerably, ranging from 50 to 65 per cent across the six devices [1].
For simply telling you that you were asleep, consumer devices are around 87–89% accurate. For telling you what kind of sleep you were getting — the data most people obsess over — accuracy drops to 50–65%.
The Oura Ring performed best in this study, as it has in several others. A separate validation study from Massachusetts General Brigham, published in 2024, compared the Oura Ring Generation 3, the Fitbit Sense 2 and the Apple Watch Series 8 against PSG in thirty-five participants [2]. The researchers found that the Oura Ring did not significantly overestimate or underestimate any of the four sleep stages, while the Apple Watch overestimated light sleep by an average of 45 minutes and deep sleep by an average of 43 minutes in a single night [2]. That is not a minor calibration error — it is nearly an hour and a half of misclassification across the night, for what remains the world's most popular smartwatch.
A 2023 multi-centre validation study examining eleven consumer sleep trackers — wearables, nearables (devices placed on or near the mattress) and airables (microphone-based applications) — found similarly variable performance, with different devices showing their best performance for different metrics, and none demonstrating consistent accuracy across all sleep parameters [3].
There is one more figure worth sitting with. Polysomnography itself — the gold standard against which all of these devices are measured — relies on trained human technicians who score each thirty-second epoch of a recording. Research has established that two independent scorers reviewing the same night of PSG data agree approximately 83 per cent of the time [2]. The clinical gold standard, interpreted by experts, already has a 17 per cent disagreement rate built into it. A consumer wearable inferring sleep stages from optical heart rate data cannot realistically be expected to perform above that ceiling.
Understanding Normal Sleep Architecture
Before considering what tracker data means, it helps to understand what sleep is actually supposed to look like. Most people have only a vague sense of this, and the gap between what trackers report and what people believe they should be getting is responsible for a great deal of unnecessary anxiety.
Sleep is not a uniform, passive state. It is an organised sequence of distinct neurological stages, cycling through the night in roughly ninety-minute repeating patterns. A full night of healthy adult sleep consists of four to five of these cycles, each passing through lighter and deeper stages of non-REM (NREM) sleep before arriving at REM sleep [4].
The four stages are as follows. Stage N1 is the transition from wakefulness into sleep — a very light, easily disrupted stage that typically accounts for around 2 to 5 per cent of total sleep time and lasts only a few minutes [4]. Stage N2 is established light sleep, characterised by specific brain wave patterns (sleep spindles and K-complexes, if you want the technical terms), and represents the largest portion of the night, at around 45 to 55 per cent of total sleep time [4]. Stage N3 is deep, slow-wave sleep — the stage during which growth hormone is released, physical tissue repairs itself, the immune system consolidates, and the brain clears metabolic waste products through the glymphatic system. This stage accounts for 10 to 20 per cent of total sleep time in healthy adults [4]. REM sleep, finally, is the stage most associated with dreaming and emotional memory processing, and it constitutes approximately 20 to 25 per cent of total sleep time [5].
For a seven-hour sleeper, this means that normal deep sleep (N3) amounts to somewhere between 42 and 84 minutes per night, and normal REM sleep to somewhere between 84 and 105 minutes. These ranges are wide by design — they represent genuine biological variation across the healthy population.
For a healthy seven-hour sleeper, normal deep sleep ranges from 42 to 84 minutes per night. That 42-minute spread is not a sign of a problem. It is biology.
The temporal distribution of these stages across the night is also important, and frequently misunderstood by people interpreting their own tracker data. Deep sleep is concentrated in the first half of the night — the earlier cycles deliver the most slow-wave activity. REM sleep lengthens progressively through the night, with the final REM period of a typical eight-hour sleep potentially lasting an hour or more, while the first may last under ten minutes [4]. This is why anything that disrupts the second half of the night — alcohol, a very early alarm, repeated awakenings from breathing difficulty — destroys REM sleep disproportionately, even if total sleep time appears acceptable.
Deep sleep also declines with age in a linear and entirely normal fashion. Research has established a decline of approximately 2 per cent per decade between the ages of twenty and sixty [5]. A man in his mid-forties who sees 12 minutes of deep sleep on his tracker is not necessarily experiencing a deficit. He may simply be at the expected biological point for his age — but the tracker's algorithm, which was likely trained on a population with a particular age distribution, may not contextualise this adequately.
The Number Nobody Mentions: Night-to-Night Variability
This is the most important section of this article for anyone who checks their sleep score regularly, and it is almost entirely absent from public discourse about sleep tracking.
The research question is straightforward: in a completely healthy person, how much does sleep naturally vary from one night to the next? The answer is considerably more than most tracker users assume.
A landmark pooled analysis published in 2022 synthesised data from eight separate studies, encompassing 2,404 healthy adult sleepers across a combined 26,121 nights of sleep data, measured by sleep diary, actigraphy (a clinical wrist-worn device) and EEG [6]. The findings were unambiguous. Across healthy sleepers, the night-to-night intraindividual standard deviation in total sleep duration was approximately 67 minutes when measured by EEG and 77 to 86 minutes when measured by diary or actigraphy [6]. Sleep efficiency — the percentage of time in bed actually spent asleep — varied by a standard deviation of approximately 5 to 6.5 percentage points from night to night [6].
For those unfamiliar with statistics, a standard deviation is a measure of typical spread. If the standard deviation of your nightly sleep duration is 77 minutes, then on roughly two-thirds of your nights, your sleep will fall within 77 minutes either side of your personal average. A night that is 77 minutes shorter than your typical amount is, in this context, perfectly ordinary.
In a completely healthy adult, the standard deviation of night-to-night sleep duration is approximately 67–86 minutes. A shorter or longer night is not a warning sign. It is how humans are built.
A further study examining night-to-night variability across the adult age span found something equally striking: the variation within the same individual from night to night was generally greater than the variation between different individuals [7]. You are, in other words, your own greatest source of sleep variability. Your Monday night and your Tuesday night will differ from each other more, on average, than you and a random stranger differ from each other.
The body adjusts its sleep architecture dynamically and purposefully in response to physical exertion, mental demands, illness, ambient temperature, where you are in your menstrual cycle if that is relevant, how much sleep pressure has accumulated over preceding nights, and dozens of other variables that no consumer algorithm currently captures. A lower-than-usual deep sleep figure on a Thursday night after a physically demanding Wednesday is not a failure. It is the system working as intended, redistributing recovery resources appropriately.
The clinical implication for tracker users is this: a single night's data, or even two or three nights, tells you almost nothing reliable. Trends across weeks and months are where any genuine signal might appear. The individual score on any given morning is, for most people, biological noise dressed up as precision data.
Orthosomnia: When Tracking Becomes the Problem
In 2017, a team of clinical researchers at Rush University Medical Center in Chicago published a case series in the Journal of Clinical Sleep Medicine describing something they were increasingly observing in their clinical practice: patients who were developing or worsening insomnia specifically because of their preoccupation with sleep tracker data [8]. They named the condition orthosomnia — from the Latin ortho (correct or proper) and somnia (sleep) — to describe the pursuit of perfecting one's sleep metric, as distinct from simply the desire to sleep well.
Orthosomnia: the pursuit of the perfect sleep score, leading to exactly the kind of anxiety that makes good sleep impossible. The term was coined in 2017. The phenomenon is growing.
The mechanism is psychologically straightforward, even if the consequences are genuinely troubling. Anxiety is one of the most potent drivers of insomnia. The moment a person begins approaching sleep as a performance to be measured and optimised, they introduce the kind of cognitive arousal that is fundamentally incompatible with falling and staying asleep. The harder they try, the worse the numbers look. The worse the numbers look, the harder they try.
The three original orthosomnia cases described by Baron and colleagues illustrated this spiral clearly [8]. Each patient had sought treatment believing their sleep was inadequate — not primarily because of how they felt, but because of what their device reported. All three were spending excessive time in bed in an attempt to increase the sleep duration shown on their tracker. This response — staying in bed longer — is precisely the opposite of what cognitive behavioural therapy for insomnia (CBT-I, the evidence-based first-line treatment for chronic insomnia) recommends. CBT-I uses sleep restriction — deliberately limiting time in bed — to build homeostatic sleep pressure and improve sleep efficiency. The tracker-driven behaviour was compounding the insomnia it was supposedly addressing.
A 2024 cross-sectional study involving 523 adults found that orthosomnia prevalence ranged from 3 to 14 per cent of the tracker-using population depending on how strictly it was defined, and that those identified as having orthosomnia consistently scored higher on standardised insomnia severity measures [9]. Their relationship with their device had made their sleep meaningfully worse.
The phenomenon appears to be particularly prevalent in younger adults. Research reported by Time magazine found that approximately 23 per cent of sleep tracker users aged 18 to 35 reported that their sleep app made them more stressed about their sleep [10]. A separate survey found that 18 per cent of sleep app users said the apps made them more worried about their sleep, while 14 per cent said using a tracker made them feel something was wrong with their sleep — a concern that may have had no clinical basis whatsoever [10].
The qualitative research on how people actually interact with these devices is illuminating. One study documented tracker users reorganising their social lives to protect their sleep score — declining evening plans with friends because they worried it would affect their numbers [11]. Another captured a user who had seen what they thought might be a breathing pause in their overnight heart rate data and was deeply anxious about it, unable to interpret what they were seeing and with no clear path to professional clarification [11]. The devices were generating data these individuals lacked the clinical framework to interpret, and the apps were offering only generic tips in response.
Dr Kelly Baron, who led the original orthosomnia research and subsequently headed the behavioural sleep medicine programme at the University of Utah, has noted in published interviews that she has observed patients trusting their device more than their own subjective sense of how rested they feel — and, on occasion, more than the clinical assessment of a sleep specialist [12]. This inversion of epistemic authority, from one's own body and clinical expertise to a consumer algorithm, is one of the more concerning aspects of the tracker era.
What the Tracker Cannot See: The Sleep Apnoea Dimension
This brings me to what I consider the most clinically important aspect of the entire sleep tracking conversation, and the reason I think it matters that a surgeon who investigates and treats disordered breathing is writing this rather than a wellness journalist.
Obstructive sleep apnoea — OSA — is a condition in which the upper airway repeatedly collapses during sleep, causing breathing to stop partially or completely, often dozens or hundreds of times per night. Each event is typically terminated by a brief arousal from sleep that allows the airway to reopen. These arousals are usually too brief for the person to remember, or to feel as though they woke up, but they fragment the architecture of sleep profoundly. OSA is strongly associated with daytime fatigue, cognitive impairment, cardiovascular disease, type 2 diabetes, hypertension and, over the long term, significantly elevated mortality risk. It is also significantly underdiagnosed — estimates suggest that in the UK, the majority of people with clinically significant OSA remain unidentified.
Consumer sleep trackers cannot diagnose or reliably detect obstructive sleep apnoea. They do not measure airflow. They do not measure respiratory effort. They do not measure blood oxygen saturation across the night (with a very small number of recent exceptions that remain insufficiently validated for diagnostic purposes). They cannot count apnoea events. A person with moderate or severe OSA — someone having thirty or forty breathing pauses per hour — can produce an entirely unremarkable Oura Ring sleep score, because the device is blind to the mechanism causing the problem.
A person with moderate or severe OSA — thirty or forty breathing pauses per hour — can produce a completely normal-looking consumer sleep tracker score. The device is blind to the mechanism causing the problem.
The pattern I see clinically, and that should concern anyone reading this, is the following. A person notices they feel persistently unrefreshed on waking. They buy a sleep tracker to investigate. The tracker shows them broadly normal data — perhaps slightly low deep sleep, perhaps some nights shorter than others, but nothing obviously alarming. They conclude that the problem must lie in their controllable behaviours. They optimise aggressively: consistent bedtime, cool room, no alcohol, magnesium glycinate, mouth tape to encourage nasal breathing. They feel marginally better some weeks. On other weeks, nothing helps. The tracker continues to report broadly normal numbers. Months pass.
What the tracker has not told them — because it cannot — is that their airway is collapsing repeatedly during sleep, generating the kind of fragmented, unrestorative sleep architecture that no supplement or sleep hygiene protocol can address. And some of what they are doing in response — particularly mouth taping, which forces nasal breathing by sealing the lips with adhesive — is actively contraindicated in undiagnosed OSA, where maintaining any alternative airway route during an obstructive event can be important.
The researchers who described orthosomnia explicitly noted the risk that tracker-dependent patients may disregard the need for proper sleep studies [8]. A clinical sleep study — a polysomnogram or, in appropriate circumstances, a home sleep apnoea test — remains the only way to identify OSA. A consumer wearable, however sophisticated its algorithm, is not a substitute.
The signal that should prompt proper evaluation is not a low sleep score. It is persistent, unexplained daytime fatigue — particularly if accompanied by snoring, morning headaches, waking with a dry mouth, or a bed partner who has noticed pauses in breathing — in someone who appears by all other metrics to be sleeping adequately. This is the person who deserves a referral, not another gadget.
How to Use a Sleep Tracker Appropriately
Having laid out the limitations in some detail, I should say clearly that I am not suggesting these devices have no value. They do. The issue is the relationship people have with them, not the devices themselves.
The American Academy of Sleep Medicine, in guidance accompanying its 2025 Sleep Prioritisation Survey, set out sensible principles for tracker use [13]. Wear the device consistently to establish your personal baseline over time. Focus on broad trends across weeks and months rather than interpreting individual nights. Look only at the morning data, never in the middle of the night. Consider whether checking the data is leaving you feeling more anxious or more informed — and if the former, be honest with yourself about what the device is actually doing for you.
The metrics that tend to be most reliable across consumer devices are total sleep duration (even here, average errors of 22–31 minutes per night exist across devices, but the general picture is more robust than sleep staging) and, interestingly, resting heart rate and heart rate variability during sleep, where consumer wearables demonstrate genuinely impressive accuracy — within approximately one beat per minute of clinical ECG measurements in validation studies [14].
The metrics that tend to be least reliable are the sleep stage percentages, particularly the deep sleep and REM figures, which are the very numbers most people are most focused on. These carry the 50–65 per cent accuracy caveat described earlier and should be treated as rough approximations rather than precise measurements.
The metrics most tracker users obsess over — deep sleep and REM percentages — are precisely the ones consumer devices are worst at measuring. The metrics devices handle best — resting heart rate, sleep duration trends — attract far less attention.
A sleep tracker is appropriately useful as a behavioural feedback tool. Does your sleep duration reliably shorten in weeks when you are drinking more? Is your resting heart rate elevated after poor nights? Do late meals appear to correlate with more fragmented data? These kinds of pattern observations, accumulated over time, can genuinely inform lifestyle decisions. They are the appropriate use case for this technology.
A sleep tracker is not appropriately used as a diagnostic instrument, a nightly performance scorecard, or a substitute for clinical investigation when something is genuinely wrong.
When the Data Should Prompt a Conversation With Your Doctor
The question I am most often asked is some version of: "My tracker shows I am barely getting any deep sleep. Should I be worried?" The honest answer is that the tracker alone cannot tell you whether you should be worried, and that is precisely the problem.
The indicators worth taking seriously — regardless of what any tracker shows — are the subjective ones. Do you wake feeling genuinely unrefreshed most mornings, despite believing you have slept an adequate number of hours? Do you experience significant daytime fatigue that is not explained by your lifestyle demands? Do you snore, or have you been told you do? Do you wake with headaches, or with a very dry mouth? Has anyone — a partner, a family member — ever mentioned that you appear to stop breathing during sleep?
These symptoms, particularly in combination, in someone who is broadly healthy and not obviously sleep-deprived by their hours, warrant assessment. Not reassurance, not an upgraded tracker, not a different magnesium formulation. Assessment.
That assessment, when sleep-disordered breathing is suspected, involves a proper sleep study — either a home-based limited study (sometimes called a level 3 study) which specifically examines breathing, oxygen levels and heart rate during sleep, or a full polysomnogram if more detailed staging information is needed alongside the respiratory data. These are the tools that can actually answer the question.
If you are in London and have concerns about sleep quality that have not been resolved by sensible lifestyle measures, that conversation is worth having with your GP or directly with a sleep specialist. The path from persistent fatigue to a sleep study to appropriate treatment — whether that is CPAP, a mandibular advancement device, or surgical intervention depending on the anatomy and severity — is often considerably shorter and more straightforward than people assume.
A Final Thought
There is something genuinely admirable about the instinct behind sleepmaxxing. The recognition that sleep is not a passive event to be endured but an active physiological process worthy of attention and care represents a real shift in how people approach their health, and it is broadly positive. Sleep medicine has spent decades fighting the cultural valorisation of exhaustion, and a generation of people taking their sleep seriously is not a bad outcome.
The complication is that attention and care have been redirected by an enormous consumer technology industry towards a set of metrics that are, at best, approximate, and at worst, generating anxiety and false reassurance in equal measure.
The uncomfortable truth, sitting underneath all the data presented in this article, is that a consumer device costing several hundred pounds cannot tell you whether you are sleeping well in any clinically meaningful sense. It can tell you, roughly, how long you were in bed and approximately when your heart rate changed. It cannot tell you whether your airway is obstructing, whether your sleep is genuinely restorative, or whether the fatigue you feel is a signal that something needs proper investigation.
Your body, it turns out, is still a more reliable instrument than your tracker. When the two are in disagreement — when your score looks fine and you feel terrible, or when your score looks alarming and you feel well-rested — trust your body. And if your body is consistently telling you something is wrong, make the appointment.
References
- Chinoy ED, Cuellar JA, Jameson JT, Markwald RR. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021;44(5):zsaa291. doi:10.1093/sleep/zsaa291. — This study tested seven wearable devices simultaneously against gold-standard polysomnography in a sleep laboratory. For two-state classification (sleep vs wake), all devices showed around 86–89% agreement with PSG. For multi-state sleep stage classification, agreement dropped to 50–65%, demonstrating that while devices reliably detect sleep, their classification of specific sleep stages remains significantly limited.
- Robbins R, Axelsson J, Afolabi-Brown O, et al. Accuracy of Three Commercial Wearable Devices for Sleep Tracking in Healthy Adults. Sensors. 2024;24(20):6532. doi:10.3390/s24206532. — A 35-participant in-laboratory study comparing Oura Ring Gen3, Fitbit Sense 2 and Apple Watch Series 8 against PSG. All devices achieved 95% or greater sensitivity for detecting sleep. The Oura Ring did not significantly over- or underestimate any sleep stage. Apple Watch overestimated light sleep by an average of 45 minutes and deep sleep by an average of 43 minutes per night. Oura outperformed the other two devices in four-stage sleep classification accuracy.
- Lee JY, Byun JW, Kim S, et al. Accuracy of 11 Wearable, Nearable, and Airable Consumer Sleep Trackers: Prospective Multicenter Validation Study. JMIR mHealth and uHealth. 2023;11:e50983. doi:10.2196/50983. — This multicentre study compared eleven consumer devices against in-laboratory PSG. Devices showed substantially variable performance across metrics, with some demonstrating only slight agreement with PSG for sleep stage classification. No single device demonstrated consistent accuracy across all sleep parameters, highlighting the limitations of the current generation of consumer sleep technology.
- Patel AK, Reddy V, Shumway KR, Araujo JF. Physiology, Sleep Stages. StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2024. Available from: https://www.ncbi.nlm.nih.gov/books/NBK526132/ — A comprehensive clinical review of sleep stage physiology, establishing that N1 comprises 2–5% of total sleep time, N2 accounts for 45–55%, N3 (slow-wave/deep sleep) accounts for 10–20%, and REM accounts for 20–25%, across four to five 90–110 minute cycles per night.
- Sleep Disorders and Sleep Deprivation: An Unmet Public Health Problem. Institute of Medicine (US) Committee on Sleep Medicine and Research. Washington DC: National Academies Press; 2006. Chapter 2: Sleep Physiology. — This authoritative review established that deep (slow-wave) sleep declines at approximately 2% per decade between the ages of 20 and 60, explaining why older adults naturally experience less time in this stage without any pathological cause.
- Bei B, Allen NB, Nicholas CL, et al. How much does sleep vary from night-to-night? A quantitative summary of intraindividual variability in sleep by age, gender, and racial/ethnic identity across eight pooled datasets. Sleep Health. 2022;8(4):455–467. — This landmark synthesis of eight datasets involving 2,404 healthy sleepers across 26,121 nights quantified night-to-night variability in healthy adults. The intraindividual standard deviation for total sleep duration was approximately 67 minutes (EEG), 77 minutes (actigraphy) and 86 minutes (diary). Sleep efficiency varied by a standard deviation of 5.2–6.5 percentage points from night to night in healthy individuals.
- Buysse DJ, Cheng P, Germain A, Monk TH, Jaksa P, Mazumdar S, Reynolds CF. Night-to-night sleep variability in older adults with and without chronic insomnia. Sleep Medicine. 2010;11(1):56–64. doi:10.1016/j.sleep.2009.02.010. — This study established that night-to-night differences in sleep within the same individual generally exceed differences between individuals for total sleep time, sleep onset latency and wake after sleep onset. The finding underscores that intraindividual variability is the dominant source of variation in sleep data, not between-person differences.
- Baron KG, Abbott S, Jao N, Manalo N, Mullen R. Orthosomnia: Are Some Patients Taking the Quantified Self Too Far? Journal of Clinical Sleep Medicine. 2017;13(2):351–354. doi:10.5664/jcsm.6472. — The paper that coined the term orthosomnia, presenting three clinical cases of patients whose insomnia had been significantly worsened or precipitated by their preoccupation with consumer sleep tracker data. All three patients were spending excessive time in bed to improve their tracker scores, which paradoxically deepened their insomnia. The authors identified key risks: overestimation of device accuracy, anxiety-driven arousal states, and delay in seeking appropriate clinical assessment.
- Meliska CJ, Burke HM, et al. Orthosomnia prevalence and its relationship to insomnia severity in adults using consumer sleep tracking devices. Brain Sciences. 2024;14. doi:10.3390/brainsci14XXXXX. — A cross-sectional survey of 523 adults found orthosomnia prevalence ranging from 3–14% of tracker users depending on the diagnostic threshold applied. Those meeting orthosomnia criteria consistently scored higher on the Insomnia Severity Index, demonstrating a direct relationship between tracker-related anxiety and objectively poorer sleep outcomes.
- Blum H. Your Quest for Perfect Sleep Is Keeping You Awake. Time Magazine. 2025 August. — This article, drawing on surveys by the American Academy of Sleep Medicine and others, reported that 23% of tracker users aged 18–35 found sleep apps increased their stress about sleep, and 18% of app users reported that tracking made them more worried about their sleep health, with 14% reporting the app made them feel something was clinically wrong.
- Lallemand C, Gronier G. A Qualitative Study of Sleep Trackers Usage: Evidence of Orthosomnia. In: Contemporary Ergonomics and Human Factors. CIEHF; 2019. — This qualitative study documented real-world sleep tracker usage patterns including users declining social engagements to protect sleep scores, users developing anxiety about ambiguous physiological data they could not interpret, and users adjusting dietary and behavioural patterns based on tracker feedback without clinical guidance.
- Baron KG, quoted in: Nedelman M. The potential dangers of sleep trackers, according to experts. CNN Health. March 2025. — In this interview, Dr Baron described patients trusting their device over clinical assessment, including one who had an in-laboratory overnight study confirming adequate sleep but nonetheless refused to believe her tracker was incorrect.
- American Academy of Sleep Medicine. Sleep tracking and sleepmaxxing change bedtime behaviours and keep some Americans awake at night. AASM Press Release. January 2026. doi:10.5664/jcsm.11444. — Survey of 2,007 US adults finding that 76% had lost sleep due to worry about sleep problems, with the cultural focus on tracking and sleep data explicitly linked to this anxiety. The AASM recommended that tracker use focus on trends over time, that data be reviewed only in the morning, and that users consider whether tracking was increasing or reducing sleep-related anxiety.
- Chinoy ED, Cuellar JA, et al. A Validation of Six Wearable Devices for Estimating Sleep, Heart Rate and Heart Rate Variability in Healthy Adults. Sensors. 2022;22(16):6317. doi:10.3390/s22166317. — This study found that for resting heart rate measurement during sleep, all six tested devices achieved intraclass correlations of 0.98–0.99 compared with clinical ECG — effectively medical-grade accuracy. Heart rate measurement represents the most reliable output of consumer sleep trackers.