Optimizing Dementia Risk Reduction: Balancing Sleep and Physical Activity Trade-offs

Optimizing Dementia Risk Reduction: Balancing Sleep and Physical Activity Trade-offs | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Optimizing Dementia Risk Reduction: Balancing Sleep and Physical Activity Trade-offs Matthew Pase, Stephanie Yiallourou, Lachlan Cribb, Beaudan Campbell-Brown, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4392320/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Engaging in regular physical activity and obtaining recommended amounts of sleep are touted as strategies to promote healthy brain aging. However, as each day is only 24 hours long, changing time spent in one activity must come at the expense or gain of another, making it necessary to understand how the whole 24-hour activity composition impacts dementia risk. We applied compositional data analysis to investigate the effect of substituting sleep duration for different levels of physical activity (i.e., inactivity, light activity, and moderate to vigorous physical activity; MVPA) on dementia risk relative to two reference compositions; a “typical” short sleeper (< 6hrs) and normal sleeper (≥ 6hrs). The study sample comprised participants from the community-based UK Biobank with 24-hour behaviors estimated using 7 days of accelerometry. The mean age of the sample was 63 years (Q1, Q3: 56, 68); 56% were women. Of the 88,654 participants, there were 718 incident all-cause dementia cases over a median follow-up of 8.2 years. For short sleepers, increasing sleep duration at the expense of inactivity or light activity was associated with a lowering of dementia risk, but not when at the expense of MVPA. For persons with normal sleep duration, the effect of increasing or decreasing sleep duration on dementia risk differed for all three substituted behaviors (i.e., inactivity, light, or MVPA). Most notably, dementia risk was higher when increasing sleep at the expense of MVPA and lower when increasing MVPA at the expense of sleep. The interpretation of the results was broadly consistent when using MRI-based outcomes (e.g., hippocampal volume) in a subset with brain imaging (n = 15,263). Our findings underscore the complexity of optimizing dementia risk reduction strategies, emphasizing the need for personalized approaches that balance trade-offs between sleep duration and differing physical activity levels based on individual circumstances, such as habitual sleep duration. Health sciences/Neurology/Neurological disorders/Dementia Health sciences/Risk factors Figures Figure 1 Figure 2 Figure 3 Figure 4 INTRODUCTION Sleep has a critical function in the clearance of amyloid beta and tau proteins, which can aggregate to form the two pathological hallmarks of Alzheimer’s disease 1,2 . Both short (≤ 6hr) 3 and long (> 9hr) 4 sleep duration have been associated with an increased risk of dementia 5 . Daytime activity levels have also been linked to dementia risk. For example, low physical activity levels and high sedentary time are associated with higher dementia risk 6,7 . Intuitively, normalizing sleep duration or increasing physical activity could be attractive targets to reduce dementia risk. However, as each day is only 24 hours long, changing time spent in one activity must come at the expense or gain of another. Consequently, the potential benefits of several lifestyle interventions are not straightforward; their efficacy may depend on the substituted behavior. For example, although increasing physical activity has been touted as one strategy to lower dementia risk, is it advisable if it comes at the expense of sleep time? Previous studies on sleep time and physical activity's link to dementia tend to overlook the fact that these activities are a quantitative component of a 24-hr composition. Traditional regression methods isolate sleep or daytime activities with only partial adjustment for time spent in other behaviors, ignoring that time-use behaviors are co-dependent (if one behavior increases, another must decrease). Consequently, the questions posed by these studies tend to be ill-defined. It is difficult to interpret the estimated effect of increasing sleep time when the precise behaviors that the additional sleep time replaces are left unspecified. In this way, observational studies generally define physical activity and sleep time exposure variables in ways that do not approximate well-defined, precise interventions, precluding meaningful causal questions; this may mean that results from such studies do not translate well to meaningful guidelines (i.e., increase physical activity but at the expense of what?). Indeed, this may also help explain recent unexpected findings, whereby associations between physical activity and cognition were only minimal 8 . The issue is of profound importance given that 1) tackling modifiable risk factors offers substantial hope in reducing the growing burden of dementia 9 , and 2) dementia risk reduction guidelines lean heavily on observational data given the challenges of conducting clinical trials with dementia endpoints 9 . Robust dementia prevention guidelines involving sleep duration or daytime physical activity levels (i.e., sedentary time, light physical activity, and moderate to vigorous physical activity; MVPA) require a complete understanding of how adding and subtracting different activity compositions affects dementia risk. Furthermore, there is a need to understand the effect of adding and subtracting sleep duration for varying levels of physical activity (or inactivity), given the role of sleep in glymphatic clearance and that the benefits of physical activity on cognitive health may be dependent on the amount of sleep 10 . Therefore, we applied compositional data analysis (CoDA) to a large community-based cohort with 24-hr behaviors estimated through 7 days of accelerometry monitoring and dementia follow-up. Unlike standard regression tools, CoDA respects the fact that time-use data are constrained to 24 hours and allows for estimating the effect of explicit time-use substitutions (e.g., adding 1 hour of sleep at the expense of 1 hour of MVPA). We aimed to determine the effect of substituting sleep duration for daytime physical activities (inactivity, light activity, and MVPA) on dementia risk and MRI volumetric outcomes. Moreover, we also aimed to determine the ideal composition of 24-hr behaviors (the absolute time spent in sleep, inactivity, light activity, and MVPA) associated with the lowest dementia risk. METHODS Study design and participants. This prospective cohort study uses data from the UK Biobank (UKB). The UK Biobank recruited over 500,000 UK adults aged 40–69 years during 2006–2010. Detailed lifestyle, environmental, and genetic details were collected at the baseline session (described in detail on the study website: https://www.ukbiobank.ac.uk/ ). Follow-up for the occurrence of several health outcomes is ongoing. Participants identified as living within a reasonable traveling distance of the 22 UKB assessment centers in the UK National Health Service patient registers were invited by mail to participate in the study. The participants who agreed to participate tended to be older, had higher socioeconomic status (SES), and were more likely to be female than non-participants, with a participation fraction of 5.5% 11 . Between February 2013 and December 2015, 236,519 UKB participants were invited to participate in a 7-day wrist-worn accelerometer study. A total of 106,053 agreed to take part 12 . All UKB participants provided informed consent to participate in the study. UK Biobank has approval from the North West Multi-Centre Research Ethics Committee as a Research Tissue Bank (RTB) approval. This approval means the present study operates under the RTB approval, satisfying the requirements of the Monash University Human Research Ethics Committee. For this study, participants with dementia or other severe neurological disease (Parkinson's disease, motor neuron disease, cerebral palsy, brain abscess, multiple sclerosis, or myasthenia gravis) at or before the time of the accelerometry study were excluded. The sample selection process from the total UKB sample is displayed in Extended Data eFigure 1. Measurement of 24-hr behaviors by accelerometry. For the objective assessment of 24-hr behaviors, participants underwent 7 days and nights of wrist-worn accelerometry (Axivity AX3). Further details of the accelerometry protocol have been published previously 12 . For the determination of time spent in sleep, inactivity, light activity, and MVPA, we used the open-source R package GGIR 13 . Raw acceleration data were processed using the Euclidean Distance Minus One procedure (ENMO) 14 . The default GGIR thresholds were used to distinguish inactivity from light activity and MVPA: < 30 mg for inactivity, ≥ 30 and < 100mg for light, and ≥ 100mg for MVPA. As a sleep diary was not available in the UKB, time in sleep was estimated using the algorithm of van Hees et al 15,16 . Low-quality accelerometer data were removed using the established UKB criteria: lack of agreement between self-reported wear time and accelerometer wear time data (5%); insufficient wear time (< 72 hours; 5%); and poor calibration (< 1%). Additionally, data were removed for participants for whom GGIR was unable to determine a sleep window (5%) and for participants providing less than two valid contiguous wear days (< 1%). Dementia case ascertainment. Incident all-cause dementia was ascertained based on hospital records, death records, and primary care 17 . As UKB record linkage with primary care providers is ongoing, primary care record linkage was only available for slightly less than half of the complete sample (approximately 45%). The algorithms used to ascertain all-cause dementia were designed to maximize positive predictive value, which has been shown to be high (> 80% for each record type) 18 . The list of Read V2 codes (primary care) and ICD-10 codes informing the identification of all-cause dementia have been published previously 19 . Brain MRI outcomes. The UKB imaging sub-study began in 2014 and aimed to acquire high-quality and consistent imaging data from 100,000 UKB participants across multiple modalities. All MRI data were acquired on 3T Siemens Skyra scanners. Imaging-derived phenotypes (IDPs) were derived using an automated pipeline described in depth previously 20 . Outcomes used for the current study included total brain volume, hippocampal volume, total grey matter volume, total white matter volume, and white matter hyperintensity volume. Volumetric IDPs of interest were normalized for head size using a head-size scaling factor 21 . Data analysis. The daily time spent in each behavior was averaged across the accelerometry wear days using a linear mixed effects model with a random intercept for participant and fixed effect for day of the week. Estimated averages were standardized over the day of the week to ensure comparability of those with and without weekend assessments. Confounder variables were selected using a causal directed acyclic graph (Extended Data eFigure 2). Primary model covariates included age, sex, education, ethnicity, APOE ε4 genotype, household income, fruit and vegetable intake, alcohol intake, antidepressant, antipsychotic, or sedative medication use, retirement status, and shift work. Missing data in confounder variables were infrequent (most < 2%) and were imputed by predictive mean matching using R package mice 22 . Imputation models included all confounder, exposure, and outcome variables (dementia status and age at dementia). Isotemporal substitution. The average time spent in each behavior (sleep, inactivity, light, MVPA) was expressed as a proportion of total daily time, summing to 1. These variables cannot be entered into traditional statistical models (e.g., multivariable regression) due to the perfect collinearity among components 23 . Accordingly, they were first transformed from the simplex into the unconstrained (Real) space using an isometric log ratio (ILR) transformation 24 . Pooled over time logistic models were then fitted with ILR coordinates and covariates as predictors 25 . For these models, follow-up was divided into discrete chunks (6 months in length) such that each participant had as many rows of data as they had discrete follow-up intervals, ending either at the time of dementia, death from non-dementia causes (competing event), or end of study follow-up. Follow-up time was modeled with a restricted cubic spline with 5 knots. Age was used as the timescale. All continuous covariates were modeled with restricted cubic splines (knots at the 10th, 50th, and 90th percentiles) to allow for departures from linearity. Quadratic terms were included for ILR variables. The gformula estimator described by Young et al. 26 was used to estimate cause-specific cumulative incidences (hereafter, “risks”) of dementia by age 76, the median age at dementia occurrence. This approach involves fitting pooled logistic models to approximate hazards for the event of interest (i.e., dementia) and the competing event (i.e., death) and calculating the risk of the event of interest as a function of these hazards, appropriately accounting for competing events. Risks were estimated for two reference compositions: the typical “short sleeper” and the typical “normal sleeper.” Risk ratios were estimated for discrete substitutions to each reference composition (e.g., adding 1 hour of MVPA at the expense of 1 hour of sleep for the typical short sleeper). The typical short sleeper and normal sleeper compositions were the geometric mean composition among those with 9 hours of sleep). This categorization was based on previous literature demonstrating that short sleep duration of < 6 hours associates with increased dementia risk 3 . This analysis used nonparametric bootstrapping with nested single imputation with 500 samples to obtain percentile-based 95% confidence intervals. All other analyses, including MRI volumetric outcomes and all sensitivity analyses, used 250 bootstrap samples to minimize computational burden. Ideal composition. To estimate the “ideal” composition, we created a grid of all compositions (in 15-minute steps) covering all possible compositions within the range of the sample data. We then fitted a multivariate Normal to all but one of the time-use variables in the sample data. Each member of the grid of compositions was then passed into the density function of this estimated Normal, and any compositions with low density (< 2.5th percentile) were excluded, thereby removing improbable compositions from consideration. Predicted dementia hazard was then estimated from the fitted pooled logistic model for each retained composition, with the composition returning the smallest estimated hazard deemed the “ideal” composition. The “worst” composition was computed analogously. The “typical” composition was chosen as the composition in the sample data that maximized the density function of the estimated Normal. MRI volumetric outcomes. A linear regression model was fitted for volumetric MRI outcomes including covariates and the ILR coordinates as predictors. Following UKB recommendations, these models were additionally adjusted for MRI assessment center, mean fMRI head motion, and head location in scanner 21 . The mean difference in volumetric MRI outcomes for a given substitution was estimated analogously to the dementia models. The “ideal” and “worst” compositions were then passed through this model to compare volumetric outcomes between the two compositions. Sensitivity analysis. In the first sensitivity analysis, we included additional adjustment for sleep fragmentation (wake after sleep onset; WASO), also assessed by accelerometry. In the second, we additionally adjusted for covariates that may be plausibly affected by time use, including history of cardiovascular disease, body mass index, blood pressure medication, systolic blood pressure, and sickness or disability (self-reported employment category). In the third, we removed participants with dementia events during the first three years of follow-up. A substantial reduction in the estimated risk ratios in this sensitivity analysis may indicate reverse causation. Finally, to assess the effect of selection bias due to the non-representativeness of the UKB (healthy cohort effect), we refitted the primary model, standardizing to the distributions of sex, retirement status, income, and smoking status of the representative UKB pseudo-population described by Schoeler et al. 27 . Product terms between each of these variables and the ILR coordinates were included in the latter model to allow for effect modification. All analysis code is available at the project GitHub repository ( https://github.com/BeaudanBrown/coda-dementia ). RESULTS Characteristics of the sample at baseline are described in Table 1 . The total sample size was 88,654. The mean age of the sample at the time of accelerometry was 63 years (Q1, Q3: 56, 68); 56% were women. There were 718 incident all-cause dementia cases over a median follow-up of 8.2 years (25th percentile 7.6; 75th percentile 8.7). The median time between accelerometry and MRI assessments was 3.4 years (25th percentile 1.9; 75th percentile 4.4). The associations between discrete time-use substitutions and dementia risk are displayed in Fig. 1 . Results are presented separately for the normal sleep duration (≥ 6 hr) and short sleep duration (< 6 hr) reference compositions. Table 1 Baseline sample characteristics Characteristic Summary Age, years, median (Q1, Q3) 63 (56, 68) Women, n (%) 49,877 (56) BMI, kg/m2, median (Q1, Q3) 26.0 (23.6, 29.0) Highest qualification, n (%) High school non-completers 12,988 ( 15 ) High school completers 5,476 (6.2) Trade qualification 10,162 ( 12 ) Graduate degree 51,799 (59) Other 7,332 (8.3) Prefer not to answer 373 (0.4) Employment, n (%) Paid employment 54,836 (62) Retired 27,628 ( 31 ) Sick or disabled 1,341 (1.5) Other 4,652 (5.2) Prefer not to answer 159 (0.2) Average total household income (thousand pounds), n (%) 100 5,800 (6.6) Don’t know/Prefer not to answer 8,479 (9.6) Ethnicity, n (%) White 81,571 (92) Asian 3,424 (3.9) Other 3,371 (3.8) Prefer not to answer 250 (0.3) Antidepressant medication 5,009 (5.7) Insomnia medication 749 (0.8) Smoking status, n (%) Never 50,561 (57) Former 31,780 ( 36 ) Current 6,076 (6.9) APOE ε4 alleles, n (%) 0 53,357 (72) 1 19,150 ( 26 ) 2 1,629 (2.2) History of cancer, n (%) 11,503 ( 13 ) History of cardiovascular disease, n (%) 35,789 ( 41 ) History of diabetes, n (%) 3,689 (4.2) Average sleep duration, hours/day* 6.5 (1.2) Average inactivity, hours/day* 12.0 (1.1) Average light activity, hours/day* 3.3 (1.3) Average moderate to vigorous activity, hours/day* 1.7 (1.5) * Geometric mean (geometric standard deviation). Baseline is defined as the time of accelerometry. Time-use substitutions for the normal sleeper reference composition. Replacing inactivity with sleep time (and vice versa) had little association with dementia risk (Fig. 1 a). In contrast, replacing light activity with sleep time was associated with a lowering of dementia risk (1b); the RR associated with reallocating 1 hour/day of light activity to sleep was 0.67 (95% CI: 0.54, 0.80), while the RR associated with reallocating 1 hour/day of sleep to light activity was 1.43 (95% CI: 1.27, 1.67). Replacing MVPA with sleep time was associated with the largest increase in dementia risk (Fig. 1 c); the RR associated with reallocating 1 hour/day of MVPA to sleep was 2.03 (95% CI: 1.70, 2.47), while the RR associated with reallocating 1 hour/day of sleep to MVPA was 0.75 (95% CI: 0.59, 0.89). Thus, the effect of increasing or decreasing sleep duration displayed a different association with dementia risk for all three substituted behaviors. Time-use substitutions for short sleeper reference composition. Unlike for the typical normal sleeper, replacing inactivity with sleep in the typical short sleeper was associated with a lowering of dementia risk (Fig. 1 d); the RR associated with reallocating 1 hour/day of inactivity to sleep was 0.89 (95% CI: 0.82, 0.96), while the RR associated with reallocating 1 hour/day of sleep to inactivity was 1.28 (95% CI: 1.18, 1.39). In the typical short sleeper, substitutions involving sleep and light activity, and sleep and MVPA appeared broadly consistent with the results observed in the typical normal sleeper, except for one crucial difference: whereas increasing MVPA at the expense of sleep was associated with a lower risk of dementia in the typical normal sleeper, this was attenuated in the typical short sleeper (Fig. 1 f). The RR associated with reallocating 1 hour/day of sleep to MVPA was 0.89 (95% CI: 0.71, 1.04). The RR associated with reallocating 1 hour/day of MVPA to sleep was 1.82 (95% CI: 1.47, 2.26). Overall, for the typical short sleeper, increasing sleep duration was associated with a lowering of dementia risk when at the expense of inactivity or light activity, but not MVPA. Ideal composition Figure 2 plots the cumulative incidence of dementia for the estimated “worst,” “typical,” and “ideal” composition. The worst composition diverged from the typical composition in the 6th decade of life (the earliest decade of life studied), with cumulative incidence increasing dramatically from age 70 onwards. In comparison, the cumulative incidence of dementia in the “typical” and “ideal” compositions was more similar and only began to diverge slightly in the 7th decade of life. MRI endophenotypes The association of the time-use substitutions with hippocampal volume are displayed in Fig. 3 . In the typical normal sleeper, replacing MVPA with sleep was associated with smaller hippocampal volume (e.g., 1 hr, mean difference [MD]: -0.15 cm 3 , 95% CI: -0.24, -0.07), while replacing sleep with MVPA was associated with little change in hippocampal volume (e.g., 1-hour, MD: 0.03 cm 3 , 95% CI: -0.00, 0.07). Replacing sleep with light activity was associated with smaller hippocampal volume (e.g., 1-hour, MD: -0.07 cm 3 , 95% CI: -0.10, -0.03). The pattern was similar for the typical short sleeper (Fig. 3 ) and other volumetric outcomes, including total brain volume, grey matter volume, white matter volume, and log white matter hyperintensities (Extended Data eFigures 3 to 6, respectively). Estimated MRI endophenotypes for the “worst,” “typical,” and “ideal” compositions are displayed in Fig. 4 . Brain volumes tended to be similar between the “ideal” and “typical” compositions and lower for the “worst” composition. Similarly, the log of white matter hyperintensities was similar between the “ideal” and “typical” compositions but higher for the ‘worst’ composition. Sensitivity analyses Results were not meaningfully different in a first sensitivity analysis adjusting for sleep fragmentation (WASO; Extended Data eFigure 7), nor in a second sensitivity analysis adjusting for chronic disease and chronic disease risk factor variables (Extended Data eFigure 8). In the third sensitivity analysis, truncating the first three years of follow-up did not substantively alter the results (Extended Data eFigure 9). The final sensitivity analysis, which corrected for selective participation in the UK Biobank, also did not substantively alter the results (Extended Data eFigure 10). DISCUSSION This study examined how sleep and physical activity trade-offs relate to brain health and dementia risk. Our findings reveal that dementia risk was influenced not only by the increasing behavior but also by the nature of the decreasing behavior and baseline sleep levels. In short sleepers, increasing sleep duration was associated with a reduction in dementia risk as long as the behavior being substituted was not MVPA, highlighting the importance of adequate sleep duration and MVPA to promote healthy cognitive aging. In normal sleepers, the association of increasing sleep duration was entirely dependent on the behavior being substituted; dementia risk decreased, increased, or remained stable when substituting out light activity, MVPA, or inactivity, respectively. Findings were similar when using brain volumes as outcomes in a subset with MRI. We also identified the most and least favorable combinations of 24-hour behaviors for dementia risk at age 76. Individuals with very short sleep, high inactivity, and low MVPA displayed the highest rates of dementia and evidence of accelerated brain aging on MRI. Overall, these results offer insights into potential behavior changes that could be targeted in interventions or guidelines to enhance brain health and prevent dementia. Previous studies have shown that short sleep duration is linked to poorer cognition 28–30 , lower brain volumes 31 , and increased dementia risk 3 . Therefore, normalizing sleep duration has been suggested as a potential way to delay the onset or reduce the risk of dementia. Findings from this study provide evidence of the specific behaviors to target in those with short sleep, namely inactivity and light activity. For those with short sleep duration, increasing sleep by 1 hour instead of engaging in inactivity or light activity was associated with an 11% and 29% reduction in dementia risk, respectively. These findings remain consistent following adjustment for measures of poor sleep quality, common comorbidities, and disease risk factors. These results suggest that it is more advantageous to increase sleep at the expense of light activity than inactivity. Although this seems counterintuitive, others have shown that increasing light activity at the expense of sedentary time was associated with reduced cognitive performance 32,33 . Our results and others may be explained by differences in cognitive load during these daytime behaviors. Some inactivity time may be spent engaging in cognitively stimulating (e.g., working at a computer, writing, reading, playing a musical instrument) or social (e.g., social dinner) activities 34 . In contrast, light activity may involve less challenging cognitive tasks (e.g., housework, a light stroll). Accordingly, in some people, replacing time spent inactive may reduce engagement in cognitive and social activities tied to a lower risk of dementia 35 . Future studies delineating the type of cognitive activity co-occurring with inactivity and light activity would be beneficial to determine if this is the case. Like short sleep, lower MVPA levels or self-report leisure time physical activity have been associated with higher dementia risk 6,7 . Previous isotemporal substitution studies have demonstrated that replacing 30 minutes/day of sleep with 30 minutes/day of MVPA was associated with better cognition in participants with self-report sleep duration > 7 hours/night 36 . Further, others found that replacing sedentary behavior with equal time engaging in different physical activities corresponded with decreased dementia risk in the UKB 37 . However, this study relied on brief questionnaires to estimate physical activity and did not consider sleep substitutions. We extend these studies and show that replacing MVPA with sleep was associated with an increase in dementia risk and poorer brain health outcomes in both normal and short sleepers. MVPA may enrich slow-wave sleep 35 , meaning that extending sleep duration at the expense of MVPA may actually decrease sleep quality. Findings from our study provide further weight to the importance of MVPA for brain health and support the premise of implementing exercise interventions for dementia prevention. Importantly, our study sheds light on potential intervention strategies. We show that increasing MVPA by 1 hour at the expense of sleep in those with normal sleep duration reduced dementia risk by 25%. So, waking up 1 hour early to exercise may benefit the brain. But , this same effect does not apply to short sleepers. In short sleepers, the benefits of increasing MVPA by 1 hour in the place of sleep were dampened, such that there was only a modest decrease in associated dementia risk. Thus, the benefits of MVPA on dementia risk depend on whether one gets adequate sleep (in this case, ≥ 6 hr). These results are consistent with findings from the English Longitudinal Study on Aging that found that the benefits of high levels of physical activity on cognitive decline were blunted in older adults who self-reported short (< 6 hours) versus normal (6–8 hours) sleep duration, with short sleepers having a more rapid decline over a 10 year follow-up 38 . These findings suggest that the neuroprotective effects of MVPA may not fully overcome the detrimental effects of short sleep. This study also explored the “worst, typical, and ideal” 24-hr activity compositions related to dementia risk. Those with high amounts of inactivity combined with low amounts of sleep and MVPA had the highest dementia risk, whereas those displaying the opposite patterns displayed the lowest risk. These compositions align well with current knowledge of these behaviors and dementia risk 6,7,39 . Our data also suggest that transitioning people from the “worst” to the “typical” composition may have a far greater impact on dementia prevention as compared to transitioning people from the “typical” to the “ideal” composition. However, it is important to note that the "ideal" composition in this study represents one ideal composition; there are likely many differing compositions resulting in similarly low estimated dementia risk. Thus, the "ideal" composition should not be considered the only one to pursue for improving brain health. Since the median follow-up duration for dementia was 8.2 years in our sample, there may be an element of reverse causation, given that several forms of dementia have a long preclinical phase 40 . Importantly, we didn’t find strong evidence to suggest this. Truncating the first three years of follow-up (during which reverse causation would be expected to be strongest) did not substantively alter the estimated risk ratios. Nevertheless, we felt it necessary to replicate findings with MRI-based subclinical endophenotypes. Hippocampal atrophy is an imaging characteristic of preclinical Alzheimer’s disease, and low hippocampal volume cross-sectionally is associated with poorer memory, worse clinical function, and higher dementia risk over the next decade 41 . We identified that those with the “ideal” composition had larger total brain, grey matter, and hippocampal volumes and less white matter injury than those with the “worst” composition. Since white matter injury is characteristic of small vessel disease and lower hippocampal volume is classical of Alzheimer’s disease, these data suggest that, combined, an “ideal” composition may protect the brain from various insults that can lead to dementia. Additional biomarker outcomes will be required to test this hypothesis further. In reference to the “ideal” composition, the amount of sleep required agrees with national guidelines for sleep recommendations, which is in the range of 7–9 hr. On the other hand, 3.3 hours/day of MVPA appears relatively high, given that the current WHO recommendations are 150–300 minutes/week to achieve health benefits. It is important to note the differences between device-worn and self-report measures of physical activity; physical activity guidelines have largely been based on self-report. It has been previously estimated that 150 min/week of self-reported MVPA equals ~ 1000 min/week of device-assessed MVPA 42 . Indeed, self-reporting could be biased only to include MVPA that is intentional or in conscious awareness. In contrast, device-assessed physical activity will also capture incidental MVPA that may go unnoticed (at work, for example). As wrist-worn devices with inbuilt activity trackers become more popular amongst the general public and are more easily implemented in large studies, our analysis captures measures of MVPA that may better reflect emerging societal and research trends. However, we acknowledge that our MVPA estimates are likely much higher than those measured by self-report or hip-worn accelerometry 43 . Nonetheless, the unique substitutions in this study overcome these nuances and are more applicable in providing a framework for personalized interventions, irrespective of ideal behavior targets. That is, an increase in MVPA by 1 hour is more achievable than simply aiming for the “ideal.” Notably, the 24-hour compositions identified in this study could be important for screening vulnerable populations that could be targeted for dementia prevention programs. Although causality cannot be assumed, several mechanisms may explain how changes in 24-hour activity compositions contribute to dementia risk. Glymphatic clearance of Alzheimer’s disease proteins (amyloid-β and tau) is maximal during sleep and thought to be coupled to slow wave sleep 1 . Shorter sleep duration may impair the clearance of these waste products that aggregate to form amyloid plaques and neurofibrillary tangles 1,2 . Short sleep can also increase blood pressure 44 and inflammatory processes 45 , potentially explaining links between short sleep and vascular brain injury and brain atrophy 46 . MVPA also has neuroprotective effects. For example, physical activity is thought to induce favorable alterations in cerebral blood flow, augment cognitive reserve via neuroplasticity processes, facilitate glymphatic clearance of amyloid-β, and reduce other dementia risk factors such as cardiovascular disease and stress 47 . Limitations The strengths of this study include the large sample size and the use of objective measures to assess 24-hour behaviors. However, we acknowledge that wrist-worn accelerometry may overestimate daytime behaviors compared to hip-worn devices. However, we argue that wrist-worn accelerometry estimates are more important to study given growing societal trends towards wrist-worn fitness trackers. We also recognize that our categorization of time use into, for instance, inactivity, does not distinguish sitting and standing, nor varied behaviors that may occur as part of those categories (e.g., reading, socializing, or watching television), some of which may plausibly be related to dementia risk in different ways. Our study only presents an association, estimated from observational data, between 24-hour behaviors and dementia risk. Therefore, we cannot rule out unmeasured confounding or reverse causation. Furthermore, as we did not have longitudinal data on 24-hour behaviors, the time when substitutions occurred, and the duration that those substitutions were maintained was unspecified 38 . Although this paper presents an important first step, future randomized trials or observational studies could be designed to clarify the effect of substitutions with a defined time course. Finally, those who participated in the UKB cohort tend to be healthier and of higher socioeconomic status than the general UK population, leading to potential issues with external validity. 11 Nevertheless, we did not find meaningfully different results in our sensitivity analysis, which at least partly corrected for selective participation in the UKB. Implications and conclusions Sleep duration has yet to receive mainstream recognition as a modifiable dementia risk factor, as evidenced by its omission from the 12 headline risk factors identified by the Lancet Commission’s dementia prevention, intervention, and care guidelines 9 . Moreover, findings for the true magnitude of benefits of physical activity on the aging brain remain contentious 8 . Accordingly, our data suggests that an ideal strategy for risk reduction should encompass a tailored approach that assesses all 24-hr behaviors and targets those behaviors that are amenable to have optimal effect. We found that, depending on whether one has short or normal sleep duration, substituting inactivity or light activity for sleep was associated with a favorable effect on brain aging and dementia risk. The same was true for substituting MVPA for sleep in those with normal but not short sleep duration. These data indicate that there could be some flexibility in setting behavior change goals to enable an achievable behavior change for the individual. For example, increasing MVPA would be an obvious choice to improve brain health, but getting enough MVPA could be challenging in some populations. Our data show that even an increase in sleep by 30 minutes/day (in place of light activity) may benefit these groups where reaching physical activity targets may be challenging. This study supports the contention that combination therapies, targeting all 24-hr behaviors, could be the next best step forward for lifestyle risk reduction for dementia. However, future intervention trials are required to confirm whether this approach would be effective. Declarations ACKNOWLEDGEMENTS: We thank the participants for giving up their time to participate in this research. We thank the UKB for making data and resources available. References Xie, L. et al. Sleep Drives Metabolite Clearance from the Adult Brain. Science 342, 373–377 (2013). Holth, J. K. et al. The sleep-wake cycle regulates brain interstitial fluid tau in mice and CSF tau in humans. Science 363, 880–884 (2019). Sabia, S. et al. Association of sleep duration in middle and old age with incidence of dementia. Nat. Commun. 12, 2289 (2021). Westwood, A. J. et al. Prolonged sleep duration as a marker of early neurodegeneration predicting incident dementia. Neurology 88, 1172–1179 (2017). Ohara, T. et al. Association Between Daily Sleep Duration and Risk of Dementia and Mortality in a Japanese Community. J. Am. Geriatr. Soc. 66, 1911–1918 (2018). Yoon, M. et al. Association of Physical Activity Level With Risk of Dementia in a Nationwide Cohort in Korea. JAMA Netw. Open 4, e2138526 (2021). Huang, S.-Y. et al. Sleep, physical activity, sedentary behavior, and risk of incident dementia: a prospective cohort study of 431,924 UK Biobank participants. Mol. Psychiatry 27, 4343–4354 (2022). Iso-Markku, P. et al. Physical Activity and Cognitive Decline Among Older Adults: A Systematic Review and Meta-Analysis. JAMA Netw. Open 7, e2354285 (2024). Livingston, G. et al. Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. The Lancet 396, 413–446 (2020). Falck, R. S. Is physical activity without good sleep enough to prevent cognitive decline? Lancet Healthy Longev. 4, e299–e300 (2023). Fry, A. et al. Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am. J. Epidemiol. 186, 1026–1034 (2017). Doherty, A. et al. Large Scale Population Assessment of Physical Activity Using Wrist Worn Accelerometers: The UK Biobank Study. PLOS ONE 12, e0169649 (2017). Migueles, J. H., Rowlands, A. V., Huber, F., Sabia, S. & Van Hees, V. T. GGIR: A Research Community–Driven Open Source R Package for Generating Physical Activity and Sleep Outcomes From Multi-Day Raw Accelerometer Data. J. Meas. Phys. Behav. 2, 188–196 (2019). Van Hees, V. T. et al. Separating Movement and Gravity Components in an Acceleration Signal and Implications for the Assessment of Human Daily Physical Activity. PLoS ONE 8, e61691 (2013). Van Hees, V. T. et al. A Novel, Open Access Method to Assess Sleep Duration Using a Wrist-Worn Accelerometer. PLOS ONE 10, e0142533 (2015). Van Hees, V. T. et al. Estimating sleep parameters using an accelerometer without sleep diary. Sci. Rep. 8, 12975 (2018). UK Biobank Algorithmically-Defined Outcomes: Version 2.0. Accessed November 1st, 2022. https://biobank.ndph.ox.ac.uk/showcase/showcase/docs/alg_outcome_main.pdf . Wilkinson, T. et al. Identifying dementia outcomes in UK Biobank: a validation study of primary care, hospital admissions and mortality data. Eur. J. Epidemiol. 34, 557–565 (2019). on behalf of Dementias Platform UK and UK Biobank et al. Identifying dementia outcomes in UK Biobank: a validation study of primary care, hospital admissions and mortality data. Eur. J. Epidemiol. 34, 557–565 (2019). Alfaro-Almagro, F. et al. Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank. NeuroImage 166, 400–424 (2018). Smith SM, Alfaro-Almagro F, Miller KL. UK Biobank Brain Imaging Documentation. Wellcome Centre for Integrative Neuroimaging (WIN-FMRIB), Oxford University on behalf of UK Biobank; 2020. https://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf . Van Buuren, S. & Groothuis-Oudshoorn, K. mice: Multivariate Imputation by Chained Equations in R . J. Stat. Softw. 45, (2011). Dumuid, D. et al. The compositional isotemporal substitution model: A method for estimating changes in a health outcome for reallocation of time between sleep, physical activity and sedentary behaviour. Stat. Methods Med. Res. 28, 846–857 (2019). Aitchison, J. The Statistical Analysis of Compositional Data. (2024). D’Agostino, R. B. et al. Relation of pooled logistic regression to time dependent cox regression analysis: The framingham heart study. Stat. Med. 9, 1501–1515 (1990). Young, J. G., Stensrud, M. J., Tchetgen Tchetgen, E. J. & Hernán, M. A. A causal framework for classical statistical estimands in failure-time settings with competing events. Stat. Med. 39, 1199–1236 (2020). Schoeler, T. et al. Participation bias in the UK Biobank distorts genetic associations and downstream analyses. Nat. Hum. Behav. 7, 1216–1227 (2023). Keage, H. A. D. et al. What sleep characteristics predict cognitive decline in the elderly? Sleep Med. 13, 886–892 (2012). Lo, J. C., Loh, K. K., Zheng, H., Sim, S. K. Y. & Chee, M. W. L. Sleep Duration and Age-Related Changes in Brain Structure and Cognitive Performance. Sleep 37, 821–821 (2014). Tworoger, S. S., Lee, S., Schernhammer, E. S. & Grodstein, F. The Association of Self-Reported Sleep Duration, Difficulty Sleeping, and Snoring With Cognitive Function in Older Women. Alzheimer Dis. Assoc. Disord. 20, 41–48 (2006). Tai, X. Y., Chen, C., Manohar, S. & Husain, M. Impact of sleep duration on executive function and brain structure. Commun. Biol. 5, 201 (2022). Mitchell, J. J. et al. Exploring the associations of daily movement behaviours and mid-life cognition: a compositional analysis of the 1970 British Cohort Study. J. Epidemiol. Community Health 77, 189–195 (2023). Whitaker, K. M. et al. Longitudinal Associations of Midlife Accelerometer Determined Sedentary Behavior and Physical Activity With Cognitive Function: The CARDIA Study. J. Am. Heart Assoc. 10, e018350 (2021). Hallgren, M., Dunstan, D. W. & Owen, N. Passive Versus Mentally Active Sedentary Behaviors and Depression. Exerc. Sport Sci. Rev. 48, 20 (2020). Wilson, R. S. Participation in Cognitively Stimulating Activities and Risk of Incident Alzheimer Disease. JAMA 287, 742 (2002). Wei, J. et al. Sleep, sedentary activity, physical activity, and cognitive function among older adults: The National Health and Nutrition Examination Survey, 2011–2014. J. Sci. Med. Sport 24, 189–194 (2021). Sun, Y. et al. Replacement of leisure-time sedentary behavior with various physical activities and the risk of dementia incidence and mortality: A prospective cohort study. J. Sport Health Sci. 12, 287–294 (2023). Hernán, M. A. Counterpoint: Epidemiology to Guide Decision-Making: Moving Away From Practice-Free Research. Am. J. Epidemiol. 182, 834–839 (2015). Raichlen, D. A. et al. Sedentary Behavior and Incident Dementia Among Older Adults. JAMA 330, 934 (2023). Jack, C. R. et al. Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade. Lancet Neurol. 9, 119–128 (2010). Weinstein, G. et al. Brain Imaging and Cognitive Predictors of Stroke and Alzheimer Disease in the Framingham Heart Study. Stroke 44, 2787–2794 (2013). Thompson, D., Batterham, A. M., Peacock, O. J., Western, M. J. & Booso, R. Feedback from physical activity monitors is not compatible with current recommendations: A recalibration study. Prev. Med. 91, 389–394 (2016). De Craemer, M. & Verbestel, V. Comparison of Outcomes Derived from the ActiGraph GT3X + and the Axivity AX3 Accelerometer to Objectively Measure 24-Hour Movement Behaviors in Adults: A Cross-Sectional Study. Int. J. Environ. Res. Public. Health 19, 271 (2021). Gangwisch, J. E. A Review of Evidence for the Link Between Sleep Duration and Hypertension. Am. J. Hypertens. 27, 1235–1242 (2014). Zhu, B. et al. Sleep disturbance induces neuroinflammation and impairment of learning and memory. Neurobiol. Dis. 48, 348–355 (2012). Kivimäki, M. et al. Physical inactivity, cardiometabolic disease, and risk of dementia: an individual-participant meta-analysis. BMJ l1495 (2019) doi: 10.1136/bmj.l1495 . Vecchio, L. M. et al. The Neuroprotective Effects of Exercise: Maintaining a Healthy Brain Throughout Aging. Brain Plast. 4, 17–52. Additional Declarations There is NO Competing Interest. Supplementary Files ExtendedDataSupplement.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4392320","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":302281406,"identity":"5ff82c4f-4107-4250-bc66-18faab2bc2bc","order_by":0,"name":"Matthew Pase","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9klEQVRIiWNgGAWjYJACZjjrA5jkAWI2vBoYm2FMxhkka2HmIUaLbvv5448LKg4zmLMff/jYNscmj7//7AGGD2WHcWoxO5PM2DzjzGEGy54cY+PcbWnFEjfyEhhnnMOj5QBQC2/bbQaDAzls0rnbDic23OAxYOZtw6Pl/GOgln9ALeefP/9tCdQy//wZA+a/+LTcANnSANRyI8GMmRGoZcOBHANmRrxaHhvO5jn2n8dyxhtjyd5taYkbb+QYHOw5l47HYYkPPvPUpMmZ86c//PBzm03ivPNnDB/8KLPGqQUGeAyQeQcIqgcBA8JKRsEoGAWjYKQCADuQW0OeQadCAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-4143-8485","institution":"Monash University","correspondingAuthor":true,"prefix":"","firstName":"Matthew","middleName":"","lastName":"Pase","suffix":""},{"id":302281409,"identity":"018b240e-41c2-4b38-a71c-cd1706732be2","order_by":1,"name":"Stephanie Yiallourou","email":"","orcid":"","institution":"Monash University","correspondingAuthor":false,"prefix":"","firstName":"Stephanie","middleName":"","lastName":"Yiallourou","suffix":""},{"id":302281410,"identity":"bd2d8851-d0a8-4318-b1f6-8cdf5a7b27c7","order_by":2,"name":"Lachlan Cribb","email":"","orcid":"","institution":"Monash University","correspondingAuthor":false,"prefix":"","firstName":"Lachlan","middleName":"","lastName":"Cribb","suffix":""},{"id":302281412,"identity":"a480d0d8-ba4d-4bd8-8d86-2031e7c9b9ac","order_by":3,"name":"Beaudan Campbell-Brown","email":"","orcid":"","institution":"Monash University","correspondingAuthor":false,"prefix":"","firstName":"Beaudan","middleName":"","lastName":"Campbell-Brown","suffix":""},{"id":302281415,"identity":"4a61f89c-30aa-4541-8699-83a37c841ecc","order_by":4,"name":"Christian Brakenridge","email":"","orcid":"","institution":"Swinburne University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Christian","middleName":"","lastName":"Brakenridge","suffix":""},{"id":302281416,"identity":"74ea2f66-4790-4e45-8fa1-0beecc3b8bc4","order_by":5,"name":"Andree-Ann Baril","email":"","orcid":"","institution":"Centre intégré universitaire de santé et de services sociaux du Nord-de-l'île-de-Montréal","correspondingAuthor":false,"prefix":"","firstName":"Andree-Ann","middleName":"","lastName":"Baril","suffix":""}],"badges":[],"createdAt":"2024-05-09 03:35:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4392320/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4392320/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":56541728,"identity":"00549475-fc84-40ec-a448-ca8aa80bf5d4","added_by":"auto","created_at":"2024-05-15 14:27:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":212066,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAll-cause dementia risk ratio (and 95% confidence interval) for time-use substitutions for normal (panels a, b, and c) and short sleepers (panels d, e, and f).\u003c/strong\u003eMVPA = moderate to vigorous physical activity. Normal sleepers are defined as persons with ≥6 hours of sleep (0.1% of the sample had \u0026gt;9 hours of sleep). Short sleepers are defined as persons with \u0026lt;6 hours of sleep.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/34e96d36edbbb672426a9eec.png"},{"id":56541729,"identity":"02126ed2-884e-4a7c-86f8-373d206f57fc","added_by":"auto","created_at":"2024-05-15 14:27:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":540374,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCumulative all-cause dementia incidence for “worst,” “typical,” and “ideal” time-use compositions\u003c/strong\u003e. Panel A shows the constituents of the “worst, typical, and ideal” compositions. The composition that was associated with the lowest estimated age 76 dementia risk, which we refer to as the “ideal” composition, was represented by 11.5 hours of inactivity, 7.3 hours of sleep, 3.3 hours of MVPA, and 2 hours of light activity (per day, respectively). The “worst” composition associated with the highest estimated dementia risk was represented by 15 hours of inactivity, 4.5 hours of sleep, 0.5 hours of MVPA, and 4 hours of light activity. The “typical” composition was represented by 12.1 hours of inactivity, 6.6 hours of sleep, 1.9 hours of MVPA, and 3.4 hours of light activity. Panel B shows the cumulative all-cause dementia incidence for each compositon. MVPA = moderate to vigorous physical activity.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/03ec4b415ced8a789edf97a1.png"},{"id":56543015,"identity":"575a6043-0aba-4f46-807f-34f029f5c8e0","added_by":"auto","created_at":"2024-05-15 14:35:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":212769,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTime-use substitutions and hippocampal volume for normal (panels a, b, and c) and short sleepers (panels d, e, and f).\u003c/strong\u003e MVPA = moderate to vigorous physical activity. Normal sleepers are defined as persons with ≥6 hours of sleep (0.1% of samples had \u0026gt; 9 hours of sleep). Short sleepers are defined as persons with \u0026lt;6 hours of sleep.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/5828da848936021d94c64021.png"},{"id":56541730,"identity":"ee4d3c09-3c9b-4b58-b4b2-e496a2980f40","added_by":"auto","created_at":"2024-05-15 14:27:25","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":162181,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eBrain MRI outcomes for “worst,” “typical,” and “ideal” time-use compositions\u003c/strong\u003e. The worst, typical, and ideal compositions are those described in Figure 2. The “ideal” composition was represented by 11.5 hours of inactivity, 7.3 hours of sleep, 3.3 hours of MVPA, and 2 hours of light activity (per day, respectively). The “worst” composition associated with the highest estimated dementia risk was represented by 15 hours of inactivity, 4.5 hours of sleep, 0.5 hours of MVPA, and 4 hours of light activity. The “typical” composition was represented by 12.1 hours of inactivity, 6.6 hours of sleep, 1.9 hours of MVPA, and 3.4 hours of light activity. For total brain volume, the average volume was lower among those with the worst composition than those with the typical composition (mean difference: -19, 95% CI: ‑30, ‑8.8), whereas the difference between the ideal and typical composition (4.6, 95% CI: -2.4, 11) was relatively smaller. A similar pattern was observed for gray matter volume (worst – typical: -14, 95% CI: -22, -8.1; ideal – typical: -1.55, 95% CI: -4.8, 2.0), hippocampal volume (worst – typical: -0.25, 95% CI: -0.43, -0.06; ideal – typical: 0.08, 95% CI: -0.03, 0.19), and log white matter hyperintensities (worst – typical: 0.3, 95% CI: 0.15, 0.5; ideal – typical: -0.02, 95% CI: -0.11, 0.082). For white matter volume, the difference between the ideal and typical composition (6.2, 95% CI: 0.92, 10) was modestly larger than that between the worst and typical composition (-4.4, 95% CI, -12, 2.7). MVPA = moderate to vigorous physical activity.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/e98ea3f1afae640b924dd8fa.png"},{"id":58140072,"identity":"8e1c29db-77b7-42bc-a45c-a0ac3b2357e6","added_by":"auto","created_at":"2024-06-11 17:08:52","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1602900,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/92c3200c-87bd-4888-8b11-82ac39e33ab3.pdf"},{"id":56541732,"identity":"206c8edd-86c2-4660-885f-f2ab40900f82","added_by":"auto","created_at":"2024-05-15 14:27:25","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1716261,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cbr\u003e\u003c/p\u003e","description":"","filename":"ExtendedDataSupplement.docx","url":"https://assets-eu.researchsquare.com/files/rs-4392320/v1/164caef2ce270322a24ecb04.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Optimizing Dementia Risk Reduction: Balancing Sleep and Physical Activity Trade-offs","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eSleep has a critical function in the clearance of amyloid beta and tau proteins, which can aggregate to form the two pathological hallmarks of Alzheimer\u0026rsquo;s disease\u003csup\u003e1,2\u003c/sup\u003e. Both short (\u0026le;\u0026thinsp;6hr)\u003csup\u003e3\u003c/sup\u003e and long (\u0026gt;\u0026thinsp;9hr)\u003csup\u003e4\u003c/sup\u003e sleep duration have been associated with an increased risk of dementia\u003csup\u003e5\u003c/sup\u003e. Daytime activity levels have also been linked to dementia risk. For example, low physical activity levels and high sedentary time are associated with higher dementia risk\u003csup\u003e6,7\u003c/sup\u003e. Intuitively, normalizing sleep duration or increasing physical activity could be attractive targets to reduce dementia risk. However, as each day is only 24 hours long, changing time spent in one activity must come at the expense or gain of another. Consequently, the potential benefits of several lifestyle interventions are not straightforward; their efficacy may depend on the substituted behavior. For example, although increasing physical activity has been touted as one strategy to lower dementia risk, is it advisable if it comes at the expense of sleep time?\u003c/p\u003e \u003cp\u003ePrevious studies on sleep time and physical activity's link to dementia tend to overlook the fact that these activities are a quantitative component of a 24-hr composition. Traditional regression methods isolate sleep or daytime activities with only partial adjustment for time spent in other behaviors, ignoring that time-use behaviors are co-dependent (if one behavior increases, another must decrease). Consequently, the questions posed by these studies tend to be ill-defined. It is difficult to interpret the estimated effect of increasing sleep time when the precise behaviors that the additional sleep time replaces are left unspecified. In this way, observational studies generally define physical activity and sleep time exposure variables in ways that do not approximate well-defined, precise interventions, precluding meaningful causal questions; this may mean that results from such studies do not translate well to meaningful guidelines (i.e., increase physical activity but at the expense of what?). Indeed, this may also help explain recent unexpected findings, whereby associations between physical activity and cognition were only minimal\u003csup\u003e8\u003c/sup\u003e. The issue is of profound importance given that 1) tackling modifiable risk factors offers substantial hope in reducing the growing burden of dementia\u003csup\u003e9\u003c/sup\u003e, and 2) dementia risk reduction guidelines lean heavily on observational data given the challenges of conducting clinical trials with dementia endpoints\u003csup\u003e9\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e Robust dementia prevention guidelines involving sleep duration or daytime physical activity levels (i.e., sedentary time, light physical activity, and moderate to vigorous physical activity; MVPA) require a complete understanding of how adding and subtracting different activity compositions affects dementia risk. Furthermore, there is a need to understand the effect of adding and subtracting sleep duration for varying levels of physical activity (or inactivity), given the role of sleep in glymphatic clearance and that the benefits of physical activity on cognitive health may be dependent on the amount of sleep\u003csup\u003e10\u003c/sup\u003e. Therefore, we applied compositional data analysis (CoDA) to a large community-based cohort with 24-hr behaviors estimated through 7 days of accelerometry monitoring and dementia follow-up. Unlike standard regression tools, CoDA respects the fact that time-use data are constrained to 24 hours and allows for estimating the effect of explicit time-use substitutions (e.g., adding 1 hour of sleep at the expense of 1 hour of MVPA). We aimed to determine the effect of substituting sleep duration for daytime physical activities (inactivity, light activity, and MVPA) on dementia risk and MRI volumetric outcomes. Moreover, we also aimed to determine the ideal composition of 24-hr behaviors (the absolute time spent in sleep, inactivity, light activity, and MVPA) associated with the lowest dementia risk.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e \u003cem\u003eStudy design and participants.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eThis prospective cohort study uses data from the UK Biobank (UKB). The UK Biobank recruited over 500,000 UK adults aged 40\u0026ndash;69 years during 2006\u0026ndash;2010. Detailed lifestyle, environmental, and genetic details were collected at the baseline session (described in detail on the study website: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.ukbiobank.ac.uk/\u003c/span\u003e\u003cspan address=\"https://www.ukbiobank.ac.uk/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Follow-up for the occurrence of several health outcomes is ongoing. Participants identified as living within a reasonable traveling distance of the 22 UKB assessment centers in the UK National Health Service patient registers were invited by mail to participate in the study. The participants who agreed to participate tended to be older, had higher socioeconomic status (SES), and were more likely to be female than non-participants, with a participation fraction of 5.5%\u003csup\u003e11\u003c/sup\u003e. Between February 2013 and December 2015, 236,519 UKB participants were invited to participate in a 7-day wrist-worn accelerometer study. A total of 106,053 agreed to take part \u003csup\u003e12\u003c/sup\u003e. All UKB participants provided informed consent to participate in the study. UK Biobank has approval from the North West Multi-Centre Research Ethics Committee as a Research Tissue Bank (RTB) approval. This approval means the present study operates under the RTB approval, satisfying the requirements of the Monash University Human Research Ethics Committee.\u003c/p\u003e \u003cp\u003eFor this study, participants with dementia or other severe neurological disease (Parkinson's disease, motor neuron disease, cerebral palsy, brain abscess, multiple sclerosis, or myasthenia gravis) at or before the time of the accelerometry study were excluded. The sample selection process from the total UKB sample is displayed in Extended Data eFigure 1.\u003c/p\u003e \u003cp\u003e \u003cem\u003eMeasurement of 24-hr behaviors by accelerometry.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eFor the objective assessment of 24-hr behaviors, participants underwent 7 days and nights of wrist-worn accelerometry (Axivity AX3). Further details of the accelerometry protocol have been published previously\u003csup\u003e12\u003c/sup\u003e. For the determination of time spent in sleep, inactivity, light activity, and MVPA, we used the open-source R package GGIR \u003csup\u003e13\u003c/sup\u003e. Raw acceleration data were processed using the Euclidean Distance Minus One procedure (ENMO)\u003csup\u003e14\u003c/sup\u003e. The default GGIR thresholds were used to distinguish inactivity from light activity and MVPA: \u0026lt; 30 mg for inactivity, \u0026ge;\u0026thinsp;30 and \u0026lt;\u0026thinsp;100mg for light, and \u0026ge;\u0026thinsp;100mg for MVPA. As a sleep diary was not available in the UKB, time in sleep was estimated using the algorithm of van Hees et al\u003csup\u003e15,16\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eLow-quality accelerometer data were removed using the established UKB criteria: lack of agreement between self-reported wear time and accelerometer wear time data (5%); insufficient wear time (\u0026lt;\u0026thinsp;72 hours; 5%); and poor calibration (\u0026lt;\u0026thinsp;1%). Additionally, data were removed for participants for whom GGIR was unable to determine a sleep window (5%) and for participants providing less than two valid contiguous wear days (\u0026lt;\u0026thinsp;1%).\u003c/p\u003e \u003cp\u003e \u003cem\u003eDementia case ascertainment.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eIncident all-cause dementia was ascertained based on hospital records, death records, and primary care\u003csup\u003e17\u003c/sup\u003e. As UKB record linkage with primary care providers is ongoing, primary care record linkage was only available for slightly less than half of the complete sample (approximately 45%). The algorithms used to ascertain all-cause dementia were designed to maximize positive predictive value, which has been shown to be high (\u0026gt;\u0026thinsp;80% for each record type)\u003csup\u003e18\u003c/sup\u003e. The list of Read V2 codes (primary care) and ICD-10 codes informing the identification of all-cause dementia have been published previously\u003csup\u003e19\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003e \u003cem\u003eBrain MRI outcomes.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eThe UKB imaging sub-study began in 2014 and aimed to acquire high-quality and consistent imaging data from 100,000 UKB participants across multiple modalities. All MRI data were acquired on 3T Siemens Skyra scanners. Imaging-derived phenotypes (IDPs) were derived using an automated pipeline described in depth previously\u003csup\u003e20\u003c/sup\u003e. Outcomes used for the current study included total brain volume, hippocampal volume, total grey matter volume, total white matter volume, and white matter hyperintensity volume. Volumetric IDPs of interest were normalized for head size using a head-size scaling factor\u003csup\u003e21\u003c/sup\u003e.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData analysis.\u003c/h2\u003e \u003cp\u003eThe daily time spent in each behavior was averaged across the accelerometry wear days using a linear mixed effects model with a random intercept for participant and fixed effect for day of the week. Estimated averages were standardized over the day of the week to ensure comparability of those with and without weekend assessments.\u003c/p\u003e \u003cp\u003eConfounder variables were selected using a causal directed acyclic graph (Extended Data eFigure 2). Primary model covariates included age, sex, education, ethnicity, \u003cem\u003eAPOE ε4\u003c/em\u003e genotype, household income, fruit and vegetable intake, alcohol intake, antidepressant, antipsychotic, or sedative medication use, retirement status, and shift work. Missing data in confounder variables were infrequent (most\u0026thinsp;\u0026lt;\u0026thinsp;2%) and were imputed by predictive mean matching using R package \u003cem\u003emice\u003c/em\u003e\u003csup\u003e22\u003c/sup\u003e. Imputation models included all confounder, exposure, and outcome variables (dementia status and age at dementia).\u003c/p\u003e \u003cp\u003e \u003cem\u003eIsotemporal substitution.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eThe average time spent in each behavior (sleep, inactivity, light, MVPA) was expressed as a proportion of total daily time, summing to 1. These variables cannot be entered into traditional statistical models (e.g., multivariable regression) due to the perfect collinearity among components\u003csup\u003e23\u003c/sup\u003e. Accordingly, they were first transformed from the simplex into the unconstrained (Real) space using an isometric log ratio (ILR) transformation\u003csup\u003e24\u003c/sup\u003e. Pooled over time logistic models were then fitted with ILR coordinates and covariates as predictors\u003csup\u003e25\u003c/sup\u003e. For these models, follow-up was divided into discrete chunks (6 months in length) such that each participant had as many rows of data as they had discrete follow-up intervals, ending either at the time of dementia, death from non-dementia causes (competing event), or end of study follow-up. Follow-up time was modeled with a restricted cubic spline with 5 knots. Age was used as the timescale. All continuous covariates were modeled with restricted cubic splines (knots at the 10th, 50th, and 90th percentiles) to allow for departures from linearity. Quadratic terms were included for ILR variables.\u003c/p\u003e \u003cp\u003eThe gformula estimator described by Young et al.\u003csup\u003e26\u003c/sup\u003e was used to estimate cause-specific cumulative incidences (hereafter, \u0026ldquo;risks\u0026rdquo;) of dementia by age 76, the median age at dementia occurrence. This approach involves fitting pooled logistic models to approximate hazards for the event of interest (i.e., dementia) and the competing event (i.e., death) and calculating the risk of the event of interest as a function of these hazards, appropriately accounting for competing events. Risks were estimated for two reference compositions: the typical \u0026ldquo;short sleeper\u0026rdquo; and the typical \u0026ldquo;normal sleeper.\u0026rdquo; Risk ratios were estimated for discrete substitutions to each reference composition (e.g., adding 1 hour of MVPA at the expense of 1 hour of sleep for the typical short sleeper). The typical short sleeper and normal sleeper compositions were the geometric mean composition among those with \u0026lt;\u0026thinsp;6 hours of sleep and among those with \u0026ge;\u0026thinsp;6 hours of sleep, respectively (0.1% of the sample had\u0026thinsp;\u0026gt;\u0026thinsp;9 hours of sleep). This categorization was based on previous literature demonstrating that short sleep duration of \u0026lt;\u0026thinsp;6 hours associates with increased dementia risk\u003csup\u003e3\u003c/sup\u003e. This analysis used nonparametric bootstrapping with nested single imputation with 500 samples to obtain percentile-based 95% confidence intervals. All other analyses, including MRI volumetric outcomes and all sensitivity analyses, used 250 bootstrap samples to minimize computational burden.\u003c/p\u003e \u003cp\u003e \u003cem\u003eIdeal composition.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eTo estimate the \u0026ldquo;ideal\u0026rdquo; composition, we created a grid of all compositions (in 15-minute steps) covering all possible compositions within the range of the sample data. We then fitted a multivariate Normal to all but one of the time-use variables in the sample data. Each member of the grid of compositions was then passed into the density function of this estimated Normal, and any compositions with low density (\u0026lt;\u0026thinsp;2.5th percentile) were excluded, thereby removing improbable compositions from consideration. Predicted dementia hazard was then estimated from the fitted pooled logistic model for each retained composition, with the composition returning the smallest estimated hazard deemed the \u0026ldquo;ideal\u0026rdquo; composition. The \u0026ldquo;worst\u0026rdquo; composition was computed analogously. The \u0026ldquo;typical\u0026rdquo; composition was chosen as the composition in the sample data that maximized the density function of the estimated Normal.\u003c/p\u003e \u003cp\u003e \u003cem\u003eMRI volumetric outcomes.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eA linear regression model was fitted for volumetric MRI outcomes including covariates and the ILR coordinates as predictors. Following UKB recommendations, these models were additionally adjusted for MRI assessment center, mean fMRI head motion, and head location in scanner\u003csup\u003e21\u003c/sup\u003e. The mean difference in volumetric MRI outcomes for a given substitution was estimated analogously to the dementia models. The \u0026ldquo;ideal\u0026rdquo; and \u0026ldquo;worst\u0026rdquo; compositions were then passed through this model to compare volumetric outcomes between the two compositions.\u003c/p\u003e \u003cp\u003e \u003cem\u003eSensitivity analysis.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eIn the first sensitivity analysis, we included additional adjustment for sleep fragmentation (wake after sleep onset; WASO), also assessed by accelerometry. In the second, we additionally adjusted for covariates that may be plausibly affected by time use, including history of cardiovascular disease, body mass index, blood pressure medication, systolic blood pressure, and sickness or disability (self-reported employment category). In the third, we removed participants with dementia events during the first three years of follow-up. A substantial reduction in the estimated risk ratios in this sensitivity analysis may indicate reverse causation. Finally, to assess the effect of selection bias due to the non-representativeness of the UKB (healthy cohort effect), we refitted the primary model, standardizing to the distributions of sex, retirement status, income, and smoking status of the representative UKB pseudo-population described by Schoeler et al. \u003csup\u003e27\u003c/sup\u003e. Product terms between each of these variables and the ILR coordinates were included in the latter model to allow for effect modification.\u003c/p\u003e \u003cp\u003eAll analysis code is available at the project GitHub repository (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/BeaudanBrown/coda-dementia\u003c/span\u003e\u003cspan address=\"https://github.com/BeaudanBrown/coda-dementia\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"RESULTS","content":"\u003cp\u003eCharacteristics of the sample at baseline are described in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The total sample size was 88,654. The mean age of the sample at the time of accelerometry was 63 years (Q1, Q3: 56, 68); 56% were women. There were 718 incident all-cause dementia cases over a median follow-up of 8.2 years (25th percentile 7.6; 75th percentile 8.7). The median time between accelerometry and MRI assessments was 3.4 years (25th percentile 1.9; 75th percentile 4.4). The associations between discrete time-use substitutions and dementia risk are displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Results are presented separately for the normal sleep duration (\u0026ge;\u0026thinsp;6 hr) and short sleep duration (\u0026lt;\u0026thinsp;6 hr) reference compositions.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eBaseline sample characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCharacteristic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSummary\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAge, years, median (Q1, Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e63 (56, 68)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWomen, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e49,877 (56)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI, kg/m2, median (Q1, Q3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e26.0 (23.6, 29.0)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHighest qualification, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh school non-completers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12,988 (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh school completers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,476 (6.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTrade qualification\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e10,162 (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGraduate degree\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e51,799 (59)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7,332 (8.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrefer not to answer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e373 (0.4)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmployment, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePaid employment\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e54,836 (62)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRetired\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e27,628 (\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSick or disabled\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,341 (1.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4,652 (5.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrefer not to answer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e159 (0.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage total household income (thousand pounds), n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026lt;18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11,642 (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e18\u0026ndash;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19,221 (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e31\u0026ndash;50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22,902 (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e52\u0026ndash;100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19,989 (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u0026gt;100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,800 (6.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDon\u0026rsquo;t know/Prefer not to answer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8,479 (9.6)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEthnicity, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWhite\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e81,571 (92)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,424 (3.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOther\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,371 (3.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrefer not to answer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e250 (0.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAntidepressant medication\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5,009 (5.7)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInsomnia medication\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e749 (0.8)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSmoking status, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNever\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e50,561 (57)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFormer\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31,780 (\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCurrent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6,076 (6.9)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cem\u003eAPOE ε4\u003c/em\u003e alleles, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e53,357 (72)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e19,150 (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,629 (2.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistory of cancer, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e11,503 (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistory of cardiovascular disease, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e35,789 (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHistory of diabetes, n (%)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3,689 (4.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage sleep duration, hours/day*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e6.5 (1.2)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage inactivity, hours/day*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e12.0 (1.1)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage light activity, hours/day*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e3.3 (1.3)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAverage moderate to vigorous activity, hours/day*\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.7 (1.5)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"2\"\u003e* Geometric mean (geometric standard deviation). Baseline is defined as the time of accelerometry.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cem\u003eTime-use substitutions for the normal sleeper reference composition.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eReplacing inactivity with sleep time (and vice versa) had little association with dementia risk (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). In contrast, replacing light activity with sleep time was associated with a lowering of dementia risk (1b); the RR associated with reallocating 1 hour/day of light activity to sleep was 0.67 (95% CI: 0.54, 0.80), while the RR associated with reallocating 1 hour/day of sleep to light activity was 1.43 (95% CI: 1.27, 1.67). Replacing MVPA with sleep time was associated with the largest increase in dementia risk (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec); the RR associated with reallocating 1 hour/day of MVPA to sleep was 2.03 (95% CI: 1.70, 2.47), while the RR associated with reallocating 1 hour/day of sleep to MVPA was 0.75 (95% CI: 0.59, 0.89). Thus, the effect of increasing or decreasing sleep duration displayed a different association with dementia risk for all three substituted behaviors.\u003c/p\u003e \u003cp\u003e \u003cem\u003eTime-use substitutions for short sleeper reference composition.\u003c/em\u003e \u003c/p\u003e \u003cp\u003eUnlike for the typical normal sleeper, replacing inactivity with sleep in the typical short sleeper was associated with a lowering of dementia risk (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed); the RR associated with reallocating 1 hour/day of inactivity to sleep was 0.89 (95% CI: 0.82, 0.96), while the RR associated with reallocating 1 hour/day of sleep to inactivity was 1.28 (95% CI: 1.18, 1.39). In the typical short sleeper, substitutions involving sleep and light activity, and sleep and MVPA appeared broadly consistent with the results observed in the typical normal sleeper, except for one crucial difference: whereas increasing MVPA at the expense of sleep was associated with a lower risk of dementia in the typical normal sleeper, this was attenuated in the typical short sleeper (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef). The RR associated with reallocating 1 hour/day of sleep to MVPA was 0.89 (95% CI: 0.71, 1.04). The RR associated with reallocating 1 hour/day of MVPA to sleep was 1.82 (95% CI: 1.47, 2.26). Overall, for the typical short sleeper, increasing sleep duration was associated with a lowering of dementia risk when at the expense of inactivity or light activity, but not MVPA.\u003c/p\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eIdeal composition\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e plots the cumulative incidence of dementia for the estimated \u0026ldquo;worst,\u0026rdquo; \u0026ldquo;typical,\u0026rdquo; and \u0026ldquo;ideal\u0026rdquo; composition. The worst composition diverged from the typical composition in the 6th decade of life (the earliest decade of life studied), with cumulative incidence increasing dramatically from age 70 onwards. In comparison, the cumulative incidence of dementia in the \u0026ldquo;typical\u0026rdquo; and \u0026ldquo;ideal\u0026rdquo; compositions was more similar and only began to diverge slightly in the 7th decade of life.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003eMRI endophenotypes\u003c/h2\u003e \u003cp\u003eThe association of the time-use substitutions with hippocampal volume are displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. In the typical normal sleeper, replacing MVPA with sleep was associated with smaller hippocampal volume (e.g., 1 hr, mean difference [MD]: -0.15 cm\u003csup\u003e3\u003c/sup\u003e, 95% CI: -0.24, -0.07), while replacing sleep with MVPA was associated with little change in hippocampal volume (e.g., 1-hour, MD: 0.03 cm\u003csup\u003e3\u003c/sup\u003e, 95% CI: -0.00, 0.07). Replacing sleep with light activity was associated with smaller hippocampal volume (e.g., 1-hour, MD: -0.07 cm\u003csup\u003e3\u003c/sup\u003e, 95% CI: -0.10, -0.03). The pattern was similar for the typical short sleeper (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and other volumetric outcomes, including total brain volume, grey matter volume, white matter volume, and log white matter hyperintensities (Extended Data eFigures 3 to 6, respectively).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eEstimated MRI endophenotypes for the \u0026ldquo;worst,\u0026rdquo; \u0026ldquo;typical,\u0026rdquo; and \u0026ldquo;ideal\u0026rdquo; compositions are displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. Brain volumes tended to be similar between the \u0026ldquo;ideal\u0026rdquo; and \u0026ldquo;typical\u0026rdquo; compositions and lower for the \u0026ldquo;worst\u0026rdquo; composition. Similarly, the log of white matter hyperintensities was similar between the \u0026ldquo;ideal\u0026rdquo; and \u0026ldquo;typical\u0026rdquo; compositions but higher for the \u0026lsquo;worst\u0026rsquo; composition.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity analyses\u003c/h2\u003e \u003cp\u003eResults were not meaningfully different in a first sensitivity analysis adjusting for sleep fragmentation (WASO; Extended Data eFigure 7), nor in a second sensitivity analysis adjusting for chronic disease and chronic disease risk factor variables (Extended Data eFigure 8). In the third sensitivity analysis, truncating the first three years of follow-up did not substantively alter the results (Extended Data eFigure 9). The final sensitivity analysis, which corrected for selective participation in the UK Biobank, also did not substantively alter the results (Extended Data eFigure 10).\u003c/p\u003e \u003c/div\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eThis study examined how sleep and physical activity trade-offs relate to brain health and dementia risk. Our findings reveal that dementia risk was influenced not only by the increasing behavior but also by the nature of the decreasing behavior and baseline sleep levels. In short sleepers, increasing sleep duration was associated with a reduction in dementia risk as long as the behavior being substituted was not MVPA, highlighting the importance of adequate sleep duration and MVPA to promote healthy cognitive aging. In normal sleepers, the association of increasing sleep duration was entirely dependent on the behavior being substituted; dementia risk decreased, increased, or remained stable when substituting out light activity, MVPA, or inactivity, respectively. Findings were similar when using brain volumes as outcomes in a subset with MRI. We also identified the most and least favorable combinations of 24-hour behaviors for dementia risk at age 76. Individuals with very short sleep, high inactivity, and low MVPA displayed the highest rates of dementia and evidence of accelerated brain aging on MRI. Overall, these results offer insights into potential behavior changes that could be targeted in interventions or guidelines to enhance brain health and prevent dementia.\u003c/p\u003e \u003cp\u003ePrevious studies have shown that short sleep duration is linked to poorer cognition\u003csup\u003e28\u0026ndash;30\u003c/sup\u003e, lower brain volumes\u003csup\u003e31\u003c/sup\u003e, and increased dementia risk\u003csup\u003e3\u003c/sup\u003e. Therefore, normalizing sleep duration has been suggested as a potential way to delay the onset or reduce the risk of dementia. Findings from this study provide evidence of the specific behaviors to target in those with short sleep, namely inactivity and light activity. For those with short sleep duration, increasing sleep by 1 hour instead of engaging in inactivity or light activity was associated with an 11% and 29% reduction in dementia risk, respectively. These findings remain consistent following adjustment for measures of poor sleep quality, common comorbidities, and disease risk factors. These results suggest that it is more advantageous to increase sleep at the expense of light activity than inactivity. Although this seems counterintuitive, others have shown that increasing light activity at the expense of sedentary time was associated with reduced cognitive performance\u003csup\u003e32,33\u003c/sup\u003e. Our results and others may be explained by differences in cognitive load during these daytime behaviors. Some inactivity time may be spent engaging in cognitively stimulating (e.g., working at a computer, writing, reading, playing a musical instrument) or social (e.g., social dinner) activities\u003csup\u003e34\u003c/sup\u003e. In contrast, light activity may involve less challenging cognitive tasks (e.g., housework, a light stroll). Accordingly, in some people, replacing time spent inactive may reduce engagement in cognitive and social activities tied to a lower risk of dementia\u003csup\u003e35\u003c/sup\u003e. Future studies delineating the type of cognitive activity co-occurring with inactivity and light activity would be beneficial to determine if this is the case.\u003c/p\u003e \u003cp\u003eLike short sleep, lower MVPA levels or self-report leisure time physical activity have been associated with higher dementia risk\u003csup\u003e6,7\u003c/sup\u003e. Previous isotemporal substitution studies have demonstrated that replacing 30 minutes/day of sleep with 30 minutes/day of MVPA was associated with better cognition in participants with self-report sleep duration\u0026thinsp;\u0026gt;\u0026thinsp;7 hours/night\u003csup\u003e36\u003c/sup\u003e. Further, others found that replacing sedentary behavior with equal time engaging in different physical activities corresponded with decreased dementia risk in the UKB\u003csup\u003e37\u003c/sup\u003e. However, this study relied on brief questionnaires to estimate physical activity and did not consider sleep substitutions. We extend these studies and show that replacing MVPA with sleep was associated with an increase in dementia risk and poorer brain health outcomes in both normal and short sleepers. MVPA may enrich slow-wave sleep\u003csup\u003e35\u003c/sup\u003e, meaning that extending sleep duration at the expense of MVPA may actually decrease sleep quality. Findings from our study provide further weight to the importance of MVPA for brain health and support the premise of implementing exercise interventions for dementia prevention. Importantly, our study sheds light on potential intervention strategies. We show that increasing MVPA by 1 hour at the expense of sleep in those with normal sleep duration reduced dementia risk by 25%. So, waking up 1 hour early to exercise may benefit the brain. \u003cem\u003eBut\u003c/em\u003e, this same effect does not apply to short sleepers. In short sleepers, the benefits of increasing MVPA by 1 hour in the place of sleep were dampened, such that there was only a modest decrease in associated dementia risk. Thus, the benefits of MVPA on dementia risk depend on whether one gets adequate sleep (in this case, \u0026ge;\u0026thinsp;6 hr). These results are consistent with findings from the English Longitudinal Study on Aging that found that the benefits of high levels of physical activity on cognitive decline were blunted in older adults who self-reported short (\u0026lt;\u0026thinsp;6 hours) versus normal (6\u0026ndash;8 hours) sleep duration, with short sleepers having a more rapid decline over a 10 year follow-up\u003csup\u003e38\u003c/sup\u003e. These findings suggest that the neuroprotective effects of MVPA may not fully overcome the detrimental effects of short sleep.\u003c/p\u003e \u003cp\u003eThis study also explored the \u0026ldquo;worst, typical, and ideal\u0026rdquo; 24-hr activity compositions related to dementia risk. Those with high amounts of inactivity combined with low amounts of sleep and MVPA had the highest dementia risk, whereas those displaying the opposite patterns displayed the lowest risk. These compositions align well with current knowledge of these behaviors and dementia risk\u003csup\u003e6,7,39\u003c/sup\u003e. Our data also suggest that transitioning people from the \u0026ldquo;worst\u0026rdquo; to the \u0026ldquo;typical\u0026rdquo; composition may have a far greater impact on dementia prevention as compared to transitioning people from the \u0026ldquo;typical\u0026rdquo; to the \u0026ldquo;ideal\u0026rdquo; composition. However, it is important to note that the \"ideal\" composition in this study represents \u003cem\u003eone\u003c/em\u003e ideal composition; there are likely many differing compositions resulting in similarly low estimated dementia risk. Thus, the \"ideal\" composition should not be considered the only one to pursue for improving brain health.\u003c/p\u003e \u003cp\u003eSince the median follow-up duration for dementia was 8.2 years in our sample, there may be an element of reverse causation, given that several forms of dementia have a long preclinical phase\u003csup\u003e40\u003c/sup\u003e. Importantly, we didn\u0026rsquo;t find strong evidence to suggest this. Truncating the first three years of follow-up (during which reverse causation would be expected to be strongest) did not substantively alter the estimated risk ratios. Nevertheless, we felt it necessary to replicate findings with MRI-based subclinical endophenotypes. Hippocampal atrophy is an imaging characteristic of preclinical Alzheimer\u0026rsquo;s disease, and low hippocampal volume cross-sectionally is associated with poorer memory, worse clinical function, and higher dementia risk over the next decade\u003csup\u003e41\u003c/sup\u003e. We identified that those with the \u0026ldquo;ideal\u0026rdquo; composition had larger total brain, grey matter, and hippocampal volumes and less white matter injury than those with the \u0026ldquo;worst\u0026rdquo; composition. Since white matter injury is characteristic of small vessel disease and lower hippocampal volume is classical of Alzheimer\u0026rsquo;s disease, these data suggest that, combined, an \u0026ldquo;ideal\u0026rdquo; composition may protect the brain from various insults that can lead to dementia. Additional biomarker outcomes will be required to test this hypothesis further.\u003c/p\u003e \u003cp\u003e In reference to the \u0026ldquo;ideal\u0026rdquo; composition, the amount of sleep required agrees with national guidelines for sleep recommendations, which is in the range of 7\u0026ndash;9 hr. On the other hand, 3.3 hours/day of MVPA appears relatively high, given that the current WHO recommendations are 150\u0026ndash;300 minutes/week to achieve health benefits. It is important to note the differences between device-worn and self-report measures of physical activity; physical activity guidelines have largely been based on self-report. It has been previously estimated that 150 min/week of self-reported MVPA equals\u0026thinsp;~\u0026thinsp;1000 min/week of device-assessed MVPA\u003csup\u003e42\u003c/sup\u003e. Indeed, self-reporting could be biased only to include MVPA that is intentional or in conscious awareness. In contrast, device-assessed physical activity will also capture incidental MVPA that may go unnoticed (at work, for example). As wrist-worn devices with inbuilt activity trackers become more popular amongst the general public and are more easily implemented in large studies, our analysis captures measures of MVPA that may better reflect emerging societal and research trends. However, we acknowledge that our MVPA estimates are likely much higher than those measured by self-report or hip-worn accelerometry\u003csup\u003e43\u003c/sup\u003e. Nonetheless, the unique substitutions in this study overcome these nuances and are more applicable in providing a framework for personalized interventions, irrespective of ideal behavior targets. That is, an increase in MVPA by 1 hour is more achievable than simply aiming for the \u0026ldquo;ideal.\u0026rdquo; Notably, the 24-hour compositions identified in this study could be important for screening vulnerable populations that could be targeted for dementia prevention programs.\u003c/p\u003e \u003cp\u003eAlthough causality cannot be assumed, several mechanisms may explain how changes in 24-hour activity compositions contribute to dementia risk. Glymphatic clearance of Alzheimer\u0026rsquo;s disease proteins (amyloid-β and tau) is maximal during sleep and thought to be coupled to slow wave sleep\u003csup\u003e1\u003c/sup\u003e. Shorter sleep duration may impair the clearance of these waste products that aggregate to form amyloid plaques and neurofibrillary tangles\u003csup\u003e1,2\u003c/sup\u003e. Short sleep can also increase blood pressure\u003csup\u003e44\u003c/sup\u003e and inflammatory processes\u003csup\u003e45\u003c/sup\u003e, potentially explaining links between short sleep and vascular brain injury and brain atrophy\u003csup\u003e46\u003c/sup\u003e. MVPA also has neuroprotective effects. For example, physical activity is thought to induce favorable alterations in cerebral blood flow, augment cognitive reserve via neuroplasticity processes, facilitate glymphatic clearance of amyloid-β, and reduce other dementia risk factors such as cardiovascular disease and stress\u003csup\u003e47\u003c/sup\u003e.\u003c/p\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThe strengths of this study include the large sample size and the use of objective measures to assess 24-hour behaviors. However, we acknowledge that wrist-worn accelerometry may overestimate daytime behaviors compared to hip-worn devices. However, we argue that wrist-worn accelerometry estimates are more important to study given growing societal trends towards wrist-worn fitness trackers. We also recognize that our categorization of time use into, for instance, inactivity, does not distinguish sitting and standing, nor varied behaviors that may occur as part of those categories (e.g., reading, socializing, or watching television), some of which may plausibly be related to dementia risk in different ways. Our study only presents an association, estimated from observational data, between 24-hour behaviors and dementia risk. Therefore, we cannot rule out unmeasured confounding or reverse causation. Furthermore, as we did not have longitudinal data on 24-hour behaviors, the time when substitutions occurred, and the duration that those substitutions were maintained was unspecified\u003csup\u003e38\u003c/sup\u003e. Although this paper presents an important first step, future randomized trials or observational studies could be designed to clarify the effect of substitutions with a defined time course. Finally, those who participated in the UKB cohort tend to be healthier and of higher socioeconomic status than the general UK population, leading to potential issues with external validity.\u003csup\u003e11\u003c/sup\u003e Nevertheless, we did not find meaningfully different results in our sensitivity analysis, which at least partly corrected for selective participation in the UKB.\u003c/p\u003e \u003c/div\u003e"},{"header":"Implications and conclusions","content":"\u003cp\u003eSleep duration has yet to receive mainstream recognition as a modifiable dementia risk factor, as evidenced by its omission from the 12 headline risk factors identified by the Lancet Commission\u0026rsquo;s dementia prevention, intervention, and care guidelines\u003csup\u003e9\u003c/sup\u003e. Moreover, findings for the true magnitude of benefits of physical activity on the aging brain remain contentious\u003csup\u003e8\u003c/sup\u003e. Accordingly, our data suggests that an ideal strategy for risk reduction should encompass a tailored approach that assesses all 24-hr behaviors and targets those behaviors that are amenable to have optimal effect. We found that, depending on whether one has short or normal sleep duration, substituting inactivity or light activity for sleep was associated with a favorable effect on brain aging and dementia risk. The same was true for substituting MVPA for sleep in those with normal but not short sleep duration. These data indicate that there could be some flexibility in setting behavior change goals to enable an achievable behavior change for the individual. For example, increasing MVPA would be an obvious choice to improve brain health, but getting enough MVPA could be challenging in some populations. Our data show that even an increase in sleep by 30 minutes/day (in place of light activity) may benefit these groups where reaching physical activity targets may be challenging.\u003c/p\u003e \u003cp\u003eThis study supports the contention that combination therapies, targeting all 24-hr behaviors, could be the next best step forward for lifestyle risk reduction for dementia. However, future intervention trials are required to confirm whether this approach would be effective.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eACKNOWLEDGEMENTS:\u003c/h2\u003e \u003cp\u003eWe thank the participants for giving up their time to participate in this research. We thank the UKB for making data and resources available.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eXie, L. \u003cem\u003eet al.\u003c/em\u003e Sleep Drives Metabolite Clearance from the Adult Brain. Science 342, 373\u0026ndash;377 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHolth, J. K. \u003cem\u003eet al.\u003c/em\u003e The sleep-wake cycle regulates brain interstitial fluid tau in mice and CSF tau in humans. Science 363, 880\u0026ndash;884 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSabia, S. \u003cem\u003eet al.\u003c/em\u003e Association of sleep duration in middle and old age with incidence of dementia. Nat. Commun. 12, 2289 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWestwood, A. J. \u003cem\u003eet al.\u003c/em\u003e Prolonged sleep duration as a marker of early neurodegeneration predicting incident dementia. Neurology 88, 1172\u0026ndash;1179 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOhara, T. \u003cem\u003eet al.\u003c/em\u003e Association Between Daily Sleep Duration and Risk of Dementia and Mortality in a Japanese Community. J. Am. Geriatr. Soc. 66, 1911\u0026ndash;1918 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoon, M. \u003cem\u003eet al.\u003c/em\u003e Association of Physical Activity Level With Risk of Dementia in a Nationwide Cohort in Korea. JAMA Netw. Open 4, e2138526 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, S.-Y. \u003cem\u003eet al.\u003c/em\u003e Sleep, physical activity, sedentary behavior, and risk of incident dementia: a prospective cohort study of 431,924 UK Biobank participants. Mol. Psychiatry 27, 4343\u0026ndash;4354 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIso-Markku, P. \u003cem\u003eet al.\u003c/em\u003e Physical Activity and Cognitive Decline Among Older Adults: A Systematic Review and Meta-Analysis. JAMA Netw. Open 7, e2354285 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLivingston, G. \u003cem\u003eet al.\u003c/em\u003e Dementia prevention, intervention, and care: 2020 report of the Lancet Commission. The Lancet 396, 413\u0026ndash;446 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFalck, R. S. Is physical activity without good sleep enough to prevent cognitive decline? Lancet Healthy Longev. 4, e299\u0026ndash;e300 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFry, A. \u003cem\u003eet al.\u003c/em\u003e Comparison of Sociodemographic and Health-Related Characteristics of UK Biobank Participants With Those of the General Population. Am. J. Epidemiol. 186, 1026\u0026ndash;1034 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDoherty, A. \u003cem\u003eet al.\u003c/em\u003e Large Scale Population Assessment of Physical Activity Using Wrist Worn Accelerometers: The UK Biobank Study. PLOS ONE 12, e0169649 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMigueles, J. H., Rowlands, A. V., Huber, F., Sabia, S. \u0026amp; Van Hees, V. T. GGIR: A Research Community\u0026ndash;Driven Open Source R Package for Generating Physical Activity and Sleep Outcomes From Multi-Day Raw Accelerometer Data. \u003cem\u003eJ. Meas. Phys. Behav.\u003c/em\u003e 2, 188\u0026ndash;196 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Hees, V. T. \u003cem\u003eet al.\u003c/em\u003e Separating Movement and Gravity Components in an Acceleration Signal and Implications for the Assessment of Human Daily Physical Activity. PLoS ONE 8, e61691 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Hees, V. T. \u003cem\u003eet al.\u003c/em\u003e A Novel, Open Access Method to Assess Sleep Duration Using a Wrist-Worn Accelerometer. PLOS ONE 10, e0142533 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Hees, V. T. \u003cem\u003eet al.\u003c/em\u003e Estimating sleep parameters using an accelerometer without sleep diary. Sci. Rep. 8, 12975 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUK Biobank Algorithmically-Defined Outcomes: Version 2.0. Accessed November 1st, 2022. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://biobank.ndph.ox.ac.uk/showcase/showcase/docs/alg_outcome_main.pdf\u003c/span\u003e\u003cspan address=\"https://biobank.ndph.ox.ac.uk/showcase/showcase/docs/alg_outcome_main.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilkinson, T. \u003cem\u003eet al.\u003c/em\u003e Identifying dementia outcomes in UK Biobank: a validation study of primary care, hospital admissions and mortality data. Eur. J. Epidemiol. 34, 557\u0026ndash;565 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eon behalf of Dementias Platform UK and UK Biobank \u003cem\u003eet al.\u003c/em\u003e Identifying dementia outcomes in UK Biobank: a validation study of primary care, hospital admissions and mortality data. Eur. J. Epidemiol. 34, 557\u0026ndash;565 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlfaro-Almagro, F. \u003cem\u003eet al.\u003c/em\u003e Image processing and Quality Control for the first 10,000 brain imaging datasets from UK Biobank. NeuroImage 166, 400\u0026ndash;424 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmith SM, Alfaro-Almagro F, Miller KL. UK Biobank Brain Imaging Documentation. Wellcome Centre for Integrative Neuroimaging (WIN-FMRIB), Oxford University on behalf of UK Biobank; 2020. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf\u003c/span\u003e\u003cspan address=\"https://biobank.ctsu.ox.ac.uk/crystal/crystal/docs/brain_mri.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Buuren, S. \u0026amp; Groothuis-Oudshoorn, K. mice: Multivariate Imputation by Chained Equations in \u003cem\u003eR\u003c/em\u003e. J. Stat. Softw. 45, (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDumuid, D. \u003cem\u003eet al.\u003c/em\u003e The compositional isotemporal substitution model: A method for estimating changes in a health outcome for reallocation of time between sleep, physical activity and sedentary behaviour. Stat. Methods Med. Res. 28, 846\u0026ndash;857 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAitchison, J. The Statistical Analysis of Compositional Data. (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD\u0026rsquo;Agostino, R. B. \u003cem\u003eet al.\u003c/em\u003e Relation of pooled logistic regression to time dependent cox regression analysis: The framingham heart study. Stat. Med. 9, 1501\u0026ndash;1515 (1990).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoung, J. G., Stensrud, M. J., Tchetgen Tchetgen, E. J. \u0026amp; Hern\u0026aacute;n, M. A. A causal framework for classical statistical estimands in failure-time settings with competing events. Stat. Med. 39, 1199\u0026ndash;1236 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchoeler, T. \u003cem\u003eet al.\u003c/em\u003e Participation bias in the UK Biobank distorts genetic associations and downstream analyses. Nat. Hum. Behav. 7, 1216\u0026ndash;1227 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKeage, H. A. D. \u003cem\u003eet al.\u003c/em\u003e What sleep characteristics predict cognitive decline in the elderly? Sleep Med. 13, 886\u0026ndash;892 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLo, J. C., Loh, K. K., Zheng, H., Sim, S. K. Y. \u0026amp; Chee, M. W. L. Sleep Duration and Age-Related Changes in Brain Structure and Cognitive Performance. Sleep 37, 821\u0026ndash;821 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTworoger, S. S., Lee, S., Schernhammer, E. S. \u0026amp; Grodstein, F. The Association of Self-Reported Sleep Duration, Difficulty Sleeping, and Snoring With Cognitive Function in Older Women. Alzheimer Dis. Assoc. Disord. 20, 41\u0026ndash;48 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTai, X. Y., Chen, C., Manohar, S. \u0026amp; Husain, M. Impact of sleep duration on executive function and brain structure. Commun. Biol. 5, 201 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMitchell, J. J. \u003cem\u003eet al.\u003c/em\u003e Exploring the associations of daily movement behaviours and mid-life cognition: a compositional analysis of the 1970 British Cohort Study. J. Epidemiol. Community Health 77, 189\u0026ndash;195 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWhitaker, K. M. \u003cem\u003eet al.\u003c/em\u003e Longitudinal Associations of Midlife Accelerometer Determined Sedentary Behavior and Physical Activity With Cognitive Function: The CARDIA Study. J. Am. Heart Assoc. 10, e018350 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHallgren, M., Dunstan, D. W. \u0026amp; Owen, N. Passive Versus Mentally Active Sedentary Behaviors and Depression. Exerc. Sport Sci. Rev. 48, 20 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWilson, R. S. Participation in Cognitively Stimulating Activities and Risk of Incident Alzheimer Disease. JAMA 287, 742 (2002).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei, J. \u003cem\u003eet al.\u003c/em\u003e Sleep, sedentary activity, physical activity, and cognitive function among older adults: The National Health and Nutrition Examination Survey, 2011\u0026ndash;2014. J. Sci. Med. Sport 24, 189\u0026ndash;194 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun, Y. \u003cem\u003eet al.\u003c/em\u003e Replacement of leisure-time sedentary behavior with various physical activities and the risk of dementia incidence and mortality: A prospective cohort study. J. Sport Health Sci. 12, 287\u0026ndash;294 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHern\u0026aacute;n, M. A. Counterpoint: Epidemiology to Guide Decision-Making: Moving Away From Practice-Free Research. Am. J. Epidemiol. 182, 834\u0026ndash;839 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRaichlen, D. A. \u003cem\u003eet al.\u003c/em\u003e Sedentary Behavior and Incident Dementia Among Older Adults. JAMA 330, 934 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJack, C. R. \u003cem\u003eet al.\u003c/em\u003e Hypothetical model of dynamic biomarkers of the Alzheimer\u0026rsquo;s pathological cascade. Lancet Neurol. 9, 119\u0026ndash;128 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeinstein, G. \u003cem\u003eet al.\u003c/em\u003e Brain Imaging and Cognitive Predictors of Stroke and Alzheimer Disease in the Framingham Heart Study. Stroke 44, 2787\u0026ndash;2794 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eThompson, D., Batterham, A. M., Peacock, O. J., Western, M. J. \u0026amp; Booso, R. Feedback from physical activity monitors is not compatible with current recommendations: A recalibration study. Prev. Med. 91, 389\u0026ndash;394 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDe Craemer, M. \u0026amp; Verbestel, V. Comparison of Outcomes Derived from the ActiGraph GT3X\u0026thinsp;+\u0026thinsp;and the Axivity AX3 Accelerometer to Objectively Measure 24-Hour Movement Behaviors in Adults: A Cross-Sectional Study. Int. J. Environ. Res. Public. Health 19, 271 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGangwisch, J. E. A Review of Evidence for the Link Between Sleep Duration and Hypertension. Am. J. Hypertens. 27, 1235\u0026ndash;1242 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu, B. \u003cem\u003eet al.\u003c/em\u003e Sleep disturbance induces neuroinflammation and impairment of learning and memory. Neurobiol. Dis. 48, 348\u0026ndash;355 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKivim\u0026auml;ki, M. \u003cem\u003eet al.\u003c/em\u003e Physical inactivity, cardiometabolic disease, and risk of dementia: an individual-participant meta-analysis. \u003cem\u003eBMJ\u003c/em\u003e l1495 (2019) doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1136/bmj.l1495\u003c/span\u003e\u003cspan address=\"10.1136/bmj.l1495\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVecchio, L. M. \u003cem\u003eet al.\u003c/em\u003e The Neuroprotective Effects of Exercise: Maintaining a Healthy Brain Throughout Aging. Brain Plast. 4, 17\u0026ndash;52.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-4392320/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4392320/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eEngaging in regular physical activity and obtaining recommended amounts of sleep are touted as strategies to promote healthy brain aging. However, as each day is only 24 hours long, changing time spent in one activity must come at the expense or gain of another, making it necessary to understand how the whole 24-hour activity composition impacts dementia risk. We applied compositional data analysis to investigate the effect of substituting sleep duration for different levels of physical activity (i.e., inactivity, light activity, and moderate to vigorous physical activity; MVPA) on dementia risk relative to two reference compositions; a \u0026ldquo;typical\u0026rdquo; short sleeper (\u0026lt;\u0026thinsp;6hrs) and normal sleeper (\u0026ge;\u0026thinsp;6hrs). The study sample comprised participants from the community-based UK Biobank with 24-hour behaviors estimated using 7 days of accelerometry. The mean age of the sample was 63 years (Q1, Q3: 56, 68); 56% were women. Of the 88,654 participants, there were 718 incident all-cause dementia cases over a median follow-up of 8.2 years. For short sleepers, increasing sleep duration at the expense of inactivity or light activity was associated with a lowering of dementia risk, but not when at the expense of MVPA. For persons with normal sleep duration, the effect of increasing or decreasing sleep duration on dementia risk differed for all three substituted behaviors (i.e., inactivity, light, or MVPA). Most notably, dementia risk was higher when increasing sleep at the expense of MVPA and lower when increasing MVPA at the expense of sleep. The interpretation of the results was broadly consistent when using MRI-based outcomes (e.g., hippocampal volume) in a subset with brain imaging (n\u0026thinsp;=\u0026thinsp;15,263). Our findings underscore the complexity of optimizing dementia risk reduction strategies, emphasizing the need for personalized approaches that balance trade-offs between sleep duration and differing physical activity levels based on individual circumstances, such as habitual sleep duration.\u003c/p\u003e","manuscriptTitle":"Optimizing Dementia Risk Reduction: Balancing Sleep and Physical Activity Trade-offs","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-05-15 14:27:21","doi":"10.21203/rs.3.rs-4392320/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5bb84192-2b0b-4dfa-9f22-8bb911103dcd","owner":[],"postedDate":"May 15th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":31893031,"name":"Health sciences/Neurology/Neurological disorders/Dementia"},{"id":31893032,"name":"Health sciences/Risk factors"}],"tags":[],"updatedAt":"2024-06-11T17:00:42+00:00","versionOfRecord":[],"versionCreatedAt":"2024-05-15 14:27:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4392320","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4392320","identity":"rs-4392320","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}