Time to regularize sleep regularity

From a quantitative view, research into sleep health has traditionally been viewed from the lens of duration: How long does sleep need to optimize performance, well-being and physical health? The focus on duration has been a key part of sleep medicine and as a cornerstone in public health campaigns promoting. For example, current guidelines, supported by scientific evidence, recommend that adults get 7–9 hours of sleep each day to maintain optimal health [1]. Other aspects of sleep, such as its quality and timing, and regularity, have increasingly been shown to be critical determinants of health outcomes [2, 3]. In some cases, these factors are even more predictive of disease risk than habitual sleep duration alone [4], underscoring the complexity of sleep’s role in health.

Less than ten years ago, the sleep regularity index, which estimates the day-to-day consistency of sleep–wake timing, was introduced as a novel metric to estimate sleep variability [5]. Evidence supporting the utility of such an index has rapidly accumulated. Reduced day-to-day consistency in sleep–wake timing has been associated with an increased risk of a wide range of adverse health outcomes, including poor academic performance [5], reduced quality of life [6], and diminished functional capacity in older adults [7]. It has also been linked to higher risks of depression and anxiety disorders [8], type 2 diabetes [9], major adverse cardiovascular events [10], dementia [11], and shorter lifespan [12].

Indeed, several large-scale studies now support the importance of focusing on sleep regularity: in one study, with data from nearly 61,000 UK Biobank participants with an average age of 63 years, sleep regularity as measured by 7-day accelerometry was a stronger predictor of all-cause mortality than sleep duration [4]. A separate analysis of the same UK Biobank accelerometry data found that pronounced day-to-day inconsistency in sleep–wake timing was associated with a higher risk of major adverse cardiovascular events, even among participants who met age-specific sleep duration recommendations [10]. Similar findings have been reported for the incidence of type 2 diabetes [9].

With these findings in mind, the field of sleep research is at an important inflection point. The growing body of evidence suggests that regularity deserves equal, if not greater, consideration than sleep duration [3, 13]. For researchers, this calls for multidimensional models of sleep health that systematically incorporate duration, quality, timing, and regularity (Figure 1). For clinicians, it underscores the importance of counseling patients not only on how much they sleep but also on when and how consistently they do so. Encouragingly, the existing arsenal of diagnostic tools, such as sleep diaries, actigraphy devices, and sleep-tracking wearables [14], can already be used to assess sleep regularity. For public health campaigns, the message should expand to emphasize both sleep duration and the regularity as equally important aspects of long-term sleep health.

Figure 1.

The four paths to sleep health. Sleep health can be conceptualized as supported by four key domains: Sleep regularity, sleep duration, sleep quality, and circadian timing of sleep. The figure was partly created using BioRender.

Despite the promising evidence supporting its potential, a major challenge for interpreting findings on sleep regularity is the lack of standardization in how this metric is defined and measured. Methods for calculating sleep regularity can vary considerably, depending on whether data are derived from self-reports, sleep diaries, or objective measures such as actigraphy or accelerometry. Even within studies using the same type of data, different analytic approaches, such as the sleep regularity index, the standard deviation of sleep timing, or other derived measures, can yield divergent estimates of regularity.

To address this issue, it is important to investigate whether commonly used algorithms for determining sleep regularity may differ in terms of how closely they correlate with one another and how strongly they relate to health outcomes previously associated with sleep regularity. In the present issue of Sleep, Czeisler et al. [15] took on this challenge by calculating the sleep regularity index using two widely applied open-source tools: sleepreg and GGIR. By analyzing more than 70,000 adults with accelerometer-derived sleep–wake data from the UK Biobank, the authors found that the two calculators produced markedly different scores. For instance, only two-fifths of participants were classified into the same sleep regularity index quintile. These discrepancies were not trivial: when applied to prospective models of health outcomes, including all-cause mortality, type 2 diabetes, and atrial fibrillation or flutter, the choice of calculator alone meaningfully altered results and interpretations. Using sleepreg, middle-aged participants with the most irregular sleep patterns had a 1.19-fold higher adjusted hazard of death compared to those with the most consistent sleep patterns over an approximately 7.5-year follow-up period. In contrast, no significant association was observed when the same data were analyzed using GGIR. For type 2 diabetes, both calculators suggested that irregular sleepers were at greater risk, but the strength of the association was notably larger and encompassed another quintile of people with sleepreg. Similarly, for atrial fibrillation or flutter, sleepreg indicated significantly higher risks among those with irregular sleep, while GGIR did not show a significant difference. In short, the method used to calculate sleep regularity could determine whether irregular sleep is identified as a meaningful health risk.

These findings have significant implications for both science and practice. From a research perspective, they underscore an urgent need for methodological harmonization. When different calculators yield divergent risk estimates— or even contradictory conclusions about whether an association exists— the field risks generating inconsistent or misleading evidence about the health effects of sleep regularity, undermining reliable comparisons and meta-analyses. From a translational perspective, clinical guidelines, public health initiatives, and patient recommendations can likewise be compromised. Standardized and transparent approaches to measuring sleep regularity are therefore essential - not only to advance research but also to ensure that future recommendations rest on solid, longitudinally robust foundations.

With these considerations in mind, Czeisler et al. [15] have proposed a 14-item Reporting Items for Regularity Indices (RIRI) statement to improve transparency, standardization, and reproducibility in future research on sleep regularity. This initiative represents a timely and necessary step for the field. By ensuring that methods for calculating sleep regularity are consistently reported and harmonized, the RIRI framework can strengthen comparability across studies, accelerate scientific progress, and provide a solid foundation for clinical translation. With rigorous methodology and actionable recommendations, the field is now well positioned to position sleep regularity as a cornerstone of not only continued sleep research, but also of sleep medicine and public health campaigns.

Funding

Supported by Novo Nordisk Foundation grants NNF24OC0101995 (J.C.) and NNF23OC0081873 (C.B.).

Disclosure statement

Financial disclosure: None.

Non-financial disclosure: None.