Abstract
INTRODUCTION
Healthy sleep may support cognitive health, but the role of weekend catch‐up sleep is unclear.
METHODS
Among 83,776 dementia‐free UK Biobank participants aged ≥50 years, sleep duration was estimated from 7‐day accelerometer data (2013–2015). Weekend catch‐up sleep was defined as the weekend–weekday sleep duration difference. Cox regression model was used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for all‐cause dementia associated with weekend catch‐up sleep.
RESULTS
During follow‐up (median: 8.0 years), 713 participants developed all‐cause dementia. Compared with ≤0.5 h, HRs for all‐cause dementia across catch‐up sleep categories were 0.91 (0.74–1.11) for > 0.5–1 h, 0.64 (0.49–0.86) for > 1–1.5 h, 0.84 (0.60–1.16) for > 1.5–2 h, and 0.83 (0.60–1.16) for > 2 h. The association was stronger among participants with weekday sleep < 8 h (HR: 0.49, 0.29–0.81 for > 1–1.5 h) but non‐significant in those with ≥8 h (p‐interaction = 0.039).
DISCUSSION
Moderate weekend catch‐up sleep was linked to lower dementia risk, especially among individuals with less weekday sleep.
Highlights
In this prospective cohort study, moderate accelerometer‐measured weekend catch‐up sleep was linked to a reduced risk of dementia, with 1–1.5 h associated with the lowest risk.
The relation was more pronounced in participants with weekday sleep duration < 8 h/day, but non‐significant among those with a longer weekday sleep.
Similar associations were observed across the subgroups stratified by sociodemographic and lifestyle factors.
Keywords: cohort, dementia, weekend catch‐up sleep
1. BACKGROUND
Dementia is a leading cause of disability among older adults and poses increasing healthcare burdens worldwide, with over 50 million cases in 2019. 1 , 2 , 3 Given that there is currently limited cure to reverse its progression, identifying modifiable lifestyle factors is crucial for effective prevention of dementia. 1 , 4 Healthy sleep duration emerged as a key component of healthy lifestyle in maintaining brain health, and both insufficient and prolonged sleep durations have been linked to cognitive decline and a higher risk of dementia. 5
In individuals with insufficient sleep during the weekdays, “weekend catch‐up sleep”, defined as the practice of extending sleep on weekends, has become a way to compensate for sleep deficits accumulated from weekday demands. 6 While prior research has linked weekend catch‐up sleep to various metabolic health status, 7 , 8 , 9 its long‐term effects on brain health are largely unexplored. A cross‐sectional study of 215 older adults reported an association between weekend catch‐up sleep and lower probability of cognitive decline. 10 However, excessive weekend catch‐up sleep may disrupt circadian rhythms in sleep–wake patterns (known as social jetlag) and, thus, not confer additional benefits for brain health. As higher day‐to‐day variability in sleep schedules has previously been associated with vascular risk factors 11 , 12 and increased dementia risk, 13 the relation of weekend catch‐up sleep with dementia remains controversial. Importantly, most existing studies employed cross‐sectional designs or relied on self‐reported sleep measures prone to reporting bias, which limit the ability to draw causal inferences about the potential protective effects of compensating for sleep debt versus the harms associated with circadian disruption.
To investigate the roles of these two potentially competing mechanisms of weekend catch‐up sleep, the current study utilizes objectively measured sleep durations to examine the association between weekend catch‐up and incident dementia. We hypothesized that moderate weekend catch‐up sleep was associated with lower dementia risk, particularly among individuals with less sleep on weekdays.
2. METHODS
2.1. Study population
This study was based on the UK Biobank, a population‐based cohort study commenced in 2006–2010. The UK Biobank recruited over 500,000 residents across the United Kingdom aged 37 years and older at 22 assessment centers and collected extensive genetic and phenotypic data. 14 , 15 Between 2013 and 2015, a random subset was invited to participate in a three‐axis logging accelerometer sub‐study for 7 days. 16 The protocol of UK Biobank was approved by the North West Multi‐Centre Research Ethics Committee (REC reference: 11/NW/0382), and all participants provided informed consent before data collection.
Among the 103,684 participants who wore the accelerometer, we included participants with ≥3 valid wearing days (> 16 h/day and at least 1 weekend day), 17 and free of all‐cause dementia before and in 2 years after participating in the sub‐study to reduce the risk of reverse causality. 18 Participants younger than 50 years at baseline were excluded, as dementia incidence is very low in this age group during follow‐up (median: 8.0 years), and the etiology of young‐onset dementia differs substantially from late‐onset dementia. This approach is consistent with previous UK Biobank studies examining sleep and dementia risk. 19 , 20 We further excluded participants who had extreme weekday sleep duration, weekend sleep duration, or weekday‐weekend difference (< 0.5th or > 99.5th percentiles). The final sample size was 83,776 (Figure 1).
FIGURE 1.
Participants inclusion flow chart.
2.2. Accelerometer‐derived sleep duration
The participants were invited to wear the Axivity AX3 triaxial accelerometer on their dominant wrist for 7 days 21 and mail the device back to the coordinating center after the 7‐day period. 16 Daily sleep duration was extracted from raw accelerometer data using a previously published machine‐learning algorithm for the UK Biobank. 22 In a validation study (n = 152), the algorithm showed high classification performance (accuracy 88%) in identifying movement behaviors, and the precision for identifying sleeping behavior was > 95%. 22
We calculated the average sleep duration for weekday (Monday to Friday) and weekend (Saturday and Sunday), respectively, and calculated the weekend‐weekday difference as weekend catch‐up sleep. 23 According to the study population distribution, the weekend catch‐up sleep duration was categorized as both multi‐class (≤0.5, > 0.5–1, > 1–1.5, > 1.5–2, or > 2 h, as the primary exposure) and binary variables (≤0.5 or > 0.5 h, as the secondary exposure). We set the threshold at 0.5 h, as it approximates the population median (0.41 h) and conservatively identifies the catch‐up sleep population 23 and ensured that it reflects compensatory sleep rather than random daily variation. We also included the weekend‐weekday difference as a continuous variable to assess the linear and non‐linear associations.
2.3. Dementia ascertainment
The outcome of interest in this study was incident all‐cause dementia. We used a linkage‐based algorithm to define incident dementia, combining information from electronic health records (EHRs) in primary care, hospital admissions, and death registry, as was described previously. 24 We used the date of the first occurrence in any of the above‐mentioned sources as the diagnosis date of dementia, which showed a positive predictive value of 82.5%. 24
RESEARCH IN CONTEXT
Systematic review: We searched PubMed for research articles published between database inception and September 7 2025, with no language restrictions, using the search term ([Weekend catch‐up] OR [Catch‐up sleep]) AND ([Dementia] OR [Cognitive function]). Weekend catch‐up sleep was defined as extending sleep duration over the weekend to compensate weekday insufficiency. While a healthy sleep pattern may benefit cognition, previous studies on weekend catch‐up sleep and brain health were limited by cross‐sectional design, subjective measurement, or both. Additionally, the optimal range of catch‐up sleep and the potential modification role of weekday sleep remain largely unknown.
Interpretation: Moderate weekend catch‐up sleep was associated with lower risk of dementia, with 1–1.5 h of catch‐up sleep being related to a 36% reduction in dementia incidence. The association was stronger among individuals with weekday sleep duration < 8 h/day.
Future directions: Our findings contributed to the evidence that weekend catch‐up sleep, as well as adequate weekday sleep, may help protect cognitive health. Future studies are warranted to validate these associations and establish optimal sleep compensation strategies for dementia prevention.
2.4. Covariates
We included multiple covariates for confounding adjustments based on the existing literature. 17 Demographic factors and socioeconomic status included age at accelerometer assessment, sex, race, and ethnicity (white or non‐white), education (below or above college), and Townsend deprivation index (as tertiles, reflecting social deprivation level). Lifestyle factors included body mass index (BMI) category (under or normal weight [< 25.0 kg/m2] /overweight [25.0–29.9 kg/m2] /obesity [≥30.0 kg/m2]), smoking status (never/former/current), alcohol drinking status (never/former/current), physical activity (measured by the International Physical Activity Questionnaire, in metabolic equivalent [MET] minutes per week), sleep duration on weekday (hours), shift work status, and chronotypes defined in a previous study. 25 Health conditions included diabetes, high blood pressure, angina, stroke, and heart attack updated until accelerometer wearing, and depressive symptom was defined according to two‐item Patient Health Questionnaire collected at recruitment, 26 which are potential confounders.
2.5. Statistical analysis
We described baseline characteristics of the participants using means (standard deviations [SDs]) for continuous variables and numbers (percentages) for categorical variables. Missing values of covariates were imputed using multiple imputation with chained equations. 27
We used Cox proportional hazard regression models to assess the association between accelerometer‐measured weekend catch‐up sleep and incident dementia. Weekend catch‐up sleep was included as multi‐class, continuous, and binary variables, in separate models. Person‐time was calculated from the accelerometer‐wearing date to the diagnosis of dementia, the ascertainment of death, or the end of follow‐up (December 2022), whichever came first. The hazard ratios (HRs) and confidence intervals (CIs) of incident dementia were sequentially adjusted for age at accelerometer assessment and sex in Model 1, education, ethnicity, Townsend deprivation index, BMI category, smoking status, alcohol drinking status, physical activity, sleep duration on weekday, shift work status, and chronotype in Model 2, and diabetes, high blood pressure, angina, stroke, heart attack, and depressive symptom in Model 3. To evaluate potential non‐linear patterns in the associations, we modelled weekend catch‐up sleep as a continuous variable using restricted cubic splines (RCSs), setting the median as the reference point and determining the optimal degrees of freedom based on the maximum Akaike information criterion. 28
To further assess whether the association differed by weekday sleep duration, we stratified the analysis by weekday sleep duration (< 8 or ≥8 h) according to the population distribution (only 4637 [5.5%] participants had weekday sleep duration < 7 h, which led to model convergence failure). Other stratified analyses were also preplanned to assess the associations in subgroups of participants defined by age (< 65 years or ≥65 years), sex, Townsend deprivation index (below or above median), BMI (obesity or non‐obesity), and smoking status (never or ever). We further stratified participants by current work status (employed or retired) and shift work status (yes or no) to examine potential heterogeneity in the associations. Likelihood ratio tests were used to calculate p‐values for the multiplicative interaction terms in the Cox regression models, comparing models with and without the interaction term.
In a secondary analysis, we explored the joint association of weekday sleep duration and weekend catch‐up sleep by categorizing participants into four groups based on a 2 × 2 classification. We further classified participants into three categories: (1) average sleep < 8 h, (2) average sleep ≥8 h without weekend catch‐up sleep (≤0.5 h), and (3) average sleep ≥8 h with weekend catch‐up sleep (> 0.5 h). To examine the potential differential associations of weekday and weekend sleep durations with dementia, we assessed the relationships of average sleep duration, weekday sleep duration, and weekend sleep duration with incident all‐cause dementia, using the same approaches as specified above. When evaluating the independent associations of weekday and weekend sleep durations, we mutually adjusted for each in the same model.
We conducted several sensitivity analyses to verify the robustness of the primary findings: (1) we excluded participants with baseline cardiovascular diseases or diabetes, who might have altered sleeping behavior because of the disease status; (2) we excluded participants with depressive symptom at baseline because sleeping disorder is often a manifestation of depression; (3) we excluded participants who developed dementia within the first 5 years of follow‐up to further reduce the impact of reverse causality; (4) to account for potential genetic susceptibility, apolipoprotein E (APOE) ε4 carrier status genotype was additionally adjusted, among participants with available genotype data in the UK Biobank; (5) we restricted to participants with insufficient weekday sleep (< 7 h), as commonly used in sleep research to define sleep deficiency; (6) the Fine–Gray sub‐distribution hazards model was applied to address the potential competing risk of non‐dementia mortality; (7) we specified participants with weekend–weekday difference ≤0 h or ≤1 h as the reference group and reassessed the association.
All analyses were performed using R version 4.3.0. Statistical significance level was set to be two‐sided p‐values < 0.05. While the subgroup analyses should be interpreted as exploratory, we applied false discovery rate (FDR) correction to account for potential inflation of Type I error and to aid interpretation. In contrast, sensitivity analyses were conducted solely to evaluate the robustness of the main results, and multiple comparison correction was therefore not applied.
3. RESULTS
3.1. Baseline characteristics
Among the 83,776 participants, the mean age at the time of accelerometer assessment was 63.3 years (SD: 6.7), 56.5% were female, and the mean weekend catch‐up sleep duration was 0.49 h (SD: 1.21, Table 1). Among the study participants, 44,520 had none or ≤0.5 h of weekend catch‐up sleep (53.1%), of whom 29,413 (35.1%) had a weekend‐weekday difference in sleep duration ≤0 h. Participants engaging in longer weekend catch‐up sleep were younger, less likely to be female, and had shorter sleep duration on weekdays (e.g., mean: 8.4 h, SD: 1.1 for > 1‐1.5 h catch‐up group), compared to those with none or ≤0.5 h of catch‐up sleep (mean: 9.0 h, SD: 1.2). Table S1 showed the mean sleep durations on each day of week.
TABLE 1.
Baseline characteristics of participants.
| Weekend catch‐up sleep | ||||||
|---|---|---|---|---|---|---|
| Variable | Overall | None or ≤0.5 h | >0.5–1 h | >1–1.5 h | >1.5–2 h | >2 h |
| N | 83776 | 44520 | 13969 | 10150 | 6458 | 8679 |
| Age at accelerometer assessment, years (mean [SD]) | 63.3 (6.7) | 64.5 (6.4) | 63.4 (6.6) | 62.4 (6.7) | 61.3 (6.7) | 59.9 (6.4) |
| Male (%) | 36478 (43.5) | 18822 (42.3) | 6159 (44.1) | 4444 (43.8) | 2962 (45.9) | 4091 (47.1) |
| Townsend deprivation index (mean [SD]) | 14.5 (11.7) | 14.3 (11.6) | 14.1 (11.3) | 14.5 (11.7) | 15.1 (12.2) | 15.8 (12.4) |
| College or university education (%) | 35667 (42.6) | 18528 (41.6) | 6200 (44.4) | 4542 (44.7) | 2766 (42.8) | 3631 (41.8) |
| White ethnicity (%) | 81565 (97.4) | 43477 (97.7) | 13629 (97.6) | 9853 (97.1) | 6262 (97.0) | 8344 (96.1) |
| BMI category (%) | ||||||
| Underweight | 2264 (2.7) | 1209 (2.7) | 393 (2.8) | 271 (2.7) | 163 (2.5) | 228 (2.6) |
| Normal weight | 30313 (36.2) | 16210 (36.4) | 5222 (37.4) | 3692 (36.4) | 2310 (35.8) | 2879 (33.2) |
| Overweight | 35016 (41.8) | 18566 (41.7) | 5846 (41.8) | 4298 (42.3) | 2709 (41.9) | 3597 (41.4) |
| Obesity | 16183 (19.3) | 8535 (19.2) | 2508 (18.0) | 1889 (18.6) | 1276 (19.8) | 1975 (22.8) |
| Smoking status (%) | ||||||
| Never | 47382 (56.6) | 24952 (56.0) | 7965 (57.0) | 5806 (57.2) | 3687 (57.1) | 4972 (57.3) |
| Former | 30966 (37.0) | 16792 (37.7) | 5136 (36.8) | 3674 (36.2) | 2352 (36.4) | 3012 (34.7) |
| Current | 5428 (6.5) | 2776 (6.2) | 868 (6.2) | 670 (6.6) | 419 (6.5) | 695 (8.0) |
| Alcohol drinking status (%) | ||||||
| Never | 2424 (2.9) | 1310 (2.9) | 391 (2.8) | 289 (2.8) | 159 (2.5) | 275 (3.2) |
| Former | 2251 (2.7) | 1195 (2.7) | 351 (2.5) | 280 (2.8) | 172 (2.7) | 253 (2.9) |
| Current | 79101 (94.4) | 42015 (94.4) | 13227 (94.7) | 9581 (94.4) | 6127 (94.9) | 8151 (93.9) |
| Physical activity—MET‐minutes/week (mean [SD]) | 2538.5 (2438.5) | 2583.4 (2448.9) | 2524.2 (2431.7) | 2490.5 (2408.4) | 2445.7 (2399.1) | 2457.0 (2455.3) |
| Sleep duration on weekday—Hours/day (mean (SD)) | 8.7 (1.2) | 9.0 (1.2) | 8.6 (1.1) | 8.4 (1.1) | 8.4 (1.2) | 8.3 (1.2) |
| Sleep duration on weekend—Hours/day (mean (SD)) | 9.2 (1.4) | 8.6 (1.2) | 9.3 (1.1) | 9.7 (1.1) | 10.1 (1.2) | 11.1 (1.4) |
| Diabetes (%) | 2372 (2.8) | 1297 (2.9) | 341 (2.4) | 279 (2.7) | 194 (3.0) | 261 (3.0) |
| High blood pressure (%) | 61461 (73.4) | 33045 (74.2) | 10231 (73.2) | 7389 (72.8) | 4609 (71.4) | 6187 (71.3) |
| Depressive symptom (%) | 3043 (3.6) | 1601 (3.6) | 464 (3.3) | 335 (3.3) | 243 (3.8) | 400 (4.6) |
| Angina (%) | 1896 (2.3) | 1052 (2.4) | 338 (2.4) | 195 (1.9) | 139 (2.2) | 172 (2.0) |
| Stroke (%) | 854 (1.0) | 483 (1.1) | 153 (1.1) | 85 (0.8) | 60 (0.9) | 73 (0.8) |
| Heart attack (%) | 1354 (1.6) | 740 (1.7) | 227 (1.6) | 130 (1.3) | 114 (1.8) | 143 (1.6) |
Abbreviations: BMI, body mass index; MET, metabolic equivalent; SD, standard deviation.
3.2. Accelerometer‐measured weekend catch‐up sleep and incident dementia
During a total of 667,928 person‐years (median follow‐up = 8.0 years), 713 participants developed all‐cause dementia. We observed a significant lower risk of dementia among individuals with a moderate level of weekend catch‐up sleep (1–1.5 h/day). The fully adjusted HRs and 95% CIs across increasing categories of weekend catch‐up sleep (none or ≤0.5, > 0.5–1, > 1–1.5, > 1.5–2, and > 2 h) were 1 (reference), 0.91 (0.74–1.11), 0.64 (0.49–0.86), 0.84 (0.60–1.16), and 0.83 (0.60–1.16), respectively (Table 2). Figure S1 illustrated the association between weekend catch‐up sleep duration and the risk of incident dementia, modeled using a RCS. The analysis demonstrated a generally decreasing hazard for dementia with increasing weekend catch‐up sleep duration, with the lowest risk around 1–1.5 h of additional weekend sleep, despite of lacking statistical non‐linearity (p‐value = 0.563).
TABLE 2.
Hazard ratios (95% confidence intervals) of incident dementia according to accelerometer‐measured weekend catch‐up sleep.
| Weekend catch‐up sleep | N | Dementia Incidence | Model 1 | Model 2 | Model 3 |
|---|---|---|---|---|---|
| None or ≤0.5 h | 44520 | 462/354120 | 1.00 (Reference) | 1.00 (Reference) | 1.00 (Reference) |
| >0.5–1 h | 13969 | 118/111199 | 0.93 (0.76–1.14) | 0.91 (0.74–1.11) | 0.91 (0.74–1.11) |
| >1–1.5 h | 10150 | 54/81260 | 0.67 (0.50–0.88) | 0.64 (0.49–0.85) | 0.64 (0.49–0.86) |
| >1.5–2 h | 6458 | 39/51750 | 0.88 (0.64–1.22) | 0.85 (0.61–1.18) | 0.84 (0.60–1.16) |
| >2 h | 8679 | 40/69600 | 0.89 (0.65–1.24) | 0.84 (0.61–1.16) | 0.83 (0.60–1.16) |
Note: Model 1 was adjusted for age at accelerometer assessment and sex. Model 2 was based on Model 1 and further adjusted for education, ethnicity, Townsend deprivation index, BMI category, smoking status, alcohol drinking status, physical activity, sleep duration on weekday, shift work status, and chronotype. Model 3 was based on model 2 and further adjusted for diabetes, high blood pressure, angina, stroke, heart attack, and depressive symptom.
Abbreviation: BMI, body mass index.
Data with p‐values below 0.05 were presented in bold type.
When we further stratified the analyses by average weekday sleep duration, the associations were stronger among participants with less weekday sleep duration (Table 3 and Figure 2). In the RCS analysis, the likelihood ratio test yielded a p‐interaction of 0.039, indicating a statistically significant interaction. For participants with a weekday sleep duration < 8 h (Figure 2A), the association was approximately linear (p‐linear = 0.003, p‐nonlinear = 0.449), with the fully adjusted HRs and 95% CIs across increasing categories of weekend catch‐up sleep (none or ≤0.5, > 0.5–1, > 1–1.5, and > 1.5 h) being 1 (reference), 0.97 (0.69–1.37), 0.49 (0.29–0.81), and 0.52 (0.34–0.82), respectively. Participants with > 0.5 h of catch‐up sleep had 0.69‐fold (0.52–0.91, p‐value = 0.009) risk of dementia compared with those with none or ≤0.5 h, and the HR of per hour increase in weekend catch‐up sleep was 0.75 (95% CI: 0.65–0.87, p‐value < 0.001, Table 3). On the contrary, there was no significant association for participants with a weekday sleep duration ≥8 h (p‐linear = 0.362, p‐nonlinear = 0.265), though associations remained in the same direction for > 1–1.5 h (Figure 2B).
TABLE 3.
Hazard ratios (95% confidence intervals) of incident dementia according to accelerometer‐measured weekend catch‐up sleep by weekday sleep duration.
| Variable | Weekday sleep duration < 8 h | p‐Value | Weekday sleep duration ≥8 h | p‐Value | p for difference |
|---|---|---|---|---|---|
| By duration | |||||
| None or ≤0.5 h | 1.00 (Reference) | 1.00 (Reference) | |||
| >0.5–1 h | 0.97 (0.69–1.37) | 0.871 | 0.85 (0.66–1.10) | 0.230 | 0.545 |
| >1–1.5 h | 0.49 (0.29–0.81) | 0.006 | 0.72 (0.51–1.01) | 0.059 | 0.221 |
| >1.5 h | 0.52 (0.34–0.82) | 0.004 | 1.01 (0.76–1.35) | 0.921 | 0.013 |
| Per hour | 0.75 (0.65–0.87) | <0.001 | 0.96 (0.89–1.04) | 0.356 | 0.003 |
| Binary | |||||
| None or ≤0.5 h | 1.00 (Reference) | 1.00 (Reference) | |||
| >0.5 h | 0.69 (0.52–0.91) | 0.009 | 0.86 (0.72‐1.04) | 0.125 | 0.197 |
Note: The Cox model was adjusted for age at accelerometer assessment, sex, education, ethnicity, Townsend deprivation index, BMI category, smoking status, alcohol drinking status, physical activity, diabetes, high blood pressure, angina, stroke, heart attack, and depressive symptom. p for difference was calculated by comparing the difference of hazards ratios between weekday sleep duration < 8 h and ≥8 h.
Abbreviation: BMI, body mass index.
Data with p‐values below 0.05 were presented in bold type.
FIGURE 2.
Associations of accelerometer‐measured weekend catch‐up sleep with risk of incident dementia modelled using restricted cubic spline model by weekday sleep duration. Weekend catch‐up sleep was defined as the weekend‐weekday sleep duration difference. The Cox model was adjusted for age at accelerometer assessment, sex, education, ethnicity, Townsend deprivation index, BMI category, smoking status, alcohol drinking status, physical activity, diabetes, high blood pressure, angina, stroke, heart attack, and depressive symptom. Reference point was set to be population median sleep duration on x‐axis. Number of knots was chosen based on AIC criteria. (A) For participants with weekday sleep duration < 8 h, p‐linear = 0.003, p‐nonlinear = 0.449. (B) For participants with weekday sleep duration ≥8 h, p‐linear = 0.362, p‐nonlinear = 0.265. The likelihood ratio test for the interaction term yielded a chi‐squared statistic of 6.46 (degrees of freedom = 2), with a p‐value of 0.039, indicating a statistically significant interaction. AIC, Akaike information criterion; BMI, body mass index.
3.3. Secondary analysis
When we dichotomized the participants according to the weekend catch‐up sleep duration, participants with > 0.5 h of catch‐up sleep had 0.81‐fold (0.70–0.95, p‐value = 0.010) risk of dementia compared with those with none or ≤0.5 h. When measured as a continuous variable, each hour increment in weekend catch‐up sleep was associated with a 9% lower risk of dementia (HR: 0.91, 95% CI: 0.84–0.98, p‐value = 0.008, Table S2).
We further categorized participants by overall sleep duration and presence of weekend catch‐up sleep (Table S3). Compared to short sleepers whose overall sleep duration < 8 h, those who had overall sleep duration ≥8 h, both with (HR = 0.67, 95% CI: 0.55–0.82, p‐value < 0.001) and without (HR = 0.77, 95% CI: 0.64–0.92, p‐value = 0.004) weekend catch‐up, were at lower risk of dementia.
These findings were confirmed by the joint associations of weekday sleep and weekend catch‐up (Table S3). Compared to participants with weekday sleep < 8 h and without weekend catch‐up, those with weekday sleep < 8 h who had weekend catch‐up showed an HR of 0.74 (95% CI: 0.56–0.98, p‐value = 0.032). Compared to those who slept < 8 h without weekend catch‐up, participants who slept ≥8 h on weekday, either with (HR = 0.62, 95% CI: 0.49–0.79, p‐value < 0.001) or without (HR = 0.72, 95% CI: 0.59–0.89, p‐value = 0.003) weekend catch‐up, had lower dementia risk.
We also explored the potential associations of sleep durations on different days with risk of dementia, and the associations for average, weekday, and weekend sleep durations all showed non‐linear patterns (Table S4 and Figure S2). Compared with participants with a daily sleep duration of > 8–9 h, those who had less sleep were at significantly higher dementia risk, with HR being 2.05 (95% CI: 1.49–2.82) for ≤7 h and 1.25 (95% CI: 1.02–1.54) for > 7–8 h. This U‐shape trend was both observed for weekday and weekend sleep durations.
3.4. Subgroup and sensitivity analysis
The associations between weekend catch‐up sleep and incident dementia were similar across the subgroups defined by the sociodemographic and lifestyle factors (Table S5 and Figure S3). Although the association for > 1–1.5 h of catch‐up sleep tended to be stronger in adults aged ≥65 years (HR = 0.38, 95% CI: 0.15–0.96) than in those aged 50–64 years (HR = 0.66, 95% CI: 0.49–0.88), the interaction was not statistically significant (p‐interaction = 0.568, FDR = 0.663). Similarly, a lower dementia risk associated with > 1–1.5 h of catch‐up sleep was observed only among non‐shift workers, with no significant interaction (p‐interaction = 0.854, FDR = 0.898).
The associations were also robust in several sensitivity analyses (Table S6). The HR (95% CI) for > 1–1.5 h of catch‐up sleep was 0.64 (95% CI: 0.47–0.88) when we excluded participants with cardiovascular diseases or diabetes, 0.60 (95% CI: 0.44–0.80) when we excluded participants with depressive symptom, and 0.64 (95% CI: 0.48–0.85) when we excluded dementia cases in the first 5‐year follow‐up. Additional adjustment for APOE ε4 carrier status and accounting for the competing risk of non‐dementia mortality did not materially alter the associations. When analyses were restricted to participants who reported < 7 h of weekday sleep, the confidence intervals widened and the associations became non‐significant, likely due to the reduced sample size (n = 12,037), but the direction of the estimates remained consistent (HR = 0.58, 95% CI, 0.23–1.42 for > 1–1.5 h group). When using catch‐up of ≤0 h or ≤1 h as the reference group, the associations were similar (Table S7).
4. DISCUSSION
In this prospective cohort study, we found that moderate weekend catch‐up sleep was associated with a lower risk of incident all‐cause dementia in later life. Participants with 1–1.5 h of weekend catch‐up sleep had the most pronounced reduction in dementia risk. The association was stronger among individuals with weekday sleep < 8 h/d. Similar associations were observed across the subgroups defined by the sociodemographic and lifestyle factors.
Our findings add to the growing evidence on sleep pattern and cognitive health. Prior studies have consistently demonstrated that inadequate sleep is associated with higher dementia risk, highlighting the importance of healthy sleep in brain health. A previous meta‐analysis reported a U‐curve association for sleep duration and dementia with a turning point at 7–8 h, 5 which was further confirmed by another study in the UK Biobank 29 and this study. However, previous studies primarily based on self‐reported sleep duration and focused on average sleep duration, with limited attention to the weekday‐weekend variations in sleep. Using data from objective sleep measurements, the current study offers new evidence on the potentially beneficial association between compensatory sleep and dementia. Our findings echoed a cross‐sectional study among older adults in Taipei city, which used sleep diaries and accelerometer to assess sleep duration and reported a 73%–74% lower odds of cognitive dysfunction among individuals with weekend catch‐up sleep. 10 The present results indicate that, not only maintaining adequate sleep on weekdays, but weekend compensatory catch‐up sleep may also play a protective role. If proven causal, these findings could contribute to the existing recommendations for dementia prevention. 30 Our findings also indicated that the association was stronger among participants with weekday sleep duration < 8 h, which suggests that moderate catch‐up sleep may primarily act as a compensatory mechanism. Those with weekday sleep ≥8 h likely already experience optimal cognitive and physiological benefits, leaving little room for further compensation.
In the RCS analysis, which was used to visually depict the overall shape of the association, the test for nonlinearity did not reach statistical significance, but the pattern of point estimates was consistent with the categorical analyses, indicating the lowest dementia risk at approximately 1–1.5 h of weekend catch‐up sleep. Existing studies have explored how adequate sleep supports long‐term brain functions, such as neural repair, synaptic plasticity, and waste clearance through the glymphatic system, 31 , 32 , 33 which are essential for preventing neurodegenerative processes associated with dementia. Moderate catch‐up sleep might restore certain cognitive and physiological processes affected by sleep debt, 34 such as improved glymphatic clearance 35 of neurotoxic waste, 31 which has been shown to occur during deep sleep. However, longer durations of weekend sleep could signal a more chronic sleep imbalance, disrupted circadian rhythms, or other health conditions that independently contribute to dementia risk. Although there is no universal optimal range, excessive weekend catch‐up may also indicate fragmented sleep or lifestyle stressors during the week that cannot be fully mitigated by longer weekend rest, suggesting diminishing returns beyond a certain threshold. Further research is needed to elucidate the underlying mechanisms and to identify the optimal range of weekend catch‐up sleep, potentially varying by weekday sleep duration, that supports cognitive resilience and informs targeted interventions.
This study has several strengths, including its large sample size, prospective design, and the objective measurements for sleep duration with a higher precision compared with self‐reported sleep duration. However, some limitations should be noted. First, while the accelerometer data provided an objective assessment, it did not capture sleep quality or specific parameters such as sleep efficiency. Although our findings are comparable with estimates from previous studies, 22 the sleep measurements in this study may include time spent in bed, and misclassify sedentary wakefulness as sleep, particularly among older adults, which could result in overestimation of actual sleep duration and bias observed associations. Further calibration using polysomnographic measurements are needed. Second, although we adjusted for a wide range of demographic, lifestyle, and health‐related factors, residual confounding cannot be completely ruled out. Unmeasured factors such as stress levels, dietary quality, cognitive reserve (e.g., occupational complexity), social engagement, and subclinical psychiatric symptoms may influence both sleep behavior and dementia risk. Our findings, therefore, may not fully represent the causal effects. Third, dementia ascertainment was based on electronic health records, which is subject to certain degree of misclassification, although such nondifferential misclassification would likely bias the associations toward the null. In addition, while accelerometer data provided detailed 7‐day sleep patterns, it was collected at a single time point, and we were unable to further investigate the associations of long‐term and time‐varying sleep behaviors. Considering the long‐term preclinical phase of dementia, this warrants further investigations. 1 Furthermore, the cohort primarily included adults of White ethnicity, which may limit the generalizability of findings to more diverse populations. Finally, the respondents to this sub‐study of the UK Biobank could be healthier than the general population, which may introduce selection bias as they could be less susceptible to dementia, although this may not fully diminish its representativeness in assessing exposure‐disease relationships. 36
In summary, weekend catch‐up sleep was associated with lower risk of dementia among middle‐aged and older adults. Our findings suggest that moderate weekend catch‐up sleep may be linked to the lowest dementia risk. The association was stronger among individuals with less sleep on weekdays. Future studies are needed to validate the findings and determine the optimal sleep compensation strategies for dementia prevention.
CONFLICT OF INTEREST STATEMENT
All authors declare no competing interests. Author disclosures are available in the Supporting Information.
CONSENT STATEMENT
Participants provided informed consent before data collection.
Supporting information
Supporting information
Supporting information
ACKNOWLEDGMENTS
This research was conducted using the UK Biobank resource under application number 55005. Data can be shared through mechanisms detailed at https://www.ukbiobank.ac.uk/. We thank for all participating individuals and staff of the UK Biobank who made the study possible. This work was supported in part by grants from the Noncommunicable Chronic Diseases‐National Science and Technology Major Project (2023ZD0508605, to Dr Yuan), the Major Research Plan of the National Natural Science Foundation of China (2022YFC2010100, to Dr Yuan), the Zhejiang University Global Partnership Fund (to Dr Yuan), and the Fundamental Research Funds for the Central Universities (226‐2025‐00178, to Dr Yuan). The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Chen H, Shen T, Zhao M, et al. Accelerometer‐measured weekend catch‐up sleep and incident dementia: A prospective cohort study. Alzheimer's Dement. 2025;21:e71001. 10.1002/alz.71001
Hui Chen and Ting Shen contributed equally as co‐first authors.
Changzheng Yuan and Tianyi Huang contributed equally as co‐senior authors.
REFERENCES
- 1. Livingston G, Huntley J, Liu KY, et al. Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. Lancet. 2024;404(10452):572‐628. doi: 10.1016/S0140-6736(24)01296-0 [DOI] [PubMed] [Google Scholar]
- 2. Knopman DS, Amieva H, Petersen RC, et al. Alzheimer disease. Nat Rev Dis Primers. 2021;7(1):1‐21. doi: 10.1038/s41572-021-00269-y [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. GBD 2019 Dementia Forecasting Collaborators . Estimation of the global prevalence of dementia in 2019 and forecasted prevalence in 2050: an analysis for the Global Burden of Disease Study 2019. Lancet Public Health. 2022;7(2):e105‐e125. doi: 10.1016/S2468-2667(21)00249-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Lourida I, Hannon E, Littlejohns TJ, et al. Association of Lifestyle and Genetic Risk With Incidence of Dementia. JAMA. 2019;322(5):430‐437. doi: 10.1001/jama.2019.9879 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Liang Y, Qu LB, Liu H. Non‐linear associations between sleep duration and the risks of mild cognitive impairment/dementia and cognitive decline: a dose–response meta‐analysis of observational studies. Aging Clin Exp Res. 2019;31(3):309‐320. doi: 10.1007/s40520-018-1005-y [DOI] [PubMed] [Google Scholar]
- 6. Pizinger TM, Aggarwal B, St‐Onge MP. Sleep extension in short sleepers: an evaluation of feasibility and effectiveness for weight management and cardiometabolic disease prevention. Front Endocrinol. 2018;9:392. doi: 10.3389/fendo.2018.00392 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Im HJ, Baek SH, Chu MK, et al. Association between weekend catch‐up sleep and lower body mass: population‐based study. Sleep. 2017;40(7):zsx089. doi: 10.1093/sleep/zsx089 [DOI] [PubMed] [Google Scholar]
- 8. Hwangbo Y, Kim WJ, Chu MK, Yun CH, Yang KI. Association between weekend catch‐up sleep duration and hypertension in Korean adults. Sleep Medicine. 2013;14(6):549‐554. doi: 10.1016/j.sleep.2013.02.009 [DOI] [PubMed] [Google Scholar]
- 9. Kim DJ, Mun SJ, Choi JS, et al. Beneficial effects of weekend catch‐up sleep on metabolic syndrome in chronic short sleepers. Sleep Medicine. 2020;76:26‐32. doi: 10.1016/j.sleep.2020.09.025 [DOI] [PubMed] [Google Scholar]
- 10. Hsiao C, Huang WC, Hsueh MC, CC Chang, Liao Y, Lin KP. Can weekend catch‐up sleep decrease the risk of cognitive dysfunction in older adults?. Sleep Breath. 2024;28(5):2303‐2310. doi: 10.1007/s11325-024-03074-5 [DOI] [PubMed] [Google Scholar]
- 11. Huang T, Mariani S, Redline S. Sleep irregularity and risk of cardiovascular events. J Am Coll Cardiol. 2020;75(9):991‐999. doi: 10.1016/j.jacc.2019.12.054 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Kianersi S, Wang H, Sofer T, et al. Association between accelerometer‐measured irregular sleep duration and type 2 diabetes risk: a prospective cohort study in the UK biobank. Diabetes Care. 2024;47(9):1647‐1655. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Yiallourou SR, Cribb L, Cavuoto MG, et al. Association of the sleep regularity index with incident dementia and brain volume. Neurology. 2024;102(2):e208029. doi: 10.1212/WNL.0000000000208029 [DOI] [PubMed] [Google Scholar]
- 14. Bycroft C, Freeman C, Petkova D, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018;562(7726):203‐209. doi: 10.1038/s41586-018-0579-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLOS Medicine. 2015;12(3):e1001779. doi: 10.1371/journal.pmed.1001779 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Doherty A, Jackson D, Hammerla N, et al. Large scale population assessment of physical activity using wrist worn accelerometers: the UK biobank study. PLOS ONE. 2017;12(2):e0169649. doi: 10.1371/journal.pone.0169649 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Raichlen DA, Aslan DH, Sayre MK, et al. Sedentary behavior and incident dementia among older adults. JAMA. 2023;330(10):934‐940. doi: 10.1001/jama.2023.15231 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Chen H, Ding Y, Dhana K, et al. Sweetened beverages and incident all‐cause dementia among older adults. JAMA Psychiatry. 2025:e251230. doi: 10.1001/jamapsychiatry.2025.1230 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Zhang Y, Yang H, Li S, dong LiW, Wang Y. Consumption of coffee and tea and risk of developing stroke, dementia, and poststroke dementia: a cohort study in the UK Biobank. PLOS Medicine. 2021;18(11):e1003830. doi: 10.1371/journal.pmed.1003830 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Huang T, Beydoun MA, Kianersi S, Redline S, Launer LJ. Multi‐dimensional sleep health and dementia risk: a prospective study in the UK Biobank. BMC Medicine. 2025;23(1):410. doi: 10.1186/s12916-025-04251-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Hedayatrad L, Stewart T, Duncan S. Concurrent validity of ActiGraph GT3X+ and Axivity AX3 accelerometers for estimating physical activity and sedentary behavior. J Meas Phys Behav. 2020;4(1):1‐8. doi: 10.1123/jmpb.2019-0075 [DOI] [Google Scholar]
- 22. Walmsley R, Chan S, Smith‐Byrne K, et al. Reallocation of time between device‐measured movement behaviours and risk of incident cardiovascular disease. Br J Sports Med. 2021;56(18):1008‐1017. doi: 10.1136/bjsports-2021-104050 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Chaput JP, Biswas RK, Ahmadi M, et al. Device‐measured weekend catch‐up sleep, mortality, and cardiovascular disease incidence in adults. Sleep. 2024:zsae135. doi: 10.1093/sleep/zsae135 [DOI] [PubMed] [Google Scholar]
- 24. Wilkinson T, Schnier C, Bush K, et al. Identifying dementia outcomes in UK Biobank: a validation study of primary care, hospital admissions and mortality data. Eur J Epidemiol. 2019;34(6):557‐565. doi: 10.1007/s10654-019-00499-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lane JM, Vlasac I, Anderson SG, et al. Genome‐wide association analysis identifies novel loci for chronotype in 100,420 individuals from the UK Biobank. Nat Commun. 2016;7(1):10889. doi: 10.1038/ncomms10889 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Harshfield EL, Pennells L, Schwartz JE, et al. Association between depressive symptoms and incident cardiovascular diseases. JAMA. 2020;324(23):2396‐2405. doi: 10.1001/jama.2020.23068 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. va Buuren, mice Groothuis‐OudshoornK. Multivariate imputation by chained equations in R. J Stat Softw. 2011;45:1‐67. doi: 10.18637/jss.v045.i03 [DOI] [Google Scholar]
- 28. Desquilbet L, Mariotti F. Dose‐response analyses using restricted cubic spline functions in public health research. Stat Med. 2010;29(9):1037‐1057. doi: 10.1002/sim.3841 [DOI] [PubMed] [Google Scholar]
- 29. Bian W, Biswas RK, Ahmadi MN, et al. Dose‐response associations of device measured sleep regularity and duration with incident dementia in 82391 UK adults. medRxiv. doi: 10.1101/2023.11.23.23298926 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Sheehan CM, Frochen SE, Walsemann KM, Ailshire JA. Are U.S. adults reporting less sleep?: findings from sleep duration trends in the National Health Interview Survey, 2004‐2017. Sleep. 2019;42(2):zsy221. doi: 10.1093/sleep/zsy221 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Raven F, Van der Zee EA, Meerlo P, Havekes R. The role of sleep in regulating structural plasticity and synaptic strength: implications for memory and cognitive function. Sleep Med Rev. 2018;39:3‐11. doi: 10.1016/j.smrv.2017.05.002 [DOI] [PubMed] [Google Scholar]
- 32. Simon KC, Nadel L, Payne JD. The functions of sleep: a cognitive neuroscience perspective. Proc Natl Acad Sci. 2022;119(44):e2201795119. doi: 10.1073/pnas.2201795119 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Lavie L. Oxidative stress in obstructive sleep apnea and intermittent hypoxia – Revisited – The bad ugly and good: implications to the heart and brain. Sleep Med Rev. 2015;20:27‐45. doi: 10.1016/j.smrv.2014.07.003 [DOI] [PubMed] [Google Scholar]
- 34. Van Dongen HPA, Rogers NL, Dinges DF. Sleep debt: theoretical and empirical issues. Sleep Biol Rhythms. 2003;1(1):5‐13. doi: 10.1046/j.1446-9235.2003.00006.x [DOI] [Google Scholar]
- 35. Spiegel K, Leproult R, Cauter EV. Impact of sleep debt on metabolic and endocrine function. Lancet. 1999;354(9188):1435‐1439. doi: 10.1016/S0140-6736(99)01376-8 [DOI] [PubMed] [Google Scholar]
- 36. Fry A, Littlejohns TJ, Sudlow C, et al. Comparison of sociodemographic and health‐related characteristics of UK biobank participants with those of the general population. A J Epidemiol. 2017;186(9):1026‐1034. doi: 10.1093/aje/kwx246 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supporting information
Supporting information