It is essential to understand the basics of sleep and sleep tracking if you are going to track your sleep or it can do more harm than good. Whilst most sleep trackers do a good job of explaining their technology to users, there are a few key aspects worth highlighting that they may not be so clear on.
In recent times sleep tracking has become ubiquitous, particularly in the health and performance space. From athletes to celebrities, everyone is tracking their sleep and flooding social media with it. This increase in awareness and (hopefully) associated increase in importance on healthy sleep is excellent for the individuals and society as a whole. But there are some fundamental concepts that need explaining and some warnings that need heeding when it comes to tracking sleep.
The gold standard of measuring sleep is what is called Polysomnography (PSG), which requires a sleep laboratory. During a PSG study many things are measured to aid in sleep staging and diagnose potential sleep disorders. These include brain wave signals (EEG), eye movement signals, cardiac signals (EKG), muscle activity. Whilst the other measures are important to the accuracy of PSG, the EEG and EOG characterise our understanding of sleep cycles and are difficult if not impossible to measure outside of a lab currently.
What are the Sleep Stages?
This depends on the level of depth you want to look at and how you want to look at it largely.
The most simple level of sleep classification is asleep or awake, often referred to as “2 stage”.
The next level of depth usually divides the sleep portion into light and deep sleep, though this is a little bit of a misnomer as in this classification, ‘deep sleep’ includes REM usually and this is a distinct phase unto itself. This is rarely talked about as it is not particularly helpful or informative.
The most common classification used in discussing sleep with the public is what is called 4 stages of sleep. These 4 stages are: awake, light, deep and REM sleep. This is done because of the difficulty delineating light forms of sleep from each other.
The actual sleep stages, as described in the scientific literature are: Awake, REM, N1, N2 and N3 (which is deep sleep). When using the above 4 stage model, N1 and N2 are combined for the aforementioned reason.
For those interested in a more scientifically based article on sleep study interpretation see this article.
What do Sleep Trackers Actually Measure?
This will differ from tracker to tracker, with more data points usually improving accuracy. That is, the more things that are tracked the easier it can be to get a good picture of sleep due to the complexity of sleep metabolically.
Generally wearables detect the following; movement, heart rate (which is then used by many for other measures to be derived from for example respiratory rate and HRV) and temperature, though this varies between devices.
It should also be noted, that this is dependent on the device and is excluding things like smart mattresses and headwear to detect sleep, it refers only to more traditional wearables like rings and watches/wrist straps.
It is using the combination of these features, with proprietary algorithms that sleep is detected.
One thing to note is that to accurately use these measures to help estimate sleep, wake or indeed sleep stage with any accuracy, the measures themselves need to be accurate. This is in itself a challenge at times. To this end, the anatomy of the blood vessels at the wrist when compared to those at the finger means wrist based optical heart rate is less accurate than finger based optical heart rate.
Accuracy of Sleep Trackers - not all trackers are equal
Given the above explanation of PSG it is probably no surprise that sleep trackers are not quite as accurate as a full lab. They are, however, impressively accurate and improving, if anything.
Generally, with the caveat of measurement accuracy from the previous section, most sleep trackers do a fairly good job of detecting sleep in 2 stages (sleep vs wake). They vary significantly in their ability to track anything more than this and the accuracy of this detection.
Whilst there have been a few studies examining the accuracy of various sleep trackers, we won’t go into too many specifics of individual devices here. The other challenge with this research is that research takes time and as a result the research often pertains to older models of current technology.
All of that said, the available evidence suggests the Oura Ring (Gen 2) and Fitbit models are the most accurate devices tested. A more recent study, looking at the new Oura Ring (Gen 3) algorithm suggests it is more accurate than previously with accuracy for 2 stage sleep detection of 96%, whilst its detection of 4 stage sleep is 96% for wake (which makes sense given the above), 80% for light sleep and 90% for deep sleep and REM.
What Sleep Measures Matter?
Given the above, you may be scratching your head and thinking, well if there is any inaccuracy, what should I be measuring if anything at all?
There are some worthwhile measures to track, even if the wearables don’t always summarize them nicely for you.
The first thing to remember here is these values show a fairly large degree of individuality and anything you are tracking is more about tracking change than the ultimate value. That is, if all is going well and you are feeling good, a ‘abnormal’ measure from a wearable may not be an issue as much as a negative change in this variable may be even if it remains in the normal ranges. This is partially due to the accuracy of these wearables and mostly due to the large inter individual difference in sleep measures.
Total Sleep Time (TST)
This is probably the one you were most interested in anyway. It is also the one that surprises most first time sleep tracker users the most. People are usually shocked by how much awake time they have in their standard ‘8 hours of sleep’. Rest assured, waking up, particularly briefly, a few times a night is very normal, especially if it is between sleep cycles.
Wake After Sleep Onset (WASO)
Whilst related to TST this is slightly different. WASO does not include periods of wakefulness before you get up, which is usually part of your TST from devices. This is a little nuanced and may not represent much of a material difference for some users but WASO is a better measure of sleep quality than TST as it reflects sleep fragmentation.
Sleep onset latency (SOL)
This is usually a metric provided by your device, and represents the time taken to get to sleep. This can be a little difficult to interpret though given differing routines in bed prior to sleep for example reading or meditation. For instance a SOL of 1min when you get straight into bed and fall asleep is not the same as one when you have read for 30 mins or even 60 mins.
For normal values on these metrics (and some other) for different ages and sexs see this website. Note, these values are for lab based sleep studies but do have relevance.
Beyond the fact that accuracy of commercial sleep trackers suggests the above may be the only measures you can take with any level of confidence, there is also support in the literature for tracking these in general using wearables and specifically good accuracy in these vs PSG when using the Oura Ring (Gen 2).
Warnings and Traps of Sleep Tracking:
It should be noted that sleep trackers, like any technology, can be used in a manner that is unhealthy. Excessive obsession with sleep trackers and the negative impacts as a result has been recognized as a pathology in the scientific literature and is termed “orthosomnia”.
Similarly there is some research suggesting a nocebo effect of negative sleep tracker information. That is, symptoms are biased by being told that one has had worse sleep.
This should serve as a warning to users to use this technology wisely. Users need to appreciate it’s limitations as well as making sure to check in with subjective experience and feelings BEFORE checking sleep trackers.
Healthy Use of Sleep Trackers:
Importantly, users should not feel as though there is something missing if not checking their sleep tracker or if the data is lost for example. To ensure this is the case, for certain people, periods of intentionally not tracking sleep may be warranted, whether these are short term or longer and whether they are regular or not is up to the user. Some examples include 1 day of not tracking a week or month, a week of not tracking each month or two etc. Alternatively, users could record but not check data in a similar pattern.
Take Home Messages:
- Healthy sleep is about consistency and minimal disturbance more than sleep stages.
- Your subjective experience is crucial, not irrelevant.
- Look for patterns and changes over time, you will need a baseline for this.
- Sleep trackers and tracking your sleep should be informative and helpful, the goal is to build better habits. They should NOT be a source of stress.
- Consider regular breaks from tracking sleep if you are concerned about your relationship with your sleep tracker.
Altini M, Kinnunen H. The Promise of Sleep: A Multi-Sensor Approach for Accurate Sleep Stage Detection Using the Oura Ring. Sensors (Basel). 2021;21(13):4302. Published 2021 Jun 23. doi:10.3390/s21134302
Longmore SK, Lui GY, Naik G, Breen PP, Jalaludin B, Gargiulo GD. A Comparison of Reflective Photoplethysmography for Detection of Heart Rate, Blood Oxygen Saturation, and Respiration Rate at Various Anatomical Locations. Sensors (Basel). 2019 Apr 19;19(8):1874. doi: 10.3390/s19081874. PMID: 31010184; PMCID: PMC6514840.
Hertenstein E, Gabryelska A, Spiegelhalder K, et al. Reference Data for Polysomnography-Measured and Subjective Sleep in Healthy Adults. J Clin Sleep Med. 2018;14(4):523-532. Published 2018 Apr 15. doi:10.5664/jcsm.7036
Shrivastava D, Jung S, Saadat M, Sirohi R, Crewson K. How to interpret the results of a sleep study. J Community Hosp Intern Med Perspect. 2014;4(5):24983. Published 2014 Nov 25. doi:10.3402/jchimp.v4.24983
Stone JD, Rentz LE, Forsey J, Ramadan J, Markwald RR, Finomore Jnr VS, Galster SM, Rezai A, Hagen JA. Evaluations of Commercial Sleep Technologies for Objective Monitoring During Routine Sleeping Conditions. Nat Sci Sleep. 2020;12:821-842
Haghayegh S, Khoshnevis S, Smolensky MH, Diller KR, Castriotta RJ. Accuracy of Wristband Fitbit Models in Assessing Sleep: Systematic Review and Meta-Analysis. J Med Internet Res. 2019 Nov 28;21(11):e16273. doi: 10.2196/16273. PMID: 31778122; PMCID: PMC6908975.
Chinoy ED, Cuellar JA, Huwa KE, et al. Performance of seven consumer sleep-tracking devices compared with polysomnography. Sleep. 2021;44(5):zsaa291. doi:10.1093/sleep/zsaa291
de Zambotti M, Goldstone A, Claudatos S, Colrain IM, Baker FC. A validation study of Fitbit Charge 2™ compared with polysomnography in adults. Chronobiol Int. 2018 Apr;35(4):465-476. doi: 10.1080/07420528.2017.1413578. Epub 2017 Dec 13. PMID: 29235907.
de Zambotti M, Rosas L, Colrain IM, Baker FC. The Sleep of the Ring: Comparison of the ŌURA Sleep Tracker Against Polysomnography. Behav Sleep Med. 2019;17(2):124-136. doi:10.1080/15402002.2017.1300587
Baron KG, Abbott S, Jao N, Manalo N, Mullen R. Orthosomnia: Are Some Patients Taking the Quantified Self Too Far? J Clin Sleep Med. 2017 Feb 15;13(2):351-354. doi: 10.5664/jcsm.6472. PMID: 27855740; PMCID: PMC5263088.
Gavriloff D, Sheaves B, Juss A, Espie CA, Miller CB, Kyle SD. Sham sleep feedback delivered via actigraphy biases daytime symptom reports in people with insomnia: Implications for insomnia disorder and wearable devices. J Sleep Res. 2018 Dec;27(6):e12726. doi: 10.1111/jsr.12726. Epub 2018 Jul 10. PMID: 29989248.