Wrist-Worn Sleep Trackers May Mislead: Researchers Flag Statistical Cracks in Landmark Actigraphy Study

When a study claims that people who say they sleep less than seven hours may actually be getting more shut-eye than they think, the finding tends to travel fast. Sleep researchers have long wrestled with a stubborn puzzle: the gap between how long people believe they sleep and what objective devices record. A recent investigation published in the Journal of Clinical Sleep Medicine by Baron and colleagues added fresh fuel to that debate, reporting on actigraphic estimates of sleep duration in individuals who reported sleeping less than seven hours per night. But a new letter to the editor, authored by Ahad Wali Khan of Jinnah Sindh Medical University and Aisha Riaz of Karachi Medical and Dental College and published in the same journal, argues that the study’s headline conclusions rest on methodological and statistical foundations that deserve far more scrutiny than they have received.

The core of the critique concerns a statistical paradox surrounding insomnia severity. Baron and colleagues built part of their interpretation on the idea of hyperarousal, a physiological state in which the brain and body remain unusually activated during sleep. Under this framework, people with insomnia symptoms may genuinely experience sleep as wakefulness, perceiving themselves as lying awake when, objectively, they have been asleep. It is an influential concept in sleep medicine, and it offers an elegant explanation for why subjective and objective measures of sleep so often diverge. Yet Khan and Riaz point out that the study’s own numbers appear to undercut this narrative. In the partially adjusted model reported in the paper’s Table 2, the Insomnia Severity Index, a standard questionnaire measure abbreviated as ISI, was a statistically significant predictor of sleep underestimation, with a beta coefficient of negative 0.0497 and a p value of 0.0077. In the fully adjusted subjective analysis, however, that significance vanished entirely, leaving a p value of 0.1503.

That shift matters because of what statistical adjustment actually does. Partially adjusted models account for only a limited set of confounding variables, while fully adjusted models attempt to isolate the independent contribution of each predictor by controlling for everything else measured in the study. When a variable that appears significant in a simpler model loses its significance once broader factors are included, it suggests that the variable’s apparent effect may have been an artifact of its relationship with those other factors. In this case, Khan and Riaz argue, the loss of significance for the Insomnia Severity Index implies that subclinical insomnia does not act independently as a cause of sleep misperception once global stress and overall sleep difficulties are taken into account. If that is true, the authors of the letter contend, the hyperarousal interpretation should be treated with caution rather than folded into the study’s finalized conclusions as though it had been confirmed.

The second major criticism strikes at something even more fundamental: the reliability of actigraphy itself. Actigraphy devices, typically worn on the wrist, use accelerometers to detect movement and apply algorithms to infer sleep and wakefulness from motion patterns. The assumption is simple enough. When the wearer is still, the device scores sleep; when the wearer moves, the device scores wakefulness. Baron and colleagues were candid about the technology’s limits, acknowledging in their own paper that actigraphy does not serve as ground truth and that it tends to overestimate sleep duration relative to polysomnography, the gold-standard laboratory method that records brain waves, eye movements, and muscle tone. Polysomnography can distinguish light sleep, deep sleep, rapid eye movement sleep, and wakefulness with far greater precision, but it is expensive, intrusive, and impractical for extended home monitoring, which is precisely why actigraphy remains so widely used.

The problem, as Khan and Riaz frame it, is a contradiction between what the original authors admitted and what they then did. If actigraphy systematically overestimates sleep, using its output as a fixed benchmark against which subjective reports are judged becomes questionable. The discrepancy between a person’s self-reported sleep and the device’s estimate could reflect genuine misperception on the part of the sleeper, or it could reflect measurement error on the part of the device. Without a more accurate reference standard, the study cannot cleanly separate the two. The letter’s authors argue that this limitation is especially consequential in the specific cohort under study, which consisted of highly stressed individuals with elevated blood pressure, conditions associated with physiological hyperarousal.

Why does hyperarousal make actigraphy particularly unreliable in this population? The answer lies in what sleep scientists sometimes call quiet or silent wakefulness. A person lying motionless in bed, mind racing but body still, produces almost no accelerometer signal. To the device, that stillness is indistinguishable from sleep. In a stressed, hypertensive cohort, the letter suggests, such silent wakefulness may be common, meaning the actigraph could be scoring substantial periods of wakefulness as sleep. If that happened, the so-called objective measure would inflate sleep duration, and the apparent underestimation by participants, who reported sleeping less than the device recorded, could actually be an artifact of device overestimation rather than a genuine perceptual error. The direction of the discrepancy, in other words, might be reversed.

Khan and Riaz raise a further technical question about how actigraphy algorithms handle the boundary between sleep and quiet wakefulness. Whether a given device and scoring algorithm records complete sleep and silent wakefulness as discrete categories, or instead merges them into a single undifferentiated signal, determines whether the resulting data can support the study’s conclusions at all. If the algorithm merges the two states, the objective estimates become unreliable for assessing misperception, because the benchmark itself is contaminated. This is not a hypothetical concern specific to one study; it is a structural weakness of movement-based sleep measurement that becomes acute precisely in the populations where sleep misperception is most likely to occur, namely people with stress, insomnia symptoms, and heightened physiological arousal.

The exchange highlights a broader tension in contemporary sleep science, one that has grown more visible as consumer wearables have flooded the market. Millions of people now check nightly sleep scores on their wrists, treating the numbers as objective fact. But the same limitations that complicate research-grade actigraphy apply to consumer devices, often with less validation. Studies comparing wearable estimates to polysomnography have repeatedly found that these devices tend to overestimate total sleep time, particularly in people with fragmented or disturbed sleep. When researchers then use such devices to study the gap between perceived and actual sleep, they risk building interpretations on a foundation that shifts depending on who is wearing the device and under what conditions. The letter by Khan and Riaz is a reminder that the word objective, when attached to a measurement, does not automatically mean accurate.

What would a more rigorous approach look like? The letter does not propose a full redesign, but its logic points toward several priorities. Studies of sleep misperception in hyperaroused populations would benefit from polysomnographic confirmation, at least in a subsample, to calibrate the actigraphic estimates against a standard that can actually distinguish silent wakefulness from sleep. Statistical modeling should present fully adjusted results as the primary basis for interpretation, rather than allowing partially adjusted findings to drive conceptual claims about mechanisms like hyperarousal. And researchers should be explicit about the direction of potential measurement bias, considering device overestimation as an alternative explanation rather than defaulting to narratives about patient misperception. None of this diminishes the value of the original study’s contribution in drawing attention to the discrepancy between subjective and objective sleep assessment in short sleepers, but it does suggest that the field’s confidence in actigraphy-based conclusions should be calibrated to the technology’s real limits.

For now, the letter stands as a concise but pointed methodological intervention, published as Volume 22, article number 112 of the Journal of Clinical Sleep Medicine, with no external funding and no declared competing interests. Its authors conceptualized the critique together, with Khan drafting the initial manuscript and Riaz revising it for intellectual content. Whether Baron and colleagues respond in kind remains to be seen, but the substance of the critique is likely to resonate beyond this single exchange. Sleep misperception is a phenomenon with real clinical consequences, shaping decisions about treatment, driving worry in people who believe they are chronically sleep-deprived, and informing public health messaging about sleep duration. If the instruments used to define objective sleep systematically blur the line between sleeping and lying quietly awake, then both researchers and the public would do well to treat wrist-based sleep numbers as estimates with known blind spots, not as ground truth. The letter’s central message is ultimately about epistemic humility: before declaring that people misperceive their sleep, science must first be certain that the devices measuring it are not misperceiving in their place.

Subject of Research: Methodological limitations of actigraphic sleep measurement and statistical analysis in studies of sleep duration misperception

Article Title: Limitations in study of actigraphic estimates of sleep

Article References: Khan, A. W., & Riaz, A. (2026). Limitations in study of actigraphic estimates of sleep. Journal of Clinical Sleep Medicine, 22(1), Article 112. https://doi.org/10.1007/s44470-026-00132-5

Image Credits: AI Generated

DOI: 10.1007/s44470-026-00132-5

Keywords: actigraphy, sleep duration, sleep misperception, insomnia, hyperarousal, polysomnography, statistical adjustment, wearable devices, sleep medicine, measurement bias, Journal of Clinical Sleep Medicine, silent wakefulness