The role of sleep in dementia prevention: building the case for causality

Targeting modifiable risk factors is a promising approach to the primary prevention of dementia [1]. Growing evidence suggests that disturbed sleep is one such modifiable risk factor, with treatment options that may hold promise for future dementia prevention efforts. In the Maastricht Study, van Baal et al. examined cross-sectional associations of accelerometer-assessed time in bed (TIB), “sleep breaks” (i.e. episodes of wakefulness), and self-reported metrics of sleep continuity and daytime sleepiness with cognitive performance and structural brain MRI outcomes [2]. The authors found that estimates of longer TIB, derived from hip-worn accelerometer data, were related to poorer cognitive performance, and that both short and long TIB were associated with lower gray matter (GM) volume, demonstrating a commonly reported inverted U-shaped relationship. Additionally, they found that participants with two or more “sleep breaks” had poorer cognitive performance and lower GM volume. These findings highlight the role of sleep in cognitive and brain health and contribute to the growing literature linking sleep to dementia. While the 2024 Lancet Commission’s report on Dementia Prevention, Intervention, and Care has acknowledged a potential role of disturbed sleep in dementia prevention, it has not yet included sleep among their 14 modifiable dementia risk factors [3]. This raises the question of how we can continue to build the most compelling case for a causal link in sleep and dementia research.

As acknowledged by van Baal et al., the cross-sectional design of this study limits the extent to which we can draw inferences about potential causal associations between sleep and brain health. Longitudinal research is an important step for establishing causality, due to the temporal requirement that sleep disturbances precede the onset of dementia. However, this is complicated by the likely bidirectional link between sleep and dementia, in which sleep disturbances may be causal risk factors for dementia, while also being common consequences of neurodegeneration [4]. Longitudinal studies with a long follow-up period can help reduce the influence of reverse causation, in which an observed sleep-dementia association may reflect the influence of neurologic changes on sleep. Several studies of this nature exist and provide a compelling evidence base [5–9].

The inclusion of structural MRI-derived neuroimaging outcomes in van Baal et al.’s study (brain volumes, white matter hyperintensities, lacunar infarcts, and cerebral microbleeds) alongside cognitive assessments highlights the value of integrating multiple indicators of brain health. In the last decade, a substantial number of studies have incorporated neuroimaging and fluid biomarkers for Alzheimer’s disease (AD) and AD-related dementias (ADRDs) into research designs and sleep disturbances have been linked to AD/ADRD biomarkers and other neural correlates of neurologic disease [10–17]. These studies help elucidate the biological mechanisms underlying the sleep-dementia association, providing another important source of evidence toward causality. In the context of observational research, studies measuring sleep in populations free of AD/ADRD pathology at baseline with subsequent repeated measures of AD/ADRD biomarkers, cognitive performance, functional status, and adjudication of mild cognitive impairment and dementia diagnoses will provide additional evidence for a causal link [18]. Such designs permit mediation analyses examining underlying AD/ADRD pathology that would further support causal reasoning.

van Baal et al. investigated both accelerometer-assessed and self-reported sleep parameters in relation to brain health. This is a strength, given self-reported and objective sleep measures (e.g. polysomnography, actigraphy) capture different dimensions of sleep, and each has its own advantages and limitations. For example, although self-reported measures are subject to various sources of bias, such as recall bias and social desirability bias, they provide important context for understanding individuals’ perceptions of their sleep (e.g. quality, satisfaction). Conversely, objective sleep measures provide data that participants are unable to accurately report (e.g. frequency and duration of arousals, and in the case of PSG with EEG, time spent in distinct sleep stages, spectral power). In addition, these complementary measurement modalities can lead to discrepant results. Indeed, in van Baal et al., accelerometer-derived measures of “sleep breaks” were associated with brain health outcomes, but a self-reported measure of sleep continuity was not.

The growing emphasis on multidimensional sleep health, exemplified by frameworks such as “RU-SATED,” underscores the value of considering a broad range of sleep domains (regularity, satisfaction, alertness, timing, efficiency, duration) in both clinical and research settings, and extends the study of sleep beyond categories of clinical disorders [19–21]. Research examining not only individual sleep parameters but also their combinations will enrich our understanding of how sleep affects brain health. This knowledge can ultimately be translated into interventions targeting the most promising dimensions.

While observational studies provide important insights regarding links between sleep and dementia and represent a critical piece of evidence, randomized controlled trials represent the gold standard for causal evidence. Interventions such as cognitive-behavioral therapy for insomnia (CBT-I) and continuous positive airway pressure (CPAP) for sleep-disordered breathing are ready for deployment in dementia-prevention trials. Separately, as finer-grained data regarding the specific aspects of sleep that most strongly modulate dementia risk accrue, they may point to treatment targets (e.g. novel aspects of sleep architecture) on which CBT-I and CPAP do not most directly intervene. Fortunately, investigators are developing and optimizing other interventions (e.g. acoustic stimulation to improve slow-wave sleep) that would help address these needs.

Given the increasing heterogeneity observed with aging [22], it is unlikely that a “one-size-fits-all” approach will be effective in targeting sleep disturbances for dementia prevention. Identifying high-risk subpopulations, such as those defined by demographics (e.g. minoritized populations) or health conditions (e.g. cardiovascular risk factors closely related to sleep and dementia) will help direct prevention efforts, which may necessitate the tailoring of interventions to these subgroups. There is also a growing recognition that multidomain interventions for dementia prevention, which target several dementia risk factors simultaneously—as in the US POINTER trial—might better address the complex and likely multifactorial etiology of dementia [1, 23]. Disturbed sleep could be included alongside other modifiable dementia risk factors (e.g. exercise for physical inactivity, hearing aids for hearing loss) in these interventions [24, 25]. To date, sleep disturbances have not typically been a component of these multidomain intervention studies.

Ultimately, the study by van Baal et al. adds to an expanding body of research highlighting links of sleep disturbances with cognitive and brain health. Although challenges remain in establishing a definitive causal link between sleep and dementia, there is now substantial evidence from both human and animal research, including from recent meta-analyses, supporting a contributory role of sleep in dementia risk [5, 10, 26, 27]. This body of work supports the inclusion of sleep as a modifiable risk factor for dementia prevention.

Disclosure statement

Financial disclosure: APS served as a paid consultant to Sequoia Neurovitality, BellSant, Inc., and Amissa, Inc.

Non-financial disclosure: None.