Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Chasing time : the specific impacts and dynamic relationships of physical activity, sedentary behaviour,… Falck, Ryan Stanley 2020

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata

Download

Media
24-ubc_2020_may_falck_ryan.pdf [ 7.58MB ]
Metadata
JSON: 24-1.0389784.json
JSON-LD: 24-1.0389784-ld.json
RDF/XML (Pretty): 24-1.0389784-rdf.xml
RDF/JSON: 24-1.0389784-rdf.json
Turtle: 24-1.0389784-turtle.txt
N-Triples: 24-1.0389784-rdf-ntriples.txt
Original Record: 24-1.0389784-source.json
Full Text
24-1.0389784-fulltext.txt
Citation
24-1.0389784.ris

Full Text

 Chasing Time: The Specific Impacts and Dynamic Relationships of Physical Activity, Sedentary Behaviour, and Sleep on Older Adults’ Cognitive Health   by Ryan Stanley Falck    A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  The Faculty of Graduate and Postdoctoral Studies (Rehabilitation Sciences)  The University of British Columbia (Vancouver)  April 2020  ©Ryan Stanley Falck, 2020ii The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:  Chasing Time: The Specific Impacts and Dynamic Relationships of Physical Activity, Sedentary Behaviour, and Sleep on Older Adults’ Cognitive Health  submitted by Ryan Stanley Falck  in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Rehabilitation Sciences  Examining Committee: Teresa Liu-Ambrose, Department of Physical Therapy Supervisor  Karim Khan, Department of Family Practice Supervisory Committee Member  Todd C. Handy, Department of Psychology Supervisory Committee Member Jordan Guenette, Department of Physical Therapy University Examiner Tricia S. Tang, Department of Medicine University Examiner   Additional Supervisory Committee Members: Jennifer C. Davis, Faculty of Management Supervisory Committee Member iii Abstract Effective lifestyle and behavioural strategies which maintain the cognitive health of older adults with Mild Cognitive Impairment (MCI) – a transition stage between healthy cognition and dementia – are greatly needed. There are three time-use activity behaviours which all humans engage in daily: physical activity (PA), sedentary behaviour (SB), and sleep. Each time-use activity behaviour is linked to cognitive health, although the magnitude of these relationships are still uncertain. There is also preliminary evidence that these time-use activity behaviours share a complex and dynamic relationship with each other and cognitive health. Thus, the aim of this dissertation was two-fold: 1) to advance the current knowledge about the dynamic relationships between time-use activity behaviours and cognitive health; and 2) to characterize potential time-use activity behaviour intervention strategies for promoting cognitive health. Using a systematic review of observational studies, I showed SB is associated with poorer cognitive function. I next conducted three cross-sectional studies which found 1) the relationships of PA and SB with cognitive function differ by MCI status; 2) PA is associated with better cognitive function independent of any sleep index, while only sleep efficiency is associated with cognitive performance independent of PA; and 3) PA is associated with greater brain cortical thickness independent of SB, but SB is not associated with cortical thickness independent of PA. I then conducted a secondary analysis of a randomized controlled trial (RCT), where I found that while the intervention significantly increased older adult PA, it did not improve cognitive function. My final thesis study was a proof-of-concept RCT to examine the effects of multimodal chronotherapy to promote better sleep among older adults with MCI and poor sleep; I found the intervention improved subjective sleep, but did not improve objective sleep or cognitive function. The results of my thesis contribute to a better understanding of how time-use activity behaviours impact olderiv adult cognitive health, and helps to refine the public health message for best promoting healthy cognitive aging through lifestyle.   v Lay Summary Effective behavioural strategies to promote healthy cognitive aging are greatly needed. Three behaviours related to cognitive health which all adults engage in daily are: physical activity (PA), sedentary behaviour (SB), and sleep. My thesis examined how PA, SB, and sleep are dynamically related to each other and cognitive health. I found: 1) SB is associated with poorer cognition; 2) the relationships of PA and SB with cognition vary by cognitive status; 3) PA and sleep are each independently associated with cognition; 4) PA is associated with better brain structure independent of SB, but SB is not associated with brain structure; 5) increasing PA does not improve cognition in highly active community-dwelling older adults; and 6) a multimodal lifestyle intervention improves sleep but not cognition in older adults at risk for dementia. My thesis contributes to a better understanding of how PA, SB, and sleep impact older adult cognitive health.   vi Preface Content from this dissertation was written and compiled (for the published chapters) by Ryan Stanley Falck. Professor Teresa Liu-Ambrose, Professor Todd C. Handy, Professor Karim Khan, and Professor Jennifer C. Davis provided comments that were taken into consideration in generating the final version of the dissertation.  The research studies in Chapters 2-7 were primarily conducted in the Aging, Mobility, and Cognitive Neuroscience Laboratory at the Research Pavilion of the Vancouver General Hospital. All magnetic resonance imaging was conducted at the UBC MRI Research Centre. Ethics approval for all studies was approved by Univeristy of British Columbia’s Clinical Research Ethics Board (H14-01301; H14-01762; H16-01029). All research presented in this thesis has been published or is submitted for publication. Details for each publication is provided below.  Chapter 2 is based on work conducted in the Aging, Mobility, and Cognitive Neuroscience Laboratory by Ryan Stanley Falck, Professor Jennifer C. Davis (supervisory committee member), and Professor Teresa Liu-Ambrose (supervisor). I was responsible for writing the first draft of the manuscript, performing the search strategy, reviewing the articles, and drafting the tables. Professor Davis provided help with the article reviews and drafting tables. Professor Liu-Ambrose and Professor Davis conceived the study concept and design, and wrote portions of the manuscript and provided critical review of the manuscript. All authors approved of the final manuscript. A version of Chapter 2 has been published:  Falck RS, Davis JC, & Liu-Ambrose T. (2017). What is the association between sedentary behaviour and cognitive function? A systematic review. British Journal of Sports Medicine, 51(10), 800-811.  Chapter 3 is based on work conducted in the Aging, Mobility, and Cognitive Neuroscience Laboratory by Ryan Stanley Falck, Dr. Glenn J. Landry, Dr. John R. Best, Professor Jennifer C. Davis, Mr. Bryan K. Chiu, and Professor Teresa Liu-Ambrose. Dr. Best, Dr. Landry, Professor Davis, Professor Liu-Ambrose, and I were each responsible for the concept and research design of the study. I wrote the first draft of the manuscript and performed all statistical analyses, and received critical review of the manuscript from Dr. Best, Professor Davis, Mr. Chiu, and Professor Liu-Ambrose. All authors approved of the final manuscript. A version of Chapter 3 has been published: Falck RS, Landry GJ, Best JR, Davis JC, Chiu BK, & Liu-Ambrose T. (2017). Cross-Sectional Relationships of Physical Activity and Sedentary Behavior With Cognitive Function in Older Adults With Probable Mild Cognitive Impairment. Physical Therapy, 97(10), 975-984.  Chapter 4 is based on work conducted in the Aging, Mobility, and Cognitive Neuroscience Laboratory by Ryan Stanley Falck, Dr. John R. Best, Professor Jennifer C. Davis, and Professor Teresa Liu-Ambrose. Professor Liu-Ambrose and I were responsible for the concept and research vii design of the study. I wrote the first draft of the manuscript and performed all statistical analyses, and received critical review of the manuscript from Dr. Best, Professor Davis, and Professor Liu-Ambrose. All authors approved of the final manuscript. A version of Chapter 4 has been published: Falck, RS, Best JR, Davis JC, & Liu-Ambrose T. (2018). The independent associations of physical activity and sleep with cognitive function in older adults. Journal of Alzheimer's Disease, 63(4), 1469-1484.  Chapter 5 is based on work conducted in collaboration between the Aging, Mobility, and Cognitive Neuroscience Laboratory and the Arthritis Research Centre by Ryan Stanley Falck, Dr. Chun Liang Hsu, Dr. John R. Best, Professor Linda C. Li, Dr. Anna R. Egbert, and Professor Teresa Liu-Ambrose. Dr. Best, Professor Liu-Ambrose, and I conceived the study concept and design. Dr. Best and I collected the data. Dr.’s Hsu and Egbert and I performed the data analyses and interpreted the results. I wrote the first draft of the manuscript, and Dr.’s Best, Hsu, and Egbert, and Professor’s Li and Liu-Ambrose wrote portions of the manuscript and provided critical review. All authors approved of the final manuscript. A version of Chapter 5 is accepted for publication in Medicine and Science in Sports and Exercise.   Chapter 6 is based on work conducted in collaboration between the Aging, Mobility, and Cognitive Neuroscience Laboratory and the Arthritis Research Centre by Ryan Stanley Falck, Dr. John R. Best, Professor Linda C. Li, Mr. Patrick C.Y. Chan, Dr. Lynne M. Feehan, and Professor Teresa Liu-Ambrose. Professor Liu-Ambrose and Professor Li were responsible for the concept and research design of the study. I wrote the first draft of the manuscript. Dr. Best and I performed the data analysis and interpreted the results. Dr. Best, Professor Li, Mr. Chan, Dr. Feehan, and Professor Liu-Ambrose wrote portions of the manuscript and provided critical review. All authors approved of the final manuscript. A version of Chapter 6 has been published: Falck RS, Best JR., Li LC, Chan PCY, Feehan LM, & Liu-Ambrose T. (2018). Can we improve cognitive function among adults with osteoarthritis by increasing moderate-to-vigorous physical activity and reducing sedentary behaviour? Secondary analysis of the MONITOR-OA study. BMC Musculoskeletal Disorders, 19(1), 447.  Chapter 7 is based on work conducted in the Aging, Mobility, and Cognitive Neuroscience Laboratory by Ryan Stanley Falck, Professor Jennifer C. Davis, Dr. John R. Best, Mr. Patrick C.Y. Chan, Professor Linda C. Li, Ms. Anne B. Wyrough, Ms. Kimberly J. Bennet, Mr. Daniel Backhouse, and Professor Teresa Liu-Ambrose. Professor Davis, Professor Liu-Ambrose, and I were responsible for the concept and research design of the study. I wrote the first draft of the manuscript and performed all statistical analyses with the help of Dr. Best. Professor Davis, Dr. Best, Professor Li, and Professor Liu-Ambrose each provided critical review of the manuscript. All authors approved of the final manuscript. Chapter 7 is currently under major revision in Journal of Alzheimer’s Disease.   viii Table of Contents Abstract ......................................................................................................................................... iii Lay Summary ................................................................................................................................ v Preface ........................................................................................................................................... vi Table of Contents ....................................................................................................................... viii List of Tables ............................................................................................................................ xviii List of Figures ............................................................................................................................. xix Acknowledgements .................................................................................................................... xxi Dedication .................................................................................................................................. xxii Chapter 1: Introduction ............................................................................................................... 1 1.1 Preamble ........................................................................................................................... 1 1.2 The cognitive health of older adults: A growing public health challenge ....................... 3 1.2.1 Defining cognitive health .......................................................................................... 4 1.2.1.1 Normal cognitive aging ..................................................................................... 4 1.2.1.2 Cognitive changes in Mild Cognitive Impairment and dementia ...................... 5 1.2.2 Measuring cognitive health ..................................................................................... 10 1.2.2.1 Neurophysiological biomarkers ....................................................................... 10 1.2.2.2 Structural neuroimaging .................................................................................. 10 1.2.2.3 Functional neuroimaging ................................................................................. 11 1.2.2.4 Neuropsychological Testing ............................................................................ 12 1.2.3 Summary ................................................................................................................. 13 1.3 Time-use activity behaviours and their impact on older adult cognitive health: Physical  activity, sedentary behaviour, and sleep......................................................................... 14 1.3.1 Physical activity and older adult cognitive health .................................................. 16 1.3.1.1 Physical activity definitions............................................................................. 16 1.3.1.2 Physical activity measurement ........................................................................ 19 1.3.1.2.A Subjective and objective measures of physical activity .............................. 20 1.3.1.2.B Important psychometric properties of physical activity measurement:  validity, reliability, population specificity, and sensitivity to change ......... 23 ix 1.3.1.2.C Measuring physical activity using MotionWatch8 wrist-worn actigraphy, the  SenseWear Mini Armband multimodal sensor, and the Community Healthy  Activities Model Program for Seniors ......................................................... 25 1.3.1.2.D Summary ...................................................................................................... 27 1.3.1.3 Current evidence examining the impact of physical activity on older adult  cognitive health................................................................................................ 27 1.3.1.3.A Impact of lifestyle physical activity on older adult cognitive health ........... 28 1.3.1.3.B Impact of exercise training on older adult cognitive health ......................... 33 1.3.1.3.C Mechanisms by which physical activity and exercise training can impact  older adult cognitive health .......................................................................... 42 1.3.1.4 Summary of the current evidence examining how physical activity impacts  older adult cognitive health ............................................................................. 50 1.3.2 Sedentary behaviour and older adult cognitive health ............................................ 51 1.3.2.1 Definition of sedentary behaviour ................................................................... 51 1.3.2.2 Sedentary behaviour measurement .................................................................. 52 1.3.2.3 Current evidence examining the impact of sedentary behaviour on older adult  cognitive health................................................................................................ 53 1.3.2.3.A Mechanism by which sedentary behaviour impacts cognitive health ......... 56 1.3.2.4 Summary of the current evidence for how sedentary behaviour can impact  older adult cognitive health ............................................................................. 56 1.3.3 Sleep and older adult cognitive health .................................................................... 56 1.3.3.1 Defining sleep .................................................................................................. 56 1.3.3.2 Sleep and circadian rhythms measurement ..................................................... 61 1.3.3.2.A Measuring sleep ........................................................................................... 61 1.3.3.2.B Measuring circadian rhythms ....................................................................... 65 1.3.3.3 Sleep and circadian rhythms in normal aging ................................................. 67 1.3.3.4 Sleep, circadian rhythms, and older adult cognitive health ............................. 69 1.3.3.5 Summary of the current evidence for how sleep and circadian rhythms can  impact older adult cognitive health ................................................................. 71 1.3.4 Summary of the current evidence for how each time-use activity behaviours  impacts older adult cognitive health ....................................................................... 71 1.4 The dynamic relationships of time-use activity behaviours and circadian rhythms with  older adult cognitive health ............................................................................................ 72 x 1.4.1 Concurrent measurement and analysis of multiple time-use activity behaviours  and/or circadian rhythms......................................................................................... 72 1.4.2 Current evidence on the dynamic relationships between time-use activity  behaviours and older adult cognitive health ........................................................... 74 1.4.3 Current evidence on the dynamic relationships between time-use activity  behaviours and circadian rhythms .......................................................................... 76 1.4.4 Summary of the current evidence examining the relationships between time-use  activity behaviours, circadian rhythms, and cognitive health ................................. 78 1.5 Current strategies to promote older adult physical activity, reduce sedentary behaviour,  and improve sleep and circadian rhythms ...................................................................... 78 1.5.1 Strategies to promote physical activity and reduce sedentary behaviour ............... 78 1.5.2 Strategies for improving sleep and circadian rhythms ............................................ 82 1.5.2.1 Cognitive behavioural therapy and sleep hygiene education .......................... 83 1.5.2.2 Physical activity and exercise training ............................................................ 84 1.5.2.3 Chronotherapy ................................................................................................. 84 1.5.2.3.A Bright light therapy ...................................................................................... 85 1.5.2.3.B Physical activity and exercise training ......................................................... 87 1.5.3 Summary of current strategies to promote physical activity, reduce sedentary  behaviour, and improve sleep and circadian rhythms ............................................. 88 1.6 Summary of the current research gaps ........................................................................... 89 1.6.1 The impact of physical activity on cognitive health ............................................... 89 1.6.2 The impact of sedentary behaviour on cognitive health ......................................... 90 1.6.3 The impact of sleep and circadian rhythms on cognitive health ............................. 90 1.6.4 The dynamic relationships of time-use activity behaviours with older adult  cognitive health ....................................................................................................... 91 1.6.5 Strategies to promote physical activity, reduce sedentary behaviour, and improve  sleep ........................................................................................................................ 92 1.7 Thesis overview.............................................................................................................. 93 1.7.1 Main thesis questions .............................................................................................. 93 1.7.2 Methodology ........................................................................................................... 94 1.7.2.1 Physical activity measures ............................................................................... 94 1.7.2.2 Sedentary behaviour measures ........................................................................ 94 1.7.2.3 Sleep measures ................................................................................................ 94 xi 1.7.2.4 Cognitive function measures ........................................................................... 96 1.7.2.5 Brain structure measures ................................................................................. 97 1.7.3 Overview of thesis chapters .................................................................................... 97 Chapter 2: What is the association of sedentary behaviour and cognitive function? A systematic review ...................................................................................................................... 102 2.1 Introduction .................................................................................................................. 102 2.2 Methods ........................................................................................................................ 103 2.2.1 Summary of search strategy .................................................................................. 103 2.2.2 Study selection ...................................................................................................... 104 2.2.3 Inclusion and exclusion criteria ............................................................................ 104 2.2.4 Data extraction ...................................................................................................... 105 2.2.5 Assessment of study quality.................................................................................. 106 2.3 Results .......................................................................................................................... 107 2.3.1 Search results and study characteristics ................................................................ 107 2.3.2 Measurement of sedentary behaviour ................................................................... 112 2.3.3 Measurement of outcomes from sedentary behaviour .......................................... 117 2.3.3.1 Assessment of memory .................................................................................. 117 2.3.3.2 Assessment of executive function ................................................................. 121 2.3.3.3 Assessment of processing speed .................................................................... 121 2.3.3.4 Assessment of cognitive impairment and all-cause dementia incidence ....... 121 2.3.3.5 Assessment of perceptual organization and planning .................................... 121 2.3.3.6 Assessment of global cognitive function ....................................................... 122 2.3.4 Quality assessment ................................................................................................ 122 2.3.5 Findings from studies on the association of sedentary behaviour with cognitive  function ................................................................................................................. 125 2.3.5.1 Cohort designs ............................................................................................... 125 2.3.5.2 Case-control designs ...................................................................................... 125 2.3.5.3 Cross-sectional designs.................................................................................. 125 2.4 Discussion .................................................................................................................... 126 2.4.1 Summary of main findings.................................................................................... 126 2.4.2 Comparison of the findings with the literature ..................................................... 126 xii 2.4.3 Assessment of sedentary behaviour ...................................................................... 127 2.4.4 Assessment of cognitive function ......................................................................... 129 2.4.5 Study quality ......................................................................................................... 130 2.4.6 Recommendations ................................................................................................. 131 2.4.7 Limitations and future directions .......................................................................... 132 2.4.8 Conclusions ........................................................................................................... 133 Chapter 3: Cross-sectional relationships of physical activity and sedentary behavior with cognitive function in older adults with probable Mild Cognitive Impairment ................... 135 3.1 Introduction .................................................................................................................. 135 3.2 Methods ........................................................................................................................ 138 3.2.1 Protocol ................................................................................................................. 138 3.2.2 Participants ............................................................................................................ 138 3.2.3 Measurement of moderate-to-vigorous physical activity and sedentary        behaviour............................................................................................................... 139 3.2.4 Data Reduction...................................................................................................... 140 3.2.5 Measurement of moderate-to-vigorous physical activity and sedentary behaviour  bouts per day ......................................................................................................... 140 3.2.6 Cognitive Function................................................................................................ 141 3.2.7 Data Analyses ....................................................................................................... 142 3.2.7.1 Participant characteristics based on probable MCI status ............................. 142 3.2.7.2 Relationship of cognitive function with moderate-to-vigorous physical activity  and sedentary behaviour based on probable Mild Cognitive Impairment        status .............................................................................................................. 143 3.3 Results .......................................................................................................................... 143 3.3.1 Participant characteristics based on probable Mild Cognitive Impairment status 144 3.3.2 Association of cognitive function with moderate-to-vigorous physical activity and  sedentary behaviour .............................................................................................. 146 3.3.3 Relationship of cognitive function with moderate-to-vigorous physical activity and  sedentary behaviour based on probable Mild Cognitive Impairment status ......... 148 3.4 Discussion .................................................................................................................... 148 3.4.1 Differences in physical activity and sedentary behaviour by cognitive status—does  activity level change because of Mild Cognitive Impairment conversion? .......... 151 xiii 3.4.2 Differences in the relationships of physical activity and sedentary behaviour with  cognitive function by Mild Cognitive Impairment status—are there underlying  differences in the Mild Cognitive Impairment brain? ........................................... 152 3.4.3 Clinical applications.............................................................................................. 153 3.4.4 Limitations and future research ............................................................................ 154 3.4.5 Conclusions ........................................................................................................... 156 Chapter 4: The independent associations of physical activity and sleep with cognitive function in older adults ............................................................................................................. 157 4.1 Introduction .................................................................................................................. 157 4.2 Methods ........................................................................................................................ 160 4.2.1 Protocol ................................................................................................................. 161 4.2.2 Participants ............................................................................................................ 161 4.2.3 Subjective measurement of sleep quality .............................................................. 161 4.2.4 Objective measurement of physical activity and sleep ......................................... 162 4.2.4.1 Data Reduction .............................................................................................. 163 4.2.5 Cognitive Function................................................................................................ 164 4.2.6 Statistical Analyses ............................................................................................... 165 4.2.6.1 Preliminary Analyses ..................................................................................... 166 4.2.6.2 Main Analyses ............................................................................................... 166 4.3 Results .......................................................................................................................... 169 4.3.1 Participant Characteristics .................................................................................... 170 4.3.2 Preliminary Analyses ............................................................................................ 170 4.3.3 Main Analyses ...................................................................................................... 170 4.3.3.1 Association of physical activity, Pittsburgh Sleep Quality Index, and  Alzheimer’s Disease Assessment Scale Plus ................................................. 170 4.3.3.2 Association of physical activity, sleep fragmentation, and Alzheimer’s Disease  Assessment Scale Plus ................................................................................... 175 4.3.3.3 Association of physical activity, sleep efficiency, and Alzheimer’s Disease  Assessment Scale Plus ................................................................................... 177 4.3.3.4 Association of physical activity, sleep duration, and Alzheimer’s Disease  Assessment Scale Plus ................................................................................... 177 4.3.3.5 Association of physical activity, sleep latency, and Alzheimer’s Disease  Assessment Scale Plus ................................................................................... 180 xiv 4.4 Discussion .................................................................................................................... 180 4.4.1 The independent relationships of physical activity and sleep with cognitive  function ................................................................................................................. 182 4.4.2 The relationship of physical activity and sleep ..................................................... 183 4.4.3 Limitations and future research ............................................................................ 185 4.4.4 Conclusion ............................................................................................................ 187 Chapter 5: Not just for joints—the associations of moderate-to-vigorous physical activity and sedentary behaviour with brain cortical thickness......................................................... 188 5.1 Introduction .................................................................................................................. 188 5.2 Methods ........................................................................................................................ 190 5.2.1 Study design .......................................................................................................... 190 5.2.2 Participants ............................................................................................................ 191 5.2.3 Measures ............................................................................................................... 193 5.2.3.1 Demographics and anthropometrics .............................................................. 193 5.2.3.2 Moderate-to-vigorous physical activity and sedentary behaviour ................. 193 5.2.3.3 Magnetic resonance imaging data acquisition and FreeSurfer analyses ....... 193 5.3 Results .......................................................................................................................... 195 5.3.1 Participant characteristics ..................................................................................... 195 5.3.2 Independent relationships of moderate-to-vigorous physical activity and sedentary  behaviour with cortical thickness .......................................................................... 197 5.4 Discussion .................................................................................................................... 197 5.4.1 Limitations ............................................................................................................ 200 5.4.2 Conclusion ............................................................................................................ 202 Chapter 6: Can we improve cognitive function among adults with osteoarthritis by increasing moderate-to-vigorous physical activity and reducing sedentary behaviour? Secondary analysis of the MONITOR-OA study ................................................................... 204 6.1 Introduction .................................................................................................................. 204 6.2 Methods ........................................................................................................................ 207 6.2.1 Study design .......................................................................................................... 207 6.2.2 Participants ............................................................................................................ 207 6.2.3 Measures ............................................................................................................... 208 6.2.3.1 Demographics ................................................................................................ 208 xv 6.2.3.2 Moderate-to-vigorous physical activity and sedentary behaviour ................. 210 6.2.3.3 Cognitive function ......................................................................................... 210 6.2.4 Randomization ...................................................................................................... 211 6.2.5 Intervention ........................................................................................................... 211 6.2.6 Statistical analyses ................................................................................................ 212 6.2.6.1 Main analyses ................................................................................................ 212 6.2.6.2 Secondary analyses ........................................................................................ 213 6.3 Results .......................................................................................................................... 213 6.3.1 Participant characteristics ..................................................................................... 213 6.3.2 Changes in cognitive function .............................................................................. 214 6.3.3 Correlation between moderate-to-vigorous physical activity and sedentary  behaviour changes with changes in cognitive function ........................................ 217 6.4 Discussion .................................................................................................................... 217 6.4.1 Clinical applications.............................................................................................. 221 6.4.2 Limitations and future research ............................................................................ 222 6.4.3 Conclusion ............................................................................................................ 223 Chapter 7: Effect of multimodal personalized chronotherapy on sleep and cognitive function in older adults with probable Mild Cognitive Impairment and poor sleep: a randomized clinical trial........................................................................................................... 225 7.1 Introduction .................................................................................................................. 225 7.2 Methods ........................................................................................................................ 226 7.2.1 Study design and setting ....................................................................................... 226 7.2.2 Recruitment ........................................................................................................... 227 7.2.3 Inclusion criteria ................................................................................................... 227 7.2.4 Exclusion criteria .................................................................................................. 229 7.2.5 Randomization and blinding ................................................................................. 229 7.2.6 Interventions ......................................................................................................... 229 7.2.6.1 Multimodal chronotherapy (INT) .................................................................. 229 7.2.6.2 Education (CON) ........................................................................................... 234 7.2.7 Measures ............................................................................................................... 234 7.2.7.1 Primary outcome: objectively-measured sleep efficiency ............................. 234 7.2.7.2 Secondary sleep outcomes: objective and subjective sleep quality ............... 235 xvi 7.2.7.3 Secondary outcomes: cognitive function ....................................................... 235 7.2.7.4 Secondary outcomes: moderate-to-vigorous physical activity and sedentary  behaviour ....................................................................................................... 236 7.2.8 Sample size ........................................................................................................... 237 7.2.9 Compliance and intervention adherence ............................................................... 237 7.2.10 Adverse events ...................................................................................................... 237 7.2.11 Statistical analyses ................................................................................................ 237 7.2.11.1 Primary analyses ............................................................................................ 238 7.2.11.2 Secondary analyses ........................................................................................ 238 7.3 Results .......................................................................................................................... 239 7.3.1 Primary outcome: sleep efficiency (objective) ..................................................... 241 7.3.2 Secondary outcomes: objective and subjective sleep ........................................... 241 7.3.3 Secondary outcomes: cognitive function .............................................................. 241 7.3.4 Relationships between changes in sleep quality and changes in cognitive       function ................................................................................................................. 244 7.3.5 Compliance and intervention adherence ............................................................... 244 7.3.6 Adverse events ...................................................................................................... 244 7.3.7 Secondary outcomes: moderate-to-vigorous physical activity and sedentary  behaviour............................................................................................................... 246 7.4 Discussion .................................................................................................................... 246 7.4.1 Limitations ............................................................................................................ 249 7.4.2 Conclusion ............................................................................................................ 250 Chapter 8: General discussion ................................................................................................. 251 8.1 Study synopses ............................................................................................................. 251 8.2 Revisiting the main research questions ........................................................................ 254 8.2.1 Question #1: How are time-use activity behaviours associated with older adult  cognitive health? ................................................................................................... 254 8.2.1.1 Relationships of time-use activity behaviours with older adult cognitive     health ............................................................................................................. 255 8.2.1.2 Cognitive status as a moderator of the relationships of time-use activity  behaviours with older adult cognitive health ................................................. 263 8.2.2 Question #2: What is the dynamic relationship between time-use activity  behaviours and older adult cognitive health? ........................................................ 266 xvii 8.2.3 Can we promote older adult cognitive health through targeted interventions on  time-use activity behaviours? ............................................................................... 272 8.3 Limitations ................................................................................................................... 280 8.3.1 General limitations ................................................................................................ 280 8.3.2 Measurement of physical activity limitations ....................................................... 281 8.3.3 Measurement of sedentary behaviour limitations ................................................. 282 8.3.4 Measurement of sleep limitations ......................................................................... 283 8.3.5 Measurement of cognitive health limitations ........................................................ 284 8.3.6 Interpretation of the results using Darwinian Medicine ....................................... 285 8.4 Future directions ........................................................................................................... 286 8.4.1 The impact of each time-use activity behaviour on older adult cognitive health . 286 8.4.1.1 Physical activity ............................................................................................. 286 8.4.1.2 Sedentary behaviour ...................................................................................... 288 8.4.1.3 Sleep and circadian rhythms .......................................................................... 289 8.4.1.4 The dynamic relationships of time-use activity behaviours with cognitive  health ............................................................................................................. 290 8.5 Conclusion .................................................................................................................... 291 References .................................................................................................................................. 292 Appendix A: Cognitive domain criteria and classification of common cognitive function measures..................................................................................................................................... 345 Appendix B: Evidence of validity and reliability for the MotionWatch8© wrist-worn actigraph for measuring physical activity, sedentary behaviour, and sleep ....................... 347 Appendix C: Description of cognitive measures included in Alzheimer’s Disease Assessment Scale Cognitive Plus score ................................................................................... 349 Appendix D: Code used for analyses (Python, R, FreeSurfer, and Excel Macros) and results of analytical models in Chapters 3-7 ....................................................................................... 351 Appendix E: Model fit statistics for structural equation models in Chapter 4 ................... 352 Appendix F: Description of compliance measures in Chapter 7 .......................................... 353 Appendix G: Differences in sleep and intraclass reliability coefficients of the MotionWatch8 for older adults with and without Mild Cognitive Impairment ................. 356 Appendix H: Chapter 4 post-hoc analyses examining the independent relationships of sedentary behaviour and sleep quality with older adult cognitive function ........................ 359  xviii List of Tables Table 2.1 Study Characteristics .................................................................................................. 109 Table 2.2 Measures and methods to classify sedentary behaviour ............................................. 113 Table 2.3 Measures and methods for outcome assessment (i.e., cognitive function) ................. 118 Table 2.4 Quality assessment for studies on the relationship of sedentary behaviour with cognitive function ....................................................................................................................... 123 Table 3.1 Participant characteristics ........................................................................................... 145 Table 3.2 Association of sedentary behaviour and physical activity with Alzheimer’s Disease Assessment Scale Plus score ....................................................................................................... 146 Table 3.3 Association of moderate-to-vigorous physical activity and sedentary behaviour with Alzheimer’s Disease Assessment Scale Plus score based on probable Mild Cognitive Impairment (MCI) status ................................................................................................................................ 149 Table 4.1 Participant characteristics ........................................................................................... 171 Table 4.2 Correlation matrix ....................................................................................................... 172 Table 4.3 Structural equation model estimates ± s.e. ................................................................. 173 Table 5.1 Participant characteristics—mean (SD) or % ............................................................. 196 Table 6.1 Baseline characteristics ............................................................................................... 214 Table 6.2 Changes in cognitive function (Baseline – 2 Months) by treatment group ................ 216 Table 7.1 Participant characteristics at baseline (N= 96); mean (SD) ........................................ 240 Table 7.2 Estimated marginal means and standard errors for changes in sleep quality, cognitive function, moderate-to-vigorous physical activity, and sedentary behaviour at baseline, 12 weeks, and 24 weeks follow-up by group ............................................................................................... 242    xix List of Figures Figure 1.1 Hypothetical model for the biological cascade of dementia pathology ........................ 8 Figure 1.2 Characteristics and relationships of older adult cognitive health .................................. 9 Figure 1.3 Time-use activity behaviours are distinct but related, and are also part of circadian regulation ...................................................................................................................................... 15 Figure 1.4 Physical activity domains and intensities of PA.......................................................... 17 Figure 1.5 Dimensions of physical activity measurement ............................................................ 20 Figure 1.6 Characteristics and considerations for choosing a physical activity measure ............. 21 Figure 1.7 Hypothesized mechanism through which IGF-1 signaling may interface with BDNF-mediated synaptic plasticity in the hippocampus during exercise ................................................ 45 Figure 1.8 Different types of sedentary behaviour ....................................................................... 52 Figure 1.9 Measures of sedentary behaviour and their classifications ......................................... 54 Figure 1.10 Dimensions of sleep and their associations ............................................................... 58 Figure 1.11 Objective and subjective quality and their relationships to the dimensions of sleep 60 Figure 1.12 Measures of sleep and their classifications ............................................................... 63 Figure 1.13 Measures of circadian rhythms and their dimensions ............................................... 66 Figure 1.14 Overview of the Brief Action Planning Approach during initial and follow-up consultations ................................................................................................................................. 80 Figure 1.15   Brief Action Planning Approach for A) promoting physical activity and B) reducing sedentary behaviour ....................................................................................................... 81 Figure 1.16 Brief introduction to the neurophysiological response to light (Panel A) and its use in bright light therapy according to the phase response curve of light (Panel B) ............................. 86 Figure 1.17 Overview of the observational studies (Chapters 2-5) in the dissertation ................. 98 Figure 1.18 Overview of the experimental studies (Chapters 6 & 7) in the dissertation .............. 99 Figure 2.1 Study selection process and sample search strategy .................................................. 108 Figure 3.1 Association of moderate-to-vigorous physical activity and sedentary behaviour with cognitive function ....................................................................................................................... 147 Figure 3.2 Association of moderate-to-vigorous physical activity and sedentary behaviour with cognitive function based on the presence of Mild Cognitive Impairment .................................. 150 Figure 4.1 Hypothesized relationship between physical activity and sleep quality ................... 168 Figure 4.2 Structural equation models exploring the relationship of physical activity with Pittsburgh Sleep Quality Index ................................................................................................... 174 xx Figure 4.3 Structural equation models exploring the relationship of physical activity with sleep fragmentation .............................................................................................................................. 176 Figure 4.4 Structural equation models exploring the relationship of physical activity with sleep efficiency..................................................................................................................................... 178 Figure 4.5 Structural equation models exploring the relationship of physical activity with sleep duration ....................................................................................................................................... 179 Figure 4.6 Structural equation models exploring the relationship of physical activity with sleep latency ......................................................................................................................................... 181 Figure 5.1 STROBE diagram ...................................................................................................... 192 Figure 5.2 Significant multiple comparison corrected clusters of greater cortical thickness associated with higher moderate-to-vigorous physical activity independent of sedentary behaviour..................................................................................................................................... 198 Figure 6.1 CONSORT diagram .................................................................................................. 209 Figure 6.2 Changes in cognitive performance by treatment group (Baseline – 2 Months) ........ 215 Figure 6.3 Relationship between intervention associated changes (i.e., Immediate Intervention= Baseline – 2 Months; Delayed Intervention= Baseline – 4 Months) in moderate-to-vigorous physical activity (minutes/day) and changes in cognitive function ............................................ 218 Figure 6.4 Relationship between intervention associated changes (i.e., Immediate Intervention= Baseline – 2 Months; Delayed Intervention= Baseline – 4 Months) in sedentary behaviour (minutes/day) and changes in cognitive function ....................................................................... 219 Figure 7.1 CONSORT diagram .................................................................................................. 228 Figure 7.2 Decision tree for bright light therapy prescription .................................................... 231 Figure 7.3 Decision tree for setting physical activity goals ........................................................ 233 Figure 7.4 Relationships between changes in sleep quality and changes in cognitive function over 24 weeks...................................................................................................................................... 245 Figure 8.1 Hypothetical model for the relationships of brain and cognitive health with sleep across the lifespan ....................................................................................................................... 262      xxi Acknowledgements This thesis dissertation would not be possible without the mentorship and guidance from my supervisor, Dr. Teresa Liu-Ambrose, as well as members of my PhD committee: Dr. Jennifer C. Davis, Dr. Karim Khan and Dr. Todd C. Handy.  I would also like to thank the current and past members of the Aging, Mobility, and Cognitive Neuroscience Laboratory: Dr. John Best, Dr. Cindy Barha, Dr. Glenn Landry, Dr. Deborah Jehu, Dr. Anna Egbert, Dr. Elizabeth Dao, Dr. Lisanne ten Brinke, Rachel Crockett, Tracy Dignum, Shirley Wang, Michelle Munkascy, Winnie Cheung, Chris Lim, Kimberly Bennet, Lauren Marcotte, Matthew Noseworthy, Daniel Backhouse, and Patrick Chang.  In addition, I would like to express my appreciation to the funding agencies that financially supported my studies.   I would also like to express my appreciation and gratitude for the support of my friends and family. The free drinks and dinners for a “struggling student” have helped me surive what seems (in retrospect) to have been an eternity of schooling.  Finally, I wish to thank my wife Jessica for having put up with me having been a student for this long. Thank you for keeping me sane, and for enjoying my awesome magic tricks.           xxii Dedication To Joyce B. Thorsten and Audrey Falck, my grandmothers who each passed away shortly before I finished this thesis. I hope to age as well and enjoy life as much as they both did.                   1  Chapter 1: Introduction  1.1 Preamble Worldwide, one new case of dementia is detected every four seconds [1]. Given the world’s aging population, the lack of effective drug therapy, and the fact a cure for dementia is likely many years away – there is an urgent need to develop interventions to prevent or at least delay dementia’s progression [2]. The societal value of identifying and developing effective behavioural strategies to help reduce dementia risk is thus an important line of scientific inquiry.   Broadly, there are three time-use activity behaviours which all humans engage in daily: physical activity (PA), sedentary behaviour (SB), and sleep [3]. Current evidence suggests PA is one important and modifiable lifestyle behaviour which can have substantial effects on older adult cognitive health [4, 5]. Emerging evidence also suggests SB is associated with poorer cognitive health [6, 7]. SB is a distinct behaviour from PA and may have different effects on cognitive health [8]. There is also a growing body of evidence which indicates poor sleep is a risk factor for cognitive decline and dementia [9]. Promoting increased PA, reduced SB, and better sleep may thus each have important benefits on older adult cognitive health.  Each of these behaviours also share a complex and dynamic relationship with each other and cognitive health [4]. Although PA and SB are independent behaviours, greater amounts of SB are consistently linked to lower PA level, and vice-versa [10]. PA has also long been thought to improve poor sleep [4], and people that are more physically active report sleeping better compared with people that are more sedentary [11]. Little is known about the impact of SB on older adult 2  sleep quality, although given the links between PA and sleep, it is plausible that SB may impact sleep too. Sleep is also inextricably linked to the function of circadian rhythms—the ~24 hour biological clock which helps align the sleep-wake cycle with the solar light-dark cycle [12]. A growing body of evidence suggests that circadian dysregulation is associated with poorer cognitive health and an increased risk of dementia [2]. PA and SB can also act as external stimuli on the biological clock, causing shifts in the timing of the clock, which may alter sleep patterns [4]. Given these complex interrelationships, it is critical to establish how PA, SB, sleep and circadian rhythms can each impact cognitive health—and whether this occurs simultaneously, in synergy, or in silos.  Thus, the overarching goals of my thesis are to: 1) advance the current knowledge about the dynamic relationships between time-use activity behaviours and cognitive health; and 2) characterize potential time-use activity behaviour intervention strategies for promoting cognitive health. This introduction provides the background knowledge and motivation for my thesis studies. Much of this background section is drawn from two book chapters I co-authored [4, 5].   Broadly, I will present this background section in six parts (Sections 1.2 – 1.7). I will first discuss the growing public health challenge of healthy cognitive aging, the definition of cognitive health, and the means by which cognitive health is measured (Section 1.2). In Section 1.3, I will review how PA (Section 1.3.1), SB (Section 1.3.2), and sleep impact older adults’ cognitive health (Section 1.3.3); definitions for each of these terms will be provided, as well as an overview of how each behaviour can be measured.   3  Section 1.4 will discuss the relationships of time-use activity behaviours with circadian rhythms, as well as the relationships of these behaviours with older adult cognitive health. In order to determine the dynamic relationships of time-use activity behaviours with cognitive health, it is important to measure these behaviours concurrently, and thus I will first discuss methods for the concurrent measurement and analysis of multiple time-use activity behaviours (Section 1.4.1). I will then review the current evidence examining the dynamic associations between time-use activity behaviours and cognitive health (Section 1.4.2), as well as the associations of these behaviours with circadian rhythms (Section 1.4.3).  Current therapies and interventions to promote PA, SB, sleep and circadian regulation will subsequently be discussed in Section 1.5. Lastly, I will review the current gaps in the literature (Section 1.6) and highlight what questions my thesis will address (Section 1.7).  1.2 The cognitive health of older adults: A growing public health challenge By 2030, there will be nearly one billion older adults worldwide [13]. Because age is the most important non-modifiable risk factor for dementia [14], an exponential increase in the number of individuals with dementia is expected [15]. There is therefore an urgent need for effective strategies which reduce dementia risk or delay disease progression.   Inextricably linked to the number of dementia cases worldwide is the prevalence of cognitive decline among older adults. An estimated 10-20% of adults over 65 years of age are living with Mild Cognitive Impairment (MCI; [16]), a transitional stage between healthy cognition and dementia [17]. MCI is defined as cognitive decline greater than expected for age and education 4  level which does not interfere with independence [18], and is associated with up to a 30% increased risk of developing dementia within five years [19]. Older adults without MCI develop dementia at a rate of 1-2% within five years [20]. Providing effective strategies to maintain cognitive health before and during this transition period might slow or prevent conversion to dementia. Thus, achieving cognitive health for all older adults is a growing public health priority which requires immediate solutions in order to effectively slow dementia rates [21].  1.2.1 Defining cognitive health A standardized definition of cognitive health is currently not established [22]. Perhaps the closest to a consensus definition is the one proposed by the National Institutes of Health’s Cognitive and Emotional Health Project [23], which described cognitive health as “not just the absence of cognitive impairment, but the development and preservation of the multi-dimensional cognitive structure that allows the older adult to maintain social connectedness, and ongoing sense of purpose, and the abilities to function independently, to permit functional recovery from illness and injury, and to cope with residual cognitive deficits.” Key elements of this definition are that cognitive health combines multiple domains of cognitive function, including traditional and measurable neuropsychological abilities (such as memory and processing speed), as well as more esoteric constructs such as wisdom and resilience. This definition also considers central the link between cognitive health and functional independence and engagement with life [22].  1.2.1.1 Normal cognitive aging While it is difficult to concisely define cognitive health at this time, there are specific aspects of cognition that change as people age. Domains of cognitive function which decline with age include: 1) attention; 2) memory; 3) executive function (i.e., planning and problem solving ability 5  [24]); and 4) perception [25]. However, some aspects of cognitive function (i.e., reading ability, general knowledge, language, etc.) remain relatively stable over the normal adult lifespan [26].  The normal aging process is also characterized by multifaceted changes in brain structure and function, which results in age-related changes in cognitive function [27]. These continuous, age-related changes begin as early as the third decade of life and include 1) declines in regional brain volume, myelin integrity, and cortical thickness [28]; 2) decreased availability of neurotransmitter receptors (e.g., serotnonin or dopamine) in the cortex [29, 30]; 3) accumulation of beta-amyloid (Aβ), neurofibrillary tangles, and other various brain metabolites [31-33]; and 4) changes in brain activation patterns at rest and during performance-based tasks [34, 35].  1.2.1.2  Cognitive changes in Mild Cognitive Impairment and dementia While some changes in cognition appear to be an unavoidable consequence of the aging process, there are some aspects of cognition which appear (at least on average) to be markedly different in individuals with MCI and dementia compared with their healthy cognitive counterparts [17-19]. Perhaps most obvious among these differences, older adults with MCI and dementia have poorer performance on neuropsychological tests of cognitive function—specifically in the domains of memory and executive function [19, 36, 37]. There also appear to be differences in age-related brain structure and function for individuals with MCI and dementia [38]. For example, compared to older adults without MCI, those with MCI have: 1) greater amounts of Aβ accumulation [39]; 2) accelerated declines in total brain volume, as well as deterioration in volume of the temporal and frontal lobes [40]; and 3) cortical thinning in the medial temporal lobe, and the frontal and parietal regions [41]. Increases in Aβ levels [42], accelerated atrophy in the medial temporal lobe 6  [43], as well as cortical thinning in the frontal, temporal and parietal lobes [44, 45] are each indicators of Alzheimer’s disease (AD) and dementia.  Older adults with MCI, when compared to their healthy peers, also exhibit disrupted functional connectivity [46], as well as changes in the activation of the temporal lobe, frontal lobe, and limbic area during performance-based tasks [47-50]. Disrupted functional connectivity, as well as altered brain activation of the frontal, temporal and parietal regions during task-based functional neuroimaging tests are each indicators of dementia [51, 52]. Interestingly, there do not appear to be differences in electrical signaling between age-matched healthy older adults and older adults with MCI [53], although changes in electrical activity appear to occur as people transition from MCI to dementia [54].  While people with MCI and dementia appear to have substantial differences compared to their healthy cognitive peers in performance on neuropsychological tests, as well as exhibiting brain structural and functional changes, the exact etiology and disease progression that occurs from healthy cognition, to MCI, to dementia is not clear. For example, autopsy studies have found that mild-to-moderate brain pathology is only modestly correlated with cognitive impairment [55, 56]. This individual variability in the trajectory of cognitive decline and manifestation of dementia symptoms is thought to be due to the existence of cognitive reserve—or the ability to tolerate age-related changes and disease related pathology of the brain without developing clinical symptoms or signs of disease [57]. Stern [58] suggested that cognitive reserve can take two forms: 1) neural reserve in which existing brain networks are more efficient or have greater capacity, and thus may be less susceptible to disruption; and 2) neural compensation in which alternate networks may 7  compensate for the pathological disruption of pre-existing networks. Mortimer [59] hypothesized that since pathological leisions can be present long before clinical symptoms of dementia occur, there are two sets of risk factors for dementia: 1) pathological risk factors; and 2) clinical expression risk factors. Irrespective of the framework by which cognitive reserve operates, there is an implicit notion within the cognitive reserve hypothesis that greater amounts of reserve will limit the clinical expression of dementia symptoms (i.e., cognitive resilience) until a threshold level of brain pathology is reached at which point cognitive reserve can no longer compensate for the underlying pathology [60].   Although the exact etiology and disease progression that occurs from healthy cognition, to MCI, to dementia is unknown, Jack and colleagues [61] proposed a hypothetical model which is worth further consideration. As described in Figure 1.1, underlying changes in older adult brain neurophysiology (due to both modifiable and non-modifiable risk factors) causes changes in brain structure and function, which leads to declines in cognitive function and clinical manifestations of dementia. Although this model was originally developed to describe the pathological cascade of AD, and thus may not be appropriate for describing the pathology of certain sub-types of dementia such as Lewy-body dementia or vascular dementia, the most common form of dementia is AD—accounting for ~50-70% of all dementia cases [62]. The model of Jack and colleagues [61] also appears to align with the concept of cognitive reserve, whereby accumulating brain pathology can be compensated for up to a certain threshold before clinical symptoms become apparent [60].   Thus, the maintenance of older adult cognitive health appears to be a complex and multi-faceted problem, whereby pathological markers of cognitive decline are necessary but not sufficient for 8  causing cognitive decline. It instead appears that neurophysiological markers, brain structure and function, as well as cognitive performance are each integral for maintaining older adult cognitive health and have a complex relationship with each other [61].   Figure 1.1 Hypothetical model for the biological cascade of dementia pathology (based on the model of Jack et al., 2010; [61]) Figure 0.1   Given the evidence which I have summarized above, I define older adult cognitive health as being composed of the following interrelated characteristics (Figure 1.2): 1) healthy levels of neurophysiological biomarkers which impact brain structure and function (e.g., Aβ42 deposition, Tau protein, IGF-1, BDNF, inflammatory markers, etc.; [61, 63]); 2) maintenance of brain structure including brain volume, cortical thickness, and white and gray matter integrity [38, 40]; 3) maintenance of brain function including functional activation, functional connectivity, and electrical activity [35, 51, 64]; and 4) maintenance of cognitive function for age and education levels, including global cognitive function and the cognitive sub-domains of memory and executive function [27, 65]. 9  Figure 1.2 Characteristics and relationships of older adult cognitive health Figure 0.2   10  1.2.2 Measuring cognitive health It is beyond the scope of this thesis to review every measure of cognitive health. However, I will provide a general overview of the methods by which each of the four main characteristics (neurophysiological biomarkers, brain structure, brain function, and cognitive function) of cognitive health are measured.   1.2.2.1 Neurophysiological biomarkers A “biomarker” is an objective measure which refers broadly to a number of medical signs that are used in diagnosing and assessing the progression of a disease or response to therapies [66, 67]. I define a neurophysiological biomarker as an objective measure which is related to the neurophysiological structure and function of the brain, can be found concentrated in the brain or biological fluids, and helps in diagnosing and assessing cognitive health. Broadly, neurophysiological biomarkers can be stratified into four main categories: cerebrospinal fluid (CSF) biomarkers, plasma or urine biomarkers, positron emission tomography (PET) markers, and genetic risk markers [68, 69]. While each of these biomarkers are important to the diagnosis and assessment of older adults’ cognitive health, my thesis studies have not included these markers.   1.2.2.2 Structural neuroimaging Structural neuroimaging uses magnetic resonance imaging (MRI) as a non-invasive method for mapping static anatomical structure [70]. Structural neuroimaging is based on the magnetization properties of atomic nuclei, wherein a powerful, uniform, external magnetic field (~1.5-7T) is employed to align hydrogen protons that are normally randomly oriented within the water nuclei of brain tissue in order to develop detailed images of brain structure [71]. Typically, T1-weighted 11  images are used in structural neuroimaging analyses of older adult cognitive health—most often by using a semi-automated pipeline on data acquired, which is then analyzed using neuroimaging software [72-76]. These analyses can include the quantification of brain regional volume, curvature, surface area, and cortical integrity. Other structural neuroimaging techniques—such as diffusion imaging, T2-weighted imaging, and fluid attenuated inversion recovery—are outside of the scope of my thesis work, but can provide other insights into older adult neuroanatomical structure and integrity.  1.2.2.3 Functional neuroimaging None of my thesis studies included functional neuroimaging, and thus I will only briefly describe these techniques for measuring brain function. Magnetic functional neuroimaging uses MRI to detect neuronal activation via changes in blood-oxygen-level-dependent (BOLD) signals [35, 51]. The BOLD signal can be used to create activation maps that describe the average engagement of brain regions during specific conditions in response to particular stimuli. There are two main methods of magnetic functional neuroimaging: task-based functional MRI (fMRI), and resting-state functional MRI (rs-fMRI). Electroencephalography (EEG) is a measurement tool for quantifying brain function that uses electrophysiological monitoring to record the electrical activity of the brain [77]. Magnetoencephalography (MEG) allows for real-time estimation of cortical activity by recording magnetic fields produced by electrical currents occurring naturally in the brain [78]. Single-photon emission computed tomography (SPECT) uses gamma rays that can be detected by a gamma camera in order to determine the biological function of regional brain metabolism [79], and can be used to diagnose and differentiate the different causal pathologies of dementia [80]. The technique is similar to PET, which can also be considered a functional 12  neuroimaging technique. Functional near-infrared spectroscopy (fNIRs) uses near-infrared light for the detection of BOLD signals in localized cerebral blood flow during specific conditions in response to particular stimuli [81].  1.2.2.4 Neuropsychological Testing As the name implies, this method for examining cognitive health uses performance-based neuropsychological tests to measure cognitive perfomance. Neuropsychological testing is a hallmark of medical research to determine changes in cognitive health and potential risk factors for cognitive decline [82]. As highlighted above, some domains of cognitive function tend to change as adults age or in response to intervention [25]. These domains of cognitive function are considered to be fluid, while cognitive domains that are far less likely to change (e.g., reading ability, general knowledge, language, etc.) are defined as crystallized and by definition remain relatively constant over the normal adult’s lifespan [26]. Thus, neuropyschological tests which are used to examine changes in cognitive function that are indicative of changes in cognitive health are typically conducted on fluid abilities.   I will broadly define the domains of cognitive function which I have examined throughout my thesis. I also include a list of common cognitive measures for each domain in Appendix A; this list was synthesized during a recent systematic review that I conducted [83]. Within this review, I defined four main cognitive domains, which I will discuss throughout my thesis: 1) global cognitive function; 2) executive function; 3) memory; and 4) processing speed. I define global cognitive function as tasks which examine multiple domains of cognitive function including memory, executive function, and processing speed. Executive function is a broad set of thinking 13  abilities which includes planning, set-shifting, working memory, and inhibition. I define memory as a broad set of thinking abilities which include spatial memory, immediate memory, and episodic memory. Processing speed includes measures of reaction time and ability to process information quickly. Of final note, it is important that any measure of cognitive function have well-established psychometric properties including evidence of validity and reliability, sensitivity to changes over time, and a minimal clinically important difference (MCID; [84]).   1.2.3 Summary Older adult cognitive health is a growing public health challenge since the number of older adults worldwide is increasing [13], and age is the greatest risk factor for cognitive impairment and dementia [14]. While a firm definition of the characteristics of older adult cognitive health are still needed, I define cognitive health as being composed of the following traits: 1) healthy levels of neurophysiological biomarkers which can impact brain structure and function [61, 63]; 2) maintenance of brain structure including brain volume, cortical thickness, and white and gray matter integrity [38, 40]; 3) maintenance of brain function including functional activation, functional connectivity, and electrical activity [35, 51, 64]; and 4) maintenance of global cognitive function for age and education levels, as well as in the cognitive sub-domains of memory and executive function [27, 65]. I have described each of the four methods for measuring cognitive health (neurophysiological biomarkers, structural neuroimaging, functional neuroimaging, and neuropsychological testing). Of note, my thesis measures cognitive health solely by neuropsychological testing and structural neuroimaging.   14  1.3 Time-use activity behaviours and their impact on older adult cognitive health: Physical activity, sedentary behaviour, and sleep Although maintaining cognitive health in later life is a growing public health priority, there is not yet a pharmaceutical therapy to improve cognitive health or treat MCI and dementia. As a result, lifestyle and behavioural strategies are becoming an increasingly popular line of research inquiry and public interest.   While there are an infinite number of behaviours which humans can engage in throughout the day, each moment of the day is spent in one of three basic activities: sleeping, SB, or engaging in PA [3]. Each of these behaviours are mutually exclusive (e.g., one cannot engage in PA while also engaging in SB or sleep), have different energy expenditures from each other [8, 85, 86], and are physiologically distinct phenomena—both in the periphery and the brain [6, 7, 87-91]. I will thus refer to these behaviours collectively as time-use activity behaviours (Figure 1.3), since each 24-hour day is spent in some assortment of sleep, SB, and PA.  Time-use activity behaviours are also linked to the circadian clock—the ~24-hour biological clock that helps to align the sleep-wake cycle with the solar light-dark cycle [12]. Specifically, time-use activity behaviours can be dichotomized into two broad and distinct types of behaviour which comprise the circadian cycle: 1) wake-based behaviour; and 2) sleep-based behaviour. Wake-based behaviour occurs while a person is awake and out of bed and consists of both PA and SB. Sleep-based behaviour occurs only while sleeping. Together, these behaviours comprise both ends of the circadian cycle (i.e., sleep and wake). Time-use activity behaviours thus have a dynamic relationship with each other and with circadian regulation.   15  Figure 1.3 Time-use activity behaviours are distinct but related, and are also part of circadian regulation Figure 0.3   METs= Metabolic Equivalents 16  Within these next sections (Section 1.3.1 – 1.3.3), I will discuss how each time-use behaviour is associated with cognitive health. I will begin by discussing the effects of PA on older adults’ cognitive health (Section 1.3.1). I will next discuss the effects of SB on older adults’ cognitive health (Section 1.3.2), and then describe how sleep impacts older adults’ cognitive health (Section 1.3.3); I will review circadian physiology in my discussion of sleep, given the link between sleep and circadian physiology [12]. In each section, I will provide important definitions, review the current methods for measuring the phenomena, and then describe how each behaviour impacts cognitive health.  1.3.1 Physical activity and older adult cognitive health  1.3.1.1 Physical activity definitions PA is defined as a wake-based behaviour of any bodily movement produced by skeletal muscles which results in energy expenditure [92]. PA can be classified by its duration (i.e., time), the frequency that it occurs (usually in days/week), the energy expenditure required to perform it (intensity), the type of PA performed (e.g., cycling, walking, etc.), and the context (location as well as social setting—that is, alone versus in a group). The different types of PA can be broadly grouped into four basic domains in which PA occurs: 1) occupational; 2) transportational; 3) household; and 4) leisure-time (Figure 1.4; [93]).  PA can also be classified by the energy expenditure which is required to perform it. Light intensity PA (LPA) involves activities wherein energy expenditure is at a level of 1.6-2.9 metabolic equivalents (METs); or 1.6-2.9 times an individual’s energy expenditure at rest [8]. LPA includes 17  activities such as slow walking, cooking food, washing dishes, and standing still; it also includes activities done from a seated or lying position, which require 1.6-2.9 METs (e.g., crafts, stretching, etc.; [94]). Moderate PA consists of activities wherein energy expenditure is between 3.0-6.0 METs and vigorous PA consists of activities >6.0 METs. Evidence suggests that >150 minutes/week of PA of ≥3.0 METs has substantial benefits on numerous health outcomes [90], and thus moderate and vigorous PA are generally examined as a single construct (i.e., moderate-to-vigorous PA [MVPA]; ≥3.0 METs). LPA and MVPA are not exclusive to any one domain of PA, and thus all domains of PA include activities, which are LPA and MVPA.  Figure 1.4 Physical activity domains and intensities of PA Figure 0.4  LPA: Light physical activity; MVPA: Moderate-to-vigorous physical activity; METs: Metabolic Equivalents  Current evidence suggests regular MVPA of greater than or equal to 150 minutes/week is important for older adult physical and cognitive health [90, 95]; unfortunately, greater than 95% 18  of older adults do not meet these recommendations [96]. Epidemiological evidence also suggests the amount of time spent engaging in each type of PA (i.e., occupational, transportational, etc.) is declining [97-100], and thus it is urgent to address the growing public health problem of physical inactivity [101].  Exercise training is a subcategory of leisure-time PA defined as planned, structured, and repetitive bodily movement done to improve or maintain one or more components of physical fitness—that is, skill- or health-related attributes which can be measured objectively [92]. Attributes of physical fitness include: 1) cardiorespiratory endurance (i.e., aerobic fitness); 2) muscular endurance; 3) muscular strength; 4) body composition; and 5) flexibility [92]. Most of the research which has examined how fitness can impact cognitive health has focused focused on aerobic fitness [102], since there are well defined tests to measure aerobic fitness such as the Balke Protocol and the Bruce Protocol [103].  Exercise training can also be classified according to its duration (usually in minutes), frequency (e.g., days/week), intensity (heart rate, repetition maximum, etc.), and type [104]. Maximizing the benefits of exercise training requires the precise prescription of these variables; however, the two most important variables for increasing physical fitness are intensity and volume (frequency x time; [105]).  The three most common forms of exercise training are aerobic training (AT), resistance training (RT), and multimodal training (MT; [104]). AT consists of repetitive movements specifically targeting the cardiovascular system. RT consists of muscle strengthening exercises typically 19  performed with free-weights or machines. MT refers to either: 1) exercise training which incorporates both AT and RT; or 2) AT and/or RT which also includes other forms of exercise training such as anaerobic, balance, agility, or flexibility training [83]. Each type of exercise training has its own distinct physiology and benefits [95, 106]. For example, AT specifically increases cardiovascular fitness (i.e., maximum oxygen uptake) whereas RT increases muscle mass and strength. Current recommendations suggest older adults regularly engage in MT, specifically: 1) moderate intensity AT five days/week for at least 30 minutes/session, or three days/week of vigorous intensity for at least 20 minutes/session; 2) moderate intensity RT at least twice per week; and 3) additional AT and RT, if possible [95]. Most forms of exercise training (including AT, RT, and MT) expend enough energy to be classified as MVPA (i.e., >3.0 METs), although some forms of exercise training can be classified as LPA (e.g., balance and flexibility training; [107]).  1.3.1.2  Physical activity measurement PA is a complex and multi-dimensional construct, and thus the measurement of PA is challenging. I broadly define the three primary dimensions (Figure 1.5) in which PA can be measured: behavioural measurement (i.e., the type and context of PA), biomechanical measurement (the movement of the body through space), and thermodynamic measurement (how much energy is expended). Each of these dimensions of PA measurement are related; for example, one minute of walking has a different rate of force production generated by skeletal muscle and total energy expenditure, than one minute of running. However, each of these dimensions has a different unit and scale of measurement. Using the previous example, walking can be measured behaviourally as minutes, distance traveled, or context (indoors vs. outside, alone vs. in a group); the 20  biomechanical measurement of walking could be in torque (i.e., force production by the muscle) or total work (force x distance); energy expenditure can be measured in METs, absolute oxygen consumption (L/min), relative oxygen consumption (ml/kg/min), or kcal. Accurately measuring each dimension of PA using a single measurement tool is thus not possible, and the consensus is that there is no one best method for measuring PA.  Figure 1.5 Dimensions of physical activity measurement Figure 0.5   1.3.1.2.A Subjective and objective measures of physical activity The complex and multidimensional structure of PA has thus led to the development of a variety of methods for measuring it. These methods for assessing PA have been categorized in a variety of ways [108-111], however PA measures can be broadly classified as either subjective or objective measures (Figure 1.6). Subjective measures assess a person’s subjective experience or recall of PA using questionnaires, surveys, diaries, logs, or ecological momentary assessment. Subjective  21  Figure 1.6 Characteristics and considerations for choosing a physical activity measure Figure 0.6   EMA: Ecological momentary assessment  22  measures can provide useful information on usual (or past) PA behaviour—specifically, the amounts, types and contexts in which an individual engages in PA. Objective measures do not require the person being measured to recall or respond, but instead use instruments (i.e., pedometers, accelerometers, heart rate monitors, multimodal sensors, calorimetry, and doubly labeled water), or systematic observation to quantify an individual’s PA.   There are benefits and disadvantages to both subjective measures and objective measures. Subjective measures are inexpensive and are easy to administer, and are thus commonly used in epidemiological and population-wide studies [112]. These methods of PA measurement also provide important information about the context in which PA occurs (i.e., where, when, and with whom); however, subjective PA measures are open to significant biases including social-desirability bias and recall bias [113]. The issue of recall bias is especially important for older adults, since older adults are more likely to experience issues with accurately recalling their past or present PA [114]. Subjective measures of PA also often fail to capture low-intensity activities (i.e., LPA; [115]). Other problems for subjective measures include issues with reliability, evidence of validity, and sensitivity to change [116]. Conversely, objective methods are considered to provide a more accurate and reliable measure of PA by eliminating recall bias; however, objective methods are more costly and require skilful administration and data interpretation [117].  Several extensive reviews can be found elsewhere on the different types of measures that exist for estimating older adult PA [118-121]. Collectively, these reviews indicate that there is no one best method for measuring PA, and each type of measure has benefits and limitations which should be carefully considered by the researcher when selecting a measure.  23   1.3.1.2.B Important psychometric properties of physical activity measurement: validity, reliability, population specificity, and sensitivity to change Accurately measuring PA requires careful consideration of the psychometric properties of the tool used. Inappropriate or crude measures of PA can have serious implications on study findings, and are likely to lead to misinterpreted results and underestimated effect sizes [112]. This is especially critical in the field of older adult PA research, as I determined in a systematic review published in 2015 [122]. In this review of older adult interventions to promote PA, I determined that only 63% of the measures used to estimate PA adhered to the principles of measurement. The use of measures with unknown psychometric properties makes the conclusions drawn from such studies questionable at best, and downright wrong at worst. I will therefore briefly describe the psychometric properties that are important for the precise and accurate measurement of PA: validity, reliability, population specificity, and sensitivity to change.  The validity of the measure (i.e., the instrument) can be best described as the soundness of the interpretations of the results of a measurement by which accurate conclusions may be drawn from the results [123]. No instrument can be “validated”, but rather evidence is collected to validate the interpretations made from an instrument. There are four different types of validity: definitional, content, criterion-based, and construct [124]; when used in summation, these may add to the evidence of validity of the instrument. Criterion-based evidence is the most common evidence of validity cited for both subjective measures and objective measures of PA [122].  24  The reliability of an instrument refers to the degree to which measurements of the same trait are reproducible under the same conditions [125]. The reliability of an instrument is an important means by which a measure can be deemed to have accuracy and consistency [126]. As with validity, an instrument can have evidence of reliability, but this does not mean that the instrument is always “reliable”. There are two broad types of reliability evidence: norm-referenced and criterion-referenced evidence of reliability.  Population specificity refers to the characteristics of the sample in which an instrument has evidence of validity and reliability. The specificity of a population can refer to a number of different characteristics such as age, ethnicity, or sex. Researchers should avoid using tools that do not have evidence of validity and reliability for the population they are presently measuring, as this greatly reduces the ability to accurately interpret findings [108]. For example, it is unlikely that a questionnaire developed for examining children’s PA would accurately reflect the PA of older adults.  Lastly, sensitivity to change (or responsiveness) refers to the ability of an instrument to detect change over time [112]. Evidence of validity and reliability are requirements in order for the instrument to have sensitivity to change. Sensitivity to change is typically quantified using the effect sizes for paired differences. For example, an instrument with evidence of validity and reliability that aims only to categorize people as “active” or “sedentary” will likely not be a sensitive instrument for detecting subtle changes over time.  25  1.3.1.2.C Measuring physical activity using MotionWatch8 wrist-worn actigraphy, the SenseWear Mini Armband multimodal sensor, and the Community Healthy Activities Model Program for Seniors I will now discuss each of the measures used in my thesis including their capabilities, limitations, and psychometric properties. The measures which I will discuss are: 1) wrist-worn accelerometry, specifically the MW8; 2) multimodal sensors (i.e., SWA); and 3) the CHAMPS questionnaire.   Wrist-Worn Accelerometry Using the MW8 Wrist-worn accelerometry (or actigraphy) is one objective field-method which is becoming increasingly popular for measuring PA [127, 128]. Accelerometry is based on basic principles of physics, namely speed and acceleration. Speed is the change in position with respect to time and acceleration is the change in speed with respect to time. Acceleration is usually measured in gravitational acceleration (i.e., 9.8 m/s2). Because acceleration is proportional to the net external forces involved in movement of a body, it is directly reflective of the energy costs of that movement [129]. Wrist-worn accelerometers, such as the MW8, therefore capture the acceleration of a body in motion which can then be translated into units of PA.   However, PA must be derived from the data outputted by a wrist-worn accelerometer. Wrist-worn accelerometers output data in an arbitrary unit of counts, which are collected and averaged over a length of time, termed an epoch. Accelerometer based studies in older adults have typically used an epoch length of 1 minute [130-132]. The data must then be translated into a rate of counts per minute (CPM). These CPM can then be categorized into a certain intensity of PA through accelerometers that have corresponding validated cut-points [96, 133]. Cut-points are used to 26  derive an estimation of PA amount, and the intensity at which the activity is performed [134]. Researchers can then evaluate the number of minutes a participant engages in PA for a given intensity, over the duration of a specific observation period (e.g., 7 days; [135]). Calibrating a wrist-worn accelerometer to provide estimates of PA from indirect calorimetry can thus provide criterion evidence of validity.  Hence, I calibrated the MW8 to provide estimates of PA for older adults [136]. Briefly, I concurrently measured indirect calorimetry and MW8 uniaxial actigraphy worn on the non-dominant wrist during 10 different activities of daily living in 23 community-dwelling older adults (aged 57-80 years). I then determined cut-points for LPA and MVPA using receiver operating characteristic (ROC) curves. In this study, I also determined the MW8 had good evidence of inter-rater reliability (r= 0.981). In a subsequent paper, I determined the MW8 had good evidence of interclass reliability for measuring PA following ≥5 days of observation [137]. I highlight the psychometric properties of the MW8 for measuring PA in Appendix B.  SWA Multimodal Sensor The SWA is a multimodal sensor which also has good evidence of validity and reliability for measuring PA across multiple populations [85, 138-143]. Briefly, the SWA integrates information from a biaxial accelerometer and other physiological sensors (heat flux, temperature, and galvanic skin cell response sensors) to provide estimates of energy expenditure in METs which can then be translated into estimates of PA. The use of multimodal sensors provides increased sensitivity for detecting subtle changes in energy expenditure associated with complex lifestyle tasks and with subtle changes in energy expenditure associated with carrying loads, walking up grades, or doing 27  non-ambulatory activities [144]. The SWA thus provides highly precise estimates of PA for older adults, which can be used to measure changes in PA over time [145].  CHAMPS PA Questionnaire The CHAMPS questionnaire is a subjective measure with evidence of validity and reliability for measuring older adult PA [146-149]. The questionnaire is also sensitive to changes in PA [150, 151]. The questionnaire consists of 40 separate questions, which asks participants to recall different activities performed in a typical week over the past four weeks.   1.3.1.2.D Summary Measuring PA is a complex and challenging issue for researchers to consider when designing a study and interpreting results. I have broadly described the different means by which PA can be measured, the psychometric properties which are important for determining the appropriateness of a PA measure, and have provided an overview of the strengths and limitations of the different measures of PA which I have used in my thesis work. I will now summarize the current body of evidence for how PA can impact older adult cognitive health.  1.3.1.3 Current evidence examining the impact of physical activity on older adult cognitive health There is a plethora of evidence which indicates that PA can promote older adult cognitive health [102, 152-154]. Broadly, the literature falls into one of three separate categories: 1) observational studies examining the associations of lifestyle PA with cognitive health (Section 1.3.1.3.A); 2) randomized controlled trials (RCTs) examining the effects of exercise training on cognitive health 28  (Section 1.3.1.3.B); and 3) animal models to determine the mechanism by which PA and exercise training impact cognitive health (Section 1.3.1.3.C). I will therefore discuss each of these areas of literature seperately.   1.3.1.3.A Impact of lifestyle physical activity on older adult cognitive health In this section I will discuss the current evidence for how lifestyle PA—that is, PA performed across all domains and intensities, both planned and unplanned—can impact older adult cognitive health. Most of the evidence for how lifestyle PA can impact older adult cognitive health comes from epidemiological studies, and is thus associative rather than causal. However, as I will highlight throughout these next three sections (Sections 1.3.1.3.A, B, and C), the current evidence meets each of the nine Bradford Hill Criteria for determining causal inference from observation studies [155]:  1) an effect size of small-to-moderate size;  2) consistent and reproducible relationships;  3) evidence of temporality in the association;  4) specificity of the relationship when controlling for confounding variables;  5) preliminary evidence of a dose-response relationship;  6) a plausible mechanism for how PA impacts cognitive health (explained in more detail in Section 1.3.1.3.C);  7) coherence between epidemiological and laboratory findings (Section 1.3.1.3.C);  8) experimental evidence from exercise training studies that PA in the form of exercise training improves cognitive health (Section 1.3.1.3.B and C); and  9) analogous associations between PA and other aspects of health. 29   The collective evidence from observational studies indicates that PA is associated with better cognitive performance and decreased incidence of dementia among older adults independent of other behaviours and prexisting factors [5]. In a meta-analysis of 16 prospective studies that examined the incidence of neurodegenerative disease based on PA at baseline, Hamer and Chida [156] determined that more PA was associated with a 28% lower risk of developing all-cause dementia and a 45% lower risk of developing AD, after adjusting for confounding variables. A second meta-analysis of 15 prospective studies among individuals without dementia determined that higher PA was associated with a 38% lower risk of cognitive decline, while low-to-moderate PA level was associated with a 35% lower risk [157]. Importantly, a more recent study which was not included in these meta-anlayses determined that moderate-to-high PA level may be especially protective against future cognitive decline among older adults with healthy cognition at baseline [158].   Epidemiological studies have also examined how PA is associated with brain structure and function. For example, Erickson and colleagues [159] examined how regular PA can impact brain structure in a sample of 299 community-dwelling older adults with normal cognitive function at baseline. At baseline, participant PA was quantified as the number of blocks walked in the past week; strucutrual MRIs were acquired 9 years later, and clinical adjudication for cognitive impairment was performed 4 years after the MRIs (i.e., 13 years since baseline). The authors determined that greater PA at baseline was associated with greater gray matter volumes in the frontal, occiptital, entorhinal, and hippocampal regions. Greater gray matter volume was also associated with a two-fold reduction in the risk of cognitive impairment. Gow and colleagues [160] 30  found that over a three year period (from age 70-73) greater amounts of self-reported PA were associated with: 1) less atrophy in total brain volume; 2) greater gray matter volume; and 3) greater integrity of white matter.  Other recent work has examined the effects of maintaining PA level over an extended period of time. Best and others [161] examined how PA maintenance over a 13-year period was associated with changes in older adults’ cognitive function and brain structure. One-hundred and forty-one  older adults aged 70-79 years at baseline (60% female) were annually queried about their self-reported time spent walking from years 1 to 10. Structural MRIs were performed at years 10 and 13. The authors determined that independent of initial time spent walking, demographics, and APOE-ε4 status (a genetic risk factor for AD [162]), better maintenance of time spent walking over the decade predicted less atrophy in hippocampal volume, better maintenance of gray and white matter integrity, and preserved cognitive function. Interestingly, PA at baseline and at year 10, as well as changes in PA over a five-year period, were less predictive of future changes in brain structure and cognition. These data thus appear to suggest that how PA levels change over longer periods of time may be an important contributor to cognitive and neural protection.  The positive association between PA and cognition in older age appears to be related to sparing of gray matter volume of brain regions susceptible to age-related atrophy, including the frontal and prefrontal lobes and the hippocampus [163]. Midlife PA level is also associated with larger total brain volume and gray matter volume of the frontal cortex 21 years later [164]. Dougherty and colleagues [165] determined that people who are more at risk for AD (defined in this study as either parental family history or carrying the APOE-ε4 allele), but have PA levels which meet or 31  exceed the current guidelines, have significantly larger temporal lobe volumes compared with individuals who did not meet the current PA guidelines.  The findings which I have described above all used subjective measures of PA; however, results from studies using objective measures also indicate PA is associated with better cognitive health. For example, Buchman and colleagues [166] reported that greater total daily PA – assessed by 10 days of hip-worn accelerometry – was associated with a two-fold reduced risk of AD over a four-year period in 716 older adults. Yuki and colleagues [167] investigated whether objectively-measured PA at baseline was associated with reduced brain atrophy (mean follow-up duration of 8.2 years) among 774 Japanese older adults. Participants in the highest quintile of PA had a significantly lower risk of frontal lobe atrophy compared with participants in the lowest quintile of PA.  There is also preliminary evidence suggesting that higher amounts of LPA, such as household chores may also help maintain brain health and cognitive function [168, 169]. Most of the literature has examined the relationships of total PA (i.e., LPA + MVPA) or MVPA with cognitive health [5, 170], and thus more research is needed as to whether LPA can also promote cognitive health.  While the collective evidence to date therefore suggests that PA can help reduce dementia risk, these findings cannot determine the optimal dose of PA to maintain cognitive health. Thus, while the dose-response relationship between PA and other health parameters is well established (i.e., the greatest benefits in health are seen in becoming physically active, with smaller health benefits for greater amounts of activity), the dose-response relationship between PA and cognitive health 32  is still relatively unclear [95]. However, Xu and colleagues [171] recently conducted a preliminary meta-analysis of five studies wherein they investigated the dose-response relationship between self-reported PA and cognitive health. The authors determined that for every 500 kcal/week of energy expenditure from PA, there was a 10% and 13% decrease in the risk of all-cause dementia and AD, repsectively. Evidence is still needed on the dose-response relationship between PA and cognitive helath using objective measures of PA.  Another key limitation of prospective cohort studies is that causality cannot be established and the potential for unmeasured confounding variables are often present. Of particular importance, the positive associations observed in prospective cohort studies may be due to reverse causaility; meaning that people who engage in more PA have a more robust genetic profile against cognitive impairment and dementia [153]. Conversely, those with very prodromal manifestations of cognitive decline may be less inclined to engage in PA. Sabia and colleagues [172] examined participants from the Whitehall II Cohort Study, finding that there was no association between PA and subsequent cognitive decline 15 years later. The authors also determined that there was no association between PA and risk of dementia over an average 27 year follow-up. Critically, PA in people with dementia began to decline up to 7 years before diagnosis, which seems to indicate that lower risk of dementia in physically active people may be attributable to a decline in PA levels in the preclinical phase of dementia.   A final and important limitation of the current literature is that most of the causal evidence for PA improving cognitive health comes from studies of exercise training [5, 173], as described in 33  Section 1.3.1.3.B. Thus, it is currently unclear whether increasing lifestyle PA (either LPA or MVPA) is sufficient to improve cognitive health.  1.3.1.3.B Impact of exercise training on older adult cognitive health In this section, I discuss exercise broadly without reference to type of the exercise (i.e., AT, RT, or another form of exercise training). The early literature on the importance of exercise for older adult cognitive health examined the importance of physical fitness—specifically aerobic fitness—for cognitive function [102, 174]. These literature reviews found conflicting results. One of the reviews found increased physical fitness was associated with impoved cognitive performance [102], while the later study did not find improved aerobic fitness was associated with greater cognitive performance [174]. It is important to note that the first review only included RCTs, while the second investigated cross-sectional and non-randomized designs, which suggests that lower quality studies may have confounded the results of the later review.  In addition to examining the effects of fitness on cognitive health, early reviews have used heterogeneous age groups and populations—from both animal and human data. Collectively, these data also suggest improved fitness can enhance older cognitive health and even prevent cognitive decline [175, 176]. Converging evidence at the cellular, molecular, behavioral, and systematic levels also suggest exercise has beneficial effects on cognitive health across all age groups [177-180]. However, most of these data are based on the findings of trials which have used AT as the exercise intervention, and there is still limited knowledge about how RT and other types of exercise training can impact cognitive health [106, 181, 182]. As such, developing and designing exercise 34  programs to promote cognitive health are not fully addressed within these review papers, and there remains an issue in translating interventions from the laboratory to the outside world [175-180].  More recent reviews have begun to examine how exercise training can promote cognitive health in people with cognitive impairment. Heyn and colleagues [183] initially determined in a meta-analysis of 30 studies among people with dementia and related cognitive impairments that exercise training significantly improved cognitive function; although the authors did not determine which domains of cognitive function were most imapcted by exercise training. In a subsequent meta-analysis of 14 RCTs among older adults with MCI, Gates and colleagues [184] determined that exercise training had a small but significant effect on verbal fluency, but did not improve response inhibition, cognitive flexibility, memory, or processing speed. Most recently, Zheng and colleagues [185] conducted a meta-analysis of 11 RCTs which examined whether AT improved cognitive function among older adults with MCI. The authors determined that AT had a large effect on global cognitive function, modestly improved immediate and delayed memory, and significantly improved executive function. While these data suggest MCI is a window of opportunity to promote cognitive health through exercise training, less is known about how other training modalities beyond AT (i.e., RT, MT, etc.) can promote cognitive health.  Several meta-analyses published in the last two years also warrant consideration. In a comprehensive systematic review and meta-analysis, Northey and colleagues [154] determined that exercise training has a modest effect (Cohen’s d= 0.29) on the cognitive funciton of older adults over the age of 50 years—both for individuals with and without cognitive impairment. The authors also determined that AT, RT, MT, and Tai chi exercise could each improve the cognitive 35  function of older adults, although it was unclear which type of exercise training was most beneficial; yoga did not significantly improve cognitive function. Moderate and vigorous intensity exercise were also significantly beneficial to cognitive function, while low intensity exercise was not. Exercise training also benefitted the cognitive domains of attention, executive function, and memory; however, the effect sizes for each domain were overlapping. The authors thus recommended that older adults should perform exercise training in the form of MT to improve cognitive function, and that improvements were independent of baseline cognitive ability. The authors also found that the effect of exercise on cognitive function was slightly larger for individuals without MCI (d= 0.36) compared with older adults with MCI (d= 0.28). However, the confidence intervals for these estimates were overlapping, and there was a significant effect of exercise on cognitive function for individuals with and without MCI.  More recently, in a systematic review of 98 studies in older adults ≥60 years of age, Gomes-Osman and colleagues [186] suggested that >52 hours of exercise training was associated with improved cognitive performance in older adults with and without cognitive impairment. However, this finding was not based on any meta-analysis, but on the median length of intervention volume (in hours) for studies which reported improvements in cognitive function. These results, thus need to be treated with extreme caution, and it is unlikely that 52 hours of exercise training represents some sort of magic number to elicit improvements in cognitive function.  There has also been increasing interest in whether there are sex-differences in the effects of exercise training on older adult cognitive function. In a systematic review and meta-analysis of 39 studies, Barha and colleagues [187] determined that exercise training interventions with a high 36  proportion of females (defined as >71% female) had significantly greater effects on executive function compared to studies with a lower proportion of females (≤71% female). A separate systematic review and meta-analysis of 17 animal studies by Barha and colleagues [188] also found sex differences in the effects of exercise training on cognition between male rodent studies and female rodent studies.  Of final note, I recently published a systematic review and meta-analysis to examine the effects of exercise training on the physical and cognitive function of older adults (60+ years; [83]). This review of 48 studies provides five important updates to the literature. First, my review provides an important update to the recommendations made in Northey and colleagues [154] by now suggesting that MT is beneficial to both cognitive function and physical function. Second, my review distinguished between primary and secondary outcomes of cognitive function; the results indicated that the effects of exercise training on cognitive function are smaller than what has been previously reported (Hedge’s g= 0.24). Third, I determined that the precise estimate of the effects of exercise training on cognitive performance appears to be hampered by use of heterogeneous samples which include people of different physical (i.e., frail vs. healthy) and cognitive stuatus (MCI vs. healthy), such that studies which include samples of mixed cognitive and physical statuses may be diluting the effects of exercise training by potential deviations from the exercise protocol. Fourth, unlike past meta-analyses which examined sex differences in exercise efficacy as a categorical variable [102, 187, 188], I examined sex as a continuous moderator (i.e., %female) and did not find that it moderated the effects of exercise training on cognitive function. While this study suggests that the sex-specific effects of exercise on cognitive function are less robust than originally suggested, meta-analyses examining the disaggregated sex-specific effects of exercise 37  are still needed. Lastly, I determined that exercise-induced improvements in physical function are associated with improvements in cognitive function at the study level, which provides support for the central benefit model—whereby cognitive and neural plasticity may be an important mechanism by which exercise training promotes mobility [189].  The evidence therefore suggests that exercise training can help reduce dementia risk. However, it is still unclear what the optimal dose of exercise is to promote cognitive health [106]. For example, little is also known about whether engaging in AT, RT, MT, or some other type of exercise training (e.g., anaerobic training, power training, etc.) will yield optimal cognitive benefits. In addition, there is preliminary evidence that genotypic differences may moderate the effects of exercise training [190], however this area of research is still under investigation and is far from conclusive. Future research is needed to 1) examine ways to ensure exercise training programs provide maximal benefits to older adult cognitive health; 2) determine the dose-response relationship between exercise and cognitive health; and 3) whether biological sex and genotypic differences moderate the effects of exercise on cognitive health.  Current interventions are now focused on assessing the effects of different types and intensities of exercise training for promoting cognitive health among people with MCI. Baker and colleagues [191] examined the efficacy of high-intensity AT on cognitive function in older adults with MCI. Participants (N= 33) were randomized to either 6 months of 4 days/week high intensity AT or a stretching control group. The high-intensity AT group trained at 75-85% of heart rate reserve for 45 to 60 minutes per session. The stretching control group carried out supervised stretching activities according to the same schedule but maintainted their heart rate at or below 50% of their 38  heart rate reserve. Interestingly, the high-intensity AT group improved cognitive function in women, but not men. While these results appear to suggest a sex-specific effect of exercise, the sample size was small—especially with regard to the sex-stratified analyses.  More recent evidence suggests AT may augment cognitive function in those with MCI by improving neural efficiency [192, 193]. Among 70 older adults with mild vascular cognitive impairment, Liu-Ambrose and colleagues [193] showed that six months of AT significantly improves global cognitive performance compared with a usual care plus education group. Examination of secondary measures showed between-group differences at intervention completion favouring the AT group in 6-minute walk distance and in diastolic blood pressure. A subset of participants participated in a functional MRI study [192], which determined that compared to the control group, AT significantly reduced activation in the left lateral occipital cortex and right superior temporal gyrus. Reduced activity in these brain regions was significantly associated with improved executive funtion performance on the flanker task at trial completion. The results of the partial correlation analysis support the notion of neural efficiency, which is defined as the level of activity a neural network requires in order to complete the task at hand [194].  Erickson and colleagues [195, 196] determined that AT can positively impact hippocampal volume and memory among healthy older adults. Subsequently, ten Brinke and colleagues [197] extended this work by demonstrating that compared with balance and tone exercises, thrice-weekly AT significantly increases left, right, and total hippocampal volumes in older adults with MCI.  39  Recent RCTs have also begun to investigate the impact of RT on older adult cognitive function, especially among individuals at risk for dementia. In 86 older women with MCI, Nagamatsu and colleagues [198] demonstrated that compared to a balance and tone control group, 6 months of twice-weekly moderate intensity RT significantly improved executive functions—specifically, the cognitive processes of selective attention and conflict resolution. The authors also found improvements in associative memory – or the ability to remember items presented simultaneously. In conjunction, regional patterns of functional plasticity were found in the RT group, whereby three key regions in the cortex showed greater functional activation during an associative memory task after 6 months training—the right lingual gyrus, the right occipital-fusiform gyrus, and the right frontal pole. Interestingly, improvements observed in executive performance occurred after only 6 months of RT in those with MCI, compared to 12 months in otherwise cognitively healthy older adults [199]. It is thus possible that the benefits of RT may be observed earlier in those with cognitive deficits [5], although more evidence is needed to confirm this hypothesis.   The SMART (Study of Mental and Resistance Training) trial [200] also demonstrated the positive impact of RT on cognitive health in older adults with MCI. One-hundred older adults with MCI were randomized to two interventions: 1) a high intensity RT group or a seated calisthenics group; and 2) a computerized cognitive training group or a sham cognitive training group. Participants completed both interventions 2-3 days/week for 6 months with an 18-month follow-up. The investigators found that RT but not computerized cognitive training significantly improved global cognitive function as well as expanded gray matter in the posterior cignulate. Interestingly, these improvements were related to each other (r= 0.25). RT also significantly reduced the progression of white matter lesions. 40   More studies are now examining the effects of exercise training among older adults with dementia. In a 12-month RCT of 210 older adults with AD [201], reseachers randomized participants to either home-based exercise, group-based exercise, or usual care for 12 months. Participants in either of the exercise programs were encouraged to engage in two 1-hour sessions of MT per week. Although the data did not support a significant effect in the most rigorous analyses, the authors suggested that home-based exercise improved executive function compared to the usual care group. The group-based exercise program produced intermediate effects and did not differ compared to either the usual care or home-based exercise participants.  Lamb and colleagues [202] recently conducted a large RCT of 494 people with dementia which examined the effects of 4 months of supervised moderate- to high-intensity MT plus unsupervised home-based exercise training on ADAS-Cog performance at 6 and 12 months. Compared with the usual care control group, the exercise group showed slight declines in ADAS-Cog performance at 12-months. However, it is important to note that rather than examining the effects of the intervention immediately following trial completion (i.e., 4 months), cognitive performance was not measured until 6 and 12 months (i.e., 2 and 8 months after the end of the trial). The delay in assessing treatment outcomes makes it very difficult to determine the true effect of the intervention as it seems unlikely that individuals with dementia would continue with the exercise program during the post-trial period.   Together with the previous studies on MCI, it is plausible that a modest degree of cognitive decline might be ameliorated by exercise training, yet more severe cognitive decline might be less 41  ammenable through exercise training. Even so, it is noteworthy that in the 2018 practice guideline summary for MCI, the American Academy of Neurology states that while no high-quality evidence exists for pharmacological treatment of MCI, exercise is likely to provide benefit [203].   Of final note, there has been some recent interest in the effects of other types of exercise training – specifically anaerobic training - on older adult cognitive health. Briefly, anaerobic training (or high intensity interval training [HIIT] as it is commonly known) uses higher intensity cardiovascular exercise which is at or above a person’s anaerobic threshold—the point at which the body can no longer meet immediate energy production needs using only beta oxidation and aerobic respiration, and begins to also produce energy through glycolysis [104]. Because of the high intensity of the exercise, it cannot be sustained continuously for long periods of time and is often performed in shorter intervals (hence the name HIIT). There is a growing interest in HIIT since 1) it might be a quicker and more time-effective means of delivering exercise as a therapy; and 2) observational data appears to suggest that higher intensity PA provides greater benefits to cognitive health than lower intensities [204, 205]. To this end, Kovacevic and colleagues [206] recently conducted a quasi-experimental study of 64 healthy older adults which examined the effects of thrice-weekly HIIT as compared with AT or a stretching control group. The results indicated that the HIIT group had significantly better memory performance than the AT and control group. While this preliminary work is interesting, these results must be treated with extreme caution since participants were not randomized at baseline. Future work using rigorous empirical methods is needed to determine the efficacy of HIIT as an intervention strategy for promoting older adult cognitive health.   42  Summary The precise characterization of optimally effective exercise training has yet to be fully elucidated. For example, it is unclear whether AT, RT, or MT is most beneficial for cognitive health. Maximizing the benefits of exercise training for cognitive health will require the precise prescription of volume (i.e., frequency*duration) and intensity that are based on the best-available evidence. It is also unclear if the effects of different modalities of exercise training are domain-specific or global. Although there has been some recent debate as to what aspects of cognitive function are impacted (or not) by exercise training [207-209], the overwhelming evidence to date suggests that exercise training can impact global cognition, processing speed, executive functions, and memory [106]. Future research is needed to determine: 1) the most beneficial intensity, frequency, and duration of exercise necessary to improve cognitive health; 2) whether AT, RT, or a combination of AT and RT is more beneficial for cognitive health; 3) how other types of exercise training (such as anaerobic and power training) can impact cognitive health; and 4) whether promoting exercise training should be done in the early, later, or mid-stages of cognitive decline.  1.3.1.3.C Mechanisms by which physical activity and exercise training can impact older adult cognitive health To date, most of the literature examining the mechanism by which PA impacts cognitive health comes from animal models of AT [179, 180]. Because exercise is a sub-category of PA, it is believed that the mechanisms by which PA and exercise improve cognitive health are similar. Broadly, the current evidence suggests that PA stimulates neurogenesis and angiogenesis, with the greatest effects being found in the dentate gyrus of the hippocampus [180]; these improvements are associated with improved memory and learning capability [177]. Animal models have also 43  found that older animals also see significant increases in neurogenesis and angiogenesis, as well as improved ability on a memory task [178]. However, these findings are broadly based on models of exercise training, wherein an animal undergoes a set volume, intensity, and type of exercise training.  As discussed previously, the two most common forms of exercise training are AT and RT, or sometimes used in combination as MT. Each type of exercise training has its own distinct physiology and benefits [104]. For example, AT specifically increases cardiovascular fitness (i.e., maximum oxygen uptake) whereas RT increases muscle mass and strength. Importantly, each type of exercise training has beneficial effects on cognitive health [106]. Given AT and RT are distinct behaviours with their own physiology, I will review the current evidence for the mechanisms by which AT and RT impact cognitive function separately. The mechanism by which MT impacts cognitive health is unknown. I will begin each of these sections by discussing the evidence for the mechanisms from animal models and then discuss the evidence from human models.   Mechanisms by which AT Impacts Cognitive Health Much of the literature examining the molecular and cellular effects of AT comes from animal models. In these studies, the exercise is manipulated by providing voluntary access to a running wheel or by forced exercise on a treadmill. Animal models have focused on three mechanisms by which AT may influence cognitive health: 1) neurogenesis or the creation of new neurons [179]; 2) cerebral angiogenesis or the creation of new blood vessels in the brain [210]; and 3) changes in inflammatory markers [180].   44  Neurogenesis  The most studied mechanism by which AT can influence cognitive health is exercise induced neurogenesis in the hippocampus, specifically the dentate gyrus [179, 180, 211-214]. As illustrated in Figure 1.7, AT stimulates neurotrophic factors in the periphery and in neuronal cells of the brain [179, 180].  There are several important neurotrophic responses which occur with AT. Perhaps most importantly, AT causes an increase in firing of pyramidal cells of the hippocampus [215], which is thought to stimulate brain derived neurotrophic factor (BDNF) production [6]. In addition to up-regulation of BDNF, vascular endothelial-derived growth factor (VEGF) increases in the periphery in response to AT and crosses the blood-brain barrier to enter the brain [216]. Insulin-like growth factor-1 (IGF-1) is also up-regulated and can also cross the blood-brain barrier to enter the brain; however, it is also centrally derived [217]. There is also some evidence suggesting the importance of serotonin for exercise neurogenesis, although little is known about the mechanism [218]. Thus, it appears that up-regulation of BDNF, VEGF, and IGF-1 are all neurotrophic responses to AT.  Both BDNF and IGF-1 are closely involved in neurogenesis in response to AT, however the relationship between IGF-1 and BDNF to modulate aspects of exercise-induced neurogenesis is complex [219]. Briefly, exercise induced expression of BDNF and IGF-1 are regulated in the hippocampus through their cognate receptors (tyrosine kinase B [TRKB] and insulin-like growth  factor I-receptor [IGF-1R], respectively; [219, 220]). Both TRKB and IGF-1R induce activation of the calmodulin kinase II (CAMKII) and mitogen-activated protein kinase (MAPKII) pathways; these pathways interface to up-regulate centrally derived IGF-1 and BDNF [221]. In addition, 45  Figure 1.7 Hypothesized mechanism through which IGF-1 signaling may interface with BDNF-mediated synaptic plasticity in the hippocampus during exercise Figure 0.7  BDNF= Brain Derived Neurotrophic Factor; CREB= Phosphorylated Cyclic AMP Response Element Binding Protein; IGF-1= Insulin-like Growth Factor 1; IGF-1R= Insulin-like Growth Factor 1 Receptor; LTP= Long Term Potentiation; p-CAMKII= Phosphorylated Calmodulin Protein Kinase II; p-MAPKII= Phosphorylated Mitogen-activated Protein Kinase II; pro-BDNF= precursor BDNF; TRKB-R= Tyrosine Kinases B Receptor; VEGF= Vascular Endothelial-derived Growth Factor. Modified from Cotman, Berchtold, & Christie 2007 [180].46  expression of CAMKII and MAPKII up-regulates Synapsin I and Synaptophysin [220]. Synapsin I tethers synaptic vesicles to each other and to the actin cytoskeletal net, thus functioning to maintain proper synaptic transmission of neurotransmitters [222]. Synaptophysin is a key protein in the biogenesis of synaptic vesicles from cholesterol and may possibly facilitate membrane retrieval during vesicle recycling [223]. Expression of these proteins may thus be an important component of maintaining cytoarchitecture during and after neurogenesis. Finally, both CAMKII and MAPKII up-regulate expression of cyclic-AMP response element binding protein (CREB; [224]). CREB is critical for activity-dependent long-term neuronal plasticity and is believed to be necessary for the formation of long-term memory [6].   There are also still gaps in the understanding of the mechanisms by which BDNF, IGF-1 and VEGF stimulate neurogenesis. For example, it is unknown whether VEGF has an independent mechanism by which it stimulates neurogenesis [180], although VEGF does appear to be necessary for neurogenesis to occur [216]. It is also unclear by what mechanism BDNF stimulates neurogenesis in the hippocampus [211], although AT increases BDNF signaling and concomitantly increases cell proliferation and neurogenesis [225, 226]. Evidence also suggests IGF-1 mediates neurogenesis in the hippocampus [217], although the precise mechanism of this process is still under investigation. What is clear, however, is AT increases cellular proliferation, dendritic complexity, and dendritic spine density in the dentate gyrus of the hippocampus [227-230]. These adaptations in the hippocampus are associated with improvements in learning and memory [179, 180, 214]. Importantly, adaptations in cytoarchitecture also occur in older and diseased animals [231-235], suggesting the importance of AT for neurocognitive health across populations.  47   Angiogenesis  The concomitant effect of AT on cerebral angiogenesis may be necessary to stimulate increased neurogenesis [236]. While the effects of BDNF on angiogenesis are largely unknown, both IGF-1 and VEGF are necessary to induce cerebral angiogenesis [237, 238]. Increased signaling of IGF-1 and VEGF are associated with endothelial cell proliferation, and increased vessel size and branching, and thus parallel generation of new neural cells in the dentate gyrus of the hippocampus [236]. As such, angiogenesis may be a necessary component of neurogenesis—increasing blood-flow to newly synthesized neural tissue in order for it to be utilized and properly maintained. Angiogenesis may thus support improvements in memory and learning through its involvement in neurogenesis [179, 180]. Importantly, this finding has been replicated in older and diseased animals [231].    Down-regulation of Inflammatory Markers AT is also linked with other improvements in neurophysiology—specifically by down-regulating inflammatory factors, which are associated with the progression of AD. For example, AT significantly reduces Aβ protein levels in the frontal cortex of transgenic mice predisposed to AD symptoms [239]. In addition to reducing Aβ load, AT in transgenic mice reduces pro-inflammatory markers including IL-1β and TNF-α [240]. Importantly, both of these inflammatory markers are associated with increased Aβ load and have been linked to AD progression [241, 242]. Immune-response proteins also significantly increase in response to AT [241, 242], and thus may help delay disease progression.  48  Evidence from Human Trials There has also been a growing amount of interest in applying the knowledge gained from animal models, to understanding the mechanisms by which exercise training can promote cognitive health in humans. AT is consistently linked with increases in neurotrophic factors [243]; however, human studies cannot directly assess central BDNF, IGF-1, or VEGF, and it is not feasible to assess the time course by which neurotrophic factors impact cognitive health.   While few studies have reported on how these growth factors impact cognition, two studies have investigated how these growth factors can impact brain plasticity and function [196, 244]. Neither study found circulating levels of BDNF, IGF-1 and VEGF changed after one year of AT. However, change in BDNF was associated with increased hippocampal volume in one study [196]; in the other study, increased serum BDNF, IGF-1, and VEGF were associated with increased functional connectivity of the bilateral parahippocampus and the bilateral middle temporal gyrus [244]. While neither study reported whether increases in growth factors were associated with improvements in cognitive function, the results appear to align with findings from animal studies which suggest these neurotrophins contribute to the positive effects of exercise on cognitive health [179].   While the precise mechanism by which AT stimulates a neurotrophic response is yet unclear, there is evidence that higher cardiorespiratory fitness is linked with greater brain volume [175, 176, 245], and there is growing evidence that AT (which also improves cardiorespiratory fitness [104]) is associated with increases in brain volume in aging humans [246, 247]. The areas which seem to be most affected by AT are the prefrontal and medial temporal areas of the brain [196, 197]. 49  Human trials cannot directly assess changes in neurotrophins as a result of AT; although it appears serum changes in BDNF are associated with changes in brain health [196].   Mechanisms by which RT Impacts Cognitive Health  Much less is known about the neurophysiological impacts of RT on cognitive health. This dearth of knowledge is due to the only recent development of animal models to examine the effects of RT [248]. Within this model, the mouse or rat is familiarized with a vertical ladder apparatus which has a shelter located at the top of the ladder. Once the animals learn to climb the apparatus, they are progressively loaded with a weight to their tail in order to mimic a progressive RT program.   The mechanism by which RT impacts cognitive function is much different than the mechanism of AT [249]. The largest difference in the mechanisms of AT and RT is the signalling of IGF-1 and BDNF [219, 250]. Paradoxically, while IGF-1 signalling is required in order to stimulate AT induced neurogenesis [219], AT seems to have little or no influence on peripheral IGF-1 levels [250]. By comparison, peripheral IGF-1 increases with RT while peripheral BDNF levels do not appear to be altered. To my knowledge, no study has examined what the effects of RT are on VEGF in animal models; however, one human study found thrice weekly RT increased peripheral VEGF [251]. Another human study using a combination of RT and AT found peripheral VEGF levels increased following the intervention [252]. Neither of these human studies examined whether VEGF changes were associated with improvements in cognitive performance.   Other important differences in the mechanism of RT, as compared with AT, are worth mentioning. One study compared an eight week RT protocol to eight weeks of AT [249]. Briefly, the results of 50  this study indicate RT and AT can both improve cognitive function, but do so through divergent mechanisms. Interestingly, hippocampal IGF-1 increased for both the RT group and the AT group; although the RT group did not appear to increase expression of BDNF. Both groups also had increased expression of Synapsin I and Synaptophysin; however, the RT group had significantly greater Synapsin I expression than the AT group. Preliminary evidence also suggests RT may stimulate neurogenesis and reduce apoptotic cell signalling in the hippocampus [253], although a recent study did not find an effect of eight weeks of RT on neurogenesis [254]. Thus, the mechanisms by which RT may impact neurophysiological health are still unclear.   Evidence from Human Trials The precise mechanisms by which RT affects cognitive health remain under investigation; however, the evidence does indicate RT can have beneficial effects on both brain plasticity and cognitive function in older adults. Specifically, RT for older adults can improve executive function and memory [198, 199, 255]. Twice weekly RT for 12 months may also have a long term effect on executive function, memory, and cortical white matter volume [256]. In addition, RT in older adults may increase cortical thickness in the posterior cingulate gyrus, decrease functional connectivity between the posterior cingulate and anterior cingulate cortex, and increase functional connectivity between the hippocampus and the left superior frontal lobe cluster [257]. These data suggest RT may help alter and improve functional connectivity in older adults, specifically the default modal network and the frontal-temporal network.   1.3.1.4 Summary of the current evidence examining how physical activity impacts older adult cognitive health 51  There is a large body of evidence which indicates that PA is important for healthy cognitive aging. Most of the evidence linking PA to healthy cognitive aging comes from observational data; however, exercise training RCTs and animal research both indicate that PA in the form of exercise can improve multiple aspects of cognitive health. While it is thus clear that PA is a pillar of healthy cognitive aging, researchers need to carefully consider how to measure PA in order to best understand it’s impacts on cognitive health. The precise dose-response relationship between PA and cognitive health also remains unclear, and it is unclear whether different types of PA will have different effects on cognitive function. A considerable amount of research has also examined how exercise training (a sub-domain of PA) can promote older adult cognitive health, specifically in the domains of memory and executive function. The mechanism by which PA promotes cognitive health is also based on animal models of exercise training; however, it is still unclear whether AT, RT, or MT is most beneficial for older adult cognitive health. Future research is still needed to address a number of areas including: 1) what types and intensities of PA and exercise training are most beneficial for older adult cognitive health; 2) which cognitive functions and areas of the brain are specifically targeted by each type of PA and/or exercise training; 3) the underlying mechanims by which PA and exercise training impacts older adult cognitive health; and 4) the dose-response relationship between PA and cognitive health.  1.3.2 Sedentary behaviour and older adult cognitive health   1.3.2.1 Definition of sedentary behaviour SB is defined as any wake-based behaviour which incurs ≤1.5 METs and includes behaviours such as sitting, television watching, and lying down [8]. A common misconception is SB is the inverse 52  of PA; however, SB is an independent behaviour with its own distinctive effects on health [258]. Like PA, SB can also be classified by it’s duration, frequency, type, and context; SB is not classified by it’s intensity since all SB requires low energy expenditure. Tremblay and colleagues proposed that SB should be classified according to the SITT formula [88], consisting of: 1. SB frequency (number of bouts of a certain duration); 2. Interruptions (e.g., getting up from the couch while watching [259]); 3. Time (duration of sitting); and 4. Type (mode of sedentary behaviour such as TV viewing, driving a car, or computer use; Figure 1.8) The context of where SB occurs (and who with) can also have important implications on health [260].   Figure 1.8 Different types of sedentary behaviour Figure 0.8   1.3.2.2 Sedentary behaviour measurement There are a few differences between measures of SB and measures of PA that are worth mentioning. Given that SB is a distinct and independent behaviour from PA, there are several 53  objective measures of PA which cannot easily estimate SB (i.e., pedometers and heart rate monitors). Unlike the measurement of PA, wherein energy expenditure above a threshold of 1.5 METs is indicative of PA, SB refers to any activity which is performed from the seated or lying position while awake. Hence, the precise objective measurement of SB also requires information about body position, which can be detected through inclinometry [8, 88]. Not all SB measurement tools are capable of determining body position, and thus these measures only estimate SB based on energy expenditure (either measured directly, or indirectly estimated from accelerometer-based devices; Figure 1.9). However, there is no SB measurement tool currently capable of accurately estimating body position and energy expenditure concurrently; measurement tools capable of estimating body position are thus only able to classify time spent sitting, lying down, or standing.  Throughout my thesis, I have measured SB using the MW8 and the SWA. Each of these measures has evidence of validity and reliability for estimating older adult SB [136, 137, 140, 141], however neither device is capable of estimating body position.  1.3.2.3 Current evidence examining the impact of sedentary behaviour on older adult cognitive health In comparison to the current level of evidence examining the impact of PA and exercise on older adult cognitive health, far less is known about how SB can impact older adult cognitive health. Nonetheless, I will review the current evidence on how SB can impact cognitive health.  There are two reviews which have suggested SB may be associated with poorer cognitive function and increased risk of cognitive impairment [6, 7]. These reviews suggest SB may 1) negatively 54  Figure 1.9 Measures of sedentary behaviour and their classifications Figure 0.9   EMA: Ecological momentary assessment55  impact the cellular mechanisms by which PA and exercise training impact cognitive health; and 2) alter the connectivity of the brain such that there is a negative impact on cognitive function. The authors of these reviews are quick to point out that epidemiological data are needed to confirm this preliminary evidence [6, 7].  Very little is known about the epidemiology of SB. Most of the evidence surrounding the effects of SB on neurophysiology is based on the control condition of exercise trials [261]; however, SB is not just the inverse of exercise or PA. Preliminary epidemiological evidence does suggest SB is associated with poorer cognitive function and a plausible mechanism is emerging by which SB is associated with cognitive decline [6, 7]. Recent data suggest prolonged sedentary time impairs glucose and lipid metabolism [88], which are both recognised as risk factors for cognitive decline and all-cause dementia [262, 263]. Only one study to date has examined the relationship between SB and brain health [264]. The authors found that higher amounts of self-reported SB were associated with decreased medial temporal lobe thickness. There is also evidence that SB is related to cognitive decline by analogy; SB is associated with many chronic diseases [265-267], which are also associated with cognitive impairment and dementia risk [268-270].   To my knowledge, there have been no RCTs conducted to examine whether reducing SB can impact older adult cognitive health. While this type of investigation is of interest, given that it is still unclear whether SB is associated with reduced cognitive function, the conducting of RCTs may need to wait until there is enough epidemiological evidence to suggest reducing SB could improve older adult cognitive health.   56  1.3.2.3.A Mechanism by which sedentary behaviour impacts cognitive health Much is also still unknown about the mechanism by which SB impacts cognitive health. At the present time, it is unclear how SB may cause changes in brain volume and functional connectivity. In addition, it is still unclear which areas of cognitive function are specifically affected by changes in SB. Two reviews have suggested SB may impact cognitive health by 1) negatively impacting the cellular mechanisms by which PA and exercise training impact cognitive health; and 2) altering the connectivity of the brain such that there is a negative impact on cognitive function [6, 7]. A more recent review by Wheeler and colleagues [271] posited that SB influences cognitive health by negatively impacting glycemic control. This hypothesis seems somewhat promising since SB is indeed linked to reduced glucose sensitivity [272], and poorer glycemic control is linked to poorer cognitive health [273]. However, there are no data to confirm this hypothesized mechanism.  1.3.2.4 Summary of the current evidence for how sedentary behaviour can impact older adult cognitive health There is only preliminary evidence at this time that SB impacts older adult cognitive health [6, 7, 271]. Research is still needed to determine whether there is a relationship between increased SB and poorer cognitive function in later life. These data would be beneficial in determining whether interventions to reduce SB—and thus improve cognitive health—may be worth pursuing. The mechanism by which SB impacts cognitive health is also unclear at this time.  1.3.3 Sleep and older adult cognitive health  1.3.3.1 Defining sleep 57  Sleep is defined as a rapidly reversible state of immobility and greatly reduced sensory responsiveness [274]. An important further criterion is that sleep is homeostatically regulated, whereby lost sleep is made up with an increased drive for sleep and a consequent “sleep rebound”. Sleep is also a behavior whose presence, quality, intensity and functions vary between species and across the lifespan. Some animals (including humans and most other primates) use sleep to maximize energy savings by reducing body and brain energy consumption, releasing hormones, and conducting a variety of recuperative processes—all of which are accomplished by finding a safe sleeping site that does not offer threat of predation [274-276]. Some species appear to be able to accomplish these processes during the waking state [277, 278]. Sleep is therefore not a universal state with the same underlying function in all species [279, 280], and human sleep is thus distinct.    Science has long recognized that human sleep is a distinct physiological and psychological phenomena (Figure 1.10; [281, 282]). Physiological sleep are the distinct physiological processes and structure of sleep (i.e., sleep architecture; [87]). The physiological processes of sleep include  hormone regulation, recuperative processes, and all other physiological processes that occur during sleep in the central nervous system (CNS) and the periphery. There are two structural types of sleep: non-rapid eye-movement (NREM) and rapid eye-movement (REM) sleep. NREM sleep is divided into stages 1, 2, 3, and 4 (or 3 stages in some more recent classification systems), representing a continuum of relative depth. Stages 3 and 4 are classified as slow-wave sleep (SWS), or deep sleep, and are also used as a marker of sleep depth and sleep quality [283]. Psychological sleep is the 1) perceptions and attitudes (e.g., anxiety, fatigue, etc.); 2) behaviours and habits (betime routines, sleep patterns, etc.); and 3) mental characteristics (mood states, psychological traits, mental disorders, etc.). 58  Figure 1.10 Dimensions of sleep and their associations Figure 0.10   59  Each component of physiological sleep (i.e., physiological processes and sleep structure) is tightly linked, and centrally controlled by the brain [87, 281, 282]. The individual components of psychological sleep are also related to each other, and physiological and psychological sleep are also dynamically associated with each other. An example of this complex interrelationship can be understood within the contexts of chronic insomnia [284]. Chronic insomnia can occur due to physiological processes (e.g., hormonal imbalances) which can alter sleep architecture due to shortened sleep cycles. These changes in sleep physiology can also affect perceptions around sleep, as well as lead to long-term changes in mood states and psychological traits. Conversely, psychological sleep changes due to anxiety and/or other external stressors can lead to changes in behaviour and perceptions of sleep, as well as changes in the physiological nature of sleep.  The term sleep quality is also widely used by researchers, clinicians, and the public in reference to how well a person sleeps (both physiological sleep and psychological sleep); however, the term sleep quality is vague and has lacked definitional consensus until recently. The National Sleep Foundation  has recently defined several aspects of sleep quality including: 1) sleep efficiency (i.e., ratio of time spent sleeping to time spent trying to sleep); 2) sleep latency (length of time in minutes it takes to transition from wake to sleep); 3) sleep duration (total time spent sleeping); 4) awakenings (number of times a person wakes after imitating sleep); 5) wake after sleep onset (WASO; the time spent awake after sleep has been initiated and before final awakening); and 6) sleep architecture [283]. Given that each of these markers of sleep quality can impact health, I define sleep quality as the distinct attributes of sleep that impact health and well-being.   60  Sleep quality is both objective and subjective (Figure 1.11; [285]). Objective sleep quality can be classified according to the physiological aspects of sleep, while subjective sleep quality is based on how an individual feels about their sleep and can thus reflect both physiological and psychological sleep. Importantly, older adults’ objective sleep quality is poorly correlated to their subjective sleep quality [286], suggesting that objective and subjective sleep quality provide different information about older adult sleep.    Figure 1.11 Objective and subjective quality and their relationships to the dimensions of sleep Figure 0.11    Sleep and sleep quality are also inextricably linked to the function of circadian rhythms. Briefly, circadian rhythms are ~24-hour cyclic changes in physiology and behavior that are governed by various biological clocks which coordinate the sleep-wake cycle with the solar light-dark cycle [287-289]. Key features of circadian rhythms include the synchronizing effect of light-dark cycles, persistence of the rhythmicity in constant darkness, and negative masking by light, leading to 61  rhythmic behaviour even in an animal that lacks the ability to maintain rhythms in constant conditions. This rhythmic behaviour in the presence of a rhythmic environment highlights the difference between a rhythm which is endogenously generated, and one which is set by the environment [290]. Rhythms that are observed in a rhythmic environment are referred to as diurnal (i.e., daily) rhythms and, thus, normal human circadian rhythms are diurnal.  The process by which the biological clock is synchronized with the solar light-dark cycle (i.e., entrainment) is controlled by the suprachiasmatic nuclei (SCN), which is located directly above the optic chiasm in the hypothalamus of the brain [291]. Under normal conditions, the SCN functions as “the master biological clock” of the CNS and interacts with the homeostatic recovery process that increases sleep need as a simple function of prior wakefulness with the function of the circadian clock. The entrainment of the biological clock is accomplished through certain external stimuli, known as zeitgebers (from the German time-givers). These time-givers help to prevent inadvertent drifting or divergence of the biological clock from the 24-hour day. Zeitgebers can be used as a chronobiotic—that is, a therapeutic agent to help realign the biological clock with the solar light-dark cycle. I will discuss two chronobiotics in detail in Section 1.5.2.3: light and activity.  1.3.3.2 Sleep and circadian rhythms measurement  1.3.3.2.A Measuring sleep Sleep is thus a complex physiological and psychological phenomena [87, 281], and the precise measurement of sleep in humans is challenging. As with PA and SB, sleep can be measured objectively and subjectively. Given that objective measures of sleep index the physiological 62  aspects of sleep, there are two distinct dimensions of sleep which are assessed using objective measurement. Physiological measurement provides information about the distinct physiological processes and architecture of sleep. Biomechanic measurement provides information about the movement of the body during sleep, and can be used to estimate sleep quantity and quality. Subjective measures can be used for understanding the dimension of psychological and behavioural measurement—that is, the distinct behaviours, perceptions and characteristics of sleep and sleep quality. Objective and subjective measures of sleep thus measure different components of sleep; hence, there is no best measure for assessing sleep (Figure 1.12). The precise measurement of sleep also requires that measures of sleep have evidence of validity and reliability, are population specific, and are sensitive to change.   The criterion measure for measuring objective sleep and sleep quality is polysomnography (PSG) [292]. PSG monitors a number of physiological processes while a subject is sleeping including brain activity (i.e., EEG), eye movements, muscle activity, heart rate, breathing functions (respiratory air flow and respiratory effort), and pulse oximetry. PSG can thus provide information about sleep architecture, sleep related breathing, and other markers of sleep quality. While PSG thus provides accurate, reliable, and sensitive information about a person’s sleep, PSG is expensive, time-consuming, and requires significant participant and researcher burden. Indeed the invasive nature of PSG—usually requiring an overnight stay in a sleep laboratory or clinic—makes long-term multi-night recordings impractical.   Estimating sleep quality using wrist-worn actigraphy is an increasingly popular alternative for objectively-measuring sleep quality, especially since these devices can be used to observe multiple 63    Figure 1.12 Measures of sleep and their classifications Figure 0.12  EMA: Ecological momentary assesment64  days of sleep under normal daily-living conditions. Wrist-worn accelerometers have also been validated for measurement of sleep parameters by comparison with PSG [293], and thus actigraphy is currently accepted as a valid, practical alternative to PSG, allowing for long-term continuous sleep assessments at home [294, 295]. While wrist-worn actigraphy does provide valid estimates of sleep, it is open to issues of validity and reliability compared to PSG—especially among individuals with chronic insomnia [296]. Wrist-worn actigraphy also does not provide information about sleep architecture. Sleep can also be objectively estimated using multimodal sensors [297], however these devices provide slightly less accurate estimates of sleep than wrist-worn actigraphy—likely because multimodal sensors cannot detect small subtle changes in movement that may indicate wakefulness. Like wrist-worn actigraphy, multimodal sensors cannot provide information about sleep architecture.  Throughout my thesis, I measured objective sleep using the MW8 wrist-worn actigraph. The MW8 has preliminary evidence of validity for estimating sleep [298]. I have also recently examined the number of days needed to reliably estimate sleep using the MW8 (under minor revision in Sleep Science and Practice, Appendix B); the data appear to indicate that ≥7 days of monitoring can provide estimates of sleep with evidence of reliability. Sleep can also be measured objectively using the SWA [297], however none of my thesis studies measured sleep using this device.  Sleep can also be measured using subjective methods such as diaries and questionnaires. Importantly, subjective measures of sleep likely measure different aspects of older adult sleep than objective measures [286]. Subjective measures of sleep are quick and easy to administer and score, and can discriminate “good” vs. “poor” sleepers, but they are not able to detect subtle but clinically 65  important changes in sleep quality due to age or disease. Throughout my thesis studies, I have measured sleep subjectively using the Pittsburgh Sleep Quality Index (PSQI; [299-301]) as well as the consensus sleep diary (CSD; [302]); each of these instruments has evidence of validity and reliability for measuring older adult subjective sleep quality.  1.3.3.2.B Measuring circadian rhythms The precise measurement of circadian rhythms is perhaps even more challenging than measuring sleep. Circadian regulation can be indexed using objective and subjective methods of measurement, and there are broadly three dimensions of circadian rhythms that can be measured in humans (Figure 1.13). Physiological measurement examines the physiological markers of circadian physiology (e.g., melatonin levels, core body temperature, etc.), while biomechanical measurement indexes the activity and movement of an individual as a less invasive and burdensome method for estimating circadian regulation; each of these methods are objective. Behavioural measurement provides information about circadian-related behaviour using subjective methods. Each of these measures has benefits and drawbacks and thus there is no single best measure for assessing circadian regulation.   Perhaps the gold standard for measuring circadian physiology is the use of physiological markers at multiple time points throughout a day(s) [288, 303]. These data can provide precise information about an individual’s daily circadian rhythm. However, the ~24-hour diurinal fluctuations of circadian rhythms makes field-measurements of circadian markers extremely difficult. Indeed, most studies of circadian physiology are lab-based studies wherein  biomarkers of circadian physiology are measured (i.e., core body temperature, melatonin levels, cortisol, etc.; [288, 303]).  66  Figure 1.13 Measures of circadian rhythms and their dimensions Figure 0.13   67  Some aspects of circadian regulation can also be measured biomechanically using actigraphy (most commonly wrist-worn actigraphy), whereby fluctuations in activity are used to estimate differences in circadian rhythms [304]. However, these measures are crude estimates of circadian timing, and cannot easily determine different phases in the circadian rhythm. Moreover, the sensitivity of these measures to change over time is unknown. It might also be possible to measure circadian rhythms using multimodal sensors such as the SWA, however no such tool has been developed to my knowledge.  Lastly, subjective questionnaires can provide some information about circadian regulation. Several questionnaires exist which can provide some information about an individual’s circadian behaviour, most specifically their chronotype—or the timing of a person’s sleep-wake cycle with the solar light-dark cycle [305-307]. However, these data are prone to self-report bias and cannot provide information about whether an individual is experiencing circadian dysregulation or is merely an early- or late-riser. It is also unclear whether these types of measures can be used to measure changes in circadian regulation over time. Given the challenges and limitations of measuring circadian regulation using field-based methods, I did not include any measures of circadian regulation as an outcome of interest in my thesis studies.   1.3.3.3 Sleep and circadian rhythms in normal aging The principal entraining zeitgeber for the human biological clock is light [308, 309], exerting its influence on retinal ganglion cells containing melanopsin [310-313]. Retinal light exposure directly stimulates the activity of the SCN, which phase delays the biological clock such that the desire for sleep decreases and wakefulness increases (or is maintained); reduced retinal light 68  exposure results in less activity of the SCN and increases the desire to sleep by phase advancing the biological clock [291]. Thus, under normal circumstances, the biological clock is entrained to the solar light-dark cycle through the regulation of the SCN—which helps maintain a regular sleep-wake cycle [314].  However, aging significantly alters the functioning of circadian rhythms. Aging is associated with the biological clock initiating sleep-promoting mechanisms earlier in the day [315, 316], and a decreased amplitude in circadian signals, which increase sleep need [317, 318]. This weakening of circadian regulation with aging likely plays a prominent role in the fragmentation of sleep-wake rhythms observed in older adults during 1) the wake maintenance zone, which occurs 2-3 hours before habitual bedtime; and 2) the sleep maintenance zone, which occurs 2-3 hours before habitual wake time [2]. In addition, older adults have reduced sensitivity to light due to age-related loss of retinal ganglion cells and axons [319], which leads to poorer functioning of the SCN and divergence of the biological clock from the solar light-dark cycle [320]. Behavioural changes in older adulthood—such as spending less time outdoors—could also further decrease bright light exposure, which may be a contributing factor to the decreased amplitude of circadian rhythms [2].  This age-associated weakening in circadian regulation may also be linked to declines in sleep quality in older adulthood. Sleep changes as a function of normal aging, both in terms of decreased quality and quantity [321, 322]. More than half of adults over 65 years report at least one chronic sleep complaint—the most common complaints being the inability to stay asleep at night, and excessive daytime sleepiness resulting in frequent daytime naps [323]. These complaints, in particular the reports of excessive daytime sleepiness (a key indicator of accumulated sleep debt 69  [324, 325]), suggest that age-related changes in sleep are likely the result of something beyond reduced need for sleep. The evidence therefore suggests that: 1) normal aging may disrupt the function of circadian rhythms; and 2) these age-related changes in the functioning of circadian rhythms may explain the declines in both sleep quality and quantity as people age.  1.3.3.4 Sleep, circadian rhythms, and older adult cognitive health The effects of poor sleep on cognitive health and functions are apparent following both acute and chronic poor sleep [326]. Neurocognitive impairments following acute sleep loss are experienced almost universally, and include impairments in attentional processing, executive function, memory, as well as emotional regulation and sensory perception [327-330]. Importantly, optimal cognitive functioning is integral for older adult quality of life since it is linked to physical function and independence [331, 332], emotional regulation [333, 334], and even eating behaviour [335]. Attentional failures or lapses due to sleep loss are considered the primary causative factor underlying fatigue-specific automobile accidents [336, 337], with a level of psychomotor impairment seen following acute sleep loss similar to that observed during alcohol intoxication [338]. Perhaps most importantly, the observed neurocognitive impairments which are a consequence of poor sleep can be attributed to sub-optimal functioning of the prefrontal cortex [326], which is the principle cortical area responsible for higher-level cognitive processes [339, 340].  While it is clear that poor sleep quality can have an immediate impact on cognitive health, the chronic effects of poor sleep can have even more sinister consequences. Indeed, poor sleep quality is recognized as an important predictor of AD [341]. Older adults diagnosed with obstructive sleep 70  apnea (OSA)—a common chronic sleep disorder characterized by frequent episodes of upper airway collapse during sleep, which results in recurrent arousals from sleep [342]—convert to MCI and AD at a younger age [343]. However, successfully treating OSA can delay the age of MCI onset [343], and improve cognitive function among adults with AD [344]. Poor sleep is also more prevalent among individuals with cognitive impairment as compared to their cognitively healthy peers [345], and epidemiological evidence indicates that poor sleep quality is associated with an increased risk of progression from MCI to dementia [346].  Poor sleep can also contribute to AD pathophysiology, and disruptions in sleep quality and circadian alignment represent typical AD biomarkers. Circadian dysregulation is also one of the hallmarks of AD progression [2]. In fact, sleep disruptions that occur in AD are often exaggerated in a way that implies a form of accelerated or hyper-aging [347], such that fragmentation of sleep-wake rhythms in adults with AD is more akin to much older adults without AD. Moreover, the SCN of older adults with AD is significantly atrophied compared to their cognitively healthy peers, which likely contributes to the fragmentation of sleep-wake cycles in AD [348-350]. Increases in cortical Aβ leads to increases in sleep fragmentation and disrupts diurnal rhythms in the APPswe/PS1δE9 mouse model of AD [351]. Chronic sleep restriction and corresponding increases in wake-time significantly escalates Aβ accumulation in transgenic Tg2576 mice [352]. Increased Aβ load is linked to disrupted NREM slow-wave sleep (i.e., stages 3 and 4) and impaired hippocampal-related memory consolidation [353]. A recent experiment randomly assigned healthy middle aged men to either 1 night of total sleep deprivation or their normal sleep routine [354]; the results indicated that 1 night of sleep deprivation significantly increased cerebrospinal fluid levels of  Aβ. A seperate PET study showed that acute sleep deprivation over one night increased 71  Aβ burden in brain regions implicated in AD [355]. It therefore appears that a vicious cycle of accelerating AD progression may occur with poor sleep—wherein poor sleep quality causes an increase in AD progression, and vice-versa [9].    While poor sleep thus appears to contribute to cognitive decline and dementia progression, good quality sleep appears to be neuroprotective. For example, NREM sleep promotes the clearance of Aβ that accumulates during wake-time [356], and combats oxidative stress (which is linked to AD pathology) by enhancing cellular restitution processes [9]. Improving sleep quality may also be an especially potent therapy for populations with a high risk for dementia, such as populations with the APOE-ε4 allele, who exhibit significant sleep deficits [357]. Given that sleep is a modifiable behaviour which can target multiple cognitive processes [358-360], promoting older adult sleep appears to be an important strategy for maintaining cognitive health in later life.  1.3.3.5 Summary of the current evidence for how sleep and circadian rhythms can impact older adult cognitive health There is thus a growing body of evidence that indicates sleep and circadian regulation are dynamically related to cognitive health. However, it is not yet clear whether improving sleep and circadian regulation can also improve cognitive health. It is also unclear whether improving circadian regulation can improve the sleep of older adults at risk for dementia.  1.3.4 Summary of the current evidence for how each time-use activity behaviours impacts older adult cognitive health 72  Current evidence thus indicates that PA, SB, and sleep may each play a critical role in maintaining older adult cognitive health. Evidence from animal models, observational studies, and RCTs indicates that PA (especially PA in the form of exercise training) can promote older adult cognitive health. Less is clear about how SB can impact older adult cognitive health, however preliminary evidence suggests that it may be associated with cognitive decline. Sleep appears to be critical for healthy cognitive aging, although it remains to be determined whether improving sleep promotes cognitive health.  1.4 The dynamic relationships of time-use activity behaviours and circadian rhythms with older adult cognitive health  1.4.1 Concurrent measurement and analysis of multiple time-use activity behaviours and/or circadian rhythms There are a few methods available for measuring PA, SB, and sleep concurrently. Presently, available methods include actigraphy, multimodal sensors, and systematic direct observation. Irrespective of the method used to capture all three time-use activity behaviours concurrently, only limited aspects of each behaviour can be estimated. For example, none of these methods can capture the psychological and behavioural aspects of sleep, and thus there is no single best measure for capturing all time-use behaviours concomittantly.  A field-method capable of capturing time-use activity behaviour within the context of the circadian cycle is not currently available. While wrist-worn actigraphy (in particular the MW8) is capable of approximating the circadian rest-activity cycle and measuring each time-use activity behaviour 73  [136, 298, 304], there is not currently an analytical framework for measuring time-use activity behaviour within the context of the circadian cycle. This is perhaps because circadian rhythms are a set of physiological processes that have little ability for self-determination [12, 288, 307, 317]; on the other hand, time-use activity behaviours are at least to some extent a choice. For example, an individual cannot control diurinal changes in melatonin and cortisol, however they can decide whether to go for a walk, watch television, or go to bed earlier (or later).  Some approaches have been suggested for analyzing the contribution of each time-use activity behaviour to health. One approach is isotemporal substitution analysis, which simultaneously models the specific activity being performed and the specific activity being displaced in an equal time-exchange manner (e.g., the impact of exchanging 1 minute of PA for 1 minute of SB, or vice-versea; [361]). Pedišić and colleagues [3, 362] recently argued for the use of compositional data analysis for examining the contributions of each time-use activity behaviour to health. In this approach, time-use activity behaviour is considered to be a composition matrix and can be analyzed according to the clustering of time-use activity behaviours at different times throughout the 24-hour day.   However, as I have highlighted above, each time-use activity behaviour is complex and multi-dimensional. Time-use allocation is not the only factor which must be considered in the analyses of these behaviours. For example, it is unlikely that 9 hours of poor sleep (i.e., low sleep efficiency, minimal stage 3 and 4 sleep, etc.) is somehow better than 7 hours of high quality sleep. It therefore seems that at the present time such analysis techniques are ill-fitted to elucidating the combined role of PA, SB, and sleep on older adult cognitive health. Furthermore, a long-standing principle 74  of scientific philosophy has been that of scientific economy—that is, complexity should not be assumed unnecessarily [363]. In this instance, I argue that examining the interactions of complex biological phenomena such as time-use activity behaviours using mathematically complex models which are difficult to interpret and/or require assumptions about the equal importance of a variable (i.e., time) seems an ill-fitted method to understanding how time-use activity behaviours can impact older adult cognitive health. Instead, I suggest that using the general linear model (GLM) can provide a simple, easily interpretable, and well-understood structure by which researchers can untangle how PA, SB, and sleep impact older adult cognitive health.   1.4.2 Current evidence on the dynamic relationships between time-use activity behaviours and older adult cognitive health Epidemiological studies have consistently found people with greater PA report sleeping better compared to more sedentary individuals [11]. While the reason for why PA and sleep are related is still unclear, current evidence suggests three possible explanations [364, 365]. First, negative affective states (i.e., depressive symptoms and anxiety) contribute to poor sleep [366], and PA counters this via its antidepressant and anxiolytic effects [367, 368]. Second, obesity is related to poorer sleep quality [369], and PA has a direct impact on weight regulation which may promote  better sleep [370]. Third, regular PA improves or maintains physical function [371]; poor physical function is associated with poorer sleep quality in older adults [372]. However, much of the evidence to date is based on self-reported PA and self-reported sleep [11, 365], which can be quite different from objectively-measured data [286, 373]. Of final importance, there is at least preliminary evidence that the relationship between PA and sleep may be bi-directional [374].  75  The relationship between PA and sleep quality may weaken with age [375]. One potential reason for this functional weakening is that we simply need less sleep as we age [376]. In addition, there is some evidence that underlying changes in older adult neurobiology (e.g., neural atrophy, nocturnal hypoxia, neuroendocrine changes, and altered neuromodulation) may reduce the potential to impact sleep quality through strategies such as PA [296].  Although there is ample evidence that both PA and sleep quality can impact older adult cognitive health, together with a growing body of evidence suggesting that PA is associated with better sleep quality, it is unclear whether PA and sleep impact cognitive health through divergent or convergent mechanisms. Recent evidence suggests that sleep efficiency may mediate the relationship between PA and cognitive function [377]; these results provide at least initial support for the restoration hypothesis, which suggests that energy expenditure (i.e., PA) stimulates a restoration process by which sleep allows the body and brain to recuperate [364].   By comparison, little is known about how SB can impact older adult sleep. A systematic review and meta-analysis of 16 studies of adults aged 18+ years indicated that higher SB was associated with an 18% and 38% increased risk for insomnia and sleep disturbances, respectively [378]; however, only 3 studies of adults over 55 years of age were included in this review. Other recent investigations among older adults also suggest SB may impact sleep. Madden and colleagues [379] determined that higher objectively-measured SB was associated with reduced objectively-measured sleep efficiency among community-dwelling older adults. Kline and colleagues [380] also found that older adults who engaged in greater amounts of self-reported SB had an increased risk of PSG-determined sleep disordered breathing. Seoul and colleagues [381] determined that 76  higher SB was associated with poorer PSQI score. Interestingly, a recent cross-sectional investigation among older adults with dementia in a long-term care facility determined that while participants were highly sedentary, they also obtained ~7 hours of sleep each night [382]. It is thus unclear how SB may impact older adult sleep, although preliminary evidence suggests that SB may have deleterious effects on sleep.   Few studies have also examined how PA and SB are associated with each other in older adulthood. A systematic review of observational studies examining the relationships between PA and SB in adults 18+ years found a negative association between PA and SB [10]. However, most of this evidence is based on self-reported PA and self-reported SB, and only one study examined these associations in middle- and older-adulthood exclusively [383]. More research is thus needed to examine how the relationships between PA and SB patterns change in older adulthood.  1.4.3 Current evidence on the dynamic relationships between time-use activity behaviours and circadian rhythms PA in the form of exercise training appears to have chronobiotic effects in humans [320, 384, 385]. The chronobiotic properties of PA have been well established in various animal models, with wheel running being the most commonly used approach in rodents [386, 387]. Briefly, the circadian rhythms of rodents can be entrained by regularly scheduled exercise [388, 389], and single episodes of running in novel running wheels can advance the circadian clock [386, 390], while presentation of a novel running wheel can accelerate re-entrainment and induce single phase advances as large as 12 hours [391]. However, human studies on the chronobiotic properties of PA are difficult, given the challenges associated with isolating the effects of exercise from other 77  factors such as light, food, and social influences [392, 393]. However, studies of blind people who lack sensitivity to light—but remain able to entrain to daily work/social schedules without the involvement of exogenous melatonin—suggest that non-photic stimuli are capable of synchronizing circadian rhythms [393]. It has yet to be determined whether it is PA, social influences, regularly scheduled mealtimes, or a combination of these (and possibly other potential zeitgebers), which provides the critical entrainment signal.  While the effects of PA as a chronobiotic have yet to be fully established, PA does appear to have a specific phase-response curve [393]. Briefly, PA performed in the morning or early afternoon does not appear to have a consistent effect on phase shifts of the biological clock; however, engaging in PA in the late afternoon causes a modest phase advance of the biological clock, while late night PA causes a modest phase delay of the biological clock [384, 385]. Importantly, the effects of PA as a zeitgeber have been found in both young adults and older adults [385]. The time-based response to how PA can impact the SCN is hypothesized to coincide with the timing of the opening of the “sleep gate”—the shift of the biological clock from generating a waking signal, which reduces sleep need, to generating a signal that facilitates sleep [12].   Very little is known about how SB may impact circadian regulation. A recent randomized cross-over study of young men (N= 16) found that a 24-hour sedentary protocol (i.e., <1,000 steps) lead to significant reductions in circadian rhythm amplitude compared to a 24-hour PA protocol (>15,000 steps) [394]. However, this appears to be the first analysis wherein the impact of SB on circadian regulation was examined.  78  1.4.4 Summary of the current evidence examining the relationships between time-use activity behaviours, circadian rhythms, and cognitive health Increasing older adult PA, reducing SB, and improving older adult sleep and circadian regulation appear to be viable strategies for maintaining older adult cognitive health. There is growing evidence that PA can positively impact older adult sleep and circadian alignment, and there is at least preliminary evidence that PA and sleep may improve cognitive health through multiple mechanisms—both convergent and divergent. However, little is known about how SB can impact older adult sleep and circadian regulation, and it is currently unknown whether SB impacts sleep through a convergent or divergent mechanism from PA.  1.5 Current strategies to promote older adult physical activity, reduce sedentary behaviour, and improve sleep and circadian rhythms In this section, I will provide a brief review of the methods for promoting each time-use activity behaviour. In-depth reviews of current methods for promoting PA [395, 396], reducing SB [397], and improving sleep [398-400] can be found elsewhere. I will instead review the methods I used for promoting PA and SB in my thesis (Section 1.5.1), as well as provide a brief discussion about the current methods for promoting older adult sleep and circadian regulation (Section 1.5.2).   1.5.1 Strategies to promote physical activity and reduce sedentary behaviour There are a wide variety of evidence-based methods for promoting PA and reducing SB [395-397].  While each of these methods may help to improve PA and SB, there is not a universal best method for promoting PA and SB change. Indeed, there is no silver bullet for promoting PA and reducing 79  SB, and if there were, one would expect that researchers would be much closer to solving the global pandemic of low PA and high SB [401].  One method for promoting PA and reducing SB is the Brief Action Planning (BAP) approach [402]. BAP is a highly structured, stepped-care, self-management support technique which is grounded in the principles and practice of motivational interviewing—a therapeutic technique designed to enhance an individual’s motivation to change behaviours and guide him or her into action (Figure 1.14; [403, 404]). BAP can be used to facilitate goal setting and action planning to build self-efficacy about PA and SB (Figure 1.16; [405-407]).   As described in Figures 1.14 and 1.15, the initial consultation using BAP involves asking three questions about an individual’s PA and then focusing on building five skills. The questions asked are meant to elicit talk about behaviour change, while the skill building opportunities are used to determine strategies which will help the person reach and obtain their behaviour change goals. During follow-up consultations, the individual is asked whether they reached their behaviour change goals, and then modifications to the goals are made accordingly.  While BAP is an evidence-based strategy for promoting PA and SB, there are some disadvantages to the use of the technique to promote PA and reduce SB. Most importantly, BAP is individually-focused; the patient (or research participant) is asked questions about whether or not they want to increase their PA and reduce their SB. According to the theory of planned behaviour [408], an individual will not change their behaviour without: 1) motivation; and 2) perceived behavioural80  Figure 1.14 Overview of the Brief Action Planning Approach during initial and follow-up consultations Figure 0.14   81  Figure 1.15   Brief Action Planning Approach for A) promoting physical activity and B) reducing sedentary behaviour Figure 0.15  82  control. If a person does not have the motivation for behaviour change, or thinks their behaviour is beyond their control, then changes to behaviour are highly unlikely. BAP may thus require multiple sessions in order to elicit talk about changing PA and SB habits. The implementation of BAP also requires extensive time on the part of the counselor and the patient. Indeed, a minimum of 20-minutes is suggested during the initial consultation [402], and bi-weekly follow-up sessions also require an agreed upon time and place for the counselor and patient to discuss the implementation of the action plan. These challenges to the implementation of BAP make it difficult for the intervention strategy to be implemented on a large-scale; however, small-scale, individually-based implementations of BAP can be highly successful for promoting PA and reducing SB [405-407].  Another strategy for promoting PA and reducing SB is consumer-available, wearable activity-monitoring technology. These devices present several distinct advantages as a PA promotion and SB reduction tool including: 1) adults typically perceive activity-monitors as useful [409]; 2) these devices incorporate multiple behavioural change strategies [410]; and 3) clinicians can readily use these devices to help promote behaviour change among their underactive patients [411]. While the use of these devices alone might not be enough to induce behaviour change [412], the delivery of BAP to promote PA and reduce SB might be enhanced by implementing wearable activity-monitors [145].  1.5.2 Strategies for improving sleep and circadian rhythms There are a number of methods available for improving sleep quality. Broadly, these can be classified as pharmacological and non-pharmacological. Pharmacological interventions to 83  improve sleep are outside the scope of my thesis and have been reviewed elsewhere [413, 414]. There are three methods of promoting sleep which I will now briefly review: 1) cognitive behavioural therapy, with a special emphasis on sleep hygiene education (Section 1.5.2.1); 2) PA and exercise training (Section 1.5.2.2); and 3) chronotherapy (Section 1.5.2.3).   1.5.2.1 Cognitive behavioural therapy and sleep hygiene education Cognitive behavioural therapy (CBT) is a powerful tool for promoting better sleep quality [284, 415-417]. Broadly, CBT is a psychological treatment designed to break patterns of maladaptive thinking and behaviour. The treatment consists of a behavioural component (stimulus control, sleep restriction, and relaxation techniques) combined with cognitive strategies (managing sleep-related worries, racing mind, and intrusive thoughts) and education (sleep hygiene). Meta-analyses indicate that CBT has moderate-to-large and durable effects on subjective sleep quality, and objective sleep efficiency, latency, and WASO; the effects of CBT on sleep are also comparable in size to pharmacological therapies [418-420].  As highlighted above, one aspect of CBT is sleep hygiene education. Sleep hygiene is a set of behavioural practices that can impact sleep quality, such that poor sleep hygiene can exacerbate or even cause poor sleep, while good sleep hygiene results in feeling more rested and alert upon awakening [421-423]. Sleep hygiene education thus teaches strategies which can enhance sleep (e.g., not watching television before bed, bedtime habit formations, or avoiding the bedroom unless one is tired and ready to sleep). It is also hypothesized that sleep hygiene might be a useful tool for promoting other behaviours which can impact sleep quality and circadian regulation—such as light exposure and PA [424]. 84   Although sleep hygiene education is currently one of the recommended strategies for promoting better sleep [425], there is insufficient evidence at this time to suggest that sleep hygiene education alone can improve sleep [421]. Current recommendations therefore suggest that sleep hygiene should be used in conjunction with other therapeutic approaches for enhancing sleep [426].  1.5.2.2 Physical activity and exercise training As discussed in Section 1.4.2, there is a growing body of evidence which indicates that PA and exercise training may each positively impact older adult sleep [427, 428]; however, it seems that the effects of PA and exercise training weaken with age [365]. While it is thus unclear whether promoting PA or exercise training should be the frontline approach for older adults experiencing poor sleep, current recommendations suggest that daytime PA and exercise training may help enhance sleep quality [429]. The precise prescription of PA and exercise training for promoting sleep is also currently unclear [11, 365]; specifically, the precise frequency, intensity, type, or duration of PA and exercise training which is most beneficial for promoting sleep. As I will discuss in Section 1.5.2.3.B, timed PA and exercise training can also be used in chronotherapy to promote sleep.  1.5.2.3 Chronotherapy Chronotherapy refers to a set of intervention strategies that use effectively timed zeitgebers in order to realign the biological clock with the solar light-dark cycle [2, 430]. There are a number of different zeitgebers which can entrain the biological clock including light, food, PA and exercise training, and social influences [291, 392, 393]. I will discuss the use of two potential 85  chronotherapuetics which I used in my thesis studies: 1) bright light therapy (BLT; Section 1.5.2.3.A); and 2) PA and exercise training (Section 1.5.2.3.B).  1.5.2.3.A Bright light therapy The principal entraining zeitgeber for the human biological clock is light [308, 309], which exerts its influence on retinal ganglion cells containing melanopsin [310-313]. Retinal light exposure directly stimulates the activity of the SCN (Figure 1.16), causing a suppression of melatonin production which phase delays the biological clock such that the desire for sleep decreases and wakefulness increases (or is maintained); reduced retinal light exposure results in less activity of the SCN (and less melatonin production) and increases the desire to sleep by phase advancing the biological clock [291]. While the importance of light is thus integral for the proper function of the SCN and the biological clock, older adults have reduced sensitivity to light which leads to poorer function of the SCN and divergence of the biological clock from the solar light-dark cycle [320]. Behavioural changes in  older adulthood—such as spending less time outdoors—can also further decrease bright light exposure, which may be a key factor in decreased amplitude of circadian rhythms [2]. Thus, older adults in particular may benefit from effectively timed bright light to strengthen the entrainment of the SCN to the solar light-dark cycle.  BLT is therefore an increasingly popular chronotherapy strategy [2]. Although the efficacy of BLT as an intervention strategy is currently inconclusive [303, 431], preliminary evidence for the use of BLT to promote sleep among older adults does appear promising. Two separate quasi-experimental interventions found that the subjective sleep quality of older women improved after six days of 1-hour morning BLT, although there were no changes in objective sleep quality 86  Figure 1.16 Brief introduction to the neurophysiological response to light (Panel A) and its use in bright light therapy according to the phase response curve of light (Panel B) Figure 0.16  Panel A: Darkness upregulates melatonin production—a hormone which helps regulate the sleep-wake cycle by increasing sleep signals. Light downregulates melatonin production, thus decreasing sleep signaling.  Panel B: Laboratory experiments indicate that when individuals are exposed to bright light in the evening or early morning prior to their core-body temperature minimum (~6 AM; CBTmin), it causes a phase delay (i.e., desire for sleep decreases and wakefulness increases). Individuals exposed to bright light in the morning after their CBTmin causes a phase advancement (i.e., desire for sleep increases and wakefulness decreases).    87   [432, 433]. Twice-daily 1-hour BLT also improves subjective sleep quality in adults with Parkinson’s disease [434]. Another study determined that 8 hours of daily BLT improved objectively-measured sleep duration and cognitive function among nursing home residents [435]. Mishima and colleagues found that 2-hours of daily morning BLT significantly improved caregiver reported daily sleep duration and nocturnal sleep duration among older adults living with dementia [436].  While these results suggest that BLT might be a useful strategy for improving the sleep of older adults, the precise prescription of BLT to promote sleep is relatively unclear. At the present time, current evidence-based guidelines for the use of BLT to promote sleep and circadian regulation do not provide a clear and concise road-map for the prescription of BLT [437, 438]. Like PA and exercise training, it is largely unknown what frequency of BLT (e.g., morning vs. evening), intensity of light (10,000 lux vs. 100,000 lux), type of light (white light vs. blue light), or duration of BLT can best promote sleep. Even more importantly, the biological clock is not equally amenable to shifts at each phase in the circadian rhythm [291]. A zeitgeber can cause the biological clock to phase advance, phase delay, or be entirely phase neutral depending on the biological clock time at which a zeitgeber is administered. It is thus highly likely that successful BLT requires an individualized approach where proper timing, intensity, frequency, duration, and type of light is essential [2].  1.5.2.3.B Physical activity and exercise training Another potential zeitgeber for use as chronotherapy is PA or PA in the form of exercise training [320, 384, 385]. Briefly, PA performed in the morning or early afternoon does not appear to have  88   a consistent effect on phase shifts of the biological clock; however, engaging in PA in the late afternoon causes a phase advance of the biological clock, while late night PA causes phase delay of the biological clock [384, 385]. The time-based response to how PA can impact the SCN is hypothesized to coincide with the timing of the opening of the sleep gate—the shift of the biological clock from generating a waking signal which reduces sleep need, to generating a signal which facilitates sleep [12].  However, the use of PA as a strategy to maintain circadian alignment may be challenging from a practical standpoint. The current evidence describing the effects of PA as a zeitgeber comes from controlled laboratory experiments, where the timing and intensity of PA in the form of exercise is tightly controlled. Conducting an intervention where participants would be asked to engage in regularly timed PA at a prescribed intensity would: 1) be burdensome to participants; and 2) require enormous resources to ensure participant adherence.   The current body of evidence therefore suggests that PA has chronobiotic effects, which may play a role in promoting good quality sleep. However, it is still unclear whether PA can improve the sleep of older adults given that circadian regulation – as well as the relationship between PA and sleep – appear to weaken with age.   1.5.3 Summary of current strategies to promote physical activity, reduce sedentary behaviour, and improve sleep and circadian rhythms There are a number of different evidence-based methods for promoting PA and reducing SB. While there is no single best method for increasing PA and reducing SB, BAP offers an effective  89   evidence-based approach to PA and SB behaviour change which may be enhanced through wearable-technology. There is also no single best method for promoting sleep. Several promising evidence-based behavioural methods for promoting sleep include 1) CBT; 2) PA and exercise training; and 3) chronotherapy. One component of CBT that might also be beneficial for improving sleep is sleep hygiene education; however, sleep hygiene education should be combined with other therapies. PA and exercise training might also improve sleep, although the precise dose of PA or exercise training needed to promote sleep is still unclear. Chronotherapy is an evidence-based approach for improving sleep which uses effectively timed zeitgebers (i.e., BLT and PA or exercise training) in order to re-align the biological clock with the solar light-dark cycle. However, the precise timing and doses of each chronotherapeutic in order to improve sleep and circadian regulation are not known.  1.6 Summary of the current research gaps The effects of PA, SB, and sleep on cognitive health are emerging; yet, much is still unknown about how these behaviours impact older adult cognitive health. Below are some of the major research gaps I have identified.  1.6.1 The impact of physical activity on cognitive health While current evidence indicates that PA is associated with better cognitive health, the precise prescription of PA for cognitive health has been elusive. The current PA guidelines suggest that all older adults should obtain at least 150 minutes/week of MVPA for overall health, as well as cognitive health [439]. However, it is unclear whether this prescription needs to be modified for  90   individuals with a higher risk for dementia. Moreover, it is unclear whether certain types of PA are more beneficial for cognitive health.  1.6.2 The impact of sedentary behaviour on cognitive health Far less is known about how SB impacts older adult cognitive health, and thus the question about how SB impacts cognitive health are preliminary. Most importantly, it is unclear if greater amounts of SB are associated with poorer cognitive function. It is also unknown what specific areas of cognition are affected by increased amounts of SB, and very little is known about how SB is associated with brain structure, function, or neurophysiological biomarkers. Finally, it is unclear if reductions in sedentary time can lead to improvements in cognitive health.   1.6.3 The impact of sleep and circadian rhythms on cognitive health While it is clear that there is a bi-directional relationship between sleep and cognitive health, it is unclear whether improving sleep can improve cognitive health. In addition, objective and subjective measures of sleep do not correlate well and may measure different aspects of sleep [286]. Thus it still remains an open question whether improving subjective sleep quality, objective sleep quality, and/or sleep architecture provides similar (or differential) benefits to cognitive health.  Far less is known about how circadian rhythms can impact cognitive health. Although sleep quality is closely tied to circadian function, it is unclear whether improving circadian regulation can promote better sleep (or vice-versa). Importantly, there have been few attempts to use chronotherapy to promote older adult sleep [2]  91    1.6.4 The dynamic relationships of time-use activity behaviours with older adult cognitive health There is an increasing amount of evidence that indicates PA, SB, and sleep are dynamically related to each other and older adult cognitive health; however, much is still unknown about these dynamic relationships and their implications on cognitive health. For example, it is unclear which time-use activity behaviour has the greatest impact on older adult cognitive health. It is also unclear whether PA, SB, and sleep impact cognitive health simultaneously, in synergy, or in silos. There is also little known about whether a combined therapeutic approach of 1) promoting PA; 2) reducing SB; 3) improving sleep quality; or 4) some combination of these strategies can improve the cognitive health of older adults.   The literature is also unclear if differences in PA and SB exist between older adults with MCI, and those without. Specifically, it is currently unknown whether differences in cognitive status can impact the relationships of PA and SB with cognitive health. Due to underlying neurobiological differences between older adults with MCI and those without [2], a functional weakening in the relationships of health behaviours with cognitive function may occur in MCI [147].   It is also unclear what the independent relationships of PA, SB, and sleep are with different aspects of cognitive health. For example, it is unknown whether PA has a stronger association with cognitive health than SB (or vice-versa). The same can also be said for the independent associations of PA (or SB) and sleep with cognitive health. Determining these relationships will  92   be a first step to grasping the independent contributions of these behaviours to older adult cognitive health.    While there is at least initial evidence that greater PA is linked with better sleep [11, 365], most of this evidence is based upon self-reported PA and self-reported sleep. Importantly, objective measures of PA and sleep often measure different things than subjective measures [122, 286]. While current evidence suggests the relationship between PA and sleep may be attenuated in older adults [365], few studies have examined why this might be. One potential reason for this apparent functional weakening in the relationship is that we simply need less sleep as we age [376]. It is also plausible that underlying changes in older adult neurobiology (e.g., neural atrophy, nocturnal hypoxia, neuroendocrine changes, and altered neuromodulation) may reduce the potential to impact sleep quality through strategies such as PA [296]. However, both of these hypotheses lack data from long-term observational studies, and it remains to be seen whether PA can modulate age-associated changes in sleep quality.   1.6.5 Strategies to promote physical activity, reduce sedentary behaviour, and improve sleep  A number of different strategies exist for promoting PA [440-443], reducing SB [397, 444], and improving sleep [398, 399]. Most strategies to increase PA and reduce SB have been only modestly effective, with improvements often being short-lived [445-447]. More research is needed about the long-term effectiveness of non-pharmacological sleep interventions [398]. There is also no evidence to date about the long-term effectiveness of PA and exercise training on sleep [428]. Importantly, it is largely unknown whether any of these strategies are effective methods of  93   behaviour change among older adults with cognitive issues (e.g., MCI), and whether these strategies can consequently improve cognitive health (either with or without cognitive impairment).  While there is preliminary evidence which supports chronotherapy as a technique for promoting sleep [2], the effectiveness of chronotherapy as a strategy to improve sleep and circadian rhythms is largely uncertain. Science has yet to determine the precise timing and prescription of BLT for promoting older adult sleep and circadian regulation, and the long-term efficacy of chronotherapy for improving sleep is hazy [448]. PA as chronotherapy is also still in its infancy [449]. While some preliminary evidence suggests that multimodal chronotherapy (i.e., BLT and PA) in conjunction with sleep hygiene might benefit sleep and cognitive health among older adults with MCI and dementia [448], larger sample sizes and more rigorous studies are needed in order to confirm these results.  1.7 Thesis overview  1.7.1 Main thesis questions The previous sections provided quality evidence that PA, SB, and sleep are each important for older adult cognitive health. My thesis aims to extend our current knowledge in these areas by characterizing how each of these time-use activity behaviours are dynamically related to each other and cognitive health, as well as elucidate how to most effectively promote these behaviours in order to improve older adult cognitive health. Specifically, my thesis aims to answer the following research questions:  94   1. How are time-use activity behaviours associated with older adult cognitive health? 2. What is the dynamic relationship between time-use activity behaviours and older adult cognitive health? 3. Can we promote cognitive health in older adults with and without cognitive impairment through targeted interventions on time-use activity behaviours?  1.7.2 Methodology Herein I outline the primary outcome measures I used throughout my thesis. I have highlighted most of these measures throuhgout Chapter 1, however I now include a brief description of each measure. Greater details about each measure are included in the relevant studies (Chapters 2-7).   1.7.2.1 Physical activity measures  I measured PA using the MW8, SWA, and CHAMPS questionnaire. Each of these measures have evidence of validity and reliabiltiy for estimating older adult PA [136, 137, 141-143, 146, 147, 149].  1.7.2.2 Sedentary behaviour measures I measured SB using the MW8 and SWA. Both of these measures have evidence of validity and reliability for estimating older adult SB [136, 140].  1.7.2.3 Sleep measures I measured sleep using the MW8,  PSQI, and CSD. The MW8 is a tri-axial accelerometer designed to observe acceleration ranging in magnitude from 0.01G to 8G, with a frequency of 3-11Hz. The  95   MW8 is the updated version of the Actiwatch7, an actigraph with evidence of validity against PSG [450, 451]. There is also initial evidence of validity against PSG for the MW8 [298]. All measurements using the MW8 used 60 second epochs [452].  The PSQI is a 19-item questionnaire that assesses subjective sleep using ratings for 7 different aspects of sleep (i.e., global sleep quality; sleep latency; sleep duration; habitual sleep efficiency; sleep disturbance; use of sleep medication; and daytime dysfunction). Participants answer the questionnaire retrospectively, as the questionnaire surveys sleep components spanning the previous month. The questionnaire has good evidence of validity and reliability [301, 453].  Although not technically an outcome measure of sleep in my thesis, participants in the studies from Chapters 3, 4,and 7 were also given the 9-item CSD and asked to complete it each morning upon waking [452]. The responses from the CSD were used to confirm sleep windows as determined by the time stamped event markers. In cases where the event marker and CSD entry disagreed for the start time of the sleep window, we used activity cessation and light sensor data from the MW8 to determine “lights out”. Similarly, when the event marker and CSD entry disagreed for the end of the sleep window, we used activity onset and “lights on” to determine the end of the sleep window. If responses from the CSD entry disagreed with the event markers entered by participants as the start of the day (i.e., finished trying to sleep and awake and out of bed), we used activity onset and light sensor data to determine the start of the day. Similarly, when the event marker and CSD entry disagreed for the end of day (i.e., time spent trying to sleep), we used activity cessation and light sensor data to determine the end of the day.   96   1.7.2.4 Cognitive function measures I measured global cognitive function in Chapters 3,4, and 7 using the ADAS-Cog Plus [454]. The ADAS-Cog Plus uses a multidimensional item response theory model which can flexibly utilize item scores from multiple cognitive assessment instruments to generate a global cognitive function score and standard error of measurement for that score. Scores are defined by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) cohort [455], wherein the mean score for cognitively healthy older adults is about -1.0, the mean for MCI is about 0.0, and the mean for dementia cases is about 1.0. Thus, higher scores indicate poorer cognitive performance. In Chapters 3 and 4, the ADAS-Cog plus score was computed using: 1) the 13-item ADAS-Cog [456]; 2) Montreal Cognitive Assessment (MoCA) [457]; 3) Trail Making Test A and B [458]; 4) Digit Span Forward and Backward [459]; and 5) verbal fluency [458]. In Chapter 7, we also included the Digit Symbol Substitution Test [460] in our computation of ADAS-Cog Plus. Appendix C provides a complete description of these measures.  In Chapter 6, I measured cognitive function using the National Institutes of Health Toolbox Cognition Battery (NIHTB-CB) [461]. Briefly, the NIHTB-CB provides a brief, convenient set of computerized and standardized measures of cognitive function. I examined two specific cognitive subdomains: 1) episodic memory using the picture sequence memory task [462]; and 2) working memory using the list-sorting task [463]. Briefly, the picture sequence memory task assesses episodic memory by having participants remember a sequence of actions embedded within a story. Participants re-arrange several pictures on the computer to match the sequence of events in the story. The list-sorting task assesses working memory by asking participants to repeat the names of orally—and visually—presented stimuli in order of size, from smallest largest. The number of  97   items per set increases from one trial to the next and is discontinued once 2 trials of the same length are failed.  1.7.2.5 Brain structure measures I measured brain structure, specifically cortical thickness, using structural MRI. Structural MRI is a neuroimaging technique that permits quantification of brain volume, curvature, and surface area. This is typically accomplished using a semi-automated analysis pipeline on data acquired from a high-resolution T1-weighted image.  1.7.3 Overview of thesis chapters To address the research questions, this thesis is comprised of six studies; each presented as a separate chapter. Figure 1.17 describes each of the observational studies included in my thesis (Chapters 2-5). Figure 1.18 describes the experimental studies included in my thesis (Chapters 6 and 7).  In Chapter 2, I investigated the current evidence examining the relationship of SB with older adult cognitive function. In this systematic review of observational studies examining the association of SB with older adult cognitive function, I determined that the current evidence suggests greater amounts of SB are associated with poorer cognitive function in later life. However, I also determined that the attributable risk of SB to dementia incidence is unclear at this time, and most of the current evidence for a relationship between SB and cognitive function was based on subjectively-measured SB.   98   Figure 1.17 Overview of the observational studies (Chapters 2-5) in the dissertation Figure 0.17   99   Figure 1.18 Overview of the experimental studies (Chapters 6 & 7) in the dissertation Figure 0.18  Chapter 6 = Dark Gray Arrows; Chapter 7 = Light Gray Arrows  In Chapter 3, I examined whether there are differences in the associations of objectively-measured PA and objectively-measured SB with cognitive function based on MCI status. I also examined whether there were differences in PA and SB levels between healthy older adults and older adults with MCI. In this cross-sectional study, I determined that older adults with MCI are less active and more sedentary than their healthy cognitive counterparts. In addition, I determined that the relationship of PA and SB with cognitive funciton is dependent on MCI status, such that there is a relationship between these behaviours and cognitive function for healthy older adults but not for older adults with MCI.   100   In Chapter 4, I assessed the independent relationships of objectively-measured PA and 1) objectively-measured sleep; and 2) subjectively-measured sleep with older adult cognitive function. In addition, I investigated the relationships of objectively-measured PA with objective and subjective sleep. In this cross-sectional study, I found that objectively-measured PA was associated with cognitive function independent of any measure of sleep. I also determined that objectively-measured sleep efficiency was associated with cognitive function independent of PA, however no other measure of sleep was associated with cognitive function. I did not find that PA was associated with any measure of sleep.  In Chapter 5, I assessed the independent relationships of objectively-measured PA and objectively-measured SB with brain cortical thickness. In this cross-sectional study, I found that PA was associated with greater cerebral cortical thickness in the left superior frontal gyrus and temporal pole, independent of SB. I did not find that SB was associated with brain cortical thickness independent of PA.  In Chapter 6, I examined whether an intervention to increase PA and reduce SB could also improve the cognitive function of older adults with knee osteoarthritis. I also examined whether increases in PA or reductions in SB were associated with improvements in cognitive performance. Using secondary cognitive outcomes from a proof-of-concept RCT, I determined that while the intervention significantly increased objectively-measured PA [145], there were no significant improvements in cognitive function. I also did not find that there was any association between changes in PA or SB and changes in cognitive function.   101   In Chapter 7, I investigated whether a multimodal chronotherapy intervention consisting of 1) individually-timed BLT; and 2) health coaching to promote PA in conjunction with general sleep hygiene education could improve the sleep and cognitive function of older adults with MCI. In this proof-of-concept RCT, I determined that multimodal chronotherapy significantly improved the subjective sleep quality of older adults with MCI, but did not significantly impact objectively-measured sleep. Furthermore, there were no significant improvements in cognitve function from the intervention, and improvements in sleep were not associated with improvements in cognitive performance.  This dissertation concludes by revisiting the proposed research questions with an integrated discussion on the dynamic relationships of PA, SB and sleep with older adult cognitive health. I will then provide an overview of the limitations of my thesis studies and also provide future directions.    102   Chapter 2: What is the association of sedentary behaviour and cognitive function? A systematic review A version of this chapter is published as FALCK RS, Davis JC, Liu-Ambrose T. What is the association between sedentary behaviour and cognitive function? A systematic review. British Journal of Sports Medicine. 2017;51(10):800-811. 2.1 Introduction Currently, one new case of all-cause dementia is detected every 4 seconds around the world [1]. All-cause dementia prevalence is also expected to rise since the number one risk factor is age [464], and the number of older adults worldwide is increasing [465]. Thus, the current lack of effective pharmaceutical treatments for all-cause dementia is creating an urgency to develop non-pharmacological strategies to prevent, or at least delay, the onset and progression of the disease [2]. As a result, lifestyle approaches have become an important line of scientific inquiry and public interest.   Increasing PA is one promising strategy to promote or maintain cognitive health in later life [197]. Accumulating empirical evidence suggests regular PA of an intensity ≥3.0 METs reduces the risk of all-cause dementia by 28% [156]. Thus, meeting current PA guidelines for older adults of 150 minutes/week of MVPA (i.e., activity of ≥3.0 METs) may help reduce all-cause dementia risk, prevent other comorbidities including type 2 diabetes and cardiovascular disease, and reduce all-cause mortality [95, 466, 467]. Since most older adults are physically inactive (i.e., do not engage in ≥150 minutes/week of MVPA) and fall short of these recommendations [96], increasing MVPA among older adults has become a public health priority. As such, it is estimated 17.7% of AD cases could be prevented by recommended amounts of MVPA [468].   103    Accumulating evidence also suggests high amounts of SB can increase morbidity and mortality risk [469]. SB is defined as any behaviour that incurs ≤1.5 METs and includes behaviours such as sitting, television watching, and lying down [8]. SB is associated with numerous health risks including type 2 diabetes [469], cardiovascular disease [266], and all-cause mortality [267]. Given the risks of SB to health, recommendations for sedentary time suggest limiting discretionary sedentary time to <2 hours/day and accumulating >2 hours/day of LPA (i.e., standing and light walking; [470, 471]). Emerging evidence also suggests SB is associated with cognitive function; however SB is a distinct behaviour from PA and thus a systematic review of the current epidemiological evidence is needed [6, 7].   While preliminary evidence suggests SB is associated with cognitive function, it is still unclear what the magnitude of this association is. For example, it is unclear if reducing SB is more important for long term cognitive health than increasing PA. Such empirical evidence is crucial to increasing our understanding of how we can best promote healthy cognitive aging through lifestyle approaches and determining whether public health should focus on reducing SB, increasing PA, or both to reduce all-cause dementia prevalence. Thus, our objective was to systematically review the epidemiological evidence regarding how SB is associated with cognitive function throughout the adult lifespan.  2.2 Methods  2.2.1 Summary of search strategy  104   We conducted a systematic review regarding the association between SB and cognitive function. In accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [472], we searched PubMed, PsychINFO, EBSCO, and Web of Science between  January 1, 1990-February 6, 2016. Included in our search terms were the following keywords: SB terms (SB, physical inactivity, television time, TV time, screen time); cognition terms (cognition, cognitive function, brain function, executive function, memory, dementia, AD); and age terms (older adults, elders, elderly, aging, aged, 40+ years). This process was repeated until all search term combinations were performed.   2.2.2 Study selection We selected peer reviewed, published, observational studies that included adults aged 40 years and older that measured SB as an exposure and cognitive function as an outcome. Articles mentioning SB and cognition in either the title or abstract were initially included for full-text review.   2.2.3 Inclusion and exclusion criteria We included studies if they were: 1) observational studies (i.e., cohort, case-control, or cross-sectional); 2) peer-reviewed; and 3) published in the English language between January 1, 1990-February 6, 2016. All studies included clearly described participants as adults aged 40 years and older at baseline assessment and measured SB at baseline assessment or over time with the purpose of assessing risk (i.e., exposure). Additionally, the studies included measured cognitive function at baseline assessment or over time with the purpose of determining change associated with increased SB (i.e., outcome).    105   We excluded articles if they were: 1) not peer reviewed articles; and 2) not published in the English language. Since we were only interested in observational studies, interventions designed to reduce SB were not included.    2.2.4 Data extraction Two authors (RSF and JCD) initially screened and identified studies based on the study title and abstract. Duplicates and articles failing to meet inclusion criteria were removed. The remaining full-text articles were reviewed by RSF and JCD to determine eligibility. Any disagreements were resolved by a third reviewer (TLA).   Two raters (RSF and JCD) independently extracted data from all articles included; discrepancies were discussed and reviewed by a third party (TLA). Data were extracted from the included articles using a custom data extraction form developed by RSF and JCD. We extracted the following categories: 1) study design; 2) participant characteristics, setting and length of follow-up; 3) measure of exposure (i.e., SB); 4) measure of outcome (i.e., cognitive function); and 5) main findings.   For exposure measures (i.e., SB), we extracted the: 1) instrument name; 2) exposure definition (e.g., SB or television time); 3) method of exposure assessment (e.g., self-report questionnaire, accelerometry, etc.); 4) data collection procedure; 5) statistical methodology; and 6) previously established validity and reliability of the instrument. For exposure definitions, SB included time spent engaging in activities with an energy cost of ≤1.5 METs and television time referred to sedentary time spent watching television.   106    For methods of assessment of cognitive function, we extracted the: 1) instrument name; 2) domain of cognitive function assessed; 3) method for assessing cognitive function; 4) statistical methodology; and 5) previously established validity and reliability for the instrument. Given the limited number of studies available and the heterogeneity of samples used in these studies, we did not perform a meta-analysis.  2.2.5 Assessment of study quality Two authors (RSF and JCD) assessed the quality of the articles via the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines [473]. The STROBE checklist contains 22 separate items to identify the quality of reporting for observational studies. In summary, we assessed study quality based on the following components: 1) an informative and balanced abstract; 2) clear scientific background, rationale, objectives, and hypothesis; 3) clear description of study design, methodology, outcomes and exposures, and statistical analyses; 4) clear description of potential biases and how these were limited; 5) clear description of study participants, incidence of loss to follow-up, and reporting of both outcomes and exposures; 6) clear reporting of all results and analyses; 7) clear summarization of study findings with reference to study objectives; 8) clear description of the limitations of the study; and 9) a cautious overall interpretation of the findings with reference to the generalizability of the findings.    Two raters (RSF and JCD) independently rated the quality of the studies and achieved consensus through discussion (Κ=0.90). Discrepancies were settled by a third author (TLA). We used a binary system (+ =Yes, - =No) for each item of interest on the STROBE checklist. High quality studies  107   were defined as studies missing fewer than three criteria of the STROBE checklist, while low quality studies were defined as studies missing three or more criteria.  2.3 Results  2.3.1 Search results and study characteristics Figure 2.1A provides a sample of the search strategy and 2.1B describes the results of the search strategy for articles examining the association of SB with cognitive function. Of the 485 articles initially identified by title and abstract screening, our final systematic review included eight articles [474-481].   Study characteristics are described in Table 2.1. Three studies used a cohort design [474-476], one was a nested case-control [477], one used a case-control design [478], and three studies used a cross-sectional design [479-481]. The average follow-up time from the cohort studies was 7.67 years [474-476], and the follow-up time for the nested case-control study was 21 years [477]. Sample sizes ranged from 125-6359 participants with samples from England, Finland, France, and the United States.   108   Figure 2.1 Study selection process and sample search strategy Figure 0.1    109   Table 2.1 Study Characteristics Table 1 Publication  & Study Design Participants,  Country,  Setting, &  Length of Follow-Up Sedentary Behaviour (exposure assessment) Cognitive Function (outcome assessment) Results Cohort designs Hamer & Stamatakis [474]  2014   Cohort design 6359 men and women from the English Longitudinal Study of Ageing  England  2 year follow-up Self-reported TV viewing considered sedentary behaviour (SB).  Immediate word recall, delayed word recall, and verbal fluency [482]. All three used to create a global cognitive function score (primary outcome). Linear inverse relationship between TV time and cognitive function. Decreased cognition from baseline (EMM= 0.39, 95% CI: [0.33, 0.45]) to follow-up (EMM= 0.25, 95% CI: [0.19, 0.31]), but no association between baseline SB and changes in cognitive function.  Kesse-Guyot et al. [475]  2014  Cohort design 2430 participants from the Supplémentation en Vitamines et Minéraux Antioxydants study  France  13 year follow-up Self-administered French version of the Modifiable Activity Questionnaire (MAQ) [483]. Participants reported average time spent at home watching TV (min/day). Digit Span forward and backward (primary outcome) [459], Delis-Kaplan trail making test [484], RI-48 cued recall test [485], Semantic fluency and phonemic fluency [486]. SB associated with decreased global cognitive function (β= -1.28; 95% CI: [-2.46, -0.11]) and decreased verbal memory (β= -1.38; 95% CI: [-2.58, -0.18]) over time. Kesse-Guyot et al [476]  2012  Cohort design 2579 participants who agreed to participate in the follow-up period of the Supplémentation en Vitamines et Minéraux Antioxydants study  France  8 year follow-up Self-administered French Modifiable Activity Questionnaire (MAQ) [483]. Participants asked about average daily time spent with SB (min/day). Phonemic and semantic fluency (primary outcome) [486], RI-48 test [485], digit span forward and backward [459],  Delis-Kaplan trail-making test [484]. Negative association observed between TV viewing and executive function cross-sectionally (β=-0.98; 95%CI: [-1.93, -0.04]), no association between executive function and SB over time.   110   Table 2.1 Continued Publication & Study Design Participants,  Country,  Setting, &  Length of Follow-Up Sedentary Behaviour (exposure assessment) Cognitive Function (outcome assessment) Results Case-control designs Kivipelto et al. [477]  2008  Nested case-control design 1449 participants from the Cardiovascular Risk Factors, Aging and Dementia study (65-79 years)   Finland  Mean follow-up time of 21 years Self-reported leisure time physical activity (PA) dichotomized into categories: active and sedentary (persons who participated in leisure time PA less than two times per week). Cognitive status determined via scores on the MMSE [487], and all-cause dementia diagnosis (primary outcome) confirmed according to the Diagnostic and Statistical Manual of Mental Disorders [488]. The odds of developing all-cause dementia were 2.07 times greater for participants who were sedentary (95% CI: [1.12-3.86]) as compared to physically active when controlling for age, sex, follow-up time, education, BMI, cholesterol, blood pressure, heart attack, stroke and diabetes. Lindstrom et al. [478]   2005   Case-control design Participants born between 1915 and 1944. 135 cases of Alzheimer’s disease 331 controls recruited from clinical settings and from the community.  United States Participants self-reported daily hours of television viewing. Diagnosed case of Alzheimer’s disease (primary outcome). Cases watched significantly more television than controls (F [1, 464]= 35.37). The odds of developing Alzheimer’s disease increased 1.32 times for every hour of daily television viewing (95% CI: [1.08-1.62]).   Cross-sectional designs   Rosenberg et al. [479]  2015  Cross-sectional design  307 older adults (67-100 years) from 11 retirement communities  United States  Self-reported sedentary behaviour assessed using a modified version of the Sedentary Behaviour Questionnaire [489]. Objective sedentary time measured using ActiGraph GT3X+ accelerometer [490]. Trail Making Test [491] Self-reported sedentary time was associated with improved performance on Trails A (β= -0.01 ± 0.01), but was not associated with improved executive performance. Objectively-measured sedentary time was not associated with Trail Making Test performance.    111   Table 2.1 Continued Publication & Study Design Participants,  Country,  Setting, &  Length of Follow-Up Sedentary Behaviour (exposure assessment) Cognitive Function (outcome assessment) Results Steinberg et al. [480]   2014  Cross-sectional design 125 healthy adults 65 or older with no clinical evidence of cognitive impairment  United States   Hours spent in SB according to the Community Health Activity Program for Seniors (CHAMPS) questionnaire [149]. CogState computerized battery measured multiple domains of cognition including: psychomotor speed, visual attention, visual recognition, and memory (primary outcome) [492].  Lower scores on executive function measures associated with increased SB (β= 0.006 ± 0.003; R2= 0.2323). Memory scores and processing speed were not associated with increased SB. Vance et al. [481]   2005  Cross-sectional design 158 participants with a mean age of 75.05 years were recruited from the Accelerate study  United States   The total amount of time spent sitting, sleeping, or lying down was used as an indicator of SB [493]. Benton Visual Retention Test [494], Trail-Making Test [491], and the Rey-Osterrieth Complex Figure Copy and Recall Tests [495]. A composite score for cognitive function was then created (primary outcome). Structural equation modeling predicted SB was associated with increased cognitive function (β= 0.34)   112   2.3.2 Measurement of sedentary behaviour Measurement of SB varied considerably with a total of eight different measures used across the eight studies, as described in Table 2.2. All eight studies measured exposure to SB via subjective  methods [474-481], and one study measured SB via an objective method (accelerometry) [479]. Five studies measured exposure as sedentary time (i.e., time spent sitting, lying down or sleeping) [476, 477, 479-481], and four studies measured exposure as TV time [474-476, 478]. One study measured the exposure as both TV time and sedentary time [476].   Five studies examined SB using a previously developed questionnaire [475, 476, 479-481]. Two studies [475, 476], used the Modifiable Activity Questionnaire (MAQ) to assess SB [483]. A single study [480], used the CHAMPS questionnaire [146, 147, 149, 150]. Another study [479], used the Sedentary Behaviour Questionnaire (SBQ; [489]). The last study [481], used an unnamed questionnaire developed from previous investigations to assess sedentary time [493]. Each of the four questionnaires showed evidence of validity and reliability, however only the MAQ, CHAMPS, and SBQ were previously validated against a criterion measure [146, 147, 149, 150, 489, 496].   A single study [479], used an accelerometer, the Actigraph GT3X+ [490], to measure SB objectively. Accelerometers show good evidence of validity [497, 498]; however there is no current minimum wear standards for reliable SB estimates. 113   Table 2.2 Measures and methods to classify sedentary behaviour Table 2 Publication Name of Measure(s) Definition of Exposure Type of Exposure  Assessment Data Collection Procedure Statistical Methods and Confounder Adjustment Validity and Reliability Cohort studies Hamer & Stamatakis, 2014 [474] Unknown TV time Subjective Measure  Questionnaire developed for measuring physical activity (PA) and television time Participants self-reported daily television time and engagement in vigorous, moderate, and low-intensity PA.  Type of regression: Linear mixed models with random effect intercept;   Covariates and confounders: age sex, smoking, alcohol, PA, social status, disability, chronic illness and body mass index (BMI). Unknown Kesse-Guyot et al., 2014 [475] Modifiable Activity Questionnaire (MAQ) [483] TV time  Subjective Measure  Questionnaire designed to assess SB and PA during past 12 months. Participants self-reported average daily time spent watching TV and leisure-time PA performed at least 10 times for at least 10 minutes per session over the past year including the frequency and duration. After multiplying the number of hours/week of each activity by the estimated metabolic equivalent (MET), a summary score was obtained.  Type of regression: structural equation modeling;  Covariates and confounders: age, gender, education, time-lag between baseline and cognitive evaluation, occupation, energy intake, number of 24-hour records, BMI, depressive symptoms, memory issues, diabetes, hypertension, and cardiovascular disease Validity: r= 0.65 [496]  Reliability: ICC= 0.77 [496]    114   Table 2.2 Continued Publication Name of Measure(s) Definition of Exposure Type of Exposure  Assessment Data Collection Procedure Statistical Methods and Confounder Adjustment Validity and Reliability Kesse-Guyot et al., 2012 [476] Modifiable Activity Questionnaire (MAQ) [483] Sedentary behaviour (SB; TV time, computer use, reading)  Subjective Measure  Questionnaire designed to assess SB and PA during past 12 months. Participants self-reported average daily time spent watching TV, using a computer or reading (min/day) Type of regression: Principal component analysis;  Covariates and confounders: interval between SB assessment and cognitive evaluation, age, gender, education, occupation, retirement status, tobacco use, BMI, depressive symptoms, health status, heart disease, diabetes, hypertension and PA. Validity: r= 0.65 [496]  Reliability: ICC= 0.77 [496] Case-control studies Kivipelto et al., 2008 [477] Unknown SB (leisure time PA <2x/week) Subjective Measure  Questionnaire developed by authors Participants self-reported leisure-time PA lasting >30 minutes and caused breathlessness and sweating. Participants dichotomized into active (>2x/week) and sedentary (<2x/week). Type of regression: Multiple logistic regressions;  Covariates and confounders: age, sex, follow-up time, education, BMI, cholesterol, blood pressure, heart attack, stroke and diabetes mellitus.  Unknown Lindstrom et al., 2005 [478] Unknown Daily hours of television viewing Subjective Measure  Questionnaire developed by authors Participants self-reported hours/month devoted to TV viewing at age 20-39 and at ages of 40-59. Daily TV viewing hours calculated from total hours/day spent watching TV. Type of regression: Unconditional logistic regression model  Covariates and confounders: age, gender, income, and education. Unknown  115   Table 2 Continued Publication Name of Measure(s) Definition of Exposure Type of Exposure  Assessment Data Collection Procedure Statistical Methods and Confounder Adjustment Validity and Reliability Cross-sectional designs Rosenberg et al., 2015 [479] Self-report measure: Sedentary Behavior Questionnaire (SBQ) [489]  Objective measure: Actigraph GT3X+ accelerometer [490]  SBQ: Hours spent in SB  Actigraph GT3X+: Hours spent in SB SBQ: Subjective Measure   Assessed time spent during typical day SB   Actigraph GT3X+: Objective Measure  Sedentary assessed using standard cutpoint of <100 counts per minute.  SBQ: Participants reported time/day spent in SB including sitting, watching TV, computer use, reading, commuting, napping, and other activities.  Actigraph GT3X+: Participants were included with at least 1 valid day of wear time and 600 minutes of accelerometer data. Sedentary time was assessed using the standard cutpoint of <100 counts per minute  Type of regression: linear mixed-effects models   Covariates and confounders: age, gender, marital and educational status. SBQ:  Validity: No significant relationship between accelerometer counts and SBQ scores; Reliability: ICC= 0.85 [489]  Actigraph GT3X+:  Validity: r= 0.59 [497, 498] Reliability: Unknown Steinberg et al., 2014 [480] Community Health Activity Program for Seniors (CHAMPS) questionnaire [148] Hours spent in SB Subjective Measure  Assessed frequency and duration of 40 different activities undertaken by older adults Participants self-reported weekly frequency and duration of 40 different activities using the CHAMPS questionnaire Type of regression: linear regression analyses   Covariates and confounders: age, sex, race, and education.  Validity: r= 0.29  Test-retest reliability: ICC= 0.76. [146, 147, 149]    116   Table 2.2 Continued Publication Name of Measure(s) Definition of Exposure Type of Exposure  Assessment Data Collection Procedure Statistical Methods and Confounder Adjustment Validity and Reliability Vance et al., 2005 [481] Unknown Total amount of time spent sitting, sleeping, or lying down used as an indicator of SB Subjective Measure  Questionnaire adapted from Paffenbarger questionnaire [493]  Participants self-reported how many hours per day spent seated, lying down, and sleeping. Types of regression: Latent growth model;  Covariates and confounders: Age, depression, and PA. Unknown  117   2.3.3 Measurement of outcomes from sedentary behaviour Table 2.3 describes the measures of cognitive function used. Thirteen different measures of cognitive function were used across the eight studies [474-481]. Studies examined the following areas of cognition: 1) five measured memory [474, 475, 480, 481]; 2) five measured executive function [474-476, 479, 481]; 3) four measured processing speed [474, 479-481]; 4) two measured  incidence of cognitive impairment or all-cause dementia [477, 478]; and 5) one measured perceptual organization and planning [481]. Three studies created scores for global cognitive function [474, 480, 481].  2.3.3.1 Assessment of memory The constructs of memory measured were non-descriptive memory (i.e., unspecified by the authors as to what construct of memory the test measured), lexical-semantic memory, working memory, visual memory and episodic memory. Non-descriptive memory was measured via delayed word recall [482] in one study [474], and the Benton Visual Retention Test [494] in another study [481]. Lexical-semantic memory was measured via phonemic and semantic fluency [486] in two studies [475, 476]. Working memory was measured by digit span forward and backward [459] in two studies [475, 476]. Visual memory was assessed using the Rey-Osterriety Complex Figure Copy and Recall Test [495] in one study [481]. Episodic memory was measured via the RI-48 test [485]in two studies [475, 476]. Among the measures used, the Benton Visual Retention Test, phonemic and semantic fluency, digit span forward and backward, Rey-Osterriety Complex Figure Copy and Recall Test and the RI-48 have evidence of validity and reliability [485, 495, 499, 500].  118   Table 2.3 Measures and methods for outcome assessment (i.e., cognitive function) Table 3  Publication Domain of Cognitive Function Assessed (Name of Measure) Data Collection Procedure Analyses Utilized Validity and Reliability Cohort designs Hamer & Stamatakis, 2014 [474] Processing speed (Immediate word recall), Memory (Delayed word recall), and Executive function (Verbal fluency) [482]. Immediate word recall: Read 10 words and recalled as many words as possible immediately after. Delayed word recall: Using same list, recalled words after they completed other cognitive function tests. Verbal fluency: Named as many animals as possible in one minute. A global cognitive function score calculated from the sum of standardized scores on each test. Type of regression: Linear mixed models with random effect intercept;   Covariates and confounders: age sex, smoking, alcohol, physical activity, social status, disability, chronic illness and body mass index (BMI). Unknown  Kesse-Guyot et al., 2014 [475] Lexical-semantic memory (Phonemic fluency and semantic fluency [486]), Episodic memory (RI-48 test [485]), Working memory (Digit span forward and backward [486]), Executive function (Delis-Kaplan trail-making test [484]). Phonemic fluency: Cited as many words as possible in 2 minutes beginning with the letter “p”. Semantic fluency: Named as many animals as possible in 2 minutes. RI-48 test: Delayed cued recall test. Digit span forward and backward: Repeated sequence of 7 digits, forward and backward. Trail making test: connecting numbers and letters alternating between the two series. Type of regression: structural equation modeling;  Covariates and confounders: age, gender, education, time-lag between baseline and cognitive evaluation, occupation, energy intake, number of 24-hour records, BMI, depressive symptoms, memory issues, diabetes, hypertension, and cardiovascular disease. Phonemic and semantic fluency: Test re-test reliability r= 0.82 [499].  RI-48: Classified 88% of people with mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) correctly [485].  Digit Span: B= 0.64 in Confirmatory Factor Analysis [500].  Trail Making: r= -0.38 compared to Stroop Color Word score [501].  119   Table 2.3 Continued Publication Domain of Cognitive Function Assessed Data Collection Procedure Analyses Utilized Validity and Reliability Kesse-Guyot et al., 2012 [476] Lexical-semantic memory (Phonemic fluency and semantic fluency [486]), Episodic memory (RI-48 test [485]), Working memory (Digit span forward and backward [486]), Executive function (Delis-Kaplan trail-making test [484]). Phonemic fluency: Cited as many words as possible in 2 minutes beginning with “p”. Semantic fluency: Named as many animals as possible in 2 minutes. RI-48 test: Delayed cued recall test. Digit span forward and backward: Repeated sequence of 7 digits, forward and backward. Trail making test: connecting numbers and letters alternating between the two series.  Type of regression: Principal component analysis; Phonemic and semantic fluency: Test re-test reliability r= 0.82 [499].  RI-48: Classified 88% of people with mild cognitive impairment (MCI) or Alzheimer’s Disease (AD) correctly [485].  Digit Span: B= 0.64 in Confirmatory Factor Analysis [500].  Trail Making: r= -0.38 compared to Stroop Color Word score [501] Case-control designs Kivipelto et al., 2008 [477] Screened for cognitive impairment and all-cause dementia. Mini-Mental State Exam (MMSE; [487]) at screening phase and for those who scored <24, neuropsychological examinations conducted to screen for all-cause dementia, according to the Diagnostic and Statistical Manual of Mental Disorders [488]. Type of regression: Multiple logistic regressions; MMSE: Test re-test reliability: r= 0.89; Validity: r= 0.78 compared to Verbal IQ & r= 0.66 compared to Performance IQ [487]. Lindstrom et al., 2005 [478] Screened for AD. Cases evaluated by neuropsychological, laboratory and neurological examinations.  Type of regression: Unconditional logistic regression model Unknown  120   Table 2.3 Continued Publication Domain of Cognitive Function Assessed (Name of Measure) Data Collection Procedure Analyses Utilized Validity and Reliability Cross-sectional designs Rosenberg et al., 2015 [479] Executive function and Processing speed (Trail Making Test [491]) Trails A was completed first, followed by Trails B. Both items were scored using completion time in seconds and scores for participants who were unable to complete the exam were set to the maximum value (300 seconds). Executive function was estimated by subtracting time of Trails A from Trails B. Type of regression: linear mixed-effects models   Covariates and confounders: age, gender, marital and educational status. Trail-Making Test: Reliability r= 0.60-0.90 [491]; Discriminat Validity: t= 16.20 (p<0.001) [501]. Steinberg et al., 2014 [480] CogState computerized battery[492]. Cognitive tests assessed: 1) Processing speed; 2) Visual attention, recognition, and memory; 4) Verbal learning and memory; 5) Immediate recall; 6) Delayed recall; 7) Working memory; and 8) Problem solving and reasoning. Participants administered the CogState tests over a 35-minute period. Type of regression: linear regression analyses   Covariates and confounders: age, sex, race, and education Validity: r= 0.49-0.83 [502]   Vance et al., 2005 [481] Memory (Benton Visual Retention Test [494]); Processing speed and Executive function (Trail Making Test [501]); Visual memory, perceptual organization and planning (Rey-Osterriety Complex Figure Copy and Recall Tests [491]). Benton Visual Retention Test: Shown a series of geometric designs and then draw from memory. Trail-Making Test: Connected 25 alternating number and letter circles in sequence as quickly as possible. Rey-Osterriety Complex Figure and Recall Tests: Reproduced complex figure while present and then from memory. Types of regression: Latent growth model;  Covariates and confounders: Age, depression, and physical activity (PA). Benton Visual Retention Test: Inter-rater reliability r= 0.80-0.90 [494].   Trail-Making Test: Reliability r= 0.60-0.90 [491]; Discriminat Validity: t= 16.20 (p<0.001) [501].   Rey-Osterriety Complex Figure Copy and Recall Tests: Reliability r= 0.90 [495].  121   2.3.3.2 Assessment of executive function Two studies  used the Delis-Kaplan Trail Making Test [484] to assess executive function [475, 476]. One study [474], used verbal fluency [482], and two studies [479, 481] used Trail Making Test [491]. The Delis-Kaplan Trail Making Test and the Trail Making Test both have evidence of validity and reliability [458, 491, 501].  2.3.3.3 Assessment of processing speed Processing speed was measured via Immediate Word Recall [482] in one study [474], and by Trail Making Test [491] in two studies [479, 481]. Only Trail Making Test has evidence of validity and reliability [458, 491, 501].  2.3.3.4 Assessment of cognitive impairment and all-cause dementia incidence One study [477] used the Mini-Mental State Exam (MMSE; [487]) to screen for cognitive impairment and then conducted neuropsychological exams to screen for all-cause dementia according to the Diagnostic and Statistical Manual of Mental Disorders [488]. The other case-control study [478], screened for cases of AD using unstated neuropsychological, laboratory and neurological examinations. Evidence of validity and reliability exists for the MMSE [487].  2.3.3.5 Assessment of perceptual organization and planning The study measuring perceptual organization and planning [481], used the Rey-Osterriety Complex Figure Copy and Recall Test [495]. Evidence of validity and reliability exists for the Rey-Osterriety Complex Figure and Recall Test [495].    122   2.3.3.6 Assessment of global cognitive function One study [480], used the CogState computerized battery to assess global cognitive function [492]. Another study [474], created a standardized global cognitive function score from the three cognitive measures used in the study: Immediate Word Recall, Delayed Word Recall, and Verbal Fluency [482]. The final study to measure global cognitive function [481], used a standardized score from the Benton Visual Retention Test [494], Trail Making Test [491], and Rey-Osterriety Complex Figure Copy and Recall Test [495]. Only the CogState computerized battery has evidence of validity and reliability as a global cognition measure [502].  2.3.4 Quality assessment  Studies varied considerably in quality as shown in Table 2.4. On average, studies met 19 of the 22 specific criteria of the STROBE checklist [474-481]. One article met all guidelines for the reporting of information [475], however three studies failed to address four or more different criteria of the STROBE checklist [478-480]. All four of the high quality studies found negative associations between SB and cognitive function [474-477].  The common issues were in failure to report: 1) study size (N= 4; [478-481]); 2) information about the design in the title and abstract (N= 4; [477, 479-481]); 3) potential biases (N= 4; [478-481]); and 4) specific objectives and hypotheses (N= 3; [474, 477, 480]). Other issues in study quality included failure to report: 1) eligibility criteria (N= 1; [478]); 2) sources of data and method of assessment (N= 1; [478]); 3) describing statistical analyses (N= 1; [476]); 4) providing a cautious overall interpretation of the study (N= 1; [479]); 5) discussing the generalizability of the findings (N= 1; [479]); and 6) outcomes, exposures and potential confounders (N= 1; [479]).  123   Table 2.4 Quality assessment for studies on the relationship of sedentary behaviour with cognitive function Table 4   Cohort designs Case-control designs Cross-sectional designs STROBE Checklist1 Hamer & Stamatakis, 2014 [474] Kesse-Guyot et al., 2014 [475] Kesse-Guyot et al., 2012 [476]  Kivipelto et al., 2008 [477]  Lindstrom et al., 2005 [478]  Rosenberg et al., 2015 [479]  Steinberg et al., 2014 [480]  Vance et al., 2005 [481]  1 + + + - + - - - 2 + + + + + + + + 3 - + + - + + - + 4 + + + + + + + + 5 + + + + + + + + 6 + + + + - + + + 7 + + + + + - + + 8 + + + + - + + + 9 + + + + - - - - 10 + + + + - - - - 11 + + + + + + + + 12 + + - + + + + + 13 + + + + + + + + 14 + + + + + + + + 15 + + + + + + + + 16 + + + + + + + + 17 + + + + + + + + 18 + + + + + + + + 19 + + + + + + + + 20 + + + + + - + + 21 + + + + + - + + 22 + + + + + + + + 1The STROBE Checklist asks the following information (+=Reported; - = Not reported): 1) Indicates study design in title and abstract and provides an informative and balance summary in the abstract  2) Gives the scientific background and rationale  3) States specific objectives and hypotheses 4) Presents key elements of study design 5) Describes setting, location, exposures, follow-up and relevant dates 6) Clearly defines eligibility criteria and methods of selecting participants  124   7) Clearly defines outcomes, exposures, potential confounders, predictors and effect modifiers 8) Gives sources of data and clearly defines method of assessment  9) Describes potential bias  10) Explains how study size was arrived at  11) Explains how quantitative variables were handled in analysis 12) Describes all statistical analyses 13) Reports number of individuals at each stage of study and gives reasons for non-participation 14) Gives characteristics of study participants and indicates number of missing data 15) Reports number of events (outcomes and/or exposures) 16) Clearly provides the main results of analyses 17) Reports all other analyses done 18) Summarizes key findings with reference to study objectives 19) Discusses limitations of the study  20) Provides a cautious overall interpretation of the study  21) Discusses the generalizability of the findings  22) Gives the sources of funding and role of the funders 125   2.3.5 Findings from studies on the association of sedentary behaviour with cognitive function In total, six studies found associations between increased SB and decreases in cognitive function [474-479]. Two studies found associations between increased SB and improved cognitive function [479, 481].  2.3.5.1 Cohort designs Among the cohort studies, one study found an association between increased SB and decreases in cognitive function over time [475]. The other two studies found associations between increased SB and lower cognitive function at baseline, but no association between SB and cognition over time [474, 476].  2.3.5.2 Case-control designs The nested case-control study found the odds of developing all-cause dementia were higher for individuals who engaged in more SB [477]. In addition, the other case-control study found that individuals who watched more hours of television per day had higher odds of developing AD in later life [478].  2.3.5.3 Cross-sectional designs The results of the cross-sectional studies were mixed. One study found associations between increased SB and lower cognitive function [480]. A second study found SB was positively associated with cognitive function [481]. The final cross-sectional study found subjectively- 126   measured SB was positively associated with processing speed; however there was no association between SB and cognitive function when measured objectively [479].   2.4 Discussion  2.4.1 Summary of main findings Our results indicate SB is associated with reduced cognitive function over the lifespan. Importantly, all four of the high quality studies found SB is associated with poorer cognitive function. However, the heterogeneity in the current methods used to quantify SB and cognitive function are the current major barriers to determining the precise magnitude of this relationship. We also found only one study used an objective measure of SB and a number of exposure measures lacked evidence of validity and reliability.   Furthermore, two of the three longitudinal studies had follow-up periods of <10 years, which may account for the significant findings at baseline but not over time [474, 476]. Changes in cognition occur gradually over the adult lifespan [27], often with detectable changes occurring after the age of 60 [503]. As such, studies with short-term follow-ups (i.e., <10 years) may not detect meaningful associations between changes in cognition and lifestyle behaviours.  2.4.2 Comparison of the findings with the literature A large body of work on the association between PA and better cognitive function exists [156], however far less is known about the association of SB with cognition. Some preliminary findings  127   have suggested that SB is associated with later life cognitive impairment, but have also noted a lack of epidemiological evidence needed to draw strong conclusions [6, 7].   Our review suggests SB is indeed associated with impaired cognitive function and all-cause dementia risk. Specifically, higher quality studies [474-477], and those of stronger epidemiological evidence (i.e., cohort or case-control design) all found associations between increased SB and poorer cognition [474-478]. Moreover, the current evidence suggests an association by meeting five of the nine epidemiological criteria for causation [155]. Specifically, the criteria met include: 1) consistent findings across persons, places, and circumstances; 2) evidence of temporality; 3) evidence of a dose-response relationship; 4) a plausible mechanism by which exposure leads to outcome; and 5) by analogy, the exposure is associated with outcome. Importantly, our systematic review found consistency in the findings [474-478, 480], evidence of temporality [475, 477, 478], and evidence of a dose-response relationship [474-476, 480].  In addition to our findings, a plausible mechanism by which SB is associated with cognitive decline is emerging. Recent data suggest prolonged sedentary time impairs glucose and lipid metabolism [88], which are both recognized as risk factors for cognitive decline and all-cause dementia [262, 263]. There is also evidence that SB is related to cognitive decline by analogy. Briefly, SB is associated with many chronic diseases [266, 267, 469, 504], which are also associated with cognitive impairment and dementia risk [268-270]. Thus, the evidence collectively suggests SB is a risk factor for later life cognitive impairment and all-cause dementia risk.  2.4.3 Assessment of sedentary behaviour  128   While our findings suggest there is now enough evidence to consider SB a risk factor for cognitive decline and dementia, the current measurement of SB still has some limitations. First, only one of the studies reviewed measured SB with an objective measure [479]. While there is no one best measure for assessing SB [505], and both objective measures and subjective measures have limitations [112], objective measures are considered to be more accurate and reliable because they eliminate recall bias [117]. This is because PA participation among older adults is often intermittent, sporadic or unstructured, which makes recall extremely difficult [506]; thus older adults may unintentionally over-report their SB [507]. However, this does not mean subjective methods of assessing SB are useless. Complete data from objective measures have an inherent selection bias which limits the generalisability of the findings, and objective assessment may miss components or dimensions of SB which may be health protective [117]. Thus, future research examining the association of SB with cognitive function should utilize both objective and subjective measures whenever possible [508].  Secondly, four of the eight studies we reviewed used measures of SB with no previous evidence of validity or reliability [474, 477, 478, 481]. Validity and reliability are important for making sound interpretations from tests and thus the lack of evidence of either calls into question the conclusions drawn from these studies [123, 125]. The continued use of measures without evidence of validity and reliability is making the conclusions drawn from these studies questionable at best, and downright wrong at worst [122].  Finally, the construct of SB was misclassified in several of the studies. For example, Kivipelto and colleagues categorized participants as sedentary based on self-reported leisure-time PA of less than  129   twice per week [477]. Yet the absence of PA does not define SB [265], and thus misclassification in the literature poses challenges to accurately assessing the association between SB and cognitive function.  2.4.4 Assessment of cognitive function While the current measurement of cognitive function in the studies reviewed appears to be more rigorous than the methods used to assess SB, there are still concerns. First, the numerous measures of cognitive function currently in use are obfuscating the relationship between SB and cognition. The eight studies we reviewed used a total of 13 measures of cognitive function. Specifically, the studies assessed memory by six different tests, executive function by four tests, processing speed by two methods, cognitive impairment by two methods, and global cognitive function by three methods. With such a wide variety of measures used to assess each domain of cognitive function, comparing study results is extremely difficult. Based on the heterogeneity of measures, we recommend future studies use the following instruments for each domain of cognition to allow comparisons across studies: 1) RI-48 for memory; 2) Trail Making Task for executive function; 3) immediate word recall for processing speed; 4) the Rey-Osterriety Complex Figure Copy and Recall Test for perceptual organization and planning; and 5) the MoCA [457] for global cognition.  In addition, the numerous domains of cognition being assessed (i.e., global cognition, memory, executive function, etc.) prevents comparisons of study results. Few studies tested similar domains of cognition, and thus it is unclear if SB is associated with decreases in global cognitive function, several different domains of cognition, or just a single domain. Future studies therefore need to first determine which domains of cognitive function decrease with increased SB. One means of  130   potentially assessing all domains of cognitive function concomitantly would be the use of the NIH toolbox [461], which could independently examine the associations of SB with memory, executive function, and so forth.  Several of the measures used to assess cognitive function in these studies also lacked evidence of validity or reliability, and thus the conclusions may not be valid for the construct the authors planned to investigate [84]. For example, Hamer and Stamatakis used a memory test (i.e., delayed word recall) without evidence of validity or reliability [156]; thus rather than measuring memory, the test may be related to another construct, such as executive function.   While there are some issues with measurement of cognitive function in these studies, our preliminary findings suggest SB is negatively associated with memory, executive function, and global cognition. These findings suggest SB has an inverse association with cognition compared to exercise training and MVPA. MVPA—as well as AT and RT—are well documented to affect multiple domains of cognitive function [106]. Furthermore, MVPA and exercise training are established as an all-cause dementia prevention measure which could reduce the incidence of all-cause dementia by as many as one million cases worldwide [509]. Given this information, SB may be adversely associated with the same neurophysiological pathways as MVPA and exercise.   2.4.5 Study quality The STROBE checklist for observational studies is designed to ensure important information on study design is available so readers of research can follow what was planned, what was done, what was found, and what conclusions can be drawn [473]. This information is an important component  131   for systematic reviews [510, 511]; however when components required by the STROBE guidelines are absent, the conclusions which can be drawn from these studies suffer.  The quality of studies we reviewed varied greatly with several of the studies showing multiple flaws in reporting. Only one study [475], met all criteria of the STROBE and several studies were missing multiple criteria. Issues such as sampling bias, selection bias, recall bias and detection bias may therefore have inflated the results of these studies. We therefore recommend future investigations on how SB is associated with cognitive function firmly adhere to the STROBE guidelines.  Finally, the lack of a sample size calculation by any of the studies we reviewed is an important concern. Sample size calculations for observational studies require a compromise between balancing the needs of power, economy and timeliness [512]. Failure to attain a sample size with enough power inevitably leads to type II error; however equally erroneous is using a sample size that is “too big” that detects an effect of little scientific importance [513]. For example, one study we reviewed included well over 6,000 participants [156], which may have accounted for the significant—albeit small—results.   2.4.6 Recommendations Current PA guidelines offer a brief policy recommendation on SB—avoid it as much as possible [258, 439]; however, in order to best promote healthy cognitive aging, an empirically derived public health message is still needed. Thus, we have developed healthy cognitive aging guidelines for SB which are in-line with current evidence and recommendations [258, 265, 471, 514]. We  132   therefore recommend all adults should: 1) avoid sedentary time wherever possible; 2) limit discretionary sitting time to <2 hours/day; 3) stand up and move after 30 minutes of uninterrupted sitting; and 4) increase LPA (i.e., standing and light walking) to >2 hours/day by substituting these activities for sedentary time (e.g., stand while watching television).  Combating a sedentary lifestyle—and associated cognitive declines—also requires an emphasis on encouraging adults to engage in ≥150 minutes/week of MVPA. Regular MVPA is a pillar of healthy cognitive aging, with current evidence suggesting ≥150 minutes/week of MVPA reduces the risk of AD by 38% [515]. Moreover, empirical evidence has found a consistent relationship between MVPA and cognitive function [156]. Given the current evidence, we recommend all adults limit discretionary SB to <2 hours/day and concomitantly engage in ≥150 minutes/week of MVPA. Meeting these recommendations may best promote healthy cognitive aging and could reduce the incidence of all-cause dementia by more than one million cases worldwide [509].   2.4.7 Limitations and future directions This review only investigated observational studies on how SB is associated with cognitive function; however, to our knowledge this is the first systematic review to evaluate the evidence. There may also be a publication bias which limits the generalisability of our findings; however this limitation is inherent in all systematic reviews. Our systematic review located only eight studies, but our findings do show a consistent relationship such that SB is associated with poorer cognitive function. Although all four high quality studies found SB is associated with poorer cognition [474-477], more high quality studies are needed before estimates can be made about the attributable risk of SB to cognitive impairment and all-cause dementia.   133    Given this area of research is still developing, our study only provides an initial platform for examining the association of SB with cognitive impairment and all-cause dementia. Our preliminary recommendations for healthy cognitive aging are therefore broadly consistent with current policy [258, 265, 471, 514], and may need to be refined as more evidence emerges.   Dementia is also a complex disease which has several forms including AD and vascular dementia, which have vastly different aetiologies. While the mechanisms may be different by which the different sub-types of dementia occur, there are certainly similarities in terms of risk factors. For example, Laurin and colleagues found increased MVPA was associated with reduced risks of cognitive impairment and dementia of any type [515]. Thus, our preliminary findings suggest reduced cognitive function and increased all-cause dementia risk are associated with a sedentary lifestyle. Future studies should determine the associations of SB with different types of dementia.  Related to this issue, different types of SB may have different associations with cognitive function. For example, there is some evidence that computer use may positively affect cognition [475-477]. However, only eight studies assessing SB were included in our review—and only three studies assessed computer use as an exposure variable [475-477]—and thus it is difficult to make comparisons and draw conclusions at this time. Future studies are needed to determine how different sedentary activities moderate the relationship between SB and cognitive function.  2.4.8 Conclusions  134   The current body of evidence suggests SB is negatively associated with cognitive function; however, the associations between SB and cognitive function are complex and largely dependent on both the exposure variable and outcomes assessed. Nonetheless, our findings suggest reducing discretionary sedentary time to <2 hours/day and concomitantly engaging in ≥150 minutes/week of MVPA may best promote healthy cognitive aging.    135   Chapter 3: Cross-sectional relationships of physical activity and sedentary behavior with cognitive function in older adults with probable Mild Cognitive Impairment A version of this chapter is published as FALCK RS, Landry GJ, Best JR, Davis JC, Chiu BK, Liu-Ambrose T. Cross-Sectional Relationships of Physical Activity and Sedentary Behavior With Cognitive Function in Older Adults With Probable Mild Cognitive Impairment. Physical Therapy. 2017; 97(10): 975-984. 3.1 Introduction By 2030, there will be nearly one billion older adults worldwide [13]. Since age is the greatest risk factor for dementia [464], the number of dementia cases is expected to increase substantially [15]. Pharmaceutical therapies to treat dementia are still in their infancy [516, 517], and thus reducing the risk of dementia—and potentially dementia incidence—requires the development of effective lifestyle-based strategies.  MCI represents a critical phase to intervene then, since it is a transitional stage between healthy cognition and dementia [17]. MCI is defined as cognitive decline greater than expected for age and education level which does not interfere with independence [18], and is associated with up to a 30% increased risk of developing dementia within 5 years [19]. By comparison, older adults without MCI develop dementia at a rate of 1% to 2% within 5 years [20]. Providing effective strategies to maintain cognitive health during this transition period might slow the conversion to dementia.  One potential strategy is PA—a behavior associated with both cognitive and physical health [106]. High levels of PA are prospectively linked to lower incidence of MCI and dementia [102, 153,  136   156, 518, 519], and an estimated 17.7% of AD cases could be prevented through PA [468]. PA also has multidimensional health benefits including reduced risk of mortality, and chronic diseases such as type 2 diabetes mellitus and cardiovascular disease [101, 520, 521]. Given the importance of PA for health, current recommendations for older adults suggest 30 minutes of MVPA (in bouts as brief as 10 minutes of moderate intensity) 5 days/week [90, 439]. Unfortunately, >95% of older adults are physically inactive (i.e., do not engage in ≥150 minutes/week of MVPA) and thus fall short of these recommendations [96].  While the importance of PA (particularly MVPA) for cognitive health is well established, less is known about the impact of SB on both physical and cognitive health. SB is any behavior which incurs ≤1.5 METs and includes activities such as sitting, television watching, and lying down; MVPA is any behavior incurring ≥3.0 METs [8]. Recent evidence suggests SB may be associated with poorer cognitive function and increased risk of cognitive impairment, although epidemiological data are needed to confirm this association [6, 7]. Accumulating evidence also suggests SB is associated with numerous chronic diseases including type 2 diabetes mellitus and cardiovascular disease [265, 266]. These chronic diseases are concomitantly linked with increased risk of cognitive decline and dementia [268, 270], providing further evidence that high SB is a risk factor for cognitive health. Due to the increasing evidence suggesting PA and SB are associated with cognitive health, older adults are recommended to limit discretionary SB to <2 hours/day; avoid sitting for longer than 30 minutes without standing; and concomitantly increase MVPA to ≥150 minutes/week [522].   137   Physical therapists are in a unique position to help influence both PA and SB. The potent position of clinicians to maximize patient compliance by influencing behavior is why the US Preventive Task Force has recommended clinicians provide PA counselling since 1989 [523, 524]. In addition, RCTs using activity counselling among older adults in a primary care setting have been highly successful at increasing patient PA [525, 526]. As such, these data further illustrate the importance of activity counselling in the clinical setting and suggest even brief questions about patient activity levels can have demonstrable improvements on patient health outcomes, such as cardiovascular disease and mortality risk [527]. Thus, as an important step to promote older adult cognitive health, clinicians should consult their older adult patients about their PA and SB.  Since the current evidence suggests PA and SB are important for healthy cognitive aging, a next step—given the window of opportunity to intervene for people with MCI—is an analysis of PA and SB differences between people with MCI and those without MCI. Importantly, it is unclear whether people with MCI engage in different amounts of PA and SB than their peers without MCI and whether the associations of PA and SB with cognitive function are the same or different between older adults with MCI and those without MCI. Indeed, because of underlying neurobiological differences between older adults with MCI and those without MCI [2], a functional weakening in the relationships of health behaviors with cognitive function may occur in MCI [296]. However, data are still needed showing the associations of PA and SB with cognitive function attenuate based on MCI status in order to confirm this hypothesis. Answering these questions may help inform clinicians concerned about the cognitive health of their older adult patients, and help determine which of their older adult patients may see the greatest benefits to cognitive health from PA and SB counselling.  138    To address these gaps in knowledge, we first investigated differences in MVPA and SB between older adults with probable MCI and those without. In addition, we determined whether the relationships of MVPA and SB with cognitive function differed based upon MCI status.  3.2 Methods All participants provided written informed consent. Ethics approval for this study was obtained from the Vancouver Coastal Health Research Institute and the University of British Columbia’s Clinical Research Ethics Board (H14-01301).   3.2.1 Protocol For this cross-sectional study, we recruited and collected data between August of 2014 and June of 2016. At study entry, we ascertained general health, demographics, socioeconomic status, and education by questionnaire. Subsequently, we screened participants for MCI using the MoCA, with a score of <26 indicating probable MCI status [457]. Participant MVPA and SB were then observed for ≥4 days using the MW8. Following MW8 observation, we measured cognitive function for all participants using the ADAS-Cog Plus [454].  3.2.2 Participants Participants were recruited from Vancouver, British Columbia by advertisements placed in local community centres, newspapers, and word of mouth referrals. Potential participants were considered eligible if they met the following 3 criteria: 1) men and women at least 55 years old and living in the greater Vancouver area; 2) scores of >24/30 on the MMSE [487]; and 3) ability  139   to read, write, and speak English with acceptable visual and auditory acuity. Participants were ineligible if they were diagnosed with dementia of any type, diagnosed with another type of neurodegenerative or neurological condition, taking medications that may negatively affect cognitive function, planning to participate in or currently enrolled in a clinical drug trial, or unable to speak as judged by an inability to communicate by phone. Of the 152 total potential participants who were recruited for this study, only 1 dropped out because of a transient ischemic attack unrelated to the study. Thus, our obtained sample size was 151 participants.  3.2.3 Measurement of moderate-to-vigorous physical activity and sedentary behaviour We measured MVPA and SB using a valid and reliable measure among older adults, the MW8 [136]. Briefly, the MW8 is a uniaxial, wrist-worn accelerometer designed to observe acceleration ranging in magnitude from 0.01g to 8g with a frequency of 3–11 Hz. The filtered acceleration signal is digitized, and the magnitude is summed over a user-specified time interval. At the end of each interval, the summed value or activity “count” is stored in memory and the integrator is reset. For the current study, we used 60-second epochs [528].  At study entry, participants were fitted with the MW8. Details of the measurement protocol used for the MW8 can be found elsewhere [137]. Consistent with established protocol for MW8, participants wore the device on the non-dominant wrist for a period of ≥4 days [137], which is enough to provide reliable estimates of MVPA (ICC=.90, 95% CI=.86–.93) and SB (ICC=.91, 95% CI=.87–.94]). After collection, stored activity counts were downloaded and saved to an IBM-compatible computer (IBM SPSS, Chicago, Illinois) for subsequent data reduction and analysis.   140   3.2.4 Data Reduction Data were analyzed using MotionWare 1.0.27 (CamNtech). Data prior to recorded wake time on the first full day of recording were manually removed in order to only investigate full 24-hour recordings of activity. Thus, a participant with 6 nights of sleep recorded had 5 full days of activity recorded. Each day of activity consisted of when the participant self-reported being awake and out of bed. Participant self-report was confirmed via event marker time stamps from MW8 or a consensus sleep diary which participants completed during the observation period.  A Microsoft Excel (Microsoft Corp, Redmond, Washington) macro written by RSF (Appendix D) was used to reduce data to the following 4 components: wake time for the participant by day; daily calculations of time the participant spent in SB (<1.5 METs) and MVPA (≥3.0 METs), as determined from the established cut-points [136]; average daily amounts of time spent in MVPA and SB; and average percentage of the day spent in MVPA (%MVPA) and SB (%SB). Non-wear time of MW8 was assessed as a period of consecutive zero counts ≥120 minutes in length [529]. In total, only 2 participants had periods of non-wear time (X̅=202 minutes/14 days; SD=107 minutes/14 days) according to this criterion. We therefore assumed participants did not remove the MW8 during observation.  3.2.5 Measurement of moderate-to-vigorous physical activity and sedentary behaviour bouts per day We also examined the average 10+ min bouts/day of MVPA and 30+ min bouts/day of SB. A Python 2.7 code written by BKC was used to analyze the data (Appendix D). The script loaded each participant’s data from the Microsoft Excel spreadsheet into a Python data table. The script  141   then cleaned and separated the data into tables, one for each type of activity level (i.e., MVPA and SB). The data loaded into the tables was a sequence of 1’s and 0’s, wherein a “1” on an epoch meant an activity level of the corresponding threshold was detected, and a “0” meant the activity level corresponding to the threshold was not detected. The script then iterated through each day’s epochs with an incrementing counter, such that when a “1” was detected, the counter incremented. Once a break (i.e., “0”) was detected, the counter was then saved to another data table which kept track of the length of the bouts (i.e., streaks of activity at the corresponding activity threshold). The counter was then reset and began counting again when the next bout of activity was detected. Once the bout lengths were counted, we determined the average 10+ min bouts/day of MVPA and the average 30+ min bouts/day of SB for each participant.  3.2.6 Cognitive Function We used the ADAS-Cog Plus to examine global cognitive function [454]. The ADAS-Cog Plus uses a multidimensional item response theory model that can flexibly utilize item scores from multiple cognitive assessment instruments to generate a global cognitive function score and standard error of measurement for that score [530]. Scores are defined by the ADNI cohort [455], wherein the mean score for cognitively healthy older adults is about −1.0, the mean for MCI is about 0.0, and the mean for dementia cases is about 1.0. Thus, higher scores indicate poorer cognitive performance. The ADAS-Cog Plus score was computed using the following 4 methods: the 13-item ADAS-Cog (validity: ICC= 0.80; test-retest: r= 0.93; [456]); Trail Making Tests A and B (validity: r= 0.36–0.93; test-retest: r= 0.67; [458]); Digit Span Forward and Backward (validity: r= 0.48–.85; test-retest: r= 0.62–0.82; [459]); and verbal fluency, consisting of animal  142   fluency and vegetable fluency (validity: r= 0.44–0.87; test-retest: r= 0.74; [459]). Detailed descriptions of the procedures used for these tests can be found elsewhere [456, 458, 459].  3.2.7 Data Analyses We performed all of our statistical analyses using R version 3.5.1. Our complete statistical analyses can be found in online in Appendix D (note: the published analyses were originally conducted in SPSS 22.0, but have been re-conducted in R version 3.5.1 for simplicity and reproducibility online). One participant was removed from our analyses as an outlier due to an ADAS-Cog Plus score which was >3 standard deviations above the mean score, suggesting the participant had possible severe cognitive impairment (MMSE=25; MoCA=17). Thus, our final sample size was 150. Given the non-normal distribution of average 10+ min bouts/day of MVPA, we transformed it by means of a natural log transformation which was used for all subsequent analyses as it most closely approximated a normal distribution. Because this was an exploratory analysis, we did not use a Bonferroni adjustment to account for multiple comparisons.  3.2.7.1 Participant characteristics based on probable MCI status Means and standard deviations were calculated for all variables of interests based upon probable MCI status. To determine demographic differences based on probable MCI status, we performed independent samples t-tests for continuous variables and chi-square tests for categorical variables, using probable MCI status (yes/no) as the grouping variable. In addition, we performed analyses of covariance (ANCOVA) to determine differences in %MVPA; 10+ min bouts/day of MVPA; %SB; 30+ min bouts/day of SB; and cognitive function based on probable MCI status. These  143   models controlled for age and sex differences while using probable MCI status as the grouping variable.  3.2.7.2 Relationship of cognitive function with moderate-to-vigorous physical activity and sedentary behaviour based on probable Mild Cognitive Impairment status We then examined the relationship of cognitive function with MVPA and SB for all participants. Multiple linear regression models were generated using ADAS-Cog Plus score as the dependent variable, while controlling for age, sex, and education. We generated four models, where the major independent variable of interest was %MVPA, 10+ min bouts/day of MVPA, %SB, or 30+ min bouts/day of SB. Beta estimates, model R2, and P values are presented for each model. We plotted the predicted relationship between each major independent variable and ADAS-Cog Plus score.  We then examined if the relationship of cognitive function with MVPA and SB differed based upon probable MCI status. Participants were stratified based upon probable MCI status and the same four multiple linear regression models were generated using ADAS-Cog Plus as the dependent variable, while controlling for age, sex, and education. We then performed z tests to determine if beta estimates for MVPA and SB differed significantly based on probable MCI status. Beta estimates, model R2, and p values are presented for each model based upon probable MCI status; z scores and p values for comparing major independent variable estimates based on probable MCI status are also presented. We then illustrated the differences in the predicted relationship between each independent variable and ADAS-Cog Plus based upon cognitive status.  3.3 Results  144    3.3.1 Participant characteristics based on probable Mild Cognitive Impairment status Participant characteristics are described in Table 3.1. The mean age was 71.11 years (SD=7.22 years) and 67.33% were female. Older adults without MCI had a mean MMSE score of 29.22 ± 0.10 (range=27–30), and a mean MoCA score of 27.19 ± 0.13 (range=26–30); older adults with probable MCI had a mean MMSE score of 28.65 ± 0.13 (range=25–30), and a mean MoCA score of 22.84 ± 0.23 (range=14–25). Older adults categorized with probable MCI were significantly older (t=2.70, df=149, p=.008) and more likely to be male (χ2=5.22, p=.022). While controlling for age and sex differences, participants with probable MCI also engaged in significantly less %MVPA (F=4.81; df=1,150; p=.030); fewer 10+ min bouts/day of PA (F=7.94; df=1,150; p=.005); and more 30+ min bouts/day of SB (F=4.04; df=1,150; p=.046); they also had poorer cognitive performance on the MMSE (F=7.36; df=1,150; p=.007), MoCA (F=215.78; df=1,150; p<.001), and ADAS-Cog Plus (F=25.22; df=1,150; p<.001).   145    Table 3.1 Participant characteristics Table 5 Participant Characteristic All participants (N=150) Non-MCI Older Adults (N=69) Probable MCI Older Adults  (N=81) p-value Age (years) 71.11 (7.22) 69.42 (6.37) 72.54 (7.62) 0.008 Females (n, %) 101, 67.33% 53, 76.81% 59.26% 0.022 Education (n, %)     High School Diploma or less 28, 18.40% 11, 14.70% 17, 20.70% 0.303 Trade School 16, 11.20% 5, 7.40% 11, 13.40%  Some University 24, 15.80% 10, 14.70% 14, 17.10%  University Diploma or Higher 82, 54.60% 43, 63.20% 39, 48.80%  Retired (n, %) 116, 77.33% 54, 78.26% 62, 76.54% 0.802 Smoking Status (n, %)     Current Smoker 3, 2.00% 1, 1.45% 2, 2.47% 0.877 Former Smoker 74, 49.33% 35, 50.72% 39, 48.15%  Never Smoked 73, 48.67% 33, 47.83% 40, 49.38%  Moderate-to-Vigorous Physical Activity and Sedentary Behaviour     %MVPA* 10.25 (6.51) 12.07 (7.19) 8.69 (5.45) 0.030|| LN(Average 10+ min bouts/day of MVPA)† 0.52 (0.44) 1.20 (1.45) 0.40 (0.36) 0.005|| %SB ‡ 59.62 (12.00) 57.24 (12.38) 61.65 (11.35) 0.161|| Average 30+ min bouts/day of SB 3.72 (1.83) 3.30 (1.73) 4.07 (1.85) 0.046|| Cognitive Function     Mini-Mental State Exam (MMSE) 28.91 (1.07) 29.22 (0.87) 28.65 (1.15) 0.007|| Montreal Cognitive Assessment (MoCA) 24.84 (2.77) 27.19 (1.10) 22.84 (2.11) <0.001|| ADAS-Cog Plus§ -0.79 (0.65) -1.11 (0.57) -0.52 (0.59) <0.001|| *Average percent of the day spent in moderate-to-vigorous physical activity (MVPA) †Natural log transformation for Average 10+ Minute bouts/day of MVPA ‡Average percent of the day spent in sedentary behaviour (SB) §Alzheimer’s Disease Assessment Scale Plus ||F-test while controlling for age and sex  146   3.3.2 Association of cognitive function with moderate-to-vigorous physical activity and sedentary behaviour Our models describing the relationship of cognitive function with MVPA and SB are described in Table 3.2. We found a significant relationship between higher %MVPA and better cognitive performance (β= −0.017, p=.024), and there was a marginal relationship between greater 10+ min bouts/day of MVPA and better cognitive performance (β= −0.203, p=.070). There was also a marginal relationship between higher %SB and poorer cognitive performance (β=0.007, p=.089). Finally, we found a significant association between greater 30+ min bouts/day of SB and poorer cognitive performance (β=0.061, p=.016). These relationships are illustrated in Figure 3.1.  Table 3.2 Association of sedentary behaviour and physical activity with Alzheimer’s Disease Assessment Scale Plus score Table 6 Independent Variable *Model R2 ΔR2 due to IV IV β value IV p-value Physical Activity (PA)     %MVPA† 0.321 0.025 -0.017 0.024 LN(Average 10+ min bouts/day of MVPA)‡ 0.312 0.016 -0.203 0.070 Sedentary Behaviour (SB)     %SB§ 0.311 0.014 0.007 0.089 Average 30+ min bouts/day of SB 0.324 0.027 0.061 0.016 DV= ADAS-Cog Plus score *Models controlling for age, sex, and education † Average percent of the day spent in MVPA ‡ Log transformed average 10+ min bouts/day of MVPA § Average percent of the day spent in SB 147    Figure 3.1 Association of moderate-to-vigorous physical activity and sedentary behaviour with cognitive function Figure 0.1  A) Association of percent of day spent in MVPA (%MVPA Time) with ADAS-Cog Plus Score; B) Association of number of 10+ minute bouts/day of MVPA with ADAS-Cog Plus Score; C) Association of percent of day spent in SB (%SB Time) with Alzheimer’s Disease Assessment Scale Plus score (ADAS-Cog Plus Score); D) Association of number of 30+ minute bouts/day of SB with ADAS-Cog Plus Score). Models controlled for age, sex, and education. 148   3.3.3 Relationship of cognitive function with moderate-to-vigorous physical activity and sedentary behaviour based on probable Mild Cognitive Impairment status Table 3.3 describes the relationship of cognitive function with MVPA and SB based on probable MCI status. For older adults without MCI, higher %MVPA and greater 10+ min bouts/day of MVPA were associated with better cognitive performance (β= −0.022 [p=.024] and β= −0.286 [p=.046], respectively). However, neither MVPA characteristic was associated with cognitive performance for older adults categorized with probable MCI. In addition, the beta estimates for %MVPA (z=2.412, p=.016) and 10+ min bouts/day of MVPA (z=1.986, p=.047) differed significantly based on probable MCI status.  We also found higher %SB was associated with poorer cognitive performance for participants without MCI (β=0.012, p=.038); there was a marginal relationship between greater +30 min bouts/day SB and poorer cognitive performance for participants without MCI (β=0.075, p=.064). By comparison, there was no relationship for either SB characteristic with cognitive performance for older adults categorized with probable MCI. Finally, the relationship of %SB with cognitive performance was marginally different based on probable MCI status (z=1.536, p=.125). These relationships are illustrated in Figure 3.2.  3.4 Discussion Our results suggest older adults with probable MCI engage in less MVPA and more SB compared with adults without MCI. In addition, the relationship of MVPA and SB (to a lesser degree) with cognitive function differs by cognitive status such that MVPA and SB are only associated with cognitive function among participants without MCI. We now provide potential explanations for 149   Table 3.3 Association of moderate-to-vigorous physical activity and sedentary behaviour with Alzheimer’s Disease Assessment Scale Plus score based on probable Mild Cognitive Impairment (MCI) status able 7 Independent Variable Non-MCI (N= 69) Probable MCI (N=81) IV β value differences based on probable MCI status  Model R2 ΔR2 due to IV IV β value IV p-value Model R2 ΔR2 due to IV IV β value IV p-value Z-score p-value Physical Activity (PA)             %MVPA† 0.209 0.066 -0.022 0.024 0.322 <0.001 <0.001 0.993 2.412 0.016 LN(Average 10+ min bouts/day of MVPA)‡ 0.195 0.052 -0.286 0.046 0.329 0.007 0.148 0.378 1.986 0.047 Sedentary Behaviour (SB)             %SB § 0.199 0.056 0.012 0.038 0.322 <0.001 <0.001 0.948 1.536 0.125 Average 30+ min bouts/day of SB 0.188 0.045 0.075 0.064 0.353 0.010 0.033 0.282 0.830 0.407 DV= ADAS-Cog Plus Score   *Models controlling for age, sex, and education   † Average percent of day spent in MVPA ‡ Log Transformed Average 10+ min bouts/day of MVPA § Average percent of day spent in SB            150   Figure 3.2 Association of moderate-to-vigorous physical activity and sedentary behaviour with cognitive function based on the presence of Mild Cognitive Impairment Figure 0.2  A) Association of percent of day spent in MVPA (%MVPA Time) with ADAS-Cog Plus Score; B) Association of number of 10+ minute bouts/day of MVPA with ADAS-Cog Plus Score. Models controlled for age, sex, and education; C) Association of percent of day spent in SB (%SB Time) with Alzheimer's Disease Assessment Scale Plus score (ADAS-Cog Plus Score); D) Association of number of 30+ minute bouts/day of SB with ADAS-Cog Plus Score). Models controlled for age, sex, and education. Blue=Non-MCI Older Adults; Red=MCI Older Adults. 151   these findings, as well as discuss how these findings can be applied to clinical practice.  3.4.1 Differences in physical activity and sedentary behaviour by cognitive status—does activity level change because of Mild Cognitive Impairment conversion? Previous studies suggest low PA and high SB are both risk factors for cognitive impairment in later life [474, 522]; however, our study is the first to show that older adults with probable MCI are less active and more sedentary than their cognitively healthy peers. One potential explanation is younger adults at high risk for MCI in later life—due to lower PA and higher SB—continue to be less active and more sedentary into older adulthood, often becoming cognitively impaired as they age. Indeed, the current evidence suggests that PA and SB patterns worsen from childhood into young adulthood and then stabilize from middle adulthood onward [531]. It is therefore plausible that older adults with MCI are less active and more sedentary throughout their adult lives, and this behavior has continued into older adulthood.  It is also plausible that a reciprocal association is present, such that conversion to MCI influences PA and SB. In particular, this effect might occur through diminished executive function [19, 36, 482]. Briefly, executive function is a broad term used to define planning and problem-solving and is known to significantly decline with age [532, 533]. Loss of executive function capability has been shown to negatively impact older adult independence and functionality [534-536]. Increasing evidence also suggests PA and SB have a bidirectional relationship with executive function such that changes in executive function can predict changes in activity levels, and vice versa [537-539]. Thus, older adults living with MCI may have impaired decision making about engaging in PA or SB, due to impaired executive function capabilities.  152    3.4.2 Differences in the relationships of physical activity and sedentary behaviour with cognitive function by Mild Cognitive Impairment status—are there underlying differences in the Mild Cognitive Impairment brain? To our knowledge, this is the first study to report differences in the relationships of MVPA and SB with cognitive function based upon cognitive status. One explanation is there might be a minimum threshold of MVPA and/or a maximum threshold of SB required to elicit a relationship with cognitive function. As discussed previously, older adults with probable MCI engaged in less PA and more SB than adults without MCI. As such, older adults with MCI may not meet a minimum threshold level of PA—or may exceed a maximum threshold of SB—which may lead to non-significant associations between these health behaviors and cognitive function. This interpretation of our findings may help explain why exercise and PA interventions for older adults with MCI can lead to significant improvements in cognitive function [185, 201]. Potentially, exercise and PA interventions for older adults with MCI help ameliorate cognitive function by providing a necessary threshold level of PA and concomitant reduction in SB. By comparison, the effects of PA and exercise interventions on older adults without MCI appears to be less substantial [540], perhaps due to higher levels of basal activity in these adults.  An alternative explanation for our findings is the relationships of MVPA and SB with cognitive performance differ by MCI status because of underlying neurobiological differences between older adults with MCI and those without [541]. For example, compared to older adults without MCI, those with MCI have greater amounts of Aβ accumulation [39], accelerated atrophy in the medial temporal lobe [40], and decreased connectivity of the posterior cingulate gyrus and hippocampus  153   with the whole brain [46]. These underlying changes in the MCI brain may potentially alter the relationships of health behaviors with cognitive function [2], leading to an attenuation—or functional weakening—of the relationships of MVPA and SB with cognitive function [296].  3.4.3 Clinical applications The findings of our study are also applicable towards improving clinical practice. First, given that PA and SB can have important implications on older adult physical and cognitive health [106, 522], all clinicians should make a serious effort to counsel their patients on PA and SB [527]. While the health care system has serious potential as a tool for promoting changes in older adult PA and SB, its potential is not being fully realized in clinical practice [542]. A first step to utilizing this untapped potential to promote behavior change is for clinicians to track their older adult patients’ PA and SB using objective monitors, such as a pedometer or a FitBit. While activity trackers may not be feasible for some clinicians to use in their practice, clinicians should at the very least ask brief questions about activity during their consultations with older adult patients since it can have important implications on both the physical and cognitive health of older adults [543].  Second, our data suggest that PA appears to have a stronger relationship with cognitive function than does SB. A recently published systematic review examining the relationship of SB with later life cognitive decline found that in order to best promote healthy cognitive aging, older adults should limit their SB and concomitantly increase their MVPA to ≥150 minutes/week [522]. In order to best promote healthy cognitive aging, it may thus be prudent at this time for clinicians to focus on ensuring their older adult patients obtain this threshold level of PA, and to advise their  154   patients to limit SB. Over 95% of older adults are underactive, and thus the greatest benefits to older adult physical and cognitive health may occur by increasing PA.  While the main findings of our study are complex and require future investigation, the recommendations highlighted above are practical for both older adults with MCI, and those without. We therefore suggest clinicians who are concerned about their older adult patients’ cognitive health should track patient activity levels with objective monitors, if possible, and at least ask simple questions about activity levels during patient visits; and encourage their older adult patients to engage in ≥150 minutes/week of MVPA and limit their SB as much as possible.  3.4.4 Limitations and future research This study was cross-sectional and therefore cannot establish whether conversion to MCI attenuates the relationship of MVPA and SB with cognitive function. While MoCA is a standard measure for determining the cognitive status of older adults, MCI diagnosis is confirmed through a physician based on several criteria, including 1) medical history; 20 assessment of independence and activities of daily living; 3) subjective memory complaints; 4) neurological assessment; and 5) laboratory tests and neuroimaging [19, 36, 482]. Given that we did not confirm the diagnosis of MCI with a physician, we cannot conclude how many of the participants we classified as living with probable MCI would actually be diagnosed with MCI by a physician.  This was a secondary analysis of a study which did not obtain information on participant comorbidities, and thus we did not account for these potential confounders within our analyses.  155   Future studies will need to determine whether the associations of MVPA and SB with cognitive performance do indeed differ by cognitive status when accounting for comorbid conditions.  Our findings may have been influenced by an overlap of the constructs of MVPA and SB. Specifically, our MVPA and SB data had a medium-sized negative correlation (r= −.515); however, a recent systematic review has also concluded that total PA and total SB have a modest to medium sized correlation [10].  While we used a measure of MVPA and SB which has good evidence of validity and reliability within this population [136, 137], given that we previously calibrated this device for MVPA and SB using this sample of older adults, the MW8 may still have over- or under-estimated participant MVPA and SB. We therefore suggest that future investigations examine the concurrent validity of the MW8 against another type of objective measurement tool (preferably a hip accelerometer; [111, 122]) and examine whether the findings reported in this manuscript can be replicated in other samples using other types of objective measures. We only measured one intensity level of PA (i.e., MVPA), and thus our conclusions about the cross-sectional relationships between PA and cognitive function in older adults with MCI may be different when accounting for LPA (i.e., 1.5-2.9 METs; [544]).  Future intervention research will also be needed to determine the minimum effective dose of PA—or SB—necessary for maintaining or improving cognitive function, and how this might differ by MCI status. In addition, future research will need to examine whether the relationships of MVPA and SB with cognitive function differ by MCI status due to underlying neurobiological changes.  156    3.4.5 Conclusions This study found that older adults with probable MCI engage in less MVPA and more SB than older adults without MCI. Our results also suggest the relationships of MVPA and SB with cognitive function differ by probable MCI status. Future investigations should examine if a threshold amount of PA and SB is required to maintain cognitive health in older adults, or if underlying neurobiological changes associated with MCI alter the relationships of PA and SB with cognitive function.  157   Chapter 4: The independent associations of physical activity and sleep with cognitive function in older adults  A version of this chapter is published as: Falck RS, Best JR, Davis JC, Liu-Ambrose T. The independent associations of physical activity and sleep with cognitive function in older adults. Journal of Alzheimer's Disease. 2018; 63(4): 1469-1484.  4.1 Introduction Maintaining older adult cognitive health is a prominent public health challenge of the 21st century, as one new case of dementia is detected every 4 seconds [1]. Age is the number one risk factor for dementia [464], and as the aging population grows [465], so too will the number of dementia cases. While there is not yet a pharmaceutical cure for dementia, non-pharmaceutical approaches may help reduce dementia risk [2]. As such, lifestyle and behavioural strategies are an important line of scientific inquiry for maintaining older adult cognitive health [545].  One strategy is increasing older adult PA, which has positive benefits on both physical and cognitive health [106]. Importantly, evidence suggests regular PA of ≥3.0 METs (i.e., MVPA) can reduce the risk of dementia by up to 28% [156].While it is promising that meeting current PA guidelines of ≥150 minutes/week of MVPA may help maintain cognitive health throughout life [95], most older adults fall short of these recommendations [96]. Thus, increasing older adult PA has become a growing public health priority since it could help prevent up to 18% of all AD cases [468].  Current evidence suggests PA can impact cognitive health through multiple mechanisms [170], including stimulating neurogenesis (i.e., the creation of new neurons) and cerebral angiogenesis  158   (the creation of new blood vessels in the brain). The precise physiological mechanism by which PA improves cognitive health is still under investigation, but it is clear that PA increases cellular proliferation, dendritic complexity, and dendritic spine density in the dentate gyrus of the hippocampus [227-230]. These adaptations in cytoarchitecture also occur in older animals [231-233, 546], suggesting the importance of PA for cognitive health throughout life. In addition to the benefits of PA for maintaining cognitive health, animal models have also experimentally shown that PA down-regulates inflammatory factors which are associated with the progression of AD [239, 240]. Specifically, transgenic mice given access to a running wheel have significantly reduced Aβ levels in their frontal cortex [239], and reduced pro-inflammatory markers IL-1β and TNF-α [240]. Both of these inflammatory markers are associated with increased Aβ load and have been linked to AD progression [242, 547].   Another promising strategy for promoting cognitive health is improving older adult sleep quality. Sleep complaints are common among older adults, with more than half of adults over 65 years reporting at least one chronic sleep complaint—the most common being the inability to stay asleep at night [323]. Changes in sleep quality are a normal consequence of aging [321]. However, poor sleep is prevalent and predictive of cognitive decline in older adults [345]. Unfortunately, while effective cognitive-behavioural interventions are available for poor sleep [366, 548], few people utilize these treatments [549].  Evidence suggests poor sleep can have serious consequences on cognitive function [326]. The observed neurocognitive impairments which are a consequence of poor sleep are attributable to suboptimal prefrontal cortical functionality [339, 340]. Importantly, the prefrontal cortex is the  159   principal cortical area involved with higher-order cognitive function [404, 550, 551]. Coupled with these data suggesting poor sleep can impair cognitive function is the growing body of evidence that poor sleep is not only more prominent among individuals with AD, but also increases the risk of developing AD [341]. Animal models have experimentally shown 1) increasing cortical Aβ, a key indicator of AD pathophysiology, increases sleep fragmentation [351]; 2) decreasing sleep quality and increasing wake-time escalates Aβ production and corresponding cortical deposition [352]; and 3) sleep promotes the clearance of extracellular Aβ which accumulates during wake-time [356]. Collectively, these results suggest sleep is a critical pathway through which the brain appears to maintain cognitive health. When this pathway is disrupted, a vicious cycle of accelerating AD progression may occur—wherein poor sleep causes an increase in AD progression, and vice-versa [9].   Importantly, PA has long been thought to help improve poor sleep [365]. Epidemiological studies have consistently found people believe PA improves their sleep, and people with higher PA level report sleeping better compared to more sedentary individuals [11]. While the reasons for why PA and sleep are related are still unclear, current evidence suggests three possible explanations [11, 364]. One theory suggests that since negative affective states (i.e., depressive symptoms and anxiety) contribute to poor sleep [366], the antidepressant and anxiolytic effects of PA explain the relationship between sleep and PA [367, 368]. Another hypothesis suggests better weight regulation through increased PA may be associated with better sleep quality [370], since obesity is related to poorer sleep quality [369]. A third hypothesis suggests that since poor physical function is associated with poorer sleep quality in older adults [372], and PA is associated with  160   improved physical function [371], PA may be related to better sleep quality through its effect on physical function.  While these preliminary hypotheses are interesting, the current evidence for a relationship between PA and sleep quality is largely based on self-reported PA and sleep quality [11, 365], which can yield vastly different measurements from objective reality [286, 373]. Since PA and sleep quality may be associated with each other and with cognition, investigating the relationships PA and sleep quality have with cognitive function should not be performed in isolation. For example, it is plausible that PA does not independently predict cognitive function, but influences cognition through sleep (or vice-versa). In order to address these questions regarding the independent associations of PA and sleep quality with older adult cognitive function, and the relationship of PA with sleep quality, it is necessary to examine objective PA and objective sleep quality concomitantly [136, 137].  This study therefore addresses the current gaps in the literature by examining whether 1) PA is associated with better cognitive performance independently of sleep quality; 2) sleep quality is associated with better cognitive performance independently of PA; and 3) whether PA is associated with sleep quality.  4.2 Methods Ethical approval for this study was obtained from the Vancouver Coastal Health Research Institute and the University of British Columbia’s Clinical Research Ethics Board (H14-01301). All participants provided written informed consent.  161    4.2.1 Protocol For this cross-sectional study, we recruited and collected data between August 27, 2014 and April 27, 2016. At study entry, we ascertained general health, demographics, socioeconomic status, and education by questionnaires. During this initial session, we measured subjective sleep quality using the PSQI [299]. Participants’ PA and sleep were then observed for 14 days using the MW8. Following MW8 observation, we measured cognitive function using the ADAS-Cog Plus [454].   4.2.2 Participants Participants (N= 157; aged 55+ years) were recruited from Vancouver, British Columbia by advertisements placed in local community centres, newspapers, and word of mouth referrals. Participants were included if they met the following criteria: 1) men and women 55+ years of age living in the Metro Vancouver area; 2) scored >24/30 on the MMSE [487]; and 3) able to read, write, and speak English with acceptable visual and auditory acuity. Participants were excluded if: 1) diagnosed with dementia of any type; 2) diagnosed with another neurodegenerative or neurological condition; 3) taking medications which may negatively affect cognition; 4) planning to participate or currently enrolled in a clinical drug trial; or 5) unable to speak as judged by an inability to communicate by phone.   4.2.3 Subjective measurement of sleep quality We measured subjective sleep quality using the PSQI [299]. This 19-item questionnaire assesses sleep quality using subjective ratings for 7 different components (i.e., sleep quality; sleep latency; sleep duration; habitual sleep efficiency; sleep disturbance; use of sleep medication; and daytime  162   dysfunction). Participants answer the questionnaire retrospectively, as the questionnaire surveys sleep components spanning the previous month. The questionnaire has good evidence of validity and reliability [301, 453].   4.2.4 Objective measurement of physical activity and sleep We measured PA and sleep using MW8, a uni-axial, wrist-worn accelerometer with evidence of validity and reliability for use among older adults [136, 137]. For the current study, we used 60 second epochs which is consistent with current guidelines for estimating both PA and sleep [452, 528, 552], and the capabilities of the MW8 to measure sleep and PA concurrently. Participants were fitted with the MW8 actigraphy unit and provided detailed information on its features (i.e., the light sensor, event marker button, and status indicator). Participants were instructed to press the event marker button each night when they started trying to sleep; and again each morning when they finished trying to sleep. Consistent with established protocol for wrist-worn actigraphy, participants wore the MW8 on the non-dominant wrist for a period of 14 days [552, 553].   Participants were also given the 9-item CSD and asked to complete it each morning upon waking [302]. The responses from the CSD were used to confirm sleep windows identified by participants, as determined by the time stamped event markers. In cases where the event marker and CSD entry disagreed for the start time of the sleep window, we used light sensor data to determine “lights out”. Similarly, when the event marker and CSD entry disagreed for the end of the sleep window, we used “lights on” and activity onset to determine the end of the sleep window. Each day of PA consisted of when the participant reported being awake and out of bed (as per responses to the CSD and confirmed via event marker time stamps from MW8). If responses from the CSD entry  163   disagreed with the event markers entered by participants as the start of the day, we used light sensor data and activity onset to determine the start of the day. Similarly, when the event marker and CSD entry disagreed for the end of day (i.e., time spent trying to sleep), we used the light sensor data to determine the end of the day.   4.2.4.1 Data Reduction  Details of our data reduction procedure can be found elsewhere [137, 286]. Briefly, data were analyzed using MotionWare 1.0.27 (camntech). Data prior to recorded wake-time on the first full day of recording were manually removed in order to only investigate full 24-hour recordings of activity. Each day of activity consisted of when the participant self-reported being awake and out of bed. Participant self-report was confirmed via event marker time stamps from MW8.  The MotionWare software was used to estimate different parameters of sleep quality including: fragmentation index, sleep efficiency (time asleep expressed as a percentage of time in bed), sleep duration (total time spent sleeping), sleep latency (time between “lights out” and falling asleep), and number of awakenings (number of times the participant woke up during the sleep period). Briefly, fragmentation index is defined by MotionWare as the sum of 1) the total time spent sleeping categorized as mobile in the epoch-by-epoch mobile/immobile categorization expressed as a percentage of the time spent asleep; and 2) the number of immobile bouts which were ≤ 1minute in length expressed as a percentage of the total number of immobile bouts during time spent sleeping. The MW8 estimates of sleep quality have evidence of validity and reliability [298].    164   A Microsoft Excel macro written by RSF (Appendix D) was used to reduce PA data to daily calculations of time spent in MVPA (≥ 3.0 METs; [136]). We then calculated the percent of each day spent in MVPA (%MVPA). For example, if for a given day a participant was asleep from 12:00-6:00 AM, and went to bed the following evening at 10:00 PM, then we assumed that the participant spent 8 hours sleeping in a 24 hour period (i.e., 480 minutes). If this individual engaged in 100 minutes of MVPA, then the estimated %MVPA for that day would be: 100 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 𝑜𝑓 𝑀𝑉𝑃𝐴960 𝑚𝑖𝑛𝑢𝑡𝑒𝑠 𝑜𝑓 𝑤𝑎𝑘𝑒 𝑡𝑖𝑚𝑒 = 10.42% of day spent in MVPA The benefit of this approach is that it controls for differences in time which participants spent awake and out of bed.  We determined MW8 non-wear time using the criterion of Hutto and colleagues [529]. Briefly, this criterion suggests periods of consecutive zero counts ≥120 minutes in length be considered as non-wear time. Only two participants had periods of non-wear time (total non-wear times of 126 minutes and 278 minutes, respectively) according to this criterion. Thus, we assumed participants did not remove the MW8 during observation.  4.2.5 Cognitive Function  We used the ADAS-Cog Plus as a global measure of cognitive function [454]. The ADAS-Cog Plus uses a multidimensional item response theory model which can flexibly utilize item scores from multiple cognitive assessment instruments to generate a global cognitive function score and standard error of measurement for that score. Higher scores indicate poorer cognitive performance. The ADAS-Cog Plus score was computed using the 13-item ADAS-Cog [456], MoCA [457], Trail  165   Making Test A and B [458], Digit Span Forward and Backward [459], and verbal fluency [458]. Detailed descriptions of the procedures used for these tests can be found in Appendix C.   4.2.6 Statistical Analyses We performed all of our statistical analyses using R version 3.3.1 using the lavaan 0.5-22 package. Our statistical code can be found in Appendix D. Because MW8 requires 14 days of continuous observation in order to provide a reliable estimate of both PA and sleep quality [137], we needed to exclude 11 participants with incomplete data (i.e., 6 participants did not complete 14 days of MW8 observation; 5 participants did not have complete cognitive data). We also excluded 9 participants from our analyses because they each had extreme outlier scores on variables of interest (i.e., >3 SD from the mean). Two participants had extreme average %MVPA (32.2% and 38.6% of day spent in MVPA), four participants had extreme average fragmentation indices (range: 66.14-71.85) and two had extreme average sleep latency (42 minutes and 72 minutes). In addition, we removed one participant from our analyses due to an extreme ADAS-Cog Plus score (1.52), which suggested possible severe cognitive impairment (MMSE = 25; MoCA = 17). Structural equation modeling is based on the general linear model, which assumes the data has a normal distribution; including these data would skew the distribution of our data and increase type I error [554]. Thus, our final sample size was 137.   Greater numbers of nocturnal awakenings decreases the accuracy of actigraphy [555], and this inaccuracy mostly affects measures of wakefulness during the night and sleep latency [452, 556]. As such, we included as a covariate the number of awakenings each night in each of our analyses of sleep efficiency (r= -0.62; p< 0.01), sleep fragmentation (r= 0.58; p< 0.01), and sleep latency  166   (r= 0.20; p< 0.01) in an effort to reduce error variance. Controlling for the number of awakenings also intuitively seems appropriate for determining the most accurate estimates of sleep fragmentation and sleep efficiency. For example, an individual could have 100 separate 1-minute bouts of wakefulness over the course of 8 hours spent in bed, or one bout of 100 minutes of wakefulness. Not controlling for the number of awakenings would provide identical estimates for sleep efficiency despite very different sleep architectures between these two hypothetical individuals. We expect there to be a similar example for sleep fragmentation, and thus controlling for the number of awakenings potentially provides the most accurate estimates of sleep fragmentation and sleep efficiency.  4.2.6.1 Preliminary Analyses We calculated mean %MVPA, PSQI score, fragmentation index, sleep efficiency, sleep duration, and sleep latency over 14 days and then estimated bivariate correlations between all variables. Given our sample size of 137 participants and a two-tailed α= 0.01, we had 80% power to detect a two-tailed correlation with a small effect size (|ρ|= 0.28; [557]).  4.2.6.2 Main Analyses Our main analyses used latent variable modeling to examine whether: 1) PA is associated with cognitive performance independently of sleep quality (i.e., PSQI score, fragmentation index, sleep efficiency, sleep duration, and sleep latency); 2) sleep quality is associated with cognitive function independently of PA; and 3) PA is associated with better sleep quality. The strength of this approach is that it allows researchers to examine latent variable models which provide separate estimates of relations of the latent construct with the measured variables used to estimate the latent  167   construct (