Advertisement

Do the relationships of physical activity and total sleep time with cognitive function vary by age and biological sex? A cross-sectional analysis of the Canadian Longitudinal Study on Aging

  • Ryan S. Falck
    Affiliations
    Aging, Mobility, and Cognitive Neuroscience Laboratory, Department of Physical Therapy, University of British Columbia, Vancouver, BC, Canada

    Djavad Mowafaghian Centre for Brain Health, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada

    Centre for Hip Health and Mobility, University of British Columbia, Vancouver, BC, Canada
    Search for articles by this author
  • John R. Best
    Affiliations
    Aging, Mobility, and Cognitive Neuroscience Laboratory, Department of Physical Therapy, University of British Columbia, Vancouver, BC, Canada

    Gerontology Research Centre, Simon Fraser University, Vancouver, BC, Canada
    Search for articles by this author
  • Cindy K. Barha
    Affiliations
    Aging, Mobility, and Cognitive Neuroscience Laboratory, Department of Physical Therapy, University of British Columbia, Vancouver, BC, Canada

    Djavad Mowafaghian Centre for Brain Health, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada

    Centre for Hip Health and Mobility, University of British Columbia, Vancouver, BC, Canada
    Search for articles by this author
  • Jennifer C. Davis
    Affiliations
    Centre for Hip Health and Mobility, University of British Columbia, Vancouver, BC, Canada

    Applied Health Economics Laboratory, University of British Columbia – Okanagan Campus, Kelowna, BC, Canada

    Social & Economic Change Laboratory, Faculty of Management, University of British Columbia – Okanagan Campus, Kelowna, BC, Canada
    Search for articles by this author
  • Teresa Liu-Ambrose
    Correspondence
    Corresponding author at: University of British Columbia, Faculty of Medicine, Aging, Mobility and Cognitive Neuroscience Laboratory, Department of Physical Therapy, Vancouver, BC, Canada.
    Affiliations
    Aging, Mobility, and Cognitive Neuroscience Laboratory, Department of Physical Therapy, University of British Columbia, Vancouver, BC, Canada

    Djavad Mowafaghian Centre for Brain Health, Faculty of Medicine, University of British Columbia, Vancouver, BC, Canada

    Centre for Hip Health and Mobility, University of British Columbia, Vancouver, BC, Canada
    Search for articles by this author

      Highlights

      • Physical activity (PA) and total sleep time (TST) are associated with cognition.
      • However, little is known about age and sex differences in these relationships.
      • We examined age and sex differences in associations of PA and TST with cognition.
      • PA is associated with better cognition irrespective of age and biological sex.
      • TST has a complex relationship with cognition which is age and sex dependent.

      Abstract

      Objectives

      Physical activity (PA) and total sleep time (TST) are each associated with cognition; however, whether these relationships vary by age and biological sex is unclear. We examined the relationships of PA or TST with cognition, and whether age and sex moderated these relationships, using baseline data from the Canadian Longitudinal Study on Aging (CLSA; 2010–2015).

      Study design

      A cross-sectional analysis of participants from the Comprehensive cohort of the CLSA with complete PA and sleep data (n = 20,307; age range 45–86 years).

      Main outcome measures

      PA and TST were measured using the Physical Activity Scale for the Elderly (PASE) and self-reported TST over the past month. Cognition was indexed using a three-factor structural equation model (i.e., memory, executive function, and verbal fluency).

      Results

      Non-linear restricted cubic spline models indicated that PA and TST explained statistically significant (p < 0.01) but modest variance of each cognitive domain (<1 % of 23–24 % variance). Age and sex did not moderate associations of PA with any cognitive domain. However, age and sex moderated relationships of TST with cognition, whereby: 1) associations of TST with memory decreased with age for males and females; and 2) males and females had different age-associated relationships of TST with executive function and verbal fluency.

      Conclusions

      PA and TST modestly contribute to multiple domains of cognition across middle and older adulthood. Importantly, the association of PA with cognition does not appear to vary across middle or older adulthood, nor does it vary by biological sex; however, TST appears to have a complex relationship with multiple domains of cognition which is both age- and sex-dependent.

      Keywords

      1. Introduction

      Aging is associated with multi-faceted changes in cognition [
      • Salthouse T.A.
      When does age-related cognitive decline begin?.
      ], affecting the domains of memory, executive function (a broad set of planning and problem solving abilities), and verbal fluency, among others [
      • Glisky E.L.
      Changes in cognitive function in human aging.
      ]. Lifestyle factors such as physical activity (PA) and sleep may be pivotal to blunting cognitive decline [
      • Liu-Ambrose T.
      • Falck R.S.
      Sleep, physical activity, and cognitive health in older adults.
      ]. Of particular importance, emerging evidence indicates that there are age and sex differences in how PA and sleep are related to cognition [
      • Barha C.K.
      • Hsu C.L.
      • Ten Brinke L.
      • Liu-Ambrose T.
      Biological sex: a potential moderator of physical activity efficacy on brain health.
      ,
      • Hajali V.
      • Andersen M.L.
      • Negah S.S.
      • Sheibani V.
      Sex differences in sleep and sleep loss-induced cognitive deficits: the influence of gonadal hormones.
      ].
      PA is a key determinant of cognitive health from midlife into older adulthood [
      • Liu-Ambrose T.
      • Falck R.S.
      Sleep, physical activity, and cognitive health in older adults.
      ]. For example, regular PA of ≥150 min/week can reduce the risk of dementia by up to 28 % [
      • Hamer M.
      • Chida Y.
      Physical activity and risk of neurodegenerative disease: a systematic review of prospective evidence.
      ]. Aging is associated with declines in PA levels [
      • Troiano R.P.
      • Berrigan D.
      • Dodd K.W.
      • Masse L.C.
      • Tilert T.
      • McDowell M.
      Physical activity in the United States measured by accelerometer.
      ]; however, it is unclear whether the association between PA and cognitive performance is moderated by age. The strongest evidence that PA promotes cognition comes from studies of older adults [
      • Erickson K.I.
      • Hillman C.
      • Stillman C.M.
      • Ballard R.M.
      • Bloodgood B.
      • Conroy D.E.
      • Macko R.
      • Marquez D.X.
      • Petruzzello S.J.
      • Powell K.E.
      Physical activity, cognition, and brain outcomes: a review of the 2018 physical activity guidelines.
      ], but far less is known about how PA is associated with cognition in middle-aged adults.
      Good sleep is also associated with lower risk of cognitive decline and dementia [
      • Liu-Ambrose T.
      • Falck R.S.
      Sleep, physical activity, and cognitive health in older adults.
      ]. Sufficient total sleep time (TST) appears to be critical for cognition, as the neurocognitive deficits following acute sleep loss are experienced almost universally and include impairments across multiple cognitive domains [
      • Lowe C.J.
      • Safati A.
      • Hall P.A.
      The neurocognitive consequences of sleep restriction: a meta-analytic review.
      ]. TST declines naturally with age [
      • Ohayon M.M.
      • Carskadon M.A.
      • Guilleminault C.
      • Vitiello M.V.
      Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan.
      ], but age is a key moderator in the association between sleep and cognition whereby older adults are particularly susceptible to the impacts of poor sleep on cognitive performance [
      • Scullin M.K.
      • Bliwise D.L.
      Sleep, cognition, and normal aging: integrating a half century of multidisciplinary research.
      ]. Importantly, both long and short TST are related to poorer cognition [
      • Devore E.E.
      • Grodstein F.
      • Schernhammer E.S.
      Sleep duration in relation to cognitive function among older adults: a systematic review of observational studies.
      ], and while adverse changes in TST are associated with declines in cognitive performance in both middle and older adulthood [
      • Yaffe K.
      • Falvey C.M.
      • Hoang T.
      Connections between sleep and cognition in older adults.
      ], it is still unclear whether there are age-associated differences in the relationships of TST with cognition.
      PA and sleep also share a complex relationship with each other and cognition [
      • Liu-Ambrose T.
      • Falck R.S.
      Sleep, physical activity, and cognitive health in older adults.
      ]. For example, Wei and colleagues [
      • Wei J.
      • Hou R.
      • Xie L.
      • Chandrasekar E.K.
      • Lu H.
      • Wang T.
      • Li C.
      • Xu H.
      Sleep, sedentary activity, physical activity, and cognitive function among older adults: the National Health and nutrition examination survey, 2011–2014.
      ] found that for older adults sleeping ≤7 h/night, higher PA was associated with better cognition. For older adults sleeping >7 h/night, PA was not consistently associated with cognitive performance; TST longer than 7 h/night was associated with worse cognitive function. It is unknown whether there are age-associated differences in the relationships of PA or TST with cognition.
      It is also unclear if the relationships of PA and TST with age-associated cognition vary by biological sex. Males are more physically active than females irrespective of age [
      • Troiano R.P.
      • Berrigan D.
      • Dodd K.W.
      • Masse L.C.
      • Tilert T.
      • McDowell M.
      Physical activity in the United States measured by accelerometer.
      ], and females experience greater age-associated deficits in TST than males [
      • Ohayon M.M.
      • Carskadon M.A.
      • Guilleminault C.
      • Vitiello M.V.
      Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan.
      ]. There are also differences in cognition between the sexes [
      • Andreano J.M.
      • Cahill L.
      Sex influences on the neurobiology of learning and memory.
      ]; females perform better on tests of verbal learning and memory, while males outperform females on tests of visuospatial ability. The strength of association between PA and cognition may also be different for females and males [
      • Barha C.K.
      • Hsu C.L.
      • Ten Brinke L.
      • Liu-Ambrose T.
      Biological sex: a potential moderator of physical activity efficacy on brain health.
      ]. For instance, females who are physically inactive at midlife experience more pronounced cognitive decline than females who are physically active at midlife, as well as males irrespective of their PA level [
      • Krell-Roesch J.
      • Syrjanen J.A.
      • Bezold J.
      • Trautwein S.
      • Barisch-Fritz B.
      • Boes K.
      • Woll A.
      • Forzani E.
      • Kremers W.K.
      • Machulda M.M.
      Physical activity and trajectory of cognitive change in older persons: Mayo Clinic study of aging.
      ]. Sleep loss also affects females more than males [
      • Hajali V.
      • Andersen M.L.
      • Negah S.S.
      • Sheibani V.
      Sex differences in sleep and sleep loss-induced cognitive deficits: the influence of gonadal hormones.
      ]. These findings highlight that we do not currently understand the complex relationships of PA and TST with age-associated cognition, and whether these phenomena vary by biological sex.
      We therefore conducted a cross-sectional analysis using baseline data from the Canadian Longitudinal Study on Aging (CLSA) to examine the relationships of PA or TST with cognition, and whether these relationships vary by age and biological sex.

      2. Methods

      2.1 Participants

      The CLSA is a Canadian multi-centre study of 51,338 participants between the ages of 45 and 85 years at the time of recruitment [
      • Raina P.S.
      • Wolfson C.
      • Kirkland S.A.
      • Griffith L.E.
      • Oremus M.
      • Patterson C.
      • Tuokko H.
      • Penning M.
      • Balion C.M.
      • Hogan D.
      The Canadian longitudinal study on aging (CLSA).
      ]. We included participants from the Comprehensive cohort (n = 30,097). During the baseline assessment (2010–2015), participants completed assessments of neuropsychological testing, in addition to providing data on demographic, lifestyle, physical, clinical, psychological, and economic measures. We excluded participants 1) with a self-reported history of cognitive impairment, dementia, Parkinson's disease, or stroke; 2) who did not complete PA or sleep questionnaires; and 3) who did not use the same language across all neuropsychological tests (i.e., some tests completed in French, some tests in English). Our final sample consisted of 20,307 participants (Fig. 1). All research conducted as part of the CLSA abides by the requirements of the Canadian Institutes of Health Research and relevant institutions for ethical conduct and privacy protection in health research. We received ethics approval from the University of British Columbia's Clinical Research Ethics Board (H19-01838). All subjects gave written informed consent.
      Fig. 1
      Fig. 1STROBE Diagram for participants from the comprehensive cohort of the Canadian Longitudinal Study on Aging (CLSA).

      2.2 Demographics and other covariates

      Self-reported age in years, biological sex, educational attainment, income, smoking status, and height and weight (used to calculate body mass index [BMI]) were assessed. The test language (English or French) during the neuropsychological testing session was also indexed. In addition, basic and instrumental activities of daily living (ADL) capability was indexed using the Older American Resources and Services (OARS) scale [
      • Fillenbaum G.G.
      • Smyer M.A.
      The development, validity, and reliability of the OARS multidimensional functional assessment questionnaire.
      ]. Depression symptoms were indexed using the Center for Epidemiologic Studies Short Depression Scale (CESD-10) [
      • Andresen E.M.
      • Malmgren J.A.
      • Carter W.B.
      • Patrick D.L.
      Screening for depression in well older adults: evaluation of a short form of the CES-D.
      ].

      2.3 Cognitive performance

      We assessed cognition using a three-factor cognitive model which we previously developed to characterize the cognitive function of participants in the CLSA [
      • Falck R.S.
      • Best J.R.
      • Davis J.C.
      • Barha C.K.
      • Khan K.M.
      • Liu-Ambrose T.
      Cardiometabolic risk, biological sex, and age do not share an interactive relationship with cognitive function: a cross-sectional analysis of the Canadian longitudinal study on aging.
      ]. The following cognitive measures collected during the CLSA were included in our model: the Mental Alternation Test, Rey Auditory-Verbal Learning Test, animal fluency, Stroop Test (Victoria version), Controlled Oral Word Association Test, and choice reaction time. Descriptions of the testing procedures can be found elsewhere [
      • Tuokko H.
      • Griffith L.E.
      • Simard M.
      • Taler V.
      Cognitive measures in the Canadian longitudinal study on aging.
      ]. Briefly, we used a structural equation model with confirmatory factor analysis wherein we compared three possible structures for characterizing the cognitive function of CLSA participants: 1) a single cognitive factor in which all six cognitive tests were loaded onto a common latent variable; 2) a two-factor model in which Rey Auditory-Verbal Learning Test was loaded onto a common latent variable of Memory, while Mental Alteration Test, animal fluency, Stroop, Controlled Oral Word Association Test, and choice reaction time were loaded onto a common latent variable of Executive Function; and 3) a three-factor model where in addition to the common latent variable of Memory, we loaded Mental Alteration Test, Stroop and choice reaction time onto a common latent variable (Executive Function), and animal fluency and Controlled Oral Word Association Test loaded onto a common latent variable (Verbal Fluency). After assessing the fit of each model, we determined that the three-factor model (i.e., Memory, Executive Function, and Verbal Fluency) had the best fit for characterizing the cognitive function of participants. Model fit statistics for the three-factor model are described in Supplementary material S1A, and indicated good model fit [
      • Falck R.S.
      • Best J.R.
      • Davis J.C.
      • Barha C.K.
      • Khan K.M.
      • Liu-Ambrose T.
      Cardiometabolic risk, biological sex, and age do not share an interactive relationship with cognitive function: a cross-sectional analysis of the Canadian longitudinal study on aging.
      ]. The standardized path estimates for the final three-factor model are described in Supplementary material S1B. Standardized scores for each domain of cognitive function were determined for each participant.

      2.4 Physical Activity Scale for the Elderly

      We measured PA using the Physical Activity Scale for the Elderly (PASE). The PASE is a questionnaire designed to assess PA among adults aged 55+ years [
      • Washburn R.A.
      • Smith K.W.
      • Jette A.M.
      • Janney C.A.
      The physical activity scale for the elderly (PASE): development and evaluation.
      ]. The questionnaire has good evidence of validity against actigraphy (r = 0.49; [
      • Washburn R.
      • Ficker J.
      Physical activity scale for the elderly (PASE): the relationship with activity measured by a portable accelerometer.
      ]), and good evidence of test-retest reliability (r = 0.75; [
      • Washburn R.A.
      • Smith K.W.
      • Jette A.M.
      • Janney C.A.
      The physical activity scale for the elderly (PASE): development and evaluation.
      ]). Scores range from 0 to 793 with higher scores indicative of greater PA levels. In the initial development of the PASE, mean scores by age group and biological sex were as follows: 1) 65–69 years: Males = 144.3, Females = 112.7; 2) 70–75 years: Males = 102.4, Females = 89.1; and 3) >75 years: Males = 101.8; Females = 62.3.

      2.5 Subjective TST

      Subjective TST was indexed using self-reported average hours of sleep. Participants were asked, “During the past month on average how many hours of actual sleep did you get at night?” Using this index is common in large epidemiological studies [
      • Lo J.C.
      • Groeger J.A.
      • Cheng G.H.
      • Dijk D.-J.
      • Chee M.W.
      Self-reported sleep duration and cognitive performance in older adults: a systematic review and meta-analysis.
      ], and has evidence of validity (r = 0.47) and reliability (r = 0.64) compared to quantitative sleep assessments such as wrist-worn actigraphy [
      • Lauderdale D.S.
      • Knutson K.L.
      • Yan L.L.
      • Liu K.
      • Rathouz P.J.
      Self-reported and measured sleep duration: how similar are they?.
      ].

      2.6 Statistical analysis

      Analyses were conducted using R version 4.0.1 (r-project.org). Our statistical analysis code can be found on GitHub (https://github.com/ryanfalck/CLSA_PA_Sleep). For each model, we adjusted p-values for false discovery rate (FDR) using a 5 % FDR significance threshold. Our structural equation model for characterizing cognitive function was conducted in lavaan version 0.6-10. All regression models were constructed using ordinary least squares regression in rms version 6.0-0. We then visualized all regression models using ggplot2 version 3.3.5.

      2.6.1 Age and sex differences in PA and TST

      We first examined age and sex differences in PA and TST by separately regressing PASE score and subjective TST onto a set of predictors (age, sex, BMI, income, educational attainment, ADL, and CESD-10). Given that PA, TST, and cognition have non-linear associations with age [
      • Salthouse T.A.
      When does age-related cognitive decline begin?.
      ,
      • Troiano R.P.
      • Berrigan D.
      • Dodd K.W.
      • Masse L.C.
      • Tilert T.
      • McDowell M.
      Physical activity in the United States measured by accelerometer.
      ,
      • Ohayon M.M.
      • Carskadon M.A.
      • Guilleminault C.
      • Vitiello M.V.
      Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan.
      ], we applied a restricted cubic spline transformation to these variables in the regression models to allow for non-linear associations [
      • Croxford R.
      ]. This transformation consists of specifying knots in the distribution of the predictor variable (e.g., age) and then allowing for distinct cubic curves between those knots. Curves are restricted to a linear form at the two tails (i.e., between the minimum score and the first and again between the last knot and the maximum score) to avoid overfitting. In the current study, three knots were specified for our continuous predictors. In addition to the main effects of all predictors, interactions with age and sex were included. A separate model was conducted for PA and TST. We calculated total R2 to determine the extent which our model explained PA and TST.
      To evaluate individual predictors, we calculated statistical significance using F-tests and effect sizes using partial R2's. All predictors were retained in each model regardless of the p-value associated with their F-test values. We then ranked variables by their order of importance (i.e., partial R2).

      2.6.2 Age and sex differences in the relationships of PA or TST with cognition

      To examine age and sex differences in the associations of PA or TST with cognition, we regressed each cognitive domain onto the same set of predictors (as well as cognitive testing language used), and included PA or TST as the major independent variable of interest. We examined age and sex differences in the relationship of PA with cognition while accounting for TST, and then separately investigated age and sex differences in the association of TST with cognition while controlling for PA. For each model, we binned participants into five-year age groups (i.e., 45–49 years, 50–54 years, etc.) by sex and then PA and TST were standardized within these age groups. This was done in order to account for age and sex differences in PASE score and subjective TST. For instance, middle-age adults likely have different low, medium, and high raw values of PASE score than older adults, and these values would also differ between males and females. Given the bi-directional relationship of PA and TST [
      • Liu-Ambrose T.
      • Falck R.S.
      Sleep, physical activity, and cognitive health in older adults.
      ], each model of PA included age-standardized TST as a covariate and each model of TST included age-standardized PA as a covariate.
      We then specified restricted cubic splines with three knots for values of PA and TST included in each model, as well as for age. Each model included two- and three-way interactions between our major independent variable of interest (i.e., PA or TST), age, and sex. The importance of age, sex, our major independent variable of interest, and their 2- and 3-way interactions, were determined by calculating partial R2's and p-values from F-tests. We graphed each model separately for males and females using simple slopes (i.e., 25th percentile and 75th percentile values) for PA or TST.

      3. Results

      3.1 Sample characteristics

      Table 1 summarizes the study sample. Participant mean age was 62 years (SD = 10 years). Fifty-one percent of the sample was female, and 80 % had a university degree or higher. Ninety-three percent of participants reported no ADL problems. Age and sex differences in cognition in Memory, Executive Function, and Verbal Fluency are described in Supplementary material S2. Mean PASE score was 164.92 (SD = 78.23) which is slightly higher than the average PASE score for older adults [
      • Washburn R.A.
      • Smith K.W.
      • Jette A.M.
      • Janney C.A.
      The physical activity scale for the elderly (PASE): development and evaluation.
      ]. Average self-reported TST was 6.81 h/night (SD = 1.19 h/night) which is slightly shorter than the recommended TST of 7 h/night for older adults [
      • Hirshkowitz M.
      • Whiton K.
      • Albert S.M.
      • Alessi C.
      • Bruni O.
      • DonCarlos L.
      • Hazen N.
      • Herman J.
      • Hillard P.J.A.
      • Katz E.S.
      National Sleep Foundation’s updated sleep duration recommendations.
      ].

      3.2 Age and sex differences in PASE score and TST

      Table 2 describes age and sex differences in PASE score, TST, and cognitive performance. Fig. 2 illustrates the variance in PASE score and TST based on age and sex differences. Partial R2 for each predictor are described in Table 3. Our models explained 29 % of the variance in PASE score and 5 % of the variance in TST.
      Table 1Participant characteristics (N = 20,307).
      Participant characteristicMean (SD)
      Age62.1 (9.9)
      Females, n (%)10,346 (50.9 %)
      Body mass index (kg/m2)27.8 (8.6)
      Educational attainment, n (%)
       Less than high school diploma876 (4.3 %)
       High school diploma1763 (8.7 %)
       Some college1466 (7.2 %)
       College degree or higher16,174 (79.8 %)
      Income level, n (%)
       <$20,000834 (4.3 %)
       $20,000–$50,0003875 (20.2 %)
       $50,000–$100,0006827 (35.5 %)
       $100,000–$150,0004036 (21.0 %)
       >$150,0003644 (19.0 %)
      Activities of daily living, n (%)
      Classified using the Older American Resources and Services (OARS) scale.
       No ADL problems18,816 (92.9 %)
       Mild ADL problems1378 (6.8 %)
       Moderate ADL problems44 (0.2 %)
       Severe ADL problems14 (0.1 %)
       Total ADL problems2 (0.1 %)
      CESD-10 score
      Center for Epidemiologic Studies Short Depression Scale.
      4.97 (4.43)
      French language cognitive testing3731 (18.4 %)
      Cognitive function (standardized scores)
       Memory0.00 (1.52)
       Executive function0.00 (4.22)
       Verbal fluency0.00 (3.30)
      Physical Activity Scale for the Elderly score164.92 (78.23)
      Subjective total sleep time (h/night)6.81 (1.19)
      a Classified using the Older American Resources and Services (OARS) scale.
      b Center for Epidemiologic Studies Short Depression Scale.
      Table 2Age and sex differences in cognitive performance, Physical Activity Scale for the Elderly (PASE) score, and total sleep time.
      Age groupMalesFemales
      NMean (SD)NMean (SD)
      Memory
      Standardized scores for each cognitive domain.
      45–54 years26550.3 (1.4)28740.85 (1.49)
      55–64 years3356−0.15 (1.32)36010.56 (1.44)
      65–74 years2445−0.65 (1.28)24070.04 (1.39)
      75–84 years1456−1.39 (1.20)1417−0.77 (1.34)
      85+ years49−1.83 (0.96)47−1.07 (1.43)
      Executive function
      Standardized scores for each cognitive domain.
      45–54 years26552.3 (3.06)28742.38 (2.92)
      55–64 years33560.53 (3.73)36010.98 (3.21)
      65–74 years2445−1.47 (3.98)2407−1.20 (3.89)
      75–84 years1456−4.25 (4.68)1417−3.60 (4.55)
      85+ years49−4.97 (3.89)47−5.32 (4.69)
      Verbal fluency
      Standardized scores for each cognitive domain.
      45–54 years26551.40 (2.91)28741.66 (2.89)
      55–64 years33560.29 (2.98)36010.82 (2.90)
      65–74 years2445−1.10 (3.00)2407−0.75 (3.02)
      75–84 years1456−2.88 (3.12)1417−2.38 (3.09)
      85+ years49−3.47 (2.54)47−3.97 (2.96)
      PASE score45–54 years2655229.46 (77.90)2874196.82 (71.82)
      55–64 years3356184.91 (75.23)3601158.07 (71.38)
      65–74 years2445147.19 (63.56)2407125.13 (56.87)
      75–84 years1456119.35 (58.53)141798.94 (47.79)
      85+ years49106.11 (70.47)4786.43 (62.32)
      Total sleep time (h/night)45–54 years26556.62 (1.08)28746.74 (1.21)
      55–64 years33566.78 (1.13)36016.81 (1.25)
      65–74 years24456.91 (1.13)24076.91 (1.24)
      75–84 years14567.01 (1.19)14176.81 (1.30)
      85+ years497.29 (1.12)477.15 (1.56)
      a Standardized scores for each cognitive domain.
      Fig. 2
      Fig. 2Summary of age and sex differences in physical activity and total sleep time (TST). A) Age and sex differences in Physical Activity Scale for the Elderly (PASE) score. B) Age and sex differences in subjective total sleep time.
      Table 3Summary of variance explained for each model examining the relationships between age, physical activity, and total sleep time, and whether these relationships are sex dependent.
      VariablePhysical activityTotal sleep time
      Partial R2p-Value
      All p-values adjusted for false discovery rate (FDR).
      Partial R2p-Value
      All p-values adjusted for false discovery rate (FDR).
      Age16 %<0.00011 %<0.0001
      Sex2 %<0.0001<1 %<0.0001
      Education<1 %0.2958<1 %0.0017
      Income1 %<0.0001<1 %0.0766
      Activities of daily living1 %<0.0001<1 %0.0005
      CESD-10 score<1 %<0.00014 %<0.0001
      Body mass index<1 %0.01566<1 %0.0013
      Age ∗ sex<1 %<0.0001<1 %<0.0001
      Physical activity measured using the Physical Activity Scale for the Elderly (PASE). Total sleep time measured using subjective total sleep time over the past month. All models controlled for body mass index (BMI), smoking status, educational attainment, income, and activities of daily living.
      a All p-values adjusted for false discovery rate (FDR).
      Age was the most important contributor to the model of PASE score (R2 = 16 %; p < 0.001); CESD-10 score was the most important contributor to the model of TST (R2 = 4 %; p < 0.001). Biological sex was the second most important contributor to the model of PASE score (R2 = 2 %; p < 0.001), and the third most important to the models of TST (R2 < 1 %; p < 0.001). The interaction of age and sex was the sixth most important contributor to PASE score (R2 < 1 %; p < 0.001) and the fifth most important contributor to TST (R2 < 1 %; p < 0.001).
      As illustrated in Fig. 2, results indicate that 1) PASE score was lower with increasing age, while TST was longer with greater age; 2) males had higher PASE score from middle to older adulthood, while females had longer TST from middle adulthood until approximately 70 years; and 3) male vs. female differences in PASE score and TST attenuated with age.

      3.3 Age and sex differences in the relationships of PASE score and TST with cognition

      Fig. 3, Fig. 4, Fig. 5 describe age by sex differences in the associations of PASE score and TST with cognition. Partial R2 for each predictor are described in Table 4. Models examining the association of PASE score with cognition explained 23–34 % of the variance. Across each of these models, age was the most important predictor (R2 range: 9–17 %; p < 0.001). Biological sex was the second most important predictor for our model of Memory (R2 = 6 %; p < 0.001), and the fifth most important predictor for our models of Executive Function (R2 = 1 %; p < 0.001) and Verbal Fluency (R2 = 1 %; p < 0.001). PASE score was the seventh most important predictor for Memory (R2 < 1 %; p = 0.004) and Executive Function (R2 < 1 %; p < 0.001), and the sixth most important predictor for Verbal Fluency (R2 < 1 %; p < 0.001). Two- and three-way interactions of PASE score with age and sex did not significantly contribute to any model of cognition. As summarized in Figs. 3A, 4A, and 5A, higher PASE score was associated with better Memory, Executive Function, and Verbal Fluency, irrespective of age or sex.
      Fig. 3
      Fig. 3Summary of age and sex differences in the associations of physical activity and subjective total sleep time with memory. A) Summary of age and sex differences in the association of Physical Activity Scale for the Elderly (PASE) score with memory. B) Summary of age and sex differences in the association of subjective total sleep time with memory.
      Fig. 4
      Fig. 4Summary of age and sex differences in the associations of physical activity and subjective total sleep time with executive function. A) Summary of age and sex differences in the association of Physical Activity Scale for the Elderly (PASE) score with executive function. B) Summary of age and sex differences in the association of subjective total sleep time with executive function.
      Fig. 5
      Fig. 5Summary of age and sex differences in the associations of physical activity and subjective total sleep time with verbal fluency. A) Summary of age and sex differences in the association of Physical Activity Scale for the Elderly (PASE) score with verbal fluency. B) Summary of age and sex differences in the association of subjective total sleep time with verbal fluency.
      Table 4Summary of variance explained for each model examining whether physical activity or total sleep time (TST) moderate the relationships between age and cognition, and whether these relationships are sex dependent.
      VariableMemoryExecutive functionVerbal fluency
      Partial R2p-Value
      All p-values adjusted for false discovery rate (FDR).
      Partial R2p-Value
      All p-values adjusted for false discovery rate (FDR).
      Partial R2p-Value
      All p-values adjusted for false discovery rate (FDR).
      Physical activity (PASE score)Age9 %<0.000117 %<0.000111 %<0.0001
      Sex6 %<0.00011 %<0.00011 %<0.0001
      PASE score<1 %0.0077<1 %<0.0001<1 %<0.0001
      Age ∗ sex<1 %<0.0001<1 %<0.0001<1 %0.0005
      Age ∗ PASE score<1 %0.1702<1 %0.3900<1 %0.8904
      Sex ∗ PASE score<1 %0.2560<1 %0.7567<1 %0.9039
      Age ∗ Sex ∗ PASE score<1 %0.4328<1 %0.7180<1 %0.9039
      Total sleep timeAge9 %<0.000117 %<0.000111 %<0.0001
      Sex6 %<0.00011 %<0.00011 %<0.0001
      TST<1 %0.0004<1 %<0.0001<1 %<0.0001
      Age ∗ sex<1 %<0.0001<1 %<0.0001<1 %<0.0001
      Age ∗ TST<1 %0.0123<1 %0.0181<1 %0.0004
      Sex ∗ TST<1 %0.2093<1 %0.0237<1 %0.0297
      Age ∗ sex ∗ TST<1 %0.1078<1 %0.0134<1 %0.0086
      All models controlled for body mass index, educational attainment, income, Center for Epidemiologic Studies Short Depression Scale, test language used, and activities of daily living.
      a All p-values adjusted for false discovery rate (FDR).
      Models examining the association of TST with cognition explained 23–34 % of the variance; age was the most important predictor in each of these models (R2 range: 9–17 %; p < 0.001), while sex was the second most important contributor for the model of Memory (R2 = 6 %; p < 0.001), and the fifth most important for Executive Function (R2 = 1 %; p < 0.001) and Verbal Fluency (R2 = 1 %; p < 0.001). TST was the seventh most important predictor for Memory (R2 < 1 %; p < 0.001) and Executive Function (R2 < 1 %; p < 0.001), and sixth for Verbal Fluency (R2 < 1 %; p < 0.001). Age by TST interactions was the ninth most important predictor for Memory (R2 < 1 %; p = 0.0123) and ninth for Verbal Fluency (R2 < 1 %; p < 0.001); the interaction of age by TST was the eleventh most important contributor to Executive Function (R2 < 1 %; p = 0.0181). Sex by TST interactions were the eleventh most important predictor for Memory (R2 < 1 %; p = 0.201), and did not contribute significantly to the model. The interactions of sex and TST significantly contributed to our models of Executive Function and Verbal Fluency, and were the thirteenth (R2 < 1 %; p = 0.024) and eleventh most important predictors (R2 < 1 %; p = 0.030), respectively. The three-way interaction of age, sex, and TST was the twelfth most important predictor for Memory (R2 < 1 %; p = 0.108), but did not significantly contribute to the model. Three-way interactions of age, sex, and TST was fourteenth most important predictor for Executive Function (R2 < 1 %; p = 0.013), and the twelfth most important predictor for Verbal Fluency (R2 < 1 %; p = 0.009). As described in Fig. 3B, longer TST was associated with better memory performance for both males and females in middle adulthood; however, this association decreased in strength with age. For females, longer TST was associated with poorer executive function (Fig. 4B) and verbal fluency performance (Fig. 5B) with increasing age. Males with longer TST had poorer executive function performance in middle adulthood, but this association attenuated with age. Longer TST was also associated with verbal fluency performance for males with increasing age; however, this relationship was more modest in strength than for females.

      4. Discussion

      The results of our study suggest three important findings about the relationships of PA and TST with age- and sex-associated differences in cognition. First, both PA and TST modestly contribute to multiple domains of cognition across middle and older adulthood. Second, the association of PA with cognition does not appear to vary across middle or older adulthood, nor does it vary by biological sex. Finally, TST appears to have a complex relationship with multiple domains of cognition which is both age- and sex-dependent.
      Greater PA was associated with better cognitive performance across middle and older adulthood for both males and females, and there were no sex-specific differences in this relationship. In the context of our results, it is possible that maintaining PA level as adults age is more important for females, given that our results also show that females are less physically active than males across middle and older adulthood. However, our results also indicate that greater PA is beneficial for cognition across middle and older adulthood irrespective of biological sex. Future work is thus still needed to determine whether PA has differential benefits for male and female cognition.
      Our study confirms and extends previous research which has used the PASE to examine the associations between PA and cognitive function [
      • Ottenbacher A.J.
      • Snih S.A.
      • Bindawas S.M.
      • Markides K.S.
      • Graham J.E.
      • Samper-Ternent R.
      • Raji M.
      • Ottenbacher K.J.
      Role of physical activity in reducing cognitive decline in older mexican-american adults.
      ,
      • Roth D.L.
      • Goode K.T.
      • Clay O.J.
      • Ball K.K.
      Association of physical activity and visual attention in older adults.
      ,
      • Eggermont L.H.
      • Milberg W.P.
      • Lipsitz L.A.
      • Scherder E.J.
      • Leveille S.G.
      Physical activity and executive function in aging: the MOBILIZE Boston study.
      ,
      • Damsbo A.G.
      • Mortensen J.K.
      • Kraglund K.L.
      • Johnsen S.P.
      • Andersen G.
      • Blauenfeldt R.A.
      Prestroke physical activity and poststroke cognitive performance.
      ]. The findings of these studies are consistent with our own, whereby higher PASE score was associated with better cognitive performance. Importantly, our study is by far the largest investigation of the relationship between PASE score and cognition (sample sizes of previous studies ranged from 140 to 1669 participants), and our study is also the first to examine whether the relationships of PA with cognition vary by age and biological sex in a large sample of middle and older aged males and females.
      Although there are known sex differences in the association between sleep and cognition [
      • Hajali V.
      • Andersen M.L.
      • Negah S.S.
      • Sheibani V.
      Sex differences in sleep and sleep loss-induced cognitive deficits: the influence of gonadal hormones.
      ], our results are unanticipated. Males and females differ in how sleep elicits memory consolidation [
      • McDevitt E.A.
      • Rokem A.
      • Silver M.A.
      • Mednick S.C.
      Sex differences in sleep-dependent perceptual learning.
      ,
      • Genzel L.
      • Kiefer T.
      • Renner L.
      • Wehrle R.
      • Kluge M.
      • Grözinger M.
      • Steiger A.
      • Dresler M.
      Sex and modulatory menstrual cycle effects on sleep related memory consolidation.
      ], and the impact of sleep on cognition for females appears to be dependent upon the phase of the menstrual cycle [
      • Baker F.C.
      • Sattari N.
      • de Zambotti M.
      • Goldstone A.
      • Alaynick W.A.
      • Mednick S.C.
      Impact of sex steroids and reproductive stage on sleep-dependent memory consolidation in women.
      ]. The adverse effects of poor sleep on cognition may also be sex-dependent [
      • Hajali V.
      • Andersen M.L.
      • Negah S.S.
      • Sheibani V.
      Sex differences in sleep and sleep loss-induced cognitive deficits: the influence of gonadal hormones.
      ], whereby men are less susceptible to the consequences of poor sleep than females. In a sample of 2413 Taiwanese older adults, prolonged TST (i.e., >8.5 h) was correlated with cognitive impairment in both men and women [
      • Chiu H.Y.
      • Lai F.C.
      • Chen P.Y.
      • Tsai P.S.
      Differences between men and women aged 65 and older in the relationship between self-reported sleep and cognitive impairment: a nationwide survey in Taiwan.
      ]. By comparison, our results suggest longer TST is associated with better memory in middle adulthood for both males and females, although this relationship weakened with age. Longer TST was also associated with poorer executive function and verbal fluency as females aged. For males, longer TST was associated with poorer executive function in middle adulthood, but this association attenuated with age. Males with longer TST had poorer verbal fluency as they aged, but this association was less pronounced in males than in females. Our results thus highlight a complex relationship between TST and age-associated cognition, which differs according to biological sex. Given the limited data exploring these phenomena, we highlight the need for more research on sex-specific differences in the association of sleep with cognition.
      Our study thus provides further evidence that PA and TST share a complex relationship with each other and cognitive performance, whereby PA and TST contribute to multiple domains of cognition across middle and older adulthood. However, the relationships which we highlight in this paper are only the tip of the iceberg. Preliminary data indicate PA and sleep quality (particularly sleep efficiency) are associated with cognition through shared or separate pathways [
      • Sewell K.R.
      • Erickson K.I.
      • Rainey-Smith S.R.
      • Peiffer J.J.
      • Sohrabi H.R.
      • Brown B.M.
      Relationships between physical activity, sleep and cognitive function: a narrative review.
      ]. There is also a growing recognition that PA and sleep are two components of a larger 24-hour activity cycle – consisting of PA, sleep, sedentary behaviour, and standing behaviour – which can impact cognitive health [
      • Rosenberger M.E.
      • Fulton J.E.
      • Buman M.P.
      • Troiano R.P.
      • Grandner M.A.
      • Buchner D.M.
      • Haskell W.L.
      The 24-hour activity cycle: a new paradigm for physical activity.
      ,
      • Falck R.S.
      • Davis J.C.
      • Khan K.M.
      • Handy T.C.
      • Liu-Ambrose T.
      A wrinkle in measuring time use for cognitive health: how should we measure physical activity, sedentary behaviour and sleep?.
      ]. We therefore suggest that future investigations should consider the roles of PA and sleep (both quantity and quality) within the context of the 24-hour activity cycle, as well how other 24-hour activity cycle behaviours (i.e., standing and sedentary behaviour) are associated with cognitive health as adults age.

      4.1 Limitations

      Our cross-sectional analysis cannot examine temporal changes in cognition, PA, or TST. We used a structural equation model to develop our cognitive scores, but this structure may only be suitable for this sample. We also assumed that PA, TST, and cognition do not have a strictly linear relationship with age or with each other, which may have influenced our results. We also excluded individuals with comorbidities closely linked with cognitive impairment (e.g., stroke or Parkinson's disease), but we did not exclude participants with other conditions that are linked with poorer cognitive health – such as diabetes or heart disease.
      While the PASE has good evidence of validity and reliability for measuring PA, there are not currently norm-referenced criterions for what constitutes “high” or “low” PASE score, and thus it is difficult to speculate about the meaning of a given PASE score. Measuring self-reported TST is a common method in epidemiological research; however, the agreement of self-reported and device-measured TST is poor [
      • Girschik J.
      • Fritschi L.
      • Heyworth J.
      • Waters F.
      Validation of self-reported sleep against actigraphy.
      ]. Measuring PA using only self-report (e.g., PASE) is also prone to issues of reporting bias and accuracy [
      • Falck R.S.
      • McDonald S.M.
      • Beets M.W.
      • Brazendale K.
      • Liu-Ambrose T.
      Measurement of physical activity in older adult interventions: a systematic review.
      ]. We are also at risk for measurement error in our estimates of PA since the PASE was not designed to measure the PA of adults under the age of 55 years [
      • Washburn R.A.
      • Smith K.W.
      • Jette A.M.
      • Janney C.A.
      The physical activity scale for the elderly (PASE): development and evaluation.
      ]; however, >70 % of our sample were aged 55+ years. Using device-based measures of PA and TST will be a critical next step to confirm our findings.

      5. Conclusion

      PA and TST share a complex relationship with each other and cognition, and these relationships are moderated by age and biological sex. Future work using device-based measures of PA and TST is needed to confirm these results.
      The following are the supplementary data related to this article.
      • Supplementary material S1

        Model fit statistics and standardized path estimates for model of cognitive function. A) Model fit statistics for single factor, two factor, and three factor model of cognitive function from the comprehensive cohort. B) Final structural equation models for characterizing the cognitive function of participants in the comprehensive cohort.

      • Supplementary material S2

        Summary of age by sex related differences for models of Memory (Panel A), Executive Function (Panel B), and Verbal Fluency (Panel C) of participants with healthy cognitive function in the comprehensive cohort (N = 20,307; 45–86 years).

      Contributors

      Ryan S. Falck helped develop the study concept and design, and performed all data analysis and wrote the first draft of the manuscript.
      John R. Best helped develop the study concept and design, and provided key edits and wrote portions of the manuscript.
      Cindy K. Barha provided key edits and wrote portions of the manuscript.
      Jennifer C. Davis provided key edits and wrote portions of the manuscript.
      Teresa Liu-Ambrose helped develop the study concept and design, and provided key edits and wrote portions of the manuscript.

      Funding

      Funding for this work was provided by the Canadian Institutes of Health Research (ACD-162982). Ryan S. Falck is a Michael Smith Foundation for Health Research BC Postdoctoral Fellow. Jennifer C. Davis is a Michael Smith Health Research BC Career Scholar and a Canada Research Chair (Tier 2) in Applied Health Economics. Cindy K. Barha is an Alzheimer's Association Postdoctoral Fellow. Teresa Liu-Ambrose is a Canadian Research Chair (Tier 2) in Physical Activity, Mobility, and Cognitive Health. This research was made possible using the data collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the CLSA is provided by the Government of Canada through the Canadian Institutes of Health Research (CIHR) under grant reference: LSA 94473 and the Canada Foundation for Innovation .

      Ethical approval

      This research was made possible using the data collected by the Canadian Longitudinal Study on Aging (CLSA). This research has been conducted using the CLSA Baseline Tracking Dataset version 3.4 and Comprehensive Dataset version 4.0 under Application ID 19CA014. The CLSA is led by Drs. Parminder Raina, Christina Wolfson and Susan Kirkland. All research conducted as part of the CLSA abides by the requirements of the Canadian Institutes of Health Research and relevant institutions for ethical conduct and privacy protection in health research. We received ethics approval from the University of British Columbia's Clinical Research Ethics Board (H19-01838). All subjects gave written informed consent.

      Provenance and peer review

      This article was not commissioned and was externally peer reviewed.

      Research data (data sharing and collaboration)

      There are no linked research data sets for this paper. These data are the property of the Canadian Longitudinal Study on Aging. Data are available from the CLSA (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.

      Declaration of competing interest

      The authors declare that they have no competing interest.

      Acknowledgements

      Funding for this work was provided by the Canadian Institutes of Health Research ( ACD-162982 ). RSF is a Michael Smith Foundation for Health Research Postdoctoral Fellow. JCD is a Michael Smith Foundation for Health Research Career Scholar and a Canada Research Chair (Tier 2) in Applied Health Economics. CKB is an Alzheimer's Association and Brain Canada Postdoctoral Fellow. TLA is a Canadian Research Chair (Tier 1) in Healthy Aging. This research was made possible using the data collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the CLSA is provided by the Government of Canada through the Canadian Institutes of Health Research (CIHR) under grant reference: LSA 94473 and the Canada Foundation for Innovation . This research has been conducted using the CLSA Baseline Tracking Dataset version 3.4 and Comprehensive Dataset version 4.0 under Application ID 19CA014. The CLSA is led by Drs. Parminder Raina, Christina Wolfson and Susan Kirkland. The opinions expressed in this manuscript are the author's own and do not reflect the views of the CLSA. Data are available from the CLSA (www.clsa-elcv.ca) for researchers who meet the criteria for access to de-identified CLSA data.

      References

        • Salthouse T.A.
        When does age-related cognitive decline begin?.
        Neurobiol. Aging. 2009; 30: 507-514
        • Glisky E.L.
        Changes in cognitive function in human aging.
        in: Brain Aging: Models, Methods, and Mechanisms. 2007: 3-20
        • Liu-Ambrose T.
        • Falck R.S.
        Sleep, physical activity, and cognitive health in older adults.
        in: Handbook of Behavioral Neuroscience. Elsevier, 2019: 665-676
        • Barha C.K.
        • Hsu C.L.
        • Ten Brinke L.
        • Liu-Ambrose T.
        Biological sex: a potential moderator of physical activity efficacy on brain health.
        Front. Aging Neurosci. 2019; 11: 329
        • Hajali V.
        • Andersen M.L.
        • Negah S.S.
        • Sheibani V.
        Sex differences in sleep and sleep loss-induced cognitive deficits: the influence of gonadal hormones.
        Horm. Behav. 2019; 108: 50-61
        • Hamer M.
        • Chida Y.
        Physical activity and risk of neurodegenerative disease: a systematic review of prospective evidence.
        Psychol. Med. 2009; 39: 3-11
        • Troiano R.P.
        • Berrigan D.
        • Dodd K.W.
        • Masse L.C.
        • Tilert T.
        • McDowell M.
        Physical activity in the United States measured by accelerometer.
        Med. Sci. Sports Exerc. 2008; 40: 181-188
        • Erickson K.I.
        • Hillman C.
        • Stillman C.M.
        • Ballard R.M.
        • Bloodgood B.
        • Conroy D.E.
        • Macko R.
        • Marquez D.X.
        • Petruzzello S.J.
        • Powell K.E.
        Physical activity, cognition, and brain outcomes: a review of the 2018 physical activity guidelines.
        Med. Sci. Sports Exerc. 2019; 51: 1242
        • Lowe C.J.
        • Safati A.
        • Hall P.A.
        The neurocognitive consequences of sleep restriction: a meta-analytic review.
        Neurosci. Biobehav. Rev. 2017; 80: 586-604
        • Ohayon M.M.
        • Carskadon M.A.
        • Guilleminault C.
        • Vitiello M.V.
        Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan.
        Sleep. 2004; 27: 1255-1273
        • Scullin M.K.
        • Bliwise D.L.
        Sleep, cognition, and normal aging: integrating a half century of multidisciplinary research.
        Perspect. Psychol. Sci. 2015; 10: 97-137
        • Devore E.E.
        • Grodstein F.
        • Schernhammer E.S.
        Sleep duration in relation to cognitive function among older adults: a systematic review of observational studies.
        Neuroepidemiology. 2016; 46: 57-78
        • Yaffe K.
        • Falvey C.M.
        • Hoang T.
        Connections between sleep and cognition in older adults.
        Lancet Neurol. 2014; 13: 1017-1028
        • Wei J.
        • Hou R.
        • Xie L.
        • Chandrasekar E.K.
        • Lu H.
        • Wang T.
        • Li C.
        • Xu H.
        Sleep, sedentary activity, physical activity, and cognitive function among older adults: the National Health and nutrition examination survey, 2011–2014.
        J. Sci. Med. Sport. 2021; 24: 189-194
        • Andreano J.M.
        • Cahill L.
        Sex influences on the neurobiology of learning and memory.
        Learn. Mem. 2009; 16: 248-266
        • Krell-Roesch J.
        • Syrjanen J.A.
        • Bezold J.
        • Trautwein S.
        • Barisch-Fritz B.
        • Boes K.
        • Woll A.
        • Forzani E.
        • Kremers W.K.
        • Machulda M.M.
        Physical activity and trajectory of cognitive change in older persons: Mayo Clinic study of aging.
        J. Alzheimers Dis. 2021; 79: 377-388
        • Raina P.S.
        • Wolfson C.
        • Kirkland S.A.
        • Griffith L.E.
        • Oremus M.
        • Patterson C.
        • Tuokko H.
        • Penning M.
        • Balion C.M.
        • Hogan D.
        The Canadian longitudinal study on aging (CLSA).
        Can. J. Aging. 2009; 28: 221-229
        • Fillenbaum G.G.
        • Smyer M.A.
        The development, validity, and reliability of the OARS multidimensional functional assessment questionnaire.
        J. Gerontol. 1981; 36: 428-434
        • Andresen E.M.
        • Malmgren J.A.
        • Carter W.B.
        • Patrick D.L.
        Screening for depression in well older adults: evaluation of a short form of the CES-D.
        Am. J. Prev. Med. 1994; 10: 77-84
        • Falck R.S.
        • Best J.R.
        • Davis J.C.
        • Barha C.K.
        • Khan K.M.
        • Liu-Ambrose T.
        Cardiometabolic risk, biological sex, and age do not share an interactive relationship with cognitive function: a cross-sectional analysis of the Canadian longitudinal study on aging.
        Appl. Physiol. Nutr. Metab. 2021; 47: 405-414
        • Tuokko H.
        • Griffith L.E.
        • Simard M.
        • Taler V.
        Cognitive measures in the Canadian longitudinal study on aging.
        Clin. Neuropsychol. 2017; 31: 233-250
        • Washburn R.A.
        • Smith K.W.
        • Jette A.M.
        • Janney C.A.
        The physical activity scale for the elderly (PASE): development and evaluation.
        J. Clin. Epidemiol. 1993; 46: 153-162
        • Washburn R.
        • Ficker J.
        Physical activity scale for the elderly (PASE): the relationship with activity measured by a portable accelerometer.
        J. Sports Med. Phys. Fitness. 1999; 39: 336
        • Lo J.C.
        • Groeger J.A.
        • Cheng G.H.
        • Dijk D.-J.
        • Chee M.W.
        Self-reported sleep duration and cognitive performance in older adults: a systematic review and meta-analysis.
        Sleep Med. 2016; 17: 87-98
        • Lauderdale D.S.
        • Knutson K.L.
        • Yan L.L.
        • Liu K.
        • Rathouz P.J.
        Self-reported and measured sleep duration: how similar are they?.
        Epidemiology. 2008; : 838-845
        • Croxford R.
        Restricted Cubic Spline Regression: A Brief Introduction. Institute for Clinical Evaluative Sciences, Toronto2016: 1-5
        • Hirshkowitz M.
        • Whiton K.
        • Albert S.M.
        • Alessi C.
        • Bruni O.
        • DonCarlos L.
        • Hazen N.
        • Herman J.
        • Hillard P.J.A.
        • Katz E.S.
        National Sleep Foundation’s updated sleep duration recommendations.
        Sleep Health. 2015; 1: 233-243
        • Ottenbacher A.J.
        • Snih S.A.
        • Bindawas S.M.
        • Markides K.S.
        • Graham J.E.
        • Samper-Ternent R.
        • Raji M.
        • Ottenbacher K.J.
        Role of physical activity in reducing cognitive decline in older mexican-american adults.
        J. Am. Geriatr. Soc. 2014; 62: 1786-1791
        • Roth D.L.
        • Goode K.T.
        • Clay O.J.
        • Ball K.K.
        Association of physical activity and visual attention in older adults.
        J. Aging Health. 2003; 15: 534-547
        • Eggermont L.H.
        • Milberg W.P.
        • Lipsitz L.A.
        • Scherder E.J.
        • Leveille S.G.
        Physical activity and executive function in aging: the MOBILIZE Boston study.
        J. Am. Geriatr. Soc. 2009; 57: 1750-1756
        • Damsbo A.G.
        • Mortensen J.K.
        • Kraglund K.L.
        • Johnsen S.P.
        • Andersen G.
        • Blauenfeldt R.A.
        Prestroke physical activity and poststroke cognitive performance.
        Cerebrovasc. Dis. 2020; 49: 632-638
        • McDevitt E.A.
        • Rokem A.
        • Silver M.A.
        • Mednick S.C.
        Sex differences in sleep-dependent perceptual learning.
        Vis. Res. 2014; 99: 172-179
        • Genzel L.
        • Kiefer T.
        • Renner L.
        • Wehrle R.
        • Kluge M.
        • Grözinger M.
        • Steiger A.
        • Dresler M.
        Sex and modulatory menstrual cycle effects on sleep related memory consolidation.
        Psychoneuroendocrinology. 2012; 37: 987-998
        • Baker F.C.
        • Sattari N.
        • de Zambotti M.
        • Goldstone A.
        • Alaynick W.A.
        • Mednick S.C.
        Impact of sex steroids and reproductive stage on sleep-dependent memory consolidation in women.
        Neurobiol. Learn. Mem. 2019; 160: 118-131
        • Chiu H.Y.
        • Lai F.C.
        • Chen P.Y.
        • Tsai P.S.
        Differences between men and women aged 65 and older in the relationship between self-reported sleep and cognitive impairment: a nationwide survey in Taiwan.
        J. Am. Geriatr. Soc. 2016; 64: 2051-2058
        • Sewell K.R.
        • Erickson K.I.
        • Rainey-Smith S.R.
        • Peiffer J.J.
        • Sohrabi H.R.
        • Brown B.M.
        Relationships between physical activity, sleep and cognitive function: a narrative review.
        Neurosci. Biobehav. Rev. 2021; 130: 369-378
        • Rosenberger M.E.
        • Fulton J.E.
        • Buman M.P.
        • Troiano R.P.
        • Grandner M.A.
        • Buchner D.M.
        • Haskell W.L.
        The 24-hour activity cycle: a new paradigm for physical activity.
        Med. Sci. Sports Exerc. 2019; 51: 454-464
        • Falck R.S.
        • Davis J.C.
        • Khan K.M.
        • Handy T.C.
        • Liu-Ambrose T.
        A wrinkle in measuring time use for cognitive health: how should we measure physical activity, sedentary behaviour and sleep?.
        Am. J. Lifestyle Med. 2021; (15598276211031495)https://doi.org/10.1177/15598276211031495
        • Girschik J.
        • Fritschi L.
        • Heyworth J.
        • Waters F.
        Validation of self-reported sleep against actigraphy.
        J. Epidemiol. 2012; 22: 462-468
        • Falck R.S.
        • McDonald S.M.
        • Beets M.W.
        • Brazendale K.
        • Liu-Ambrose T.
        Measurement of physical activity in older adult interventions: a systematic review.
        Br. J. Sports Med. 2015; 50: 464-470