Citation
The Influence of Structured, Long Term Exercise on Objectively Measured Daily Activity in Older Adults

Material Information

Title:
The Influence of Structured, Long Term Exercise on Objectively Measured Daily Activity in Older Adults
Creator:
Wanigatunga, Amal Asiri
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Epidemiology
Committee Chair:
MANINI,TODD M
Committee Co-Chair:
YAGHJYAN,LUSINE
Committee Members:
LU,XIAOMIN
WOODS,ADAM J

Subjects

Subjects / Keywords:
accelerometry
activity
adults
exercise
older
physical

Notes

General Note:
While exercise, defined as moderate-to-vigorous intensity physical activity (MVPA), has been observed to provide substantial cardiometabolic benefits in older adults, this subpopulation of US adults aged 70+ years remains the most physically inactive. Previous literature has consistently observed physical inactivity to predict adverse health outcomes, such as physical disability, which leads to increased risks in morbidity and mortality. Thus, large-scale interventions such as the Lifestyle Interventions and Independence for Elders (LIFE) study have developed exercise interventions which have been observed to delay physical disability and increase independence towards daily living. However, the effect of long-term exercise on daily lifestyle physical activity, especially those activities which require low amounts of energy expenditure and constitute most daily physical activity, remains unknown. Sedentary behaviors, defined as waking activities which involve sitting or reclining and require less than or equal to 1.5 metabolic equivalents (METs), make up the most engaged activity level by US older adults. These sedentary pursuits have been identified as a potential health risk factor for age-related morbidities and mortality, independent of exercise participation. The second most engaged daily activity level, light-intensity activity, has been independently associated with improved health markers in older adults and is defined as activities requiring between 1.5-3 METs. Yet, with the infusion of daily MVPA, the question remains whether attributes of these activity levels are influenced. In addition, hospitalizations-acute events which displace activity patterns- among older adults negatively affect daily lifestyle yet previous literature on the topic is sparse. Therefore, the intervening effect of hospitalizations on the association between exercise and daily activity accumulation is unclear, yet important to understand. The LIFE study provides a unique opportunity to examine the effect of exercise and daily activity accumulation because it is the largest physical activity intervention to date, which also collected objectively measured physical activity through accelerometry. Therefore, this study was to examine accelerometer-based patterns of activity and inactivity in terms of volume, intensity, and duration and how these variables were influenced by long-term moderate-intensity exercise. Additionally, the effect of exercise post hospitalization on daily physical activity, a novel and not well understood relationship, was examined.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
12/31/2018

Downloads

This item has the following downloads:


Full Text

PAGE 1

THE INFLUENCE OF STRUCTURED, LONG TERM EXERCISE ON OBJECTIVELY MEASURED DAILY ACTIVITY IN OLDER ADULTS By AMAL ASIRI WANIGATUNGA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016

PAGE 2

2016 Amal Asiri Wanigatunga

PAGE 3

To my parents, late grandfather, family, and friends

PAGE 4

4 ACKNOWLEDGMENTS I thank my parents for their continual and unyielding support in all areas of my life. I thank my la te grandfather who helped shape my understanding of service to our fellow humans. I express gratitude to my family and friends who inspire me to be the best version of myself. I am profoundly grateful to Dr. Todd Manini who continues to be a role model to me and is the best mentor I cou ld ever have. I gratefully acknowledge my dissertation committee members, Drs. Adam Woods, Xiaomin Lu, and Lusine Yaghjyan for their support, positiv e influence and guidance during this journey. I am indebted to Dr. Cindy Prins who had provided immense su pport at all times of need. Finally, I thank the University of Flori da, a top tier academic institution of excellence for allowing me to continue my studies as a graduate student

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 12 CHAPTER 1 INTRODUCTI ON ................................ ................................ ................................ .... 14 Aging and Public Health ................................ ................................ .......................... 14 Physical Activity ................................ ................................ ................................ ...... 15 Sedentary Behavior ................................ ................................ ................................ 16 Light Inte nsity Physical Activity ................................ ................................ ............... 17 Measuring Physical Activity ................................ ................................ .................... 18 Hospitalizations ................................ ................................ ................................ ....... 20 Rationale ................................ ................................ ................................ ................. 21 2 APPROACH ................................ ................................ ................................ ............ 24 Overall Strategy ................................ ................................ ................................ ...... 24 Analytic Sample ................................ ................................ ................................ ...... 24 Accelerometry ................................ ................................ ................................ ......... 27 Hospitalizations ................................ ................................ ................................ ....... 32 Intervention s ................................ ................................ ................................ ........... 33 Covariates ................................ ................................ ................................ ............... 35 3 EFFECTS OF A LONG TERM PHYSICAL ACTIVITY PROGRAM ON SEDENTARY TIME IN OLDER ADULTS ................................ ............................... 42 Background ................................ ................................ ................................ ............. 42 Methods ................................ ................................ ................................ .................. 43 Trial Design and Participants ................................ ................................ ............ 43 Interventions ................................ ................................ ................................ ..... 44 Accelerometer Measurements ................................ ................................ .......... 45 Se dentary Variables ................................ ................................ ......................... 46 Physical Function Measures ................................ ................................ ............. 47 Statistical Analysis ................................ ................................ ............................ 47 Results ................................ ................................ ................................ .................... 48 Population Sample ................................ ................................ ........................... 48

PAGE 6

6 Sedentary Variables by Intervention ................................ ................................ 49 Associations with Change in Physical Function ................................ ................ 50 Discussion ................................ ................................ ................................ .............. 50 4 EFFECTS OF A LON G TERM MODERATE INTENSITY EXERCISE PROGRAM ON PHYSICAL ACTIVITY COMPOSITION IN MOBILITY IMPAIRED OLDER ADULTS ................................ ................................ .................. 60 Background ................................ ................................ ................................ ............. 60 Methods ................................ ................................ ................................ .................. 63 Trial Design and Participant Population ................................ ........................... 63 Interventions ................................ ................................ ................................ ..... 64 Accelerometer Measurements ................................ ................................ .......... 65 Physical Activity Characterization ................................ ................................ ..... 65 Other Measurements ................................ ................................ ........................ 67 Statistica l Analysis ................................ ................................ ............................ 67 Results ................................ ................................ ................................ .................... 68 Sample Characteristics ................................ ................................ ..................... 68 Activity Intensity and Bout Lengths at Baseline ................................ ................ 68 Intervention Effects on Intensities in Total Activity Time ................................ ... 69 Intervention Effects on Time Spent at Intensities and Bout Lengths ................ 69 Discussion ................................ ................................ ................................ .............. 70 5 INVESTIGATING THE INTERVENING EFFECT OF HOSPITALIZATIONS ON PHYSICAL ACTIVITY PATTERNS MEASURED BY ACCELEROMETRY ............. 81 Background ................................ ................................ ................................ ............. 81 Methods ................................ ................................ ................................ .................. 83 Trial Design and Participant Population ................................ ........................... 83 Interventions ................................ ................................ ................................ ..... 84 Accelerometer Measurements ................................ ................................ .......... 86 Accelerometer Variables ................................ ................................ .................. 87 Hospitalization Variable ................................ ................................ .................... 89 Covariates ................................ ................................ ................................ ........ 90 Statistical Analys is ................................ ................................ ............................ 90 Results ................................ ................................ ................................ .................... 92 Sample Characteristics ................................ ................................ ..................... 92 Baseline Accelerometer Metrics ................................ ................................ ....... 94 Hospitalization Effect on Accelerometer Metrics ................................ .............. 95 ...... 97 Discussion ................................ ................................ ................................ .............. 97 6 CONCLUSIONS ................................ ................................ ................................ ... 111 Summary ................................ ................................ ................................ .............. 111 Future Directions ................................ ................................ ................................ .. 115

PAGE 7

7 APPENDIX A R CODE USED TO GENERATE ACCELEROMETER VARIABLES .................... 117 B R CODE USED TO GENERATE HOSPITAL VARIABLES ................................ ... 129 C HOSPITALIZATIONS BY INTERVENTION ................................ .......................... 132 D HOSPITALIZATION DESCRIPTIVES BY TIME INTERVAL ................................ 134 E CORRELATION ANALYSIS BETWEEN HOSPITALIZATION AND ACCELEROMETRY ................................ ................................ .............................. 135 LIST OF REFERENCES ................................ ................................ ............................. 138 BIOGRAPH ICAL SKETCH ................................ ................................ .......................... 148

PAGE 8

8 LIST OF TABLES Table page 2 1 Baseline participant characteristics by accelerometry collection status .............. 36 2 2 Accelerometry data collection by visit ................................ ................................ 37 2 3 Accelerometry characteristics ................................ ................................ ............. 38 2 4 Accelerometry cut points denoting intensity ranges ................................ ........... 39 2 5 Staff contact by intervention group ................................ ................................ ..... 40 3 1 Baseline participant characteristics by intervention ass ignment ......................... 55 3 2 Changes in sedentary behavior patterns and physical function by baseline SPPB a functional status ................................ ................................ ...................... 56 4 1 Descriptions of accelerometer derived metrics of physical activity composition ................................ ................................ ................................ ........ 76 4 2 Baseline participant characteristics by intervention group ................................ .. 77 4 3 Baseline daily activity level and bouts by intervention group .............................. 78 5 1 Accelerometer cut points used to derive metrics of physical inactivity and activity ................................ ................................ ................................ .............. 106 5 2 Baseline participant characteristics by intervention group and hospitalization status over 24 months ................................ ................................ ...................... 107 5 3 Hospitalizations by intervention group among participants who collected 3+ days of 10+ hours/day of accelerometry ................................ ........................... 108 5 4 Descriptive baseline dai ly activity bouts by intervention group and hospitalization^ ................................ ................................ ................................ 109 5 5 Effect of hospital duration on daily accelerometer pattern c hanges (daily minutes) in mobility limited older adults ( beta coefficient (SEM))^ .................... 110 C 1 Reason for hospitalization across interv ention group ................................ ....... 132 C 2 Baseline participant characteristics by hospitalization status and intervention over 24 months ................................ ................................ ................................ 133

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Conceptual framework of the interaction between moderate intensity exercise, hospitalizations, and daily waking activity.. ................................ ......... 23 2 1 Tri axial accelerometer. The vertical plane is the x axis; medial lateral is the y axis; and anterior posterior is the z axis. ................................ ......................... 41 3 1 Flow of participants through the study.. ................................ .............................. 57 3 2 Intervention effect across sedentary measures. ................................ ................. 58 3 3 Association between standardized changes in sedentary measures and physical function.. ................................ ................................ ............................... 59 4 1 Compositional changes of total daily physical activity by intensity and interven tion status. ................................ ................................ ............................. 79 4 2 Intervention effects on activity intensity levels by bout pattern. .......................... 80 D 1 Flow diagram of participants hospitalized within each time interval according to intervention. Note: LOS length of stay (average daysSD) ....................... 134 E 1 Correlation between days from hospitalization and subsequent accelerometer collection and accelerometer outcome metrics ......................... 135 E 2 The physical activity intervention effect (beta coefficient, SEM) on sedentary and activity bouts ................................ ................................ .............................. 136 E 3 The physical activity intervention effect (beta coefficient, SEM) on intensity specific activity bouts ................................ ................................ ........................ 137

PAGE 10

10 LIST OF ABBREVIATIONS 3MSE ADLs Modified Mini Mental State Examination Activities of Daily Living CHAMPS c m DMAQC ECG G g HE HLPA HMPA Hz IADLs kg LIFE LIFE P LLPA LMPA LPA MB MedRA m METs min Community Health Activities Model Program Centimeter Data Management and Quality Check Electrocardiogram Gravity Grams Health Education High Light Physical Activity High Moderate Physical Activity Hertz Instrumental Activities of Daily Living Kilogram Lifestyle Interventions and Independence for Elders Lifestyle Interventions and Independence for Elders Pilot Low Light Physical Activity Low Moderate Physical Activity Light Physical Activity Megabytes Medical Dictionary for Regul atory Activities Meter Metabolic Equivalents M inute

PAGE 11

11 MMD mmHg MVPA NHANES PA RCT RPE SB SBRN SD SPPB TPA Major Mobility Disability Millimeters of Mercury Moderate to Vigorous Intensity Physical Activity National Health and Nutrition Examination Survey Physical Activity Randomized Controlled Trial Rating of Perceived Exertion Sedentary Behavior Sedentary Behaviour Research Network Standard Deviation Short Physical Performance Battery Total Physical Activity US VO 2 Max United States Maximal Oxygen Consumption

PAGE 12

12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TH E INFLUENCE OF STRUCTURED, LONG TERM EXERCISE ON OBJECTIVELY MEASURED DAILY ACTIVITY IN OLDER ADULTS By Amal A. Wanigatunga December 2016 Chair: Todd M. Manini Major: Epidemiology Whil e exercise, defined as moderate to vigorous intensity physical activity (MVPA) has been observed to provide substantial cardiometabolic benefits in older adults, this subpopulation of US adults aged 70 years and older remains the most physically inactive. Previous literature has consistently observed physical inactivity to predict adverse health outcomes, such as physical disability, which leads to increased risks in morbidity and m ortality. Thus, large scale intervention studies such as the Lifestyle Interventions and Independence for Elders (LIFE) study have developed exercise interventions which have been observed to delay physical disability and increase independence towards dail y living. However, the effect of long term exercise on daily lifestyle physical activity especially those activities which require low amounts of energy expenditure and constitute most daily physical activity, remains unknown. Sedentary behaviors, define d as waking activities which involve sitting or reclin ing activity level by US older adults. These sedentary pursuits have been identified as a potential health risk factor for age re lated morbidities and mortality, independent of exercise participation. The second most engaged daily activity level, light intensity

PAGE 13

13 activity, has been independently associated with improved health markers in older adults and is defined as activities requ iring between 1.5 3 METs. Yet, with the infusion of daily MVPA the question remains whether attributes of these activity levels are influenced. In addition, hospitalizations acute events which displace activity patterns among older adults negatively aff ect daily lifestyle yet previous literature on the topic is sparse. Therefore, the intervening effect of hospitalizations on the association between exercise and daily activity accumulation is unclear, yet important to understand. The LIFE study provides a unique opportunity to examine the effect of exercise and daily activity accumulation because it is the largest and longest physical activity intervention to date which also collected objectively measured physical activity throug h accelerometry. Therefor e, this study was to examine accelerometer based patterns of activity and in activity in terms of volume, intensity, and duration and how these variables were influenced by long term moderate intensity exercise. Additionally, the effect of exercise post hos pitalization on daily physical activity, a novel and not well understood relationship, was examined

PAGE 14

14 CHAPTER 1 INTRODUCTION Aging and Public Health The population across the world is aging due to increases in lifespan ( 22 ) In 2014, the World Health Organization projected that the amount of adults 60 years and older a round the globe will double from 2000 to 2050 ( 85 ) Among this age group, those aged 80 years and older are projected to almost quadruple from 2000 to 2050. Also by 2050, the number of older adults is projected to outnumber children under the age 14 years old ( 85 ) In the United States, the population of older adults aged 65 years and older increased 24.7% from 2003 to 2013, reaching 44.7 million and is projected to more than double to 98 million in 2040 ( 109 ) Furthermore, the population of older adults 85 years and older is projected to rapidly increase from 6 million in 2013 to 14.6 million in 2040. This has become a public health challenge because older adults carry the highest burden of chronic disease with a large proportion of individuals living with 2 or more chronic conditions ( 117 ) Consequences of living with chronic conditions lead to decreased functionality, threatening the ability to live independently ( 39 ) The National Center for Health Statistics estimated that a large portion of US older adults who reside in the community are depend ent in either activities of daily living (ADLs) or instrumental activities of daily living (IADLs) ( 58 ) Due to the high vulnerability of physical disability in a rapidly growing US older adult population, one of the national public health objectives of the Healthy People 2020 goals is to increase the health, function, and quality of li ving among older US adults. To achieve this public health goal, prevention of physical disability and maintenance of independence is critical to reduce risks of illness, injury, institutionalization, and even death ( 10 61 84 )

PAGE 15

15 Physical Activity Age related physiological changes such as declines in aerobic capacity and reductions in muscle mass (sarcopenia) negatively affect basic daily functioning ( 35 60 ) In regards to aerobic capacity, findings indicate that maxima l oxygen consumption (VO 2 max) decreases approximately 5 15% per decade, while decreases in maximal heart rate, stroke volume, and muscle oxidative capacity affect oxygen delivery mechanisms which reduce oxygen consumption ( 118 ) Sarcopenia coupled with reductions in neural control have been associated with age related reductions in muscle strength ( 21 ) Thoug h these aging effects are inevitable, epidemiological studies provide evidence that a strong relationship exists between moderate intensity exercise and functionality later in life ( 68 ) As age and disease occurrence increase, physi cal function tends to decrease ( 68 ) However, physical activity can attenuate the negative impact of aging and disease onset at different stages of functioning, preserving function and delaying physical disability ( 68 ) Moreover, a large amount of literature supports the beneficial effect of physical activity toward delaying all cause cardiovascular disease and cancer related mortality among older adults ( 36 63 ) In addition, a systematic review of the health benefits of physical activity reported a culmination of evidence supporting exercise to prevent coronary heart disease, type 2 diabetes, stroke, hypertension, and cancer ( 116 ) Furthermore, older adults engaging in purposeful physical activity have improved bone mineral density whi ch reduces the risk of falls and fractures, improved cognition which reduces the risk of depression and dementia, and increased reserve capacity and aerobic fitness ( 46 68 1 16 ) Because of the large benefits gained by structured physical activity, the US Department of Health and Human Services published the first national guidelines in 2008, which state that some physical

PAGE 16

16 activity is better than none and that adults should engag e in 150 minutes of moderate to vigorous intensity physical activity (MVPA) weekly ( 50 83 ) Sedentary Behavior The Federal guidelines of 150 min utes of weekly MVPA among adults translates to approximately 20 minutes of MVPA per day or 2% (16 waking hours/day) of daily activity potentially committed to exercise. However, the scope of daily activity independent of MVPA is constantly changing with ad vancements in technology. Energy expenditure required for daily activity has decreased due to innovations developed to increase task automation, provide instant access to information, and ease effort when engaging in daily routines. These technological adv ances span across communication, transportation, occupation, and entertainment ; which effectively chang es the human environment to wards promot ing sitting like behaviors ( 87 ) These sitting like or sedentary behavi ors (SBs) are distinct from lack of MVPA participation, emerging as an independent predictor of adverse health outcomes such as heart disease, metabolic syndrome, type 2 diabetes, cancer, hospitalization, and mortality ( 7 49 113 ) According to the Sedentary Behaviour Research Network (SBRN), SBs are defined as waking reclining posture ( 103 ) Examples of SBs include television watching, computer use, vehicle driving, and video game playing. Older adults in the United States have been observed to engage in SBs for nearly 60% of waking hours, equating to around 7.7 hours/day ( 74 ) Not surprisingly, older adults in the United States are the most sedentary age group while also measured to have the lowest participation in MVPA ( 62 74 ) Yet, controversy still remains towards understanding whether adverse health consequences are uniquely attributable to high SB engagement or if observed evidence

PAGE 17

17 can be acc ounted for by engaging in too little daily light/moderate/vigorous intensity physical activity. Light Intensity Physical Activity Participation in SBs involves prolonged sitting or reclining activities, which have been linked poor health outcomes ( 87 ) Yet, emerging evidence supporting the relationship between breaks in prolonged daily SB and improved health markers has sparked interest in understanding accumulation of daily activity at different levels ( 51 ) Breaks in SBs incorporate a wide spectrum of activity ranging from light to vigorous intensity physical activity. Examples of activities at light intensities (1.5 2.9 METs) include walking slowly, making a be d, washing dishes, preparing food, and using light hand tools ( 1 ) Moderate intensity physical activity includes activities such as walking at a brisk pace, heavy cleaning, m owing a yard, and bicycling, all of whic h require between 3.0 6.0 METs ( 1 ) Vigorous intensity physical activity demands high energy expenditure (>6.0 METs) such as runnin g, digging, and playing sports ( 1 ) In 2008, researchers published findings that higher breaks in sedentary activity were associated with i mproved metabolic risk profile ( 51 ) This is important because these breaks in SB occur differently between individuals and can impact health markers differently ( 30 ) While this research highlights the importance of accounting for accumulation of total daily SB understand ing the type of activity bouts in daily life independent of MPVA remains unclear. Interestingly, the body of evidence supporting the health benefits of light intensity physical activity (LPA) is growing. One study found that higher time spent in LPA was as sociated with significantly lower 2 hour plasma glucose ( 52 ) The effect was observed among individuals with an average age of 53 years and independent of time spent in sedent ary activity or MVPA. Another study found that high engagement in LPA

PAGE 18

18 in adults aged 60+ years old was negatively associated with arterial stiffening, a predictor of cardiovascular disease ( 40 ) In relation to physical function, researchers detected a positive relationship between higher durations of LPA with improved lower extremity performance among community dwelling Japanese older adults ( 86 ) In a cross sectional analysis among mobility limited older adults a population at high risk for physical disability higher engagement in LPA was found to be associated with lower body mass index ( 4 ) In addition, older men in the same sample who spent more time in higher intensities of l ight activity had higher grip strength. Another study examining mobility limited older adults found a negative association between objectively measured duration in LPA and 10 year risk of hard coronary heart disease ( 34 ) The overall implications of these studies indicate favorable effects of LPA on health among an older adult population. This is particularly important among individuals in this age group who has been observed to engage in higher levels of daily LPA than those in other age groups. Measuring Physical Activity There is an emerging push to incorporate activity back into the daily lifestyle to combat the effects of physical inactivity and sedentary lifestyles. To do so, measuring daily physical activity patterns objectively is essential. Traditionally, questionnaires are used to assess daily activity such as the Community Health Activities Model Program for Seniors physical activity self r eport assessment (CHAMPS) ( 56 ) This questionnaire was developed to capture physical activity by type and intensity among older adult s However, moderate correlations between self report and objective measures were observed with activities such as leisure walking, volunteer work, and light gardening while low correlations were found with more sedentary behaviors such as television

PAGE 19

19 watching, reading, and riding in a car were detected. Since the validity of questionnaires is subject to biases in recall and social desirability ( 100 ) objective measurement tools were developed for implementation in physical activity research. Directl y observing activity or using tools which measure activity from a physiological standpoint remove bias relating to self reported physical activity. However, correlations between direct observation methods versus self reported measures in adults were low to moderate, and that the variability of these correlations was wide ( 102 ) Currently, free living indirect calorimetry or oth erwise known as doubly labelled water is considered the gold standard to measure energy expenditure through stable isotopes of oxygen and hydrogen evaluation from water intake and elimination ( 102 ) However, this method is expensive and imposes a significant time burden on the participant. Other measures which collect physiological responses to movement such as heart rate and p ulmonary gas exchange devices are subject to confounding effects such as temperature, humidity, posture, and stress may confound the results ( 13 ) Sensor mon itors (e.g., pedometers and accelerometers ) are portable and in expensive compared to other objective measures. These devices excel at measuring duration of physical activity but do so in different ways ( 13 ) While pedometers are the least expensive of the sensor devices, they only co llect daily step counts or volume of physical activity. Accelerometers collect movement in terms of acceleration which enables the collection of intensity, volume, and frequency domains of physical activity while also collecting daily step counts Though b oth devices can be used in studies to determine associations and predictive factors, developers of large, longitudinal

PAGE 20

20 epidemiological studies or randomized contro l trials for physical activity are now turn ing to accelerometers to objec tively measure physi cal activity Hospitalizations Extant literature has shown physical inactivity to be a large contributor towards subsequent physical disability among older adults ( 110 ) Though e xercise confers large health benefits, older adults remain the least physically active age group in the US ( 114 ) Physical inactivity can occur due to multiple reasons such as low fitness levels, inconvenience, exacerbations in pain, and comorbidity related impairments ( 36 ) One major risk factor for physical inactivity is hospitalization among older adult s ( 25 ) Despite the intentions of hospitalizations to mend ailments, unintended adverse health consequences tend to occur indepen dent of the initial condition. Hospitalizations tend to impose a degree of immobility through bed rest and have been associated with declines in functionality ( 25 ) The interaction between aging and hospitalization effects have been observed to negatively impact muscle strength, aerobic capacity, respiratory characteristics, bone density, continence, and sensory systems ( 25 ) Among older adults 70 years and older, hospitalization is a strong risk factor for declines in ADL function, independent of the reason for hospitalization ( 24 ) Without a return to pre hospitalization functional status, hospitalized older adults tend to shift into inactive lifestyles which accelerate the likelihood of being physical disability and losing the ability to be independent. However, little is known whether purposeful physical activity can reduce the risk of becoming hospitalized, and whether hospitalizations may affect the benefits of exercise on daily lifestyle activity in older adults.

PAGE 21

21 Rationale Understanding the distributions of daily ph ysical activity and related determinants among the population of US older adults still remains unclear. This is particularly important among mobility limited older adults who are at high risk of physical disability and loss of independence. Since the emerg ence of accelerometers and the ability to objectively capture daily routines of activity, we can now examine the influence of a long term exercise intervention on daily lifestyle activity. In addition, the interceding effect of hospitalizations on the rela tionship of exercise and daily total activity is equally important due to the vulnerability of accelerated health declines in older adults. Since the goal of independence in free living settings is critical to public health the impact of exercise, defined as moderate intensity physical activity for older adults on daily activity must be understood. The interaction between exercise, daily lifestyle activity, and hospitalizations can be conceptualized according to Figure 1 1 While it is obvious moderate in tensity physical activity increases with the effect of exercise, we hypothesize that all other activity levels will improve ( e.g meaningful activity increases while SBs will decrease). This conceptual framework also captures the intervening effect of hos pitalization on the association between exercise and daily activity. Figure 1 1 depicts that when hospitalizations occurs, the benefits of exercise on daily activity are negatively impacted, however the relationship between exercise and improved daily acti vity remain. This study examine d how long term exercise affects daily activity characteristics and functionality among older adults with mobility impairments The central hypothesis is that long term exercise beneficially influences patterns of daily phys ical activity by frequency, intensity, and duration More specifically, time spent in light intensity is

PAGE 22

22 thought to increase while engagement in sedentary activities would decrease. Also, long term exercise is hypothesized to reduce the impact of hospitali zations, events that negatively affect lifestyle and independence among a highly vulnerable population of older adults. With the rapid surge of accelerometer use in many physical activity interventions, daily activity can now be assessed in a variety of wa ys including duration, frequency, intensity, and accumulation. Exploring these characteristics of total daily physical activity will fill a knowledge gap on the longitudinal effects of long term exercise on daily physical activity among older adults, a pop ulation which independent living must be maintained.

PAGE 23

23 Figure 1 1 Conceptual framework of the interaction between moderate intensity exercise, hospitalizations, and daily waking activity. Moderate intensity exercise increases the time spent in other le vels of physical activity intensities such as light intensity physical activity (LPA) and moderate to vigorous physical activity (MVPA). Additionally, moderate intensity exercise reducing time spent in sedentary activity. However, hospitalizations play an intervening role on benefits of chronic, moderate intensity exercise on daily waking activity.

PAGE 24

24 CHAPTER 2 APPROACH Overall Strategy The goal of this proposal is to examine the effects of a long term moderate intensity exercise on objectively measured physical inactivity and activity levels among older adults with mobility impairments Long term exercise has been observed to reduce morbidity and mortality while improve quality of life ( 83 ) This is important in older adults, a population already at risk for a multitud e of adverse health conditions. However, the effects of long term exercise tailored for mobility limited older adults on total daily activity are not well known. This study compare d the effect of a 2 year exercise program to a health education program usin g objective accelerometer metrics of physical inactivity and activity Because older adults are at high risk of adverse health events, the role of hospitalizations was examined between the relationship of exercise and accelerometry measures of daily activi ty. Analytic Sample This analysis utilized the Lifestyle Interventions and Independence for Elders (LIFE; n=1,635) The LIFE study is a multi center, single blinded, parallel rand omized controlled trial (RCT) which evaluate d the efficacy of a structured, l ong term exercise program (PA; average of 2.7 years) in reduc ing the risk of major mobility disability (MMD) as compared to a health education program ( HE ) among older adults ( 89 ) The inability to walk a quarter mile (400 meters) within 15 minutes without sitting, leaning against a wall, or the assistance of another person or walker was utilized as the definition of MMD. In addition, secondary outcomes such as cognition, falls, self reported disability and cost effectiveness, and tertiary outcomes including physical

PAGE 25

25 performance, sleep wake distur bances, dyspnea and ventilatory capacity, and hospitalizations or cardiopulmonary events were examined ( 32 ) The LIFE study examine d whether PA prevent ed MMD in order to address the emerging public health issue of continually increasing rates of physical disability in a majorly sedentary and aging US population. Currently, the LIFE study is the longest and largest physical activity RCT focused on fu nctio nal outcomes ( 72 ) To accomplish the goals of this study, eight participating centers across the US participated in recruitment and data collection. The se centers included the University of Florida at Gainesville and Jacksonville, Florida; Northwestern University at Chicago, Illinois; Pennington Biomedical Research Center at Baton Rouge, Louisiana; University of Pittsburgh at Pittsbur gh, Pennsylvania; Stanford University at Stanford, California; Tufts University at Boston, Massachusetts; Wake Forest School of Medicine at Winston Salem, North Carolina; and Yale University at New Haven, Connecticut. Th is recruitment approach aimed to ens ure a representative sample of US older adults highly susceptible of MMD within rural, suburban, and urban communities. The LIFE study accomplished this by recruiting within a 21 month phase using screening methods to target adults aged 70 89 years old, se dentary, at risk of mobility disability yet able to walk mile, without any adverse health problem which may exacerbate during increased physical activity, and were willing to be randomized to either the intervention or health education. T hese screening criteria was applied in order to recruit older adults who did not engage in structured and planned physical activity, who were at high risk of mobility disability and willing to participate in the LIFE trial. The decision to recruit adults aged

PAGE 26

26 70 89 was b ecause this segment of the US represented one of the fastest growing in the US ( 70 ) Additionally, this subpopulation is at high risk of illness, institutionalization, reduced quality of life, and mortality ( 10 ) Older adults with a sedentary lifestyle were operationally defined as self reporting engaging in < 20 minutes per week of regular exercise and reporting 125 minutes per week of MVPA on the 18 item CHAMPS physical activity questionnaire. Sedentary older adults were important to recruit because high ly activ e older adults who already engage in chronic physical activity may not benefit from a physical activity intervention. In addition, highly active older adults may not be at h igh risk of mobility disability. This is typically defined as scoring less than a 10 (of 12) on the Short Physical Performance Battery (SPPB) ( 48 ) a test of physical performance. Additionally, the LIFE study recruited a subset of participants who scored less than an 8 (45% of the sample) to enhance the total sample of participants at high risk of mobility disability Also, eligi bility required participants to complete a 400 meter walk test within 15 minutes without any assistant devices, or the help of another person. Lastly, participants had to be willing to accept randomization into either the intervention group (PA) or the com parison group ( HE ). Exclusions to the LIFE study largely encompassed criteria which compromised safety of the participant. Specifically, those who had speech problems, had a diagnosis of schizophrenia, other psychotic disorders, or bipolar disorder, repor ted consumption of more than 14 alcoholic drinks per week, were unable to walk across a room, had no major cognitive impairment below education and race specific norms, had severe arthritis, underwent cancer treatments within the past 3 years except for no n melanoma skin cancers and those cancers with excellent prognosis, had lung disease requiring

PAGE 27

27 disease or other progressive neurological disorders, had renal disease requi ring dialysis, had chest pains/shortness of breath/other safety concerns during the 400 meter walk, had medical/psychiatric/behavioral concerns that may interfere with study participation, had illness severity which yields life expectancy to less than 12 months, and had other clinical judgements concerning safety or noncompliance of study procedures. Other exclusion criteria included not giving informed consent, not willing or unable to be randomized, plans to relocate outside of study area within the next 2 years or plans to be out of the study for 6 consecutive weeks during the year, nursing home residence, a participant of the LIFE P study (pilot study), or another member of the household was a participant of the LIFE study. Eligible individuals may expe rience temporary exclusion criteria which must be managed before participation. These included uncontrolled hypertension defined as systolic blood pressure greater than 200 mmHg and/or diastolic blood pressure greater than 110 mmHg ; uncontrolled diabetes w ith recent weight loss, diabetic coma, or frequent insulin reaction; myocardial infarction, major heart surgery, deep vein thrombosis, pulmonary embolus, stroke, hip fractures, hip or knee replacements, or spinal surgery within the past 6 months; ECG deter mined conduction disorder, uncontrolled arrhythmia, new Q waves or ST segment depressions; going through physical therapy or cardiopulmonary rehabilitation; or enrolled in another trial focused on lifestyle or therapeutic intervention. Accelerometry The L IFE study collected measures of daily physical activity and inactivity using cm x 3.7 cm x 1.8 cm), hip worn device which collected movement in terms of acceleration.

PAGE 28

28 As the device measured accelerations, it recorded within the magnitude range of 0.05 to digitized signal passed through a band filter which limited the frequency range to 0.25 to 2.5 Hz to eliminate non human motion. Each sample was summed over an epoch, defined as a user specified interval of time. Preprocessing the accelerometer data outputs summarized activity counts which are comparable between individuals. The GT3X is capable of storing over 3 weeks of data on 16 MB of flash memory. This type of memory retained stored contents even if the device lost power. The device was built to house a rechargeable Lithium Polymer battery. When full charged, the battery was able to was primarily chosen for the LIFE study because it has been widely used in previous research studies ( 23 65 74 114 ) Further, this type of accelerometer has been effective to capture physical activity patterns in older adults ( 23 ) All participants randomize d to the LIFE study were asked to collect accelerometry at baseline and at 6 12 and 24 month follow up visits. The study coordinators explained the accelerometry procedures to each participant, including how to adjust and wear the elastic belt which he ld the accelerometer. Then, the participant was monitored as they self attached the device onto the hip. The participant was instructed to wear the device before and during the 400 meter walk test, and then for the remainder of the day. Then instructions f or the participant were to wear the accelerometer for the next seven days while performing normal daily routines of activity. Participants were given written instructions to not wear the accelerometer during sleep times and water related activities such as bathing, showering, and swimming.

PAGE 29

29 Additionally, participants were given a picture depicting how the device should be placed when it was not worn. On the second or third day of data collection, the participant was called to assess if there were any additio nal questions and how they were doing. The participant was also reminded to either mail the accelerometer back with a pre addressed and stamped envelope given during the clinic visit or return the accelerometer to the clinic after the seven day acceleromet ry collection period. Once the study staff received the accelerometer, the data was downloaded and evaluated for at least 5 days of 10+ hours of time where the monitor was worn and the participant was awake. If there was no data, study staff asked the part icipant to wear the monitor again for another 7 days on a different accelerometer or the same accelerometer if the cause of collecting no data was determined to be an uncharged battery. Valid accelerometry data was transferred to the Data Management and Qu ality Check (DMAQC) committee. Study staff accomplished this by uploading the accelerometry data to the LIFE website. Standardized protocols specific to the clinical sites were developed to track accelerometers, repair records, and how accelerometers were returned to the clinics. Tri axial activity monitors collect data on three planes of space or axes (Figure 2 1 ). While LIFE accelerometers collected on three axes, only data collected on the vertical axis was processed. The vertical axis data have observe d to be closely correlated with locomotion and stepping movements through numerous validity studies ( 37 73 ) The LIFE activity monitors were set to collect data at a rate of 30 Hz per second. Each sample was then expressed as an activity count, a quantitative m easurement of movement over time. The hip worn accelerometers continuously collected data whether it was on the body or not. Participants were given a visualization

PAGE 30

30 which depicted how to orient the device when it was off the hip. Six or more seconds in thi s orientation assisted investigators to estimate whether the device was not worn, where Z since the X axis (vertical) data was used to examine activity, validated non wear verti cal axis algorithms were used to detect and remove non wear times. The LIFE study used the criteria determined by Choi and colleagues ( 18 ) scanning for at least 90 minute intervals of consecutive zero counts. The algorithm also built in a scanning up stream and down stream window of 30 minutes to identify allowances of 2 minute artifactual non zer o counts. Thus, any zero counts not meeting the Choi criteria was considered as valid non movement and non zero counts not considered artifactual movement was valid movement otherwise known as valid wear time The LIFE Accelerometry Working Group and DMAQ C found 1, 411 of 1,635 participants at baseline had satisfactory accelerometry data. This was defined as participant data successfully received, imported into the LIFE database, verified by study staff, and met t he 10 hour daily wear time minimum criteria. This study further refine d the inclusion criteria to only those with 3 + days of 10 + hours of daily wear time. Though there is missing accelerometry data at baseline, we assume d data is missing completely at random. However, we examine d differences between participants with and without valid accelerometry data at baseline across age, sex, race/ethnicity, education, body mass index, cognition, and walk speed (Table 2 1). There was > 95% missing values for self reported race/ethnicity and education for those with missing accelerometry data but across all factors but, in general, there were no differences ( Table 2 1 ) Table 2 2 shows a ccelerometry collected at three additional time points (6

PAGE 31

31 12 24 month follow ups) where missing data at these follow ups was < 10% ( except for the 24 month visit; 10.5% ) Accelerometers of the LIFE study collected 1 sec ond epochs of activity or inactivity and generated vast amounts of data. To thoroughly explore and interpret these large datasets, the study derive d and analyze d various novel variables of accelerometer data of physical inactivity and activity. Table 2 3 describes characteristics of accelerometry that were generated by the investigator. The most traditional variable of accelerometry is volume or the time (e.g. minutes, hours, and days) spent within a specific activity or inactivity intensity range. This was generally reported as a daily average and compared between groups. Accumulation of accelerometry data was calcu lated as bouts of activity or inactivity, defined as events of consecutive minutes registering within a specified intensity range. For example, bouts of 10+ sedentary minutes are events with at least 10 or more sequential minutes within activity intensitie s below 100 counts per minute. Time spent in these bouts reveal a type of activity/inactivity accumulation throughout the day. Intensity of accelerometry gives an approximation of how much energy is being expended by the body. Intensity was examined in two different ways: average activity intensity of total daily accelerometry wear time, and absolute intensity thresholds. Average activity intensity provides an approximate measure of individual daily activity yield, where absolute intensities were pre determ ined using previous literature ( 23 37 73 ) Time spent in these thresholds will be evaluated and combines both volume and intensity. Table 2 4 presents accelerometer cut points used to capture the range of intensities across physical activity Physical inactivity, labeled sedentary activity, was

PAGE 32

32 explored at an < 100 counts/minute Physical activity was considered as 100+ counts/minute broken up into light (low & high) and moderate intensity (low & high) physical activity. To tal physical activity (TPA) was also examined, defined as a combination of all activity levels 100 + counts/minute This study explore d traditional thresholds of intensity levels which have been typically used in national studies such as the National Health and Nutrition Examination Survey (NHANES) study aimed to target all adult age ranges. Also, lowered cutoff levels was explored to bet ter understand the impact of exercise on activities which require lower intensity accelerations generally generated by older adults. This novel approach was possible because of the richness of the LIFE data and the large sample size of participants providi ng valid accelerometry. To our knowledge, only one study using LIFE data has used lower light activity variations (100 counts/minute ) to examine the relationship with cardiovascular disease prediction ( 34 ) but effects of exercise and hospitalization on these accelerometer measures remain unclear. Hospitalizations Accelerometry in relation to the intervening effect of hospitalizations on the benefits of exercise focus ed within time intervals characterized as baseline to 6 month, 6 to 1 2 month, and 12 to 24 month intervals Hospitalizations were collected via an outcome events questionnaire where participants were asked whether a hospitali zation event occur since the last assessment period Previous literature has validated the use of self reported hospitalizations as way to ascertain the occurrence of these events ( 6 55 106 ) The outcome evaluators were blinded to intervention assignment. Medical record technicians and adjudicators could also identify hospitalizations while reviewing medical records. Additionally, participants were asked to notify study staff (unmasked)

PAGE 33

33 about a ny hospitalization or serious illness in between follow up interviews where events were reported via adverse event forms O nly hospitalizations reported to masked staff at the scheduled follow up visits were considered for this dissertation This was to e qualize the group comparisons because the PA intervention had more contact with unmasked interventionists who could report serious adverse events. Characteristics of hospitalizations were assumed to be equal between intervention groups but were examined by participants hospitalized, number of hospitalizations, and days hospitalized across intervention group. Though we assume d the number of hospitalizations >1 day are equal between intervention groups but baseline participant characteristics were examined A time varying hospitalization variable was derived by calculating whether participants had any hospital days during each time interval Because hospital days are positively skewed, we binned hospital days into short (1 3 days) and long (4+ days) and then c reated three categories: no hospitalizations, short cumulative hospital duration, and long cumulative hospital duration. This was used at the primary predictor of changes in accelerometry Hospitalization events were also examined for any potential overlap with accelerometer collection periods. Interventions The main predictor for this dissertation is a long moderate intensity physical activity intervention. This PA intervention was developed from the theory of a social cognitive model of acquisition and maintenance of health behaviors ( 2 ) This model considers health behavior as acquirable and maintainable through behavioral, cogni tive, physiological, and environmental conditions. This concept was coupled with strategies developed from the Transtheoretical Model, a conceptualization of intentional behavioral

PAGE 34

34 change ( 94 ) These strategies involve raising awareness of health behavior and using cognitive processes to first prepare, stimulate, and then reinforce, and manage physical activity. These approaches were tailored as needed and applied at an individual level by stud y staff. The goal of the PA intervention was to administer 150 minutes/week of moderate intensity physical activity consisting of aerobic, strength, flexibility, and fi tness level, the PA intervention was modifiable towards illness, injury, and physical symptoms. Walking was the primary component of the PA intervention, due to its ease of administration and widespread popularity among US adults ( 108 ) Each session involved a brief warm up and cool down period, and flexibility exercises were performed after each bout of walking. Strengthening e xercises targeting lower extremity muscles were performed using variable ankle weights. Balance training was combined with the walking and strengthening exercises. For the first 2 3 weeks of training, participants were introduced to lighter intensity physi cal activity and then progressively trained at moderate intensity physical activity. To achieve moderate intensity physical activity, ( 8 9 ) This scale ranged from 6 to 20, where participants were asked to achieve an intensity of 13 to exercise or below. These levels were maintained during walking exercises. During strengthening exercises, 2 sets of 10 repetitions were performed at intensi ties of 15 to 16 using the

PAGE 35

35 At baseline, participants were randomized to either the PA or HE intervention group. The purpose of the HE intervention was to control for the interaction between the study staff and participant and other secular/s easonal effects which could influence outcomes of interest. Staff contact by intervention group is detailed in Table 2 5 The purpose of the HE intervention arm was to provide a control measurement to the PA intervention without affecting the study outcome s. The HE intervention was an educational program which provided information relevant to older adults. Examples of the topics include discussing the healthcare system, traveling safely, preventive services, and how/where to gather reliable health informati on. Combined with the educational lectures, an instructor led a 5 to 10 minute upper extremity stretching plus relaxation component administered at each session. Covariates There are a multitude of baseline characteristics which may be examined to asses s differences based on ho spitalization. Sociodemographic factors include d sex race/ethnicity, income, education, and marital status. Behavioral factors includ ing smoking, alcohol use, and body mass index were examined Self reported comorbidities and hospitalizations were also assessed. However, the strength of using the LIFE study is the randomization of participants into intervention groups stratified by clinical site and gender. This was assumed to eliminate any baseline differences between interven tion groups. For all aims, accelerometer wear time and baseline accelerometer time were used as a covariate. For aims 1 and 2, sex and clinical site were added (randomization factors) while aim 3 will include sociodemographic, behavioral, medical, and acc elerometer covariates

PAGE 36

36 Table 2 1. Baseline participant characteristics by accelerometry collection status Accelerometry Valid (n = 1,341) Invalid (n = 98) Missing (n = 196) Age in years, mean (SD) 78.7 (5.3) 79.4 (5.2) 79.6 (4.9) Female, n(%) 892 (66.5) 67 (68.4) 139 (70.9) Non Hispanic White, n(%) 1,017 (75.8) 69 (70.4) 6 (3.1) b College education, n(%) 847 (63.2) 62 (63.3) 5 (2.6) b Body mass index (kg/m 2 ), mean( SD ) 30.3 (6.0) 29.5 (6.2) 33.1 (5.4) Modified Mini Mental State Examination score a mean(SD) 91.7 (5.4) 89.3 (5.3) 88.7 (4.7) 400 meter walk speed (m /sec), mean(SD ) 0.8 (0.2) 0.8 (0.2) 0.8 (0.2) a score range: 0 100 where higher scores indicate better performance b factor has > 95% missing values

PAGE 37

37 Table 2 2. Accelerometry data collection by visit 6 month (n=1,556) 12 month (n=1,510) 24 months (n=1,417) Valid, n(%) 1,303 (83.7) 1,310 (86.8) 1,164 (82.0) Invalid, n(%) 118 (7.6) 105 (7.0) 104 (7.3) Missing, n(%) 95 (6.3) 95 (6.3) 149 (10.5)

PAGE 38

38 Table 2 3 Accelerometry characteristics Type Outcomes Description Volume Total time, minutes Time within specific activity/inactivity intensity range Accumulation Time spent in bouts, minutes Time spent within specific activity/inactivity intensity range for at least the amount of consecutive minutes specified Intensity Average activity, counts/min Time spent in average counts/min across total daily accelerometry wear time Absolute threshold, counts/min Time spent in pre determined (sedentary, total activity, light, lifestyle, moderate intensity) count/min thresholds

PAGE 39

39 Table 2 4 Accelerometry cut points denoting intensity ranges Type Intensity Counts/minute Physical inactivity Sedentary <100 Physical activity L ow light 100 759 High light 760 1,040 Low moderate 1,041 2,019 High moderate 2,020+

PAGE 40

40 Table 2 5 Staff contact by intervention group Physical activity group Week Center based Home based Adoption: w ee k 1 to 52 2 times a w ee k 1 time/w ee k (w ee k 1 4) 2 times/w ee k (w ee k 4 8) Up to 3 4 times/w ee k (w ee k 8 52) Maintenance: w ee k 53 to end 2 times a w ee k Up to 3 4 times/w ee k Successful aging group Week Center based Telephone contact Adoption: w ee k 1 to 26 1 time a w ee k After one missed visit, contact was made to problem solve barriers to attendance Maintenance: w ee k 27 to end Once a month, while offered twice a month After first session missed, contact was made

PAGE 41

41 Figure 2 1 Tri axial accelerometer. The vertical plane is the x axis; medial lateral is the y axis; and anterior posterior is the z axis.

PAGE 42

42 CHAPTER 3 EFFECTS OF A LONG TERM PHYSICAL ACTIVITY PROGRAM ON SEDENTARY TIME IN OLDER ADULTS Background Excessive daily sedentary behaviors (SB) among older years of age has recently emerged as a national public health issue, averaging approximately 9 hours/day or 66% of daily waking time in SB ( 74 ) SB, characterized as ( 103 ) include lying awake quietly, watching television, using a computer, and driving a vehicle. Epidemiological evidence suggests these behaviors are associated with adverse hea lth outcomes such as cardiovascular disease, diabetes, and the metabolic syndrome, regardless of participation in physical activity (PA) ( 3 53 54 ) Coupled with aging, the increased risk of these negative health conditions can adversely affect independence later in life ( 39 110 ) Beyond total sedentary time, the manner in which sedentary time accumulates is also important to health. Daily accrual of long, uninterrupted sedentary bouts is linked to poorer cardiometabolic health and death when compared to patterns displaying frequent breaks from SB ( 5 14 30 ) Recently, patterns of sedentary time were examined in a large sample of US middle aged and older adults (45 years and older); the results suggested that almost half of total daily sedentary time was accumulated in continuous bouts of 30 minu tes or greater ( 27 ) This growing body of literature supports further study of ways to link PA guidelines with strategies to reduce sedentary time, particularly in regard to its accumulation patterns. A large amount of evidence suggests that daily participation in moderate to vigorous intensity physical activity (MVPA) delays risks of m ortality, morbidity, and

PAGE 43

43 ( 116 ) Additionally, this age segment ha s the lowest participation of MVPA among the entire US population, which exacerbates the public health impact ( 114 ) To address this issue, the Lifestyle Interventions and Independence for Elders (LIFE) randomized trial was conducted to compare structured, moderate intensity PA and health education (HE) interventions among older adults (70 89 years old) with functional limitations to reduce the risk of major mobility disability ( 32 ) The main result demonstrated a s ignificant increase in moderate intensity PA and mobility benefit ( 89 ) but it was not clear whether patterns of daily sedentary engagement were also changed in the process. The LIFE study provided a unique opportunity to address this knowledge gap with the collection of objective daily activity and sedentary time using accelerometers over 24 months. The primary aim of this paper was to compare d ifferences in sedentary time between PA and HE. It was hypothesized that a 24 month PA intervention would lead to a decrease in sedentary time compared to HE. The second aim of this paper was to determine whether changes in sedentary time were associated w ith changes in physical function. It was hypothesized that increased sedentary time would be associated with declines in physical function, independent of intervention assignment. Methods Trial Design and Participants Details of the design a nd methods of the LIFE study ( 32 ) and our primary results ( 89 ) have been published elsewhere. In summary, this was a Phase 3, multicenter, investigator blinded, randomized controlled trial to test whether a long term moderate intensity PA intervention would reduce the risk of major mobility disa bility as compared to HE. The trial included 1,635 mobility limited men and women, aged 70 89 years old.

PAGE 44

44 Mobility limitation was defined as scoring <10 (out of 12) on the Short Phys ical Performance Battery (SPPB) ( 48 ) but able to walk 400 meters within 15 minutes without sitting, leaning against a wall, o r getting any assistance. Participants were eligible for the trial if they were sedentary (self reported spending <20 min/w ee k in the past month getting regular PA and <125 min/w ee k of moderate intensity PA) and had no major cognitive impairment (score >1. 5 standard deviations below education and race specific norms on the Modified Mini Mental State Examination [3MSE] ( 111 ) ). Participants were recruited from eight field centers across the nation within a 21 month period, where the primary recruitment strategy was targeted community mass mailings. The LIFE study was registered with www.clinicaltrials.gov before trial enrollme nt (NCT01072500). The institutional review boards approved the study protocol at all participating sites and is available on request at https://www.thelifestudy.org/public/index.cfm Written informed consents were obtained from all participants. Intervent ions At baseline, participants were randomized to either the PA intervention or HE program. Both arms promoted behavioral change based on principles of social cognitive theory and strategies formulated around the Transtheoretical Model (e.g., raising healt h behavior consciousness, preparing for action, and providing support for maintenance through cognitive approaches) ( 93 ) The PA intervention consisted of an individually tailored plan to initiate and mainta in 150 min/week of moderate intensity PA. The intervention included aerobic exercise consisting of a walking regimen in parallel with lower extremity strength, flexibility, and balance training. The PA program format consisted of 2 center based sessions/w e e k and 3 4 home based sessions/w ee k. To

PAGE 45

45 was used ( 8 ) This scale ranges from 6 to 20, where participants were asked to achieve 2 sets of 10 repetitions scale. The HE group participated in workshops specific to health and well being for older adults, with intentional avoidance of topics related to PA. The program includes a 5 10 minute session of u pper extremity stretching or flexibility exercises. The HE arm was expected to meet weekly during the first 26 weeks of the intervention and at least once monthly thereafter. Accelerometer M easurements Objective daily PA and sedentary time were collected v ia accelerometers at the accelerations within a magnitude range of 0.05 to 2.5 units of gravity. The analog signal was digitized at a rate of 30 times per second (30 Hz) and passed through a band filter that eliminated non human motion. Each sample was summed over 1 second epochs and converted to an activity count. An activity count is a unit less digitized quantity of movement expressed over time (i.e., activity counts/minut e). The GT3X is capable of collecting and storing data up to 20 days on a single charge of battery. Participants were asked to wear the accelerometers on their right hip at all times for at least 7 consecutive days except during sleep and water related act ivities (e.g., swimming or showering). Though the accelerometer collected tri axial data, only data collected from the vertical axis were processed. Vertical axis accelerometry collected from the hip has been validated to assess locomotion type movements i n a number of studies ( 37 73 )

PAGE 46

46 Sedentary V ariables The accelerometer continuously logs movement and non movement so there is no distinction between wear and non wear time. To distinguish between the two, PA data were first processed for valid wear time periods using a non wear algorithm developed by Choi and colleagues ( 18 ) This algorithm was chosen because it improves on the NHANES algorithm to detect low activity levels and high SB, particularly in those who are expected to have prolonged bouts of sedentary time ( 114 ) Briefly, data were binned into 1 minute epochs and scanned for consecutive zero counts lasting at least 90 minutes. Two minute intervals for non zero counts within 30 minute upstream and downstream zero count windows were allowed as an exception; any oth er non zero count registered by the accelerometer was considered wear time. Additionally, outliers were identified (minutes where activity counts exceeding 10,000, 3,500 counts over median of the 2nd highest activity counts/min across all days, or 1,000 co unts over the 2nd highest activity count/min of the same day) and were classified as outliers and re labeled as non wear time. Additionally, the data were visually inspected for isolated patterns or unexplainable phenomenon (e.g., maintaining the same acti vity level for extended periods of time at extremely high levels of activity), which were removed ad hoc. In total, 34 days from 18 participants were removed: 1 day from baseline, 7 at 6 months, 5 at 12 months, and 21 at 24 months. Each valid wear minute was labeled as sedentary if counts summed to <100 over 1 minute ( 114 ) Accumulated sedentary time was characterized in three ways. First, bouts of sedentary time were calc ulated as consecutive sedentary minutes of at least 1+, 10+, 30+, or 60+ minutes. Second, an individualized measure of sedentariness was derived as a bout length at which 50% of total daily sedentary time was accumulated (50% bout

PAGE 47

47 length). This was done by creating a regression model for each participant as follows: % sedentary time = bout length + (bout length) 2 The equation was solved for each person to predict bout length at 50% sedentary time. Third, sedentary breaks were calculated as an interruption of consecutive sedentary minutes where one sedentary minute transitioned to a non sedentary minute (>99 counts/min). Physical F unction M easures Measures of physical function considered in this analysis were walking speed (m eter /s econd ) and lower extremity physical performance measured at baseline and 6, 12, and 24 months. Walking speed was assessed during a 400 meter walk where participants were asked to complete 10 laps on 20 m eter course at their usual pace. Participants were allowed to stop and take a br eak for up to 1 minute without sitting during the assessment. If a participant stopped before the 10 laps were completed, the distance traveled was measured. Walking speed was calculated as distance/time in meters/second. The SPPB was used to assess physic al performance using balance, usual gait speed, and chair stand tests. Each component was scored from 0 to 4, with higher scores indicating better performance. The component scores were summed to yield a total score from 0 to 12, where 12 was the highest a ttainable score of physical performance. Statistical Analysis Only participants with valid accelerometry (10+ hours/day of accelerometry for at least 3 days at all time points) were included in this analysis. The distribution of the sedentary bout data were evaluated for normality prior to conducting analyses. The ti me spent in 1+ minute bouts had a large range and was normally distributed. Baseline characteristics and total daily average wear time metrics of participants with valid data

PAGE 48

48 were examined by intervention. Mean differences in sedentary outcomes between int ervention arms were estimated using linear mixed effects modeling. Visit (time) was treated as a repeated factor, and the baseline sedentary specific metric, age, visit, and wear time were included as fixed effects. Factors used to stratify randomization ( sex and clinical site) were also included in the model as fixed effects. The association between changes in sedentary time and changes in physical function were evaluated after pooling the effects across intervention group s. Individual slopes for sedentary measures were derived from mixed effects linear regression over four time points while adjusting for wear time at each measurement. Individual slopes for physical function (400 m eter walk speed and SPPB score) were calculated similarly. Because there are multiple determinants of changes in physical function, we included in exploratory analyses the following covariates to understand the independent effect of changes in each sedentary metric: age, sex, intervention assignment, clinical site, race, comorbidit y index >2, education level, 3MSE score, self reported depression, body mass index (kg/m 2 ), reported having smoked 100+ cigarettes during lifetime, and daily activity (accelerometer minutes registering 760+ counts/min ute ). Additionally, we included an int eraction term of baseline SPPB category (<8 vs. 8+ score as the reference) by sedentary change to examine effect modification by baseline functional status. Accelerometer data were processed using R ( www.r project.org ) ( 18 115 ) and statistical analyses were performed in STATA v13 (STATA Corp.). Results Population Sample Of the 1,341 participants from whom we obtained valid accelerometry data at baseline, 669 participants were randomized to the PA group and 672 randomized to the

PAGE 49

49 HE group (Figure 3 1). Baseline participant characteristics described in Table 3 1 were similar between group s. Average age was 78.75.3 (SD) years old, 66.5% were women, 75.8% were Non Hispanic white, mean body mass index (kg/m2) was 30.36.0, 44.4% had an SPPB score < 8, and 42.3% had a 400 m walking speed < 0.8 m eter /sec ond Also, average wear time (837.0111.0 minutes/day ) and total daily sedentary time (1+ min bouts; 646.8116.4 minutes/day ) were similar between group s. On average, 488.3131.8, 296.0132.7, and 145.4102.9 min/day were spent in 10+ min ute 30+ min ute and 60+ min ute sede ntary bouts, respectfully. Those without valid accelerometry data (n = 294) were similar in regard to age (79.65.0 years old) and sex (70.1% women) as compared to the sample who had valid accelerometry data. Sedentary V ariables by I ntervention Changes in daily sedentary variables over 2 years by intervention group are illustrated in Figure 3 2. While there was an overall 0.80.1 minutes/day increase per month in total daily sedentary time within the sample (p<0.001), the PA group experienced an overall red uction of 8.82.9 total sedentary minutes/day compared to the HE group over 24 months (p=0.002). These group differences appeared to occur within the first 6 months and were maintained throughout the study (p=1.00 for intervention group difference at 6 mo nths vs. 24 months). Daily sedentary time in 10+, 30+, and 60+ minute bouts increased over time (1.40.1, 1.60.2, 1.30.1 minutes/day respectively; p<0.001 for all) where significant intervention differences were detected for 10+ and 30+ minute bouts (P A vs HE: 11.53.9 minutes/day p=0.003 & 9.34.1 minutes/day p=0.02, respectively). Average 50% bout length at baseline was 28.113.2 minutes and increased 0.20.03 (p<0.001) minutes monthly, with no significant differences between intervention group s ( p=0.25). Daily average sedentary breaks

PAGE 50

50 taken during baseline was 70.817.9 and decreased by 0.20.02 times per month, with no significant intervention differences detected (p=0.90). Associations with C hange in P hysical F unction Analyses were conducted to examine individual changes in physical function and sedentary variables by baseline functional status category (i.e., SPPB <8 vs. 8 9). Results in Table 3 2 demonstrate that sedentary time patterns increased whereas sedentary bre aks declined over time. These changes were greater among those with low baseline function, except for 1+ min ute bouts (p=0.08). With regards to physical function, walking speed decreased over time ( 0.0020.003 m eter /s econd per month) and this decrease was greater in those with lower compared to higher function (p<0.001). SPPB scores increased in the higher function group (0.040.03 SPPB units per month), but declined in the lower function group ( 0.010.04 SPPB units per month). Figure 3 3 illustrates cor relations between standardized changes in sedentary variables with predicted standardized changes in physical function measures after full covariate adjustment. Changes across sedentary variables explained 3 7% (p<0.001) of changes in walking speed and 2 4 % (p<0.001) of changes in physical performance. Those who accumulated more sedentary time (1+, 10+, 30+, 60+ minute and 50% bout length) or decreased sedentary breaks were more likely to experience declines in walking speed (p<0.05 for all). Furthermore, those with lower function at baseline appear to be more susceptible to declines in walking speed as sedentary time was increased (p interaction <0.05). Similar results were observed with changes in SPPB (Figure 3 3 ). Discussion Mobility limited older adults randomized to a two year moderate intensity PA intervention not only increased their PA levels ( 89 ) but also reduced their time spent in

PAGE 51

51 acceler ometry derived sedentary time compared to those in a HE program. The PA intervention attenuated increases in time spent in short bouts (1+, 10+, & 30+ minutes); i.e., the largest reduction was observed in 10+ minute sedentary bouts, resulting in 1.4% less daily sedentary time. While these results support our original hypothesis that a PA program decreases sedentary time compared to the HE program, we found no intervention differences in prolonged sedentary metrics (60 + min ute bouts, bout length representin g 50% of daily sedentary accumulation, daily sedentary breaks). An additional finding was that changes in all sedentary bouts and bout lengths representing 50% of total daily sedentary time were negatively correlated with changes in physical function measu res. This effect was stronger in participants with lower physical function at baseline. In support of these findings, decreases in sedentary breaks were associated with d ecreases in physical function. These results are unique to the sedentary behavior lite rature as they suggest that a traditional activity program can impact daily sedentary behavior metrics that appear to align with indicators of improved physical function. At baseline, vulnerable older adults who were screened for low activity engaged in hi gh rates of sedentary time across all bout lengths. In 2008, Matthews and colleagues used accelerometers to show that US adults 70 85 years of age spend about 9 hours, or 66% of daily waking time, in SB ( 74 ) Another study demonstrated that older women spent 65% of total sedentary time in 10+ minute bouts, 31% in 30+ minute bouts and 12% in 60+ minute bouts ( 105 ) These women also averaged 9 sedentary breaks/hour. In comparison, the current study sho wed that mobility limited older women and men spent 11 waking hours/day, or 77% of daily waking time, being sedentary at

PAGE 52

52 baseline. Almost 75% of the total daily sedentary time was accumulated through 10+ minute bouts, 46% through 30+ minute bouts, 22% thro ugh 60+ minute bouts and averaged 4 breaks/hour daily. Furthermore, half of their sedentary time was accumulated with 28+ min bout lengths. These rates were similar to a recent cross sectional study examining accelerometry determined sedentary time among U S adult aged 60 years and older with mobility disability (defined as any difficulty walking mile or up 10 steps) ( 69 ) A large population based study of black and white adults aged 45 + years spent 77%, 48%, and 26% in 10+, 30+, and 60+ minute sedentary bouts respectively ( 27 ) While these findings are similar to ours, caution must be taken when comparing results between these two studies, due to differences in accelerometer devices, activity count threshold criteria, screening criteria, and age of participants ( 90 ) Overall, it appears older adults at high risk for mobility disability are more sedentary than what others have reported in the literature. Our results suggest long term moderate intensity PA was most effective at decre asing short bouts of daily sedentary time, though the effects were relatively modest. The effects occurred within the first 6 months, and thereafter sedentary bout time steadily increased for both intervention group s. There were fewer intervention differen ces in long bouts of sedentary time. Longer bouts may be more resistant to change because they represent a relatively small amount of the total accumulated sedentary time. Further, results suggest that increases in prolonged sedentary time among mobility l imited older adults remained unchanged despite reporting high levels of moderate intensity PA (218 min utes /w ee k) ( 89 ) Thus, older adults with mo bility impairments who participate in a structured moderate intensity PA program appear to

PAGE 53

53 reallocate short, but not long, bouts of sedentary time (about 8 10 minutes / day) into active time. We found a significant although relatively weak association between longitudinal changes in sedentary time and changes in physical function that was independent of intervention assignment. Our results suggest that increases in time spent across all sedentary patterns were correlated with decreases in lower extremit y physical performance and usual pace walking speed. Similar longitudinal analyses reported by Seguin and colleagues showed that greater amounts of self reported sedentary time were linked with greater declines in self reported physical function among post menopausal women aged 50 79 years old ( 104 ) They showed that women reporting greater amounts of sedentary/sitting time had lowe r physical function at baseline and eventually declined more rapidly than those accumulating less sedentary/sitting time. These observations also support our findings that lower functioning participants at baseline were more susceptible to declines in phys ical function resulting from increases in sedentary time. Therefore, changes in sedentary time appear to play a role in changes in physical function among older adults with mobility limitations. The strengths of this study include objective collection o f daily PA via accelerometers in a randomized controlled trial design multiple follow up assessments over 24 months a large sample of mobility limited older adults, and high adherence to both interventions (median 24 month on site attendance: PA[71%]; HE [82%]). Limitations include the inability to detect posture, sitting versus idle standing, sleep time, type of sedentary behaviors (e.g., educational reading versus television for

PAGE 54

54 entertainment), and possible wear/non wear time misclassification. However, these effects are likely to be negligible because this population is less likely to stand for among older adults ( 19 ) In conclusion, these results suggest that mobility limited older adults randomized to a 24 month moderate intensity PA program have a small but potentially beneficial reduction in short bouts of sedentary time. The effect did not transfer to prolonged sedentariness (60+ min ute bout time, 50% bout length, sedentary breaks), although we found that 60+ min ute bout time was relatively low in the sample to begin with. Further work is needed to explore whether moderate intensity PA replaces sedentary time and how the type of PA influences sedentary time. Developing ways to increase PA among older adults remains a key public health goal and future interventions should consider integrating a sedentary behavior reduction component to maximize health benefits in this vulnerable population.

PAGE 55

55 Table 3 1. Baseline participant characteristics by intervention assignment Physical activity (n = 669) Health education (n = 672) Age in years, mean( SD ) 78.5 (5.3) 79.0 (5.3) Female, n(%) 431 (64.4) 461 (68.6) Non Hispanic White, n(%) 497 (74.3) 520 (77.4) College or higher education, n(%) 417 (62.3) 430 (64.0) Body mass index (kg/m 2 ), mean( SD ) 30.2 (5.8) 30.4 (6.3) Smoked 100+ cigarettes ever, n(%) 339 (50.7) 297 (44.2) Modified Mini Mental State Examination score a mean( SD ) 91.7 (5.4) 91.7 (5.3) Sought medical advice for depression in past 5 years, n(%) 97 (14.5) 85 (12.7) Comorbidities > 2, n(%) 172 (25.7) 177 (26.3) Short Physical Performance Battery score b mean( SD ) Score < 8, n(%) 7.5 (1.6) 286 (42.8) 7.3 (1.6) 309 (46.0) 400 meter walk speed (m/sec), mean( SD ) Speed < 0.8 m/sec, n(%) 0.8 (0.2) 271 (40.5) 0.8 (0.2) 296 (44.1) Wear days, mean( SD ) 8.0 (3.3) 7.9 (3.1) Wear minute/day, mean(SD) 839.1 (112.7) 835.0 (109.4) 27.6 (22.8) 27.9 ( 25.9) Daily sedentary time spent in 1+ minute bouts, mean(SD) 648.9 (120.3) 644.7 (112.4) Daily sedentary time spent in 10+ minute bouts, mean(SD) 490.6 (136.4) 486.1 (127.0) Daily sedentary time spent in 30+ minute bouts, mean(SD) 298.3 (136.5) 293.8 (128.8) Daily sedentary time spent in 60+ minute bouts, mean(SD) 146.4 (105.8) 144.3 (100.1) Average bout length spent in 50% of daily sedentary minutes (min ute ), mean(SD) 28.2 (13.4) 28.0 (13.2) Daily number of sedentary breaks, mean(SD) 70.7 (17.6) 70.8 (18.1) a score range: 0 100 where higher scores indicate better performance b score range: 0 12 where higher scores indicate better performance; scoring < 8 indicates poor functioning

PAGE 56

56 Table 3 2 Changes in sedentary behavior patterns and physical functio n by baseline SPPB a functional status Total (n=1,341) Score <8 (n=595) Score 8 or 9 (n=746) p value b Sedentary variables c,d : Mean (SD) 1+ min sedentary bouts (min ute /day) 0.76 (0.25) 0.77 (0.25) 0.74 (0.25) 0.08 10+ min sedentary bouts (min ute /day) 1.28 (0.41) 1.31 (0.43) 1.26 (0.40) 0.03 30+ min sedentary bouts(min ute /day) 1.60 (0.78) 1.66 (0.83) 1.55 (0.73) 0.007 60+ min sedentary bouts (min ute /day) 1.39 (0.80) 1.51 (0.89) 1.30 (0.71) <0.001 50% bout length (min ute ) 0.20 (0.01) 0.20 (0.02) 0.20 (0.01) <0.001 Sedentary breaks ( n ) 0.21 (0.10) 0.22 (0.11) 0.20 (0.10) 0.003 Physical function measures c : Mean (SD) 400 meter walk speed (m/s) 0.002 (0.003) 0.003 (0.003) 0.002 (0.002) <0.001 SPPB score 0.016 (0.041) 0.011 (0.038) 0.038 (0.027) <0.001 a Short Physical Performance Battery b Significance at p<0.05 c Values represent beta coefficients (changes per month) derived from mixed effects linear regression over four time points d Wear time was considered as a time varying covariate

PAGE 57

57 Figure 3 1. Flow of participants through the study a Data pre processed for at least 10 hours of accelerometer data which were screened for data key in errors, accelerometer malfunctions, participant operation errors, visual identification of unex plained patterns or phenomenon, and duplicate files. b Data were screened for outliers and processed for valid wear/non wear times based on a accelerometer wear time classification algorithm ( 18 )

PAGE 58

58 Figure 3 2. Intervention effect across sedentary measures All models adjusted for baseline sedentary measure, age, gender, clinical site, and wear time *p < 0.05; **p<0.01; ***p <0.001

PAGE 59

59 Figure 3 3. Association between standardized changes in sedentary measures and physical function a Short Physical Performance Battery All models adjusted for baseline sedentary measure, baseline SPPB category ( <8 vs. 8+ score as the reference), sedentary slope x SPPB category interaction term (included if significant at p<0.05 ), age, sex, intervention, clinical site, race/ethnicity, comorbidities > 2, education, modified mini mental state examination score, self reported depression, body mass index, smoked 100+ cigare ttes during lifetime, daily activity (accelerometer minutes registering 760+ counts/min ute ).

PAGE 60

60 CHAPTER 4 EFFECTS OF A LONG TERM MODERATE INTENSITY EXERCISE PROGRAM ON PHYSICAL ACTIVITY COMPOSITION IN MOBILITY IMPAIRED OLDER ADULTS Background Adults aged 70+ years of age represent the most physically inactive age segment of the US population ( 114 ) and they are linked to a high risk for physical disability and lo ss of independence ( 59 ) Large scale physical activity interventions, such as the Lifestyle Interventions and Independence for Elders (LIFE) study ( 32 ) were developed to evaluate combating the negative effects of physical inactivity on mobility and Federally recomme nded amount of 150 minutes per week of structured moderate to vigorous intensity physical activity (MVPA; otherwise considered as exercise) ( 15 ) Specifically, the Federal guidelines recommend engaging in at least 30 minutes of continuous, structured physical activity for most, if not all, days. For those wh o have difficulty reaching 30 minute physical activity bouts, the guidelines recommend performing multiple 10+ minute bouts of activity. Despite these recommendations, there is little work to verify that physical activity interventions influence engagement in these types of activity bouts particularly in vulnerable older adults who may have impediments to performing longer duration exercise. Previous research has identified that exercise interventions impact not only MVPA, but also lighter intensity activit ies in older adults ( 44 ) Several studies in younger populations demonstrated that long term exercise increased total daily physical activity suggesting that lighter inte nsity, non exercise activities are preserved ( 20 79 95 99 ) Meijer and colleagues demonstrated that adults 28 41 years of age who trained 5 months for a 1/2 marathon competition increased total daily physical activity ( 79 )

PAGE 61

61 Therefore, the exercise training program did not affect non exercise activity, resulting in the observed increase of total amount of physical activity accumulated throughout the day. However, evidence in older adults suggests that exercise training has no impact on total daily physical activity, which may indicate that lighter intensity activities reduces with long term exercise ( 44 77 78 82 ) Older adults who undertook vigorous endurance training demon strated no changes in total daily energy expenditure ( 44 ) This suggests a compensation effect whereby lighter intensity activities reduced to accommodate increased engag ement in higher intensity activities. However, the compensation effect remains largely unconfirmed in long term, large scale randomized trials and particularly among vulnerable older adults with mobility limitations. Furthermore, there has been little work to examine this compensation effect of physical activity according to frequency, duration, and intensity ( 120 ) Measuring physical activity is challenging because it is difficult to precisely describe intensity patterns in free living conditions. There are a multitude of measurement tools to assess physical activity including ques tionnaires, doubly labeled water, and electronic sensors, all of which capture physical activity in different ways. However, some of these methods are limited in their ability to simultaneously capture all the complex components of daily physical activity such as frequency, duration, and intensity. Questionnaires are cost effective measurement tools that excel at qualitatively capturing types and contexts of activities and are easily deployed in large scale research. However, validity (when compared to more objective methods) of these instruments is considered poor ( 57 ) Doubly labeled water is considered the gold standard method to indirectly measure free living total caloric energy expenditure ( 101 )

PAGE 62

62 but this method is costly for large scale research studies and unable to capture the type or intensity of specific activities performed during the monitored tim e period ( 121 ) Fortunately, innovations in engineering in the past couple decades have provided an objective method to assess movements continuously in free living environments ( 38 ) Body worn accelerometers offer a non invasive, reasonably cost effective, and continuous measure of movement patterns throughout a waking day while also being well suited for evaluating behavioral studies such as physical activity inter ventions. Accelerometers have greatly expanded the measurement of physical activity by objectively capturing the composition of activity by intensity, duration, and frequency patterns ( 91 ) The primary aim of this analysis was to examine how a long term, moderate intensity physical activity intervention modified time spent at different intensities of physical activity as measured using an accelerometer. We hypothesized that the physical activity intervention would increase total daily time spent in levels within moderate intensity physical activity (low and high+) by replacing time spent in light intensity physical activity (low and high) when compared to a general health education program. Second, we aimed to deter mine whether the physical activity intervention modified patterns of engagement within different intensities. We hypothesized that the physical activity intervention would increase total daily time spent in 10+ minute bouts at all physical activity intensi ties when compared to a health education group. These hypotheses were tested in the LIFE study, which provides a unique opportunity to analyze accelerometer data collected at multiple time points over 2 years. The LIFE study enrolled a large sample of mob ility limited older adult population who were

PAGE 63

63 randomized into either a moderate intensity physical activity (PA) or a health education (HE) intervention ( 32 ) Methods Trial Design and Participant Population Details outlining the study design and methods ( 32 ) and primary results ( 89 ) can be found elsewhere. In short, the LIFE stud y was a Phase 3, multicenter, investigator blinded, randomized clinical trial testing the capability of a physical activity intervention to reduce the risk of major mobility disability among mobility limited older adults. A total of 1,635 men and women wer e recruited Eligible participants were 70 89 years of age and sedentary ( defined as self reporting < 20 minute/week getting regular physical activity within the last month ) and were tested for functional limitations operationally defined as scoring < 10 o n the Short Physical Performance Battery (SPPB), where 12 was the highest physical performance achievable. Moreover, eligibility required the ability to walk 400 meters within 15 minutes without sitting, leaning, or receiving any assistance. Participants w ere excluded for the following reasons: not willing to be randomized, self reported inability to walk across a room, diagnosis of psychotic related conditions, low mental health, progressive neurological conditions, alcohol abuse, speech or hearing problem s, severe arthritis, plans to relocate in the near future, living in a nursing home, LIFE pilot intervention participation, active treatment for cancer, regular use of corticosteroids or supplemental oxygen needed for lung disease, cardiovascular disease, dialysis required for renal disease, safety concerns such as chest pains or shortness of breath during the 400 meter walk, life expectancy was less than 12 months due to severe illness, and any clinical judgements concerning safety or study non compliance. The LIFE study was registered with www.clinicaltrials.gov before

PAGE 64

64 trial enrollment (NCT01072500). The study protocol was approved by the institutional review boards at all participating sites and is available on request at https://www.thelifestudy.org/publ ic/index.cfm All participants provided written informed consents. Interventions Participants were randomized either to the PA intervention or the HE program at baseline using a block algorithm (random block lengths) stratified by field center (8 centers t hroughout the United States) and sex. After randomization, participants received an individual face to face introductory session with a health educator who described the intervention, expectations, and fielded any questions. Both programs promoted behavior al change based on social cognitive theory principles and strategies were developed around the Transtheoretical Model ( 94 ) The PA intervention comprised of an individually tailored plan to increase physical activity levels with a goal of achieving 150 minutes/week of moderate intensity physical activity through th e promotion of multiple 10+ minute bouts. This involved aerobic exercise through a walking regimen that also included lower extremity strength exercises, flexibility, and balance training. The PA intervention design consisted of 2 center based sessions and 3 4 home exercise sessions that accompanied the in for rating perceived exertion was utilized to determine whether participants attained moderate intensity physical activity ( 8 ) Participants were asked to reach an intensity of 20 range. Strength exercises consisted of 2 sets of 10 repetitions, where participants were asked to reach a 15 to 16 conversational workshops focused on older adult health and well being, while

PAGE 65

65 intentionally avoiding topics related to physical activity. A few examples of the topics were the US healthcare system, how to tr avel safely, access to preventive services, and reliable health information sources. Additionally, participants in the HE group were led through a 5 to 10 minute light intensity upper extremity stretch and relaxation component during each session. Partici pants in the HE group were expected to meet weekly during the first 26 weeks of the intervention and at least once monthly thereafter. Accelerometer M easurements Tri at baseline and then at 6, 12, and 24 month follow up visits. These devices detect accelerations within a magnitude range of 0.05 to 2.5 units of gravity, digitize the analog signals at a rate of 30 Hertz (Hz), and pre process data through a band filter to eliminate n on human motion. Samples were summed over a 1 second interval, otherwise termed as epoch, and converted to activity counts. An activity count is a unit less quantity of overall movement expressed as a rate (e.g., counts/minute). Participants were instructe d to wear the accelerometers on their right hip at all times for 7 consecutive days, except during sleep and water related activities (e.g., swimming or showering). While tri axial accelerometer data were collected on the hip, only vertical axis data (most sensitive to ambulatory movements) were used in the current analysis ( 37 73 ) Physical A ctivity Characterization The accelerometer provides no information about whether a participant was wearing the device. Therefore, accelerometer data were first processed to classify valid wear algorithm ( 18 ) This algorithm improves on previous methods at detecting low intensity activity, particularly in those who are expected to accumulate prolonged bouts of inactivity ( 114 ) In short, data were binned into 1 minute

PAGE 66

66 epochs and scanned for at least 90 minutes of consecutive zero counts. Any other non zero count registered by the accelerometer was considered wear time. Non zero counts were only allowed for up to two minute intervals but within at least 30 minute upstream and downstream zero count windows. Outliers were identified as minutes where activity counts exceeded 10,000 counts/min, 3,500 counts over median of the 2nd highest ac tivity counts/minute across all days, or 1,000 counts over the 2nd highest activity counts/minute of the same day. Minutes classified as outliers were re labeled as non wear time and not included in the analysis. Further, the data were visually inspected f or unexplainable phenomenon (e.g., maintaining the same activity level for extended periods of time at extremely high levels of activity), which were removed ad hoc. In total, 34 of the 3,102 days (0.01%) from 18 participants were removed: 1 day from basel ine, 7 at 6 months, 5 at 12 months, and 21 from 24 months. Only participants who provided valid accelerometry (10+ hours/day of accelerometer data for at least 3 days at all time points ; n=1,341 ) were included in the analytic sample. Physical activity patt erns were characterized by intensity and bout lengths. Each valid wear minute was labeled as activity (of any intensity) if counts summed to 100+ activity counts over 1 minute. Accelerometer variables and accelerometer cut points were calculated according to Table 4 1. The sum of all activity labeled minutes was calculated as total activity (TPA) and categorized into low light physical activity (LLPA), high light physical activity (HLPA), low moderate physical activity (LMPA) and high moderate and greater p hysical activity (HMPA). These cut points were chosen because they have been specifically evaluated among older adults ( 23 37 114 ) The percent contribution of intensity specific activity to TPA was derived as follows: (m inutes in

PAGE 67

67 intensity specific activity) / (TPA) 100. For example, the percent contribution of HLPA to TPA was calculated as: total HLPA / (TPA) 100. Bouts of activity time were calculated as consecutive minutes of at least 1+, 2+, 5+, and 10+ minutes where the accelerometer registered within cut points specified in Table 4 1. Time spent in 1+ minutes of activity bouts represent the total volume of inte nsity specific activity, where 2+, 5+, and 10+ minute bout lengths represent consecutively smaller segments of the total activity volume. We chose to examine overlapping categories of activity (e.g., 2+ minute bouts contain 5+ and 10+ minute bouts) because physical activity recommendations are typically disseminated with no upper limit. However, mutually exclusive categories can be calculated; for example, if 5+ minute bouts were subtracted from 2+ minute bouts, the subsequent segment would represent 2.0 4. 9 minute bouts. Other Measurements Participants were assessed at the clinical site at baseline and every 6 months thereafter by study staff masked to intervention group assignment. Information on age, sex, along with other sociodemographic factors, medica l history, hospitalizations, medications, quality of well being, and functional limitation were collected via self report at baseline Statistical A nalysis Activity bout time distributions were assessed for normality within each intensity range prior to p erforming analyses. Mixed effects (random and fixed) linear regression models were constructed for each bout length and intensity level. Visit (time) was treated as a repeated factor, and the baseline activity specific metric, age, visit, and wear time wer e included as fixed effects. Factors used to stratify randomization (sex

PAGE 68

68 and clinical site) were also included in both models as fixed effects. Also, the interaction between intervention and visit was tested for each model and included if found to be signi ficant. Alpha level for all analyses was set to 0.05. Accelerometer data were processed using R ( www.r project.org ) ( 18 115 ) and statistical analyses were performed in STATA v13 (STATA Corp.). Results Sample Characteristics Baseline participant characteristics according to intervention group are described in Tabl e 4 2. Randomized groups were similar across demographics, behavioral factors, and medical history. Participants were 78.75.3 (SD) years and predominately females (66.5%) and non Hispanic white (75.8%). Mean wear time was 13.91.8 wear hours/day for 7.93.2 days. Those with invalid accelerometer data (n=294) were similar in age (79.65.0 years old) and sex (70.1% women) compared to the sample with valid accelerometer data. Activity Intensity and Bout Lengths at Baseline Baseline daily activity patter ns expressed in minutes per day ( minutes/day ) by specific bout length are presented in Table 4 3. On average, participants spent 189.769.9 minutes/day in TPA, which made up 22.88.2% of total daily wear time. 86.78.0% of daily TPA was spent in LLPA, 6.1 2.9% in HLPA, 5.94.3% in LMPA, and 1.32.8% in HMPA. Within LLPA, 72.67.2%, 23.78.6%, and 4.74.0% was spent in 2+, 5+, and 10+ minute bouts, respectively. 21.611.0%, 1.24.1%, and 1.01.2% HPLA time was spent in 2+, 5+, and 10+ minute bouts, respectiv ely. Of total LMPA time, 35.919.0%, 8.113.9%, and 2.59.0% was spent in 2+, 5+, and 10+ minute bouts, respectively. For HMPA, 29.732.0%, 8.820.3%, and 3.312.8% was spent in 2+, 5+,

PAGE 69

69 and 10+ minute bouts, respectively. No differences were noted between randomized groups. Most participants engaged in at least 1 minute of 1+, 2+, 5+ and 10+ minute bouts in LLPA (>85%) during baseline accelerometry collection which was also true for 1+ and 2+ minute bouts at HLPA and LMPA. However, in general, participan ts engaged in less activity with longer bout lengths and relatively higher intensity. Less than 15% of participants engaged in 10+ minute bouts at HLPA, LMPA and HMPA. Intervention E ffects on Intensities in Total Activity Time Figure 4 1 illustrates chang es in intensity patterns within TPA per month. The PA group decreased daily time spent in LLPA when compared to the HE group, with the highest impact observed at 6 months ( 3.30.4 %) but reduced by 24 months ( 1.80.4 %; p interaction < 0.001). In comparis on, the HE group increased time spent in LLPA by 0.80.3% (p=0.008) over 24 months. In contrast to LLPA, the PA group increased time spent in HLPA, LMPA, and HMPA when compared to the HE group. Intervention differences for HLPA and LMPA w ere 0.30.1% (p<0. 001) and 1.30.2 % (p<0.001), respectively. For HMPA, intervention differences we re highest at 6 months (1.40.2 %) but re duced by 24 months (0.40.2 %; p interaction < 0.001). Intervention E ffects on Time Spent at Intensities and Bout Lengths Figure 4 2 de picts changes in time spent at intensity by bout lengths per month. Both groups decreased TPA time in 1+ bouts ( 0.90.1 minutes/day ), 2+ bouts ( 0.60.1 minutes/day ), 5+ bouts ( 0.20.0 minutes/day ), and 10+ minute bouts ( 0.10.0 minutes/day ). However, the PA intervention attenuated TPA decreases across 1+, 2+, and 5+ minute bouts while increasing time in 10+ minute bouts when compared to the HE program (1+ bouts: 6.42.1 minutes/day ; 2+ bouts: 5.41.6 minutes/day ; 5+ bouts:

PAGE 70

70 3.60.9 minutes/d ay ; 10+ bouts: 3.30.4 minutes/day ). After separating TPA by intensity, no intervention differences were detected for all bout lengths in LLPA. For HLPA, the PA intervention attenuated decreases in 1+ bouts (0.80.3 minutes/day ) but not 2+ minute bouts (p= 0.46) while increased time in longer bouts (5+ minute bouts: 0.20.04 minutes/day ; 10+ minute bouts: 0.10.02 minutes/day ). Intervention differences found in 1+, 5+, and 10+ minute HLPA bouts were maintained for 24 months. The PA intervention increased tim e in all LMPA bout lengths. Increases in 1+ bouts (2.70.4 minutes/day ), 2+ bouts (2.40.3 minutes/day ), and 10+ minute bouts (1.10.1 minutes/day ) were maintained for 24 months whereas 5+ minute bout increases were highest at 6 months (2.00.2 minutes/day ), maintained at 12 months (1.70.3 minutes/day p 6vs12 diff =0.12), and reduced by 24 months (12 vs 24 months: 0.360.3 minutes/day p<0.001). For HMPA, the impact of the intervention was highest at 6 months where the PA intervention increased time in all bouts (1+ bouts: 2.80.4 minutes/day ; 2+ bouts: 2.50.3 minutes/day ; 5+ bouts: 2.10.3 minutes/day ; 10+ minute bouts: 1.70.2 minutes/day ). All increases in HMPA bouts reduced by 24 months (p interaction <0.001 for all). Discussion The results provide an i mportant understanding of the activity pattern changes that occur in response to a long duration physical activity program in vulnerable older adults. First, the PA intervention changed total daily activity by in creasing HLPA, LMPA, and HMPA. These increas es appeared to be at the expense of a decrease in LLPA. Despite increased HLPA, LMPA, and HMPA, the PA group continued to experience an overall decline in total daily activity, albeit less than the HE group. Additionally, the PA intervention increased enga gement in long duration bouts of 10+ minutes of activity

PAGE 71

71 within HLPA, LMPA, and HMPA (but not LLPA) when compared to the HE program. This finding was also supported by an increased time in short bouts of HLPA (1+ and 5+), LMPA (1+, 2+, and 5+), and HMPA (1 +, 2+, and 5+). It is important to note that group differences at lighter intensity bouts were not a sole result of increased physical activity in the PA group, but an attenuated decline when compared to the HE group. To our knowledge, this is the first study to characterize activity composition among older adults aged 70 89 years old who were considered to be inactive (<20 min ute /week of structured physical activity) and had low physical function according to a standardized assessment battery ( 47 ) At baseline, LIFE participants accrued lower a mounts of physical activity and performed less activity in longer bouts when compared to other studies on older adults. In 2008, Troiano and colleges reported from NHANES (National Health and Nutrition Examination Survey) data that older adults (70+ years) spent ~7 minutes/day of HMPA whereas our sample of mobility limited older adults spent 3 minutes/day ( 114 ) In a subsequent study by Evenson et al. using updated data, older adults accumulated ~52 minutes/day in HLPA, which is considerably higher than the 13 min/day in LIFE study participants ( 31 ) Additionally, in these same reports, older adults spent 3 7 minutes/day in 10+ min ute bouts of HMPA, whereas the LIFE sample spent 0.5 min utes /day ( 31 114 ) More recently, Buman and colleges reported average time spent in LLPA + HLPA (257 minutes/day ), LMPA (19 minutes/day ) and HMPA (10 min utes /day) in 70+ year olds were substantially higher than participants in the LIFE study ( 12 ) These compariso ns suggest that LIFE study participants were indeed less active across activity intensities and accumulated their activity in shorter bouts compared to the general population and previous reports

PAGE 72

72 The PA intervention was most effective at increasing long term, moderate intensity physical activity but at the cost of decreasing time spent at light intensities, confirming our first hypothesis. These results support previous findings that older populations do not simply add exercise into their daily lifestyle but compensate by reducing low intensity activities to accommodate higher intensity activity. This finding of a compensation effect is supported by previous studies implementing exercise training programs in older adults ( 44 77 78 ) In 1992, investigators observed no change in total physica l activity, despite increases in vigorous intensity exercise measured through energy expenditure ( 44 ) Authors concluded that though older adults did not change total ene rgy expenditure through a possible compensatory reduction in non exercise activity, the high intensity training may have fatigued the participants afterwards. In 1999, a study using accelerometers assessed how a 3 month moderate intensity training program impacted daily physical activity in older adults aged 55 68 years old ( 78 ) Participant s trained twice a week and accelerometry data were collected on training and non training days. Results showed that training days showed a significant decrease of non training physical activity when compared to physical activity accumulated on non training days even with a lower intensity exercise program. Our results add to these findings by confirming the replacement effect occurring among older adults at high risk of mobility disability. U nique to the literature is that we determined compositional change s in intensity levels of daily physical activity using accelerometers. At baseline, LIFE participants spent approximately 20% of total waking time in some form of physical activity. Of this 20%, LLPA was most predominant, reaching >80% of TPA. Over 24 mont hs, the PA group experienced decreases in LLPA

PAGE 73

73 when compared to HE; the most observed at 6 months ( 3%). At 6 months, the PA group concurrently increased time in HLPA, LMPA, and HMPA by 3% percent. Though intervention differences significantly reduced over time, the observed replacement effect was preserved over 24 months. It appears that modifying the intensity composition of total daily activity can impact the maintenance of mobility and independence among vulnerable, older adults. In 2015, the LIFE stud y demonstrated that the PA intervention conferred mobility and health benefits to mobility limited older adults ( 89 ) This appears to occur by replacing LLPA with higher activity intensities. Other studies have shown improvements in p hysical function and health markers, but not increases in total physical activity, associated with exercise training ( 44 77 78 89 ) In contrast, the HE group increased time in LLPA by replacing time in HLPA, LMPA, and HMPA and supports previous evidence of declines in physical activity after reaching the age of 60 years old ( 119 ) Daily intensity shifts within TPA were also experienced by the HE group but with highe r activity intensities being replaced by LLPA. These results indicate that changes in the composition of physical activity intensities may be important for understanding transitions in disability and/or disease progression states ( 28 ) particularly among older adults with mobility impairments. Participants in the PA intervention averaged 32 44 more minutes registering at or above 760 counts/min ute per week than the HE grou p, which was similar to accelerometer data reported by Pahor and colleagues ( 89 ) points ( 23 ) we found these int ervention differences consisted of 6 min utes /week spent in HLPA, 19 min utes /week in LMPA and 20 min utes /week in HMPA at 6 months, 14

PAGE 74

74 min utes /week at 12 months and 7 min utes /week at 24 months. For LLPA, no group differences were observed. When combining all intensity levels, PA intervention accrued an average of 45 more minutes/week of TPA than the HE program. However, both groups experienced overall decreases in time spent in TPA over 24 months ( 6 min utes /week of TPA per month), which was primarily driven by decreases in LLPA ( 4 min utes /week of LLPA per month). Additionally, we were able to examine intervention effects on volumes and bout patterns according to intensity level. In general, the PA intervention influenced bout patterns similar to their respec tive volumetric times. Intervention effects observed within HLPA and LMPA occurred similarly across longer bout times. However, intervention effects reduced over time for LMPA 5+ minute bouts and all HMPA bouts. The results support our second hypothesis th at the PA intervention increased time spent in 10+ minute bouts at physical activity at all intensities when compared to the HE group. However, bouts of LLPA were unchanged. Additionally, the PA intervention attenuated but did not eliminate total daily phy sical activity declines despite increasing 10+ minute bouts of physical activity at higher intensity levels. We were able to objectively evaluate a long term physical activity intervention designed to increase amount of time and Federally recommended phys ical activity guidelines. This capability was not restricted to the laboratory but expanded into free living settings, revealing a novel and potentially useful way to assess unsupervised adherence. This is particularly important among older adults who have low adherence to unsupervised exercise programs while at home ( 16 97 ) To do this, we combined commonly used accelerometer cut points calibrated for adult populations (20+ years of age) to examine how the PA intervention influenced time spent at different activity

PAGE 75

75 in tensities and bout lengths ( 23 37 114 ) This approach to examine intervention effects on daily activity accumulation patt erns is unique to the accelerometer literature, particularly in vulnerable older adults at high risk of mobility disability. Strengths of this study include objective collection of daily PA via accelerometers, repeated measures of accelerometry over 24 mo nths in a randomized controlled trial design among a large sample of older adults at h igh risk of mobility disability The study had a few limitations. Specifically, lack of knowledge regarding posture ability to detect standing versus sitting, sleep, and type of activity (e.g. structured exercise versus gardening) may misclassified activity as inactivity, potentially underestimating intervention effects However, this population is less likely to stand for extended periods of time and the wear time classi fication algorithms reduces wear time misclassification among older adults ( 18 ) In conclusion, a structured, moderate intensity physical activity increased time in physical activity bouts in higher intensities at the expense of very light intensity physical activity. These results highlight that the PA intervention increased physical activity by shifting the composition of activity toward higher intensity activity in longer duration bouts. However, the PA intervention did not eliminate overall declines in total daily activity patterns experienced by mobility impaired older adults. Furt her work is needed to understand the declines in total activity because it appears that increasing longer bouts of high intensity physical activity does not completely eliminate declines in total activity experienced by older adults highly vu lnerable to mo bility disability.

PAGE 76

76 Table 4 1. Descriptions of accelerometer derived metrics of physical activity composition Intensity Bouts Light physical activity (LLPA): 100 759 counts/min ute High light physical activity (HLPA): 760 1,040 counts/min ute Low moderate physical activity (LMPA): 1,041 2019 counts/min ute High moderate physical activity (HMPA): 2,020+ counts/min ute Total physical activity (TPA): 100+ counts/min ute 1. Time spent in 1+ min bouts (min utes /day; total activity time) 2. Time spent in 2+ m in bouts (min utes /day) 3. Time spent in 5+ min bouts (min utes /day) 4. Time spent in 10+ min bouts (min utes /day)

PAGE 77

77 Table 4 2. Baseline participant characteristics by intervention group Physical activity (n = 669) Health education (n = 672) Age in years, mean(SD) 78.5 (5.3) 78.9 (5.2) Female, n(%) 431 (64.4) 461 (68.6) Non Hispanic White, n(%) 497 (74.3) 520 (77.4) College or higher education, n(%) 417 (62.3) 430 (64.0) Body mass index (kg/m 2 ), mean(SD) 30.2 (5.8) 30.5 (6.3) Smoked 100+ cigarettes ever, n(%) 339 (50.7) 297 (44.2) Modified Mini Mental State Examination score a mean(SD) 91.7 (5.4) 91.7 (5.3) Sought medical advice for depression in past 5 years, n(%) 97 (14.5) 85 (12.7) Comorbidities > 2, n(%) 172 (25.7) 177 (26.3) Short Physical Performance Battery score < 8 b n(%) 286 (42.8) 309 (46.0) 400 meter walk < 0.8 m/sec, n(%) 271 (40.5) 296 (44.1) Wear days, mean(SD) 8.0 (3.3) 7.9 (3.1) Wear minute/day, mean(SD) 839.1 (112.7) 835.0 (109.4) 27.6 (22.8) 27.1 (25.9) a score range: 0 100 where higher scores indicate better performance b score range: 0 12 where higher scores indicate better performance; scoring < 8 indicates poor functioning

PAGE 78

78 Table 4 3. Baseline daily activity level and bouts by intervention group P hysical activity (n=669) H ealth education (n=672) mean (SD), min utes /day n (%) a mean (SD), min utes /day n (%) a Low light minutes (100 759 counts/min ute ) 1+ min ute bouts 161.9 (54.4) 669 (100) 162.1 (54.5) 672 (100) 2+ min ute bouts 120.1 (47.5) 669 (100) 120.3 (47.2) 672 (100) 5+ min ute bouts 41.7 (25.8) 668 (99.9) 41.3 (24.7) 672 (100) 10+ min ute bouts 8.5 (8.8) 597 (89.2) 8.5 (8.6) 595 (88.5) High light minutes (760 1,040 counts/min ute ) 1+ min ute bouts 12.6 (9.3) 669 (100) 12.6 (9.3) 672 (100) 2+ min ute bouts 3.1 (3.1) 634 (94.8) 3.1 (3.1) 625 (93.0) 5+ min ute bouts 0.1 (0.5) 94 (14.1) 0.2 (0.6) 83 (12.4) 10+ min ute bouts 0.01 (0.2) 3 (0.5) 0.02 (0.2) 7 (1.0) Low moderate minutes (1,041 2,019 counts/min ute ) 1+ min ute bouts 12.3 (11.5) 668 (99.9) 12.6 (13.5) 671 (99.9) 2+ min ute bouts 5.4 (6.8) 622 (93.0) 5.9 (8.9) 621 (92.4) 5+ min ute bouts 1.2 (2.9) 306 (45.7) 1.7 (4.6) 302 (44.9) 10+ min ute bouts 0.4 (1.6) 72 (10.8) 0.7 (2.9) 82 (12.2) High moderate minutes (2,020+ counts/min ute ) 1+ min ute bouts 2.7 (5.7) 577 (86.3) 2.7 (6.4) 577 (85.9) 2+ min ute bouts 1.6 (4.6) 331 (49.5) 1.7 (5.0) 320 (47.6) 5+ min ute bouts 0.8 (3.6) 127 (19.0) 0.9 (3.7) 112 (16.7) 10+ min ute bouts 0.4 (2.6) 45 (6.7) 0.5 (2.6) 54 (8.0) a Frequency of those who spent any time in that activity level (min/day > 0)

PAGE 79

79 Figure 4 1. Compositional changes of total daily physical activity by intensity and intervention status. All models adjusted for baseline activity variable age, sex clinical site, and wear time Note: LLPA low light physical activity; HPLA high light physical activity; LMPA low moderate physical activity; HMPA high moderate and greater physical activity. *p < 0.05; **p<0.01; ***p <0.001

PAGE 80

80 Figure 4 2. Intervention effects on activity intensity levels by bout pattern. All models adjusted for baseline activity variable age, sex clinical site, and wear time Note: TPA total physical activity; LLPA low light physical activity; HPLA high light physical activity; LMPA low moderate physical activity; HMPA high moderate and greater physical activity. Physical activity vs health education. p < 0.05; **p<0 .01; ***p <0.001

PAGE 81

81 CHAPTER 5 INVESTIGATING THE INTERVENING EFFECT OF HOSPITALIZATIONS ON PHYSICAL ACTIVITY PATTERNS MEASURED BY ACCELEROMETRY Background A hospitalization event is an independent predictor of poorer musculoskeletal, autonomic, circulatory, ur inary, sensory, and metabolic health systems, all of which lead to subsequent functional decline and changes in daily physical activity patterns ( 25 45 ) Despite the fact that hospitalizations are intended for restoration from adverse health conditions, the degree of imposed immobility and dispa rate complications to the cause of hospitalization only p artially explain accelerated declines in physical function following a hospitalization in older adults ( 25 43 ) This has become a critical public of age ), a rapidly growing segment of the US population at high risk for physical disability ( 17 70 ) Further more older adults at risk for disability who experience a h ospitalization from intervening illness or injury are more likely to develop outright disability when compared to their less vulnerable counterparts ( 41 ) Induced physical inactivity, which accompan ies and follows a hospitalization, is associated with mortality, morbidity, and disability ( 81 88 ) However, the relationship between hospitalizations and subseque nt changes in both physical inactivity and activity is not well known particularly in free living settings This is important because older adults represent the most sedentary segment in the US ( 74 ) and excess sedentary volumes have been linked to adverse health outcomes ( 53 62 75 ) independent of moderate to vigorous physical activity. Regular p hysical activity is known to delay mortality, morbidity, and disability ( 83 ) but is also known to decline immediately after hospitalization ( 96 ) Hospitalizations could partially describe the large activity declines

PAGE 82

82 that occur during the aging process, placing those hospitalized on a new trajectory of increased physical inactivity and declining physical activity. Additionally, the manner in which inactivity and activity are accrued after hospitalizations remains uncl ear but has drawn focus For example, prolonged sedentary bouts is linked to poorer cardiometabolic profiles when compared to taking frequent breaks in sedentary behavior ( 51 52 ) Federal guideline s of regular p hysical activity promote physical activity participation in continuous bouts of 10 minutes or more for older adults but the temporal effect of hospitalizations on daily activity patterns remains unclear. There remains a public health need to examine the role that hospitalizations have in daily activity changes characterized across volume, duration and intensity, particularly among older adults vulnerable to disability and hospitalizations In the Lifestyle Interventions and Independence for E lders (LIFE) study, a long term structured moderate intensity physical activity (PA) program reduced the risk of major mobility disability in 1,635 sedentary men and women aged 70 + years of age with mobility limitations, when compared to a health education (HE) program ( 89 ) Though h ospitalizations within the PA group were s lightly higher compared to HE, the PA intervention did not statist ically increase the risk of hospitalizations by comparison ( 71 ) Despite having slightly higher rates of hospitalizations, the PA group accumulated 32 44 more minutes per week of accelerometry derived physical activity (minutes registering at 760+ counts) ( 89 ) M ore interventions are being developed to promote recovery of hospitalizations through exercise training but they are limited to short study periods ( 26 ) T here remains a knowledge gap of the degree to which a long term structured physical activity intervention improves regular physical activity levels and decreases

PAGE 83

83 sedentary times, particularly among vulnerable older adults who experience at least one hospitalization The LIFE study provides a unique opportunity to examine the com plex relationship between intervening hospitalizations, structured physical activity intervention, and daily accrual of sedentary and activity times through objective measures using a hip worn accelerometer collected periodically over 2 years. As such we are able to characterize changes in patterns of inactivity and activity by volume, duration, and intensity. Additionally, we can determine hospitalizations not only by occurrence but also by cumulative length of stay in a hospital setting ( hospital duratio n ) Therefore, t he primary aim was to examine the impact of an intervening hospitalization on daily accelerometry derived levels of physical inactivity and activity. We hypothesize d that an intervening hospitalization is associated with increases in sedent ary time and decreases in physical activity when compared to those without a hospitalization Additionally, we hypothesized that long hospital durations had greater associations than short durations. The secondary aim was to assess whether the is different between intervention groups. W e hypothesize d that those randomized to PA have attenuated increases in sedentary time while experiencing attenuated decreases in physical activity patterns when compared to HE A mong those with long hospital durations, the beneficial PA effects are greater than those with short hospital durations. Methods Trial Design and Participant Population St udy design and methods details ( 32 ) and primary results ( 89 ) are outlined elsewhere. Briefly, the LIFE study was a Phase 3, multicenter, investigator blinded,

PAGE 84

84 randomized clinical trial examining the ability of a structured, moderate intensity physical activity intervention to reduce the risk of major mobility disab ility among mobility limited older adults. A total of 1,635 men and women were enrolled Those eligible were 70 89 years of age and sedentary ( defined as self reporting < 20 minutes/week getting regular physical activity within the last month and < 125 min utes/week of moderate intensity physical activity ) Eligible participants were screened for functional limitation, defined as scoring < 10 (of 12) on the Short Physical Performance Battery (SPPB), where 12 was the highest physical performance achievable. F urther, eligibility required participants to possess the ability to walk 400 meters within 15 minutes without sitting, leaning, or receiving any assistance. Exclusion criteria were as follows: unwilling to be randomized, self reported not able to walk acro ss a room, diagnosis of psychotic related conditions, low mental health, progressive neurological conductions, alcohol abuse, hearing or speed problems, severe arthritis, plans to relocate in the near future, living in a nursing home, LIFE pilot interventi on participation, active treatment for cancer, regular use of corticosteroids or supplemental oxygen needed for lung disease, cardiovascular disease, dialysis required for renal disease, safety concerns such as chest pains or shortness of breath during the 400 meter walk, life expectancy < 12 months due to severe illness, and any clinical judgments concerning safety or study non compliance. Study protocols were approved by the institutional review boards at all participating sites and is available on reques t at www.thelifestudy.org/public/index.cfm. All participants gave written informed consents. Intervention s Baseline randomization of participants was performed using a block algorithm (random block lengths) stratified by field center (8 centers throughout the United States)

PAGE 85

85 and sex to either the PA intervention or the HE program. An individual face to face introductory session with a health educator was held for each participant where the intervention and expectations were described and questions were answ ered. Behavioral change was promoted through both programs using social cognitive theory principles and strategies developed around the Transtheoretical Model ( 94 ) The PA intervention consisted of an individually tailored plan to increase physical activity with a goal of achieving 150 minutes/week of moderate inte nsity physical activity through the promotion of multiple 10+ minute bouts. This plan was comprised primarily of aerobic exercise through a walking regimen but also included lower extremity strength exercises, flexibility, and balance training. The PA inte rvention design comprised of 2 center based sessions accompanied by 3 4 home exercise sessions. Rate of perceived ensure compliance to a program of moderate intensity physical activity ( 8 ) Participants were asked to reach an intensity of 20 range. Strength exercises comprised of 2 sets of 10 repetitions, where participants were asked to reach a 15 Participants were placed on medical leave if the participant missed or voluntarily or temporarily withdr ew from physical activity treatment for 4+ consecutive sessions due to self reported hospitalization, injury, or other health reasons. Additionally, a participant may be classified as on medical leave if their primary care physician orders withdrawal from physical activity. Participants on medical leave were periodically contacted for a status update while also provided with support and develop ment of plans to restart the intervention when appropriate. Suspended physica l activity was restarted when the

PAGE 86

86 participant was able to leave home and independently walk a minimum of 4 meters T he participant had to have any prescribed therapy completed prior to restarting the PA program. All participant s underwent individualized re evaluation s to modify the physical activity prescription to cope with any conditions caused or attained during the hospitalization The HE intervention consisted of conversational workshops focused on older adult health and well begin (e.g., US healthca re system, how to travel safely, reliable health information sources, etc.), while intentionally avoiding topics related to physical activity. Additionally, participants in the HE group were led through a 5 10 minute light intensity upper extremity strengt h and relaxation component during each session Participants in the HE intervention were expected to meet weekly during the first 26 weeks of the intervention and at least once monthly thereafter. Accelerometer M easurements A h ip worn accelerometer (Actig was used to collect daily activity patterns at baseline and then at 6 12 and 24 month follow up visits. In the LIFE study, accelerometers were utilized detect accelerations within a magnitude range of 0.05 2.5 units of gravity These signal s were digitized from analog signals at a rate of 30 Hertz (Hz) and pre process ed through a band pass filter to eliminate non human motion noise Samples were summed into 1 second epochs (time intervals) and converted to activity counts. An activity count is defined as a unit less quantity of overall movement expressed as a rate (e.g., counts/minute) that is correlated with the intensity (i.e. speed) and number of movements over a given time period Participants were instructed to wear the accelerometer on their right hip at all waking times for 7 consecutive days, except during sleep and water related activities (e.g., swimming or

PAGE 87

87 showering). While accelerometer data were collected on three axes, only vertical axis data (most sensitive to ambulatory moveme nt s) were used in this paper ( 37 ) Accelerometer V ariables The monitor continuously logs movement and non movement, but does not provide a specific distinction between wear and non wear times. Therefore, accelerometer data was first processed for valid wear time periods using a non wear algorithm developed by Choi and colleagues ( 18 ) This algorithm was chosen because it improves on the National Health and Nutrition Examination Survey (NHANES) algorithm to determine low activity levels fr om sedentary behavior, particularly in those who are expected to have prolonge d bouts of sedentary time ( 19 ) Briefly, 1 second epoch data were binned into 1 minute epochs and scanned for consecutive zero counts lasting at least 90 minutes. Exceptions f or non zero counts within non wear periods were only allowed for up to 2 minute intervals but within at least a 30 minute upstream or downstream zero count window. Any other non zero count registered by the accelerometer were considered wear time. Outliers were identified as minutes where ac tivity counts exceeded 10,000 counts/min, 3,500 counts over median of the 2 nd highest activity counts/min ute across all days, or 1,000 counts over the 2 nd highest activity count s /min ute of the s ame day. Minutes classifie d as outliers were re labeled as non wear time and not included in the analysis Further the data w ere visually inspected for unexplainable phenomenon (e.g. maintaining the same activity level for extended periods of time at extremely high levels of activity), which were removed ad hoc. In total, 34 of the 3,102 days (0.01%) from 18 participants were removed: 1 day from baseline, 7 at 6 months, 5 at 12 months and 21 from 24 months. Only participants

PAGE 88

88 who collected valid accelerometry defined as 10+ hours/day of accelerometer data for at least 3 days at all time points ( n=1,341 ) were included in the analytic sample. Accelerometer variables and accelerometer c ut points were calculated according to Table 5 1. Each valid wear minute was labeled as sedentary if counts summed to <100 over 1 minute ( 74 75 114 ) Each valid wear minute was labeled as activity (of any intensity) if counts summed to 100+ acti vity counts over 1 minute. Total physical activity (TPA) was defined as the sum of all activity labeled minutes and categorized into low light physical activity (LLPA), high light physical activity (HLPA), low moderate physical activity (LMPA) and high mod erate and greater physical activity (HMPA). These cut points were chosen because they have traditionally been used in the literature or that they have been specifically determined for o lder adults Bouts of activity were studied to understand the how sedentary and activity times were patterned throughout the day. Sedentary bouts were calculated as consecutive sedentary minutes of at least 1+, 10+, 30+, or 60+ minutes described in Table 5 1 Activity bouts were calculated as consecutive minutes of at least 1+, 2+, 5+, and 10+ minutes where the accelerometer registered within cut points specified in Table 5 1. Time spent in 1+ minutes of activity bouts represent the total volume of intensity specific activity, where longer bout lengths represent cons ecutively smaller segments of the total activity volume. We chose to examine overlapping categories of activity (e.g., 2+ minute bouts contain 5+ and 10+ minute bouts) because physical activity recommendations typically have no upper limit. However, mutual ly exclusive categories can be derived; for example, if 10+ minute bouts were subtracted from 5+ minute bouts, the subsequent segment would represent a 5.0 9.9 minute bout length.

PAGE 89

89 Hospitalization V ariable The h ospitalization event, date and reason were collected continuously throughout the study. M asked assessors collected this information during scheduled clinic visits using an outcome events questionnaire. Medical recor d technicians and adjudicators verified the reason and date of hospitalization by reviewing medical records. For this analysis, only hospitalizations reported to masked staff at the scheduled follow up visits were considered as an event Unmasked hospitalizations were not considered to reduce ascertainment bias of the PA participant s who were permitted to freely report events to unmasked interventionists during intervention sessions (<10% out of total hospitalization reported were unmasked ). The variable for hospitalization combined the event and the length of stay. To do this, hosp ital length of stay (L O S ; in days ) was summed between each time interval of baseline to 6 months, 6 to 12 months, and 12 to 24 months. Those without a hospital stay received a zero. T he LOS of variable was categorized into short and long durations by split ting the variables at the median of 4 days spent in a hospital setting among those hospitalized The variable was expressed as no hospitalizations (0 hospital days), short cumulative hospital duration (1 3 hospital days), and long cumulative hospital duration (4+ hospital days) within each time interval per participant. To verify hospitalizations that occurred before accelerometry collection o verlap between hospitalizations and accelerometer collection periods were examined. O f 3,176 follow up observ ations 10 (0.003%) were found to overlap Observations with overlap were not considered in this analysis

PAGE 90

90 Covariates At baseline, information on age, sex, race/ethnicity, education, marital status, income, and smoking status were collected via self repor ted along with medical/hospitalization history, physical examinations, cognitive testing (Modified Mini Mental State Examination; 3MSE), body weight, and height. Participants function was assessed at the clinical site at baseline and every 6 mont hs thereafter by study staff masked to intervention group assignment. Walk ing speed was assessed during a 400 meter walk where participants were asked to complete 10 laps on 20 meter course at their usual pace. Participants were allowed to stop and take a break for up to 1 minute without sitting during the assessment. If a participant stopped before the 10 laps were completed, the distance traveled was measured. Walk ing speed was calculated as distance/time (meters/second) The SPPB was used to assess physi cal performance via balance, usual gait speed, and chair stand tests. Each component was scored from 0 to 4, with higher scores indicating better performance. Summing the components yield ed a total score from 0 to 12, where 12 was the highest score of physical performance. Statistical A nalysis Baseline differences between hospitalization duration categories were tested by either t tests (means) or chi square tests (frequency) within each intervention group T o assess changes in accelerometer outcomes, each metric was subtracted by baseline values for each participant. Distributions of accelerometer change metrics were assessed for normality within each accelerometer cut point range and time interval prior to c onducting analyses. Mixed effects (random and fixed) linear regression models were constructed for each accelerometer metric ( cut point and bout length specific) where

PAGE 91

91 visit (time interval ) within a participant was treated as a repeated factor. The unadju sted regression model for each outcome was as follows: ( 5 1 ) w here is the outcome (accelerometer metric) for the i th participant at the j th visit, and are dummy variables for the categories of the hospitalization variable (e.g., = 1 if the participants spends 1 3 days in a hospital setting, = 0 otherwise), and is the time interval (in y) from baseline for the i th subject at the j th visit. The random intercept is (i.e., the deviation of the i ) and the random slope is population slope ). For this analysis the parameters of interest were and whic h tested whether the change in accelerometer metric outcomes were different for participants who spent 1 3 or 4+ days in a hospital setting, respectively, when compared to those who spent no time in a hospital setting ( reference). Also and te sted whether post hospital changes (either 1 3 days or 4+ days) in accelerometer metric outcomes were different by intervention group when compared to the reference of participants who did not spent time in the hospital in each intervention group. For both aims, the interaction of time with other parameters of interest was tested but found not to be significant for any of the models (p<0.05 for all). After running unadjusted models, final models excluded non significant interaction terms and were fully adju sted by demographic, behavioral, and medical history covariates. A priori contrast statements were used to estimate the difference between short and long hospital durations on each

PAGE 92

92 accelerometer outcome. There was a potential for hospitalizations which occurred closer to accelerometer collection to have more impact on change in accelerometer patterns Therefore, regression models were used to test whether days from the most recent hospitalization discharge date to subsequent accelerometer collect ion was correlated with changes in any accelerometer parameter. Two tailed alternative hypotheses and type 1 error rate of 0.05 were used for all analyses. Accelerometer data were processed using R ( www.r project.or g ) ( 18 115 ) and statistical analyses were performed in STATA v13 (STATA Corp.). Results Sample C haracteristics Baseline participant characteristics according hospitalization status within intervention assignment are described in Table 5 2. A total of 1,341 participants collected at least 10 hours of valid accelerome ter data for at least 3 days at baseline. Of those, 669 were randomized to the PA group and 672 were randomized to the HE group. Those without valid accelerometry data (n = 294) were similar in regard to age (79.6 5.0 years old ; mean SD ) and sex (70.1% women) as compared to the sample who had valid accelerometry data but differed by cognition ( 1.7 3MSE score), walking speed ( 0.03 m/s), physical performance ( 0.3 SPPB score) and LOS among those hospitalized (+1.8 hospital days) Randomized groups were s imilar across demographics, behavioral factors, and medical history where participants were 78.75.3 (SD) years, predominately females (66.5%), non Hispanic white (75.8%) and collected an average of 13.91.8 wear hours/day for 7.93.2 days. Furthermore, pa rticipants in the PA group who experienced at least one hospitalization (n=187) were similar to those not hospitalized (n=482)

PAGE 93

93 across demographics, behavioral factors, cognition, depression medical conditions, physical function, and accelerometer wear met rics but not for self reported hospitalizations in the past 6 months from baseline (11% vs. 5%, respectively, p=0.01) Those hospitalized during the HE program (n=171) were older, had lower body mass index, and had higher rates of 2+ comorbidities compared to those who did not experience a hospitalization (n=501; p<0.05 for all). Additionally, those in HE group who had experienced a hospitalization were more likely to have had a previous hospitalization in the past 6 months at baseline lower physical funct ion, lower walk speed, and, on average, spent less time in activity minutes registering greater than or equal to 760 counts (p<0.05 for all). Figure D1 describes the number of hospitalized participants, the number of hospitalizations and the average leng th of stay per hospitalizations by intervention group within each time interval. The first time interval (baseline to 6 months), 10% of the participants experienced at least one h ospitalization in the PA group. There was a rate of 1.4 hospitalizations per hospitalized PA participants, who averaged 5 LOS per hospitalization. Similarly, 8% of the HE participants experienced at least one hospitalization, had a rate of 1.2 hospitalizations among those hospitalized, and averaged 4 LOS per hospitalization. The s econd and third time intervals were similar in all hospital characteristics by intervention. Overall the cumulative characteristics are presented in Table 5 3. Of the total sample, 358 (27%) participants of the analytic sample experienced at least one hos pitalization and were similar between randomized groups. Among those hospitalized, 18%, 5.4%, and 3.3% experienced 1, 2, and 3 or more hospitalizations, respectively. Participants who experienced at least one

PAGE 94

94 hospitalization spent an average of 5.46.2 day s in the hospital across all assessment intervals Participants who experienced a short (1 3 days) and long (4+ days) hospitalization duration averaged 2 .00.8 days and 7.96.8 days across all assessment intervals respectively. Baseline A ccelerometer M et rics Between randomized groups, accelerometer metrics were similar for sedentary bouts and categories of TPA bouts (LLPA, HLPA, LMPA, and HMPA) showed in Table 5 4. Participants averaged 646.8 116.4 min/day total daily sedentary time (1+ min bouts; 77.2% of total daily wear time) at baseline. On average, 488.3 131.8, 296.0 132.7, and 145.4 102.9 min/day were spent in 10+ min, 30+ min, and 60+ min sedentary bouts respectfully. On average, participants spent 189.769.9 min /day in daily TPA ( 22.8 8.2% of to tal daily wear time), where 86.7 8.0% of daily TPA was spent in LLPA, 6.1 2.9% in HLPA, 5.9 4.3% in LMPA, and 1.3 2.8 % in HMPA Among the PA group, baseline accelerometer metrics were different across hospitalization categories where there were higher sed entary and lower activity patterns with longer hospitalization lengths except for total sedentary time and 10+ LLPA minute bouts (Table 5 4). Additionally, there were no differences found for 5 10+ HLPA minute bouts, 5+ LMPA minute bouts, and 5 10+ HMPA bo uts; all of which participants spent 2 minutes daily (Table 5 4). T here were differences in baseline accelerometer metrics a cross hospitalization categories in the HE group where there were higher sedentary and lower activity patterns with longer hospita lization lengths except for 5 10+ HLPA & LMPA minute bouts, and all HMPA minute bouts (Table 5 4)

PAGE 95

95 Hospitalization E ffect on A ccelerometer M etrics The correlation between days from most recent hospitalization discharge to subsequent accelerometer collect ion was not correlated with the change in any accelerometer parameter (Figure E 1). The effect of hospitalization on changes in daily accelerometer metrics over 24 months are presented in Table 5 5. For the PA group, a positive association with both short and long hospital durations was found for 1+ sedentary minute bout change when compared to no hospitalizations (p<0.05 ). A positive association was found in only long durations and changes in bout patterns of 10+ and 30+ sedentary minutes (p<0.05 for all). There was a significant trend of increasing sedentary time across higher hospital duration s for all bout lengths among the PA group (except 60+ sedentary minute bouts; trend p>0.05). For the HE group, positive associations with long durations were found f or all sedentary bout lengths when compared to no hospitalizations (p<0.01 for all). For HE, there was a significant trend of increasing sedentary time with longer hospital durations (trend p<0.05). When the groups were combined, a positive association wit h both short and long durations was found for daily total sedentary time (+7.53.1 daily minutes, p<0.05 for short duration; +16.212.2, p<0.001 for long duration). Only long durations were associated with 10+, 30+, and 60+ sedentary minute bout changes (p <0.001 for all). A significant trend of increasing sedentary time as hospital duration became longer was found for all sedentary bout lengths among the pooled groups (p<0.05 for all). Both short and long hospital durations were associated with 1+ and 5+ TPA minute bout declines when compared to no hospitalizations in the PA group (p<0.05 for both) where only long durations were associated with decreased time spent in 2+ TPA minute bouts (p<0.001). A significant trend of dec lining TPA time was observed as

PAGE 96

96 hospital durations became longer for 1+, 2+, and 5+ minute TPA bouts (p<0.05 for all). For the HE group, only long durations were associated with 1+ and 2+ minute TPA bout declines when then those without a hospitalization (p<0.05 for both). There was a tr end of TPA decline with longer hospital durations for these bout lengths (p<0.05 for both). The pooled group who experienced short and long durations had lower time spent in 1+, 2+, and 5+ TPA minute bouts than those without a hospitalization (p<0.05 for a ll) and experienced a significant trend of decreasing TPA time in these bouts as hospital durations became longer (trend p<0.05 for all). Different intensity levels were examined by hospitalization status and interventions Those with a short duration hosp italization demonstrated a decline in 1 +, 2+, and 5+ LLPA minute bout s (p<0.05 for all) and 1 + and 2+ HLPA minutes bouts (p<0.05 for both) for the PA group. This decline was more severe in those with a long duration hospitalization for 1 2+ LLPA & HLPA min ute bout (p<0.05 for all ) and 1 +, 2+, and 5+ LMPA minute bout (p<0.05 for all) For the HE group, only the 1 + and 2+ LLPA minute bout s were negatively impacted by a long duration hospitalization (p<0.01 for both). In the pooled group, s hort durations were associated with declines in 1 5+ LLPA minute bouts (p<0.05 for all) 1 2+ HLPA minute bouts (p<0.05), and 1 5+ LMPA minute bouts. For long durations, negative associations were found for changes in 1 5+ LLPA minute bouts (p<0.05), 1 2+ HLPA minute bouts (p<0.001), and 1 5+ LMPA minute bouts (p<0.05) All bout lengths significantly impacted by a hospitalization in PA, HE, and the pooled group showed a significant trend towards greater declines in activity as hospital durations became long er (trend p<0.05 for all). For the HE group, 2+ HLPA bouts and 1 + minute L M PA bouts showed a significant trend towards greater activity

PAGE 97

97 decline with longer hospital durations (trend p<0.05) though no significant hospitalization impact was observed for eith er short or long hospital duration s H I mpact on A ccelerometer M etrics between I nterventions Table 5 5 presents the p value testing whether the impact of hospitalization on daily inactivity/activity patterns was different between interventi on groups in the far right column Across all accelerometer metrics, those hospitalized for either short or long duration s did not differ from those who did not experience hospitalization (p PA vs HE >0.05) independent of intervention effects on accelerometer outcomes Additionally, there was no detectable interaction between hospitalization and intervention by time for all metrics (p interaction >0.05 for all). Figure E 2 describes the intervention effect over 2 years, independent of hospitalizatio n status. T he PA group experienced a decrease in sedentary time for 1+ minute bouts ( 8 minutes, p=0.005) and 10+ minute bouts ( 10 minutes, p=0.009) compared to the HE group Also, the PA group experienced increased daily time in TPA across all bouts (3 6 minutes, p<0.01 for all) Figure E 3 further describes intervention effects on TPA by intensity category. The both hospitalized and non hospitalized PA participants experienced increases in 10+ minute LLPA bouts (1 minute, p=0.03), all bouts lengths in HL PA (0.1 0.8 minutes, p<0.01), LMPA (1.1 2.8 minutes, p<0.001), and HMPA (1.2 1.9 minutes, p<0.001) when compared to HE As bout lengths became longer, increases appear to be lower. D iscussion This is the first study to examine the intervening effect of hos pitalizations on objectively measured sedentary and physical activity patterns in mobility limited older adults. Among LIFE participants, we found that a n intervening hospitalization is associat ed with subsequent increases in short bouts of sedentary time (<10 minutes)

PAGE 98

98 and de clines in short bouts of physical activity ranging from LLPA (<10 minutes) to HLPA (<5 minutes) Further, t hose who accumulated more than 3 hospital days experienced magnified increases in all sedentary bouts while experiencing greater declines in total daily physical activity primarily due to declines in L LPA bouts (< 10 minutes) HLPA bouts (<5 minutes), and LMPA bouts (<10 minutes) W e found that those who experience 4+ cumulative hospital days but not 1 3 days, had increases in s edentary bouts > 10 minutes, and declines in shorter LMPA bouts (< 10 minutes) While these results support our primary hypothesis we found that long bouts of physical activity (10+ minutes) were largely unaffected by hospitalizations across all activity l evels. Also, an overall intervening hospitalization did not impact bouts lasting 5 minutes or longer in HLPA Additionally, hospitalization did not impact HMPA whatsoever regardless of the length of cumulative hospital days. Finally, a structured moderate intensity physical activity intervention did not attenuate the negative, intervening effect of hospitalization on daily activity patterns compared to health education which was contrary to our second hypothesis. Additionally, there was no evidence to show that the PA exacerbated the activity patterns after hospitalization. Yet, intervention effects on accelerometer metrics, such as decreases in sedentary patterns or increases in activity patterns, were still experienced by the PA group, regardless of hospi talization status. These findings increase understanding of the degree to which hospitalizations influence patterns of free living sedentary and physical activity and its relation to a long term structured physical activity intervention in mobility limite d older adults. There is sparse literature characteriz ing sedentary patterns in free living conditions among older adults who have experienced a hospitalization. Previous

PAGE 99

99 observational studies show that > 80% of a hospitalized stay is spent in sedentary tim e determined by activity monitors ( 11 76 ) In recent observational studies hospitalized older adults who survived a stroke had higher total sedentary time after hospitalization when compared to matched counterparts with no existing medical problems which remained even 6 months after the hospitalization ( 80 112 ) Our results are similar by showing that older adults at risk for mobility disability increased their total daily sedentary time after a hospitalization. The current work extends this knowledge by showing the effect is primarily seen in short rather than long bouts of sedentary time. Notably all patterns of sedentary time, particularly long bouts, were increased when an intervening hospitalization extended to 4+ days spent in a hospital setting Thus, a n overall intervening hospitalization (short or long durations) appear s to increase short sedentary bouts w here only longer hospital durations extend s this effect to longer sedentary bouts I ncreases in sedentary time did not dimi nish in hospitalized LIFE participants, suggesting a new permanent and new trajectory towards overwhelming inactivity beyond normal aging effects. The coupling effect of hospitalization and aging on sedentary behavior may h elp describe the accelerated traj ectory towards functional loss and transitions into disability states ( 42 43 ) Existing literature demonstrates that physical activity patterns are reduced among hospitalized older adults when compared to disability free and non hospitalized counter part s ( 66 92 98 ) A recent observational study showed that older adults who were h ospitalized for a stroke had reduced objectively measured walking patterns (short and long distances) 40 months later when compared to older adults who did not experience stroke and lived without disability ( 98 ) Our results similarly show no

PAGE 100

100 indication that pre hospital patterns of activity were regained, suggesting that hospitalizations are associated with permanent and deleterious effects on daily activity patterns. A dditionally we showed that hospitalizations in LIFE participants largely affected short (<5 minutes), rather than long bouts of activity which lasted 5 minutes or more Furthermore, bouts of activity lasting 10+ minutes at all intensity levels were not impacted by hospitalizations This discrepancy between low and high activity intensities may be due to a longer amount of time spent in lowe r intensit y physical activity (100 759 counts/min ute ) Yet, h ospitalization did not impact 10+ TPA minute bouts in which LIFE participants participated in approximately one 10 mi nute bout of activity per week and even less periodic participation in l ong bouts at higher intensities ( e.g., biweekly or monthly). As previous literature suggest s this bout length may describe an activity that is rout ine in an independent lifestyle (e.g. grocery shopping, retrieving mail, walking to a bank, visiting a doctor, etc.) and thus critical to maintain in life after a hospital event ( 80 98 ) Also, l ittle to no time was spent in long bouts of activity at intensities higher than LLPA particularly in bouts of LMPA and HMPA. Individuals engage in these bout lengt hs less frequent ly in daily life routines that are often variable in length and may be more resistant to events such as hospitalizations O ur results suggest that older adults maintain the capability to perform physical activity following a hospitalization despite spending long durations in the hospital but may be at a more vulnerable state due to shrinking activity patterns and higher sedentary times This is an important public health priority as hospitalized mobility limited older adults are a vulnerab le population in which interventions could be designed to preserve and perhaps bolster the capability to perform physical activity that has proven to reduce the risk of MMD.

PAGE 101

101 The increases in sedentary patterns and decreases in physical activity may be a continuation of imposed bed rest often seen during hospitalization. Even short duration b ed rest reduces muscle mass and function ( 64 ) and may wor sen due to co occurring sarcopenia ( 33 ) Not only doe s an intervening hospitalization affect activity patterns but more time spent in the hospital leads to greater physical inactivity ( 25 45 76 ) We show that mobility limited older adults do not regain pre hospital activity levels which suggests that hospitalizations increase the risk of functional decline potentially through a lo ss in physical activity ( 25 42 43 ) This in turn negatively impacts the ability to maintain independence ( 24 ) Although, h ospitalized participants appear to be on a new trajectory towards disability and functional loss they paradoxically maintain activity at high er intensities which may reflect preservation of routine activities necessary for self care and independence in a community dwelling setting. F uture work is needed to implement efficient and immediate strategies to promote a sustained reduction in a sedentary lifestyle and provide support to increase physical activity back to pre hospital levels in older adults on the cusp of being disabled There was no evidence that a long term structured physical activity program reduced the negative effect that hospitalizations have on activity patterns. This is contrary to our hypothesis, which stated that a moderate intensity structured physical activity intervention shown to reduce the risk of major mobility disability would attenuate the effect that a hospitalization is expected to have on activity This is particularly puzzling because the LIFE study implemented a post hospital re engagement protocol for tho se randomized to the PA group. Though the effect of hospitaliz ation did not change between PA and HE, w e observed that there were significant intervention

PAGE 102

102 effects (8 10 minute decrease in short sedentary bouts and 4 6 minute increases in activity bouts) experienced by the PA group vs H E, independent of hospitalization This suggest s that the PA intervention did not change the m impact on bout patterns but improved recovery of activity afterwards. Even though hospitalized LIFE participants experienced a deleterious hospital effect, those hospitalized still increased levels of activity and managed sede ntary behaviors better than those without structured physical activity. This supports post hospitalization strategies to promote the integrati on of structured physical activity to combat functional decline and improve mobility after a hospitalization ( 26 6 7 107 ) W hile structured physical activity improves mobility, effects on daily lifestyle inactivity and activity patterns due to a h ospitalization appears to be challenging to circumvent because those hospitalized in the PA intervention were still at lower levels of activity than those not hospitalized in PA Future studies should fo cus on strategies within structured physical activity plans to improved recovery of post ho spital activity patterns which may require earlier post hospital care specifically tailored for patients to return back to regular daily activity levels as soon as safely able Strengths of this study include a large sample of racial ly diverse mobility limited men and women from eight field centers across the US for an average of 2.6 years with high retention and adherence rates. Patterns of physical inactivity and a ctivity w ere measured objectively using a hip worn accelerometer and repeated 3 more times over 2 years. Free living accelerometer patterns were examined by volume, frequency, and intensity to reveal associations with intervening hospitalizations. Further a ccelerometer cut points used in this study combined traditional cut points used in the literature

PAGE 103

103 ( calibrated with adults 20+ years and older ) and cut points calibrated in older adult populations. There are limitations to our study. We did not consider t he reason for hospitalization in our analysis However, the reason for hospitalization was diverse and no single type accounted for more than 15% of total hospitalizations in the LIFE study ( 71 ) Those who collected valid accelerometry tended to experience shorter stay hospitalizations compared to the entire sa mple of LIFE participants who experienced any hospitalization so our results represent a smaller sample of mobility limited older adults who liv e with less severe health conditions and had better physical function The reason of hospitalization was not ass essed in our study which may impact the effect by conditions specific to physical activity engagement and ability. Table C 1 shows that the reason for hospitalization (classified by organ class) in our sample was heterogeneous between both intervention gro ups except for hospitalizations for autonomic nervous system disorders (PA vs HE: 8 to 1, p<0.04) However, this class of hospitalization consisted of less than 3% of total hospitalizations in this analysis. Though measuring accelerometry periodically is o ne of the main strength s of this study, hospitalization effects on activity patterns lasting for less than 6 months may not have been captured in this analysis and may underestimate the negative hospitalization effect. L ack of knowledge regarding posture (e.g., sleep, sitting, etc.) may overestimate sedentary estimates by including standing and strength exercise behaviors. We also acknowledge the potential for residual confounding specifically in regards to reasons for h ospitalizations The LIFE study is a randomized control trial but this analysis was designed as a prospective cohort with valid accelerometer to examine effect of

PAGE 104

104 intervening hospitalizations on activity patterns To account for confounding, we adjusted fo r numerous factors which may affect both hospitalization reason ( co morbidities) and inactivity/activity patterns longitudinally. Another limitation is the potential im balance between groups where t h ose in PA had more contact to study staff and therefore gi ven more opportunity to report health events To reduce this bias, we excluded hospitalizations which were not blinded to assessment staff, but it remains possible that participants were more likely to remember hospitalizations in the PA group Additionall y, the PA intervention incorporated a re engagement protocol with strategies to bring hospitalized participants back into the intervention The HE program did not implement these same strategies In conclusion, hospitalizations negatively impact post hosp ital patterns of daily activity. Further, long er durations spent in a hospital setting leads to increased prolonged inactivity following a hospitalization Those in the PA intervention decreased sedentary behavior in short bouts and increased activity towa rds longer and higher intensity bouts, regardless of being hospitalized or not. Still, engagement in a structured physical activity intervention, known to improve mobility, did not change the negative effect of hospitalization on patterns of daily activity These findings help describe age related declines in activity and further describe the irreversible trajectory towards inactivity that contributes to functional loss after hospitalizations Future work is needed to develop interventions to promote immediate and sustained recovery of activity levels after hospitalization to help prevent mobility loss in the future Additionally, f uture physical activity interventions should incorporate strategies to combat sedentary behaviors, while prioritizing redu ction in prolonged inactivity immediately after a

PAGE 105

105 hospitalization. Finally, the results suggest that post hospital home services should integrat e strategies to increase participation in both short and long activity patterns at home to help prevent mobility decline particularly among mobility limited older adults who are at high risk of disability

PAGE 106

106 Table 5 1. Accelerometer cut points used to derive metrics of physical inactivity and activity Intensity Bouts Physical inactivity Sedentary behavior (SB): < 100 counts/min ute 1. Time spent in 1+ min ute bouts (min ute /day; total activity time) 2. Time spent in 10+ min ute bouts (min ute /day) 3. Time spent in 30+ min ute bouts (min ute /day) 4. Time spent in 60+ min ute bouts (min ute /day) Physical activity Light physical activity (LLPA): 100 759 counts/min ute High light physical activity (HLPA): 760 1,040 counts/min ute Low moderate physical activity (LMPA): 1,041 2019 counts/min ute High moderate physical activity (HMPA): 2,020+ counts/min ute Total physical activity (TPA): 100+ counts/min ute 1. Time spent in 1+ min ute bouts (min ute /day; total activity time) 2. Time spent in 2+ min ute bouts (min ute /day) 3. Time spent in 5+ min ute bouts (min ute /day) 4. Time spent in 10+ min ute bouts (min ute /day)

PAGE 107

107 Table 5 2. Baseline participant characteristics by intervention group and hospitalization status over 24 months PA (n=669) HE (n=672) Hosp. (n=187) No hosp. (n=482) p Hosp. (n=171) No hosp. (n=501) p Age, mean(SD) >= 80, n(%) 79.0 (5.4) 79 (42.3) 78.3 (5.2) 192 (39.8) 0.14 0.27 79.8 (5.2) 87 (50.9) 78.7 (5.2) 206 (41.1) 0.01 0.03 Female, n(%) 126 (67.4) 305 (63.3) 0.32 107 (62.6) 354 (70.7) 0.05 Non Hispanic white, n(%) 139 (74.3) 358 (74.3) 0.68 144 (84.2) 376 (75.1) 0.04 College or higher education, n(%) 112 (65.5) 318 (63.5) 0.65 111 (59.4) 306 (63.5) 0.49 Married, n(%) 79 (42.3) 174 (36.1) 0.24 57 (33.3) 184 (36.7) 0.70 Annual income < $25,000 54 (28.9) 136 (28.2) 0.95 45 (26.3) 145 (28.9) 0.51 Body mass index, mean (SD) 30.3 (5.8) 30.1 (5.8) 0.68 30.1 (6.1) 30.6 (6.3) 0.44 Smoked 100+ cigarettes ever, n(%) 102 (54.6) 237 (49.2) 0.41 84 (49.1) 213 (42.5) 0.32 Modified Mini Mental State Examination score a mean(SD) 91.4 (5.3) 91.9 (5.5) 0.33 91.8 (5.4) 91.6 (5.3) 0.60 Sought medical advice for depression in past 5 years, n(%) 28 (15.0) 69 (14.3) 0.75 28 (16.4) 57 (11.4) 0.17 Self reported hospitalization in the past 6 months n(%) 21 (11.2) 26 (5.4) 0. 01 22 (12.9) 30 (6.0) 0.004 Comorbidities > 2, n(%) 57 (30.5) 115 (23.9) 0.18 63 (36.8) 114 (22.8) 0.001 Short Physical Performance Battery score < 8 b n(%) 88 (47.1) 198 (41.1) 0.16 103 (60.2) 206 (41.1) <0.0 01 400 meter walk < 0.8 m/sec, n(%) 111 (59.4) 287 (59.5) 0.97 77 (45.0) 299 (59.7) 0.001 Wear days, mean(SD) 8.0 (3.4) 8.0 (3.2) 0.89 8.1 (3.2) 7.8 (3.1) 0.87 Wear minute/day, mean (SD) 843.5 ( 120.2 ) 837.4 ( 109.8 ) 0.53 827.1 (85.8) 837 ( 116.3 ) 0.27 (SD) 25.6 (20.2) 28.3 (23.7) 0.18 22.4 (20.4) 29.8 (27.3) 0.001 Note: PA physical activity; HE health education; hosp. hospitalized a score range: 0 100 where higher scores indicate better performance b score range: 0 12 where higher scores indicate better performance; scoring < 8 indicates poor functioning

PAGE 108

108 Table 5 3. Hospitalizations by intervention group among participants who collected 3+ days of 10+ hours/day of accelerometry Physical activity (n=669) Health education (n=672) Participants hospitalized, n(%) 187 (28.0) 171 (25.4) Hospitalization events, n(rate per participant) 0, n(%) 1, n(%) 2, n(%) 3+, n(%) Baseline to 6 months, n(%) 6 to 12 months, n(%) 12 to 24 months, n(%) 299 (0.45) 482 (72.1) 120 (18.0) 41 (6.1) 26 (4.0) 77 (11.5) 77 (11.5) 145 (21.7) 247 (0.37) 501 (74.6) 121 (18.0) 32 (4.8) 18 (2.7) 55 (8.2) 73 (10.9) 119 (17.7) Days hospitalized per participant, mean(SD) Days hospitalized per hospitalized participant, mean(SD) 1.6 (4.3) 5.7 (6.4) 1.3 (3.8) 5.2 (6.0) 1 3 days, n(rate per participant) 4+ days, n(rate per participant) Missing 94 (2.0) 93 (6.7) 0 93 (2.0) 75 (6.8) 3

PAGE 109

109 Table 5 4. Descriptive baseline daily activity bouts by intervention group and hospitalization ^ Physical activity Health education None 1 3 days 4+ days p None 1 3 days 4+ days p Bout Daily minutes, m ean ( SD ) Sedentary 1+ 639.9 (110.1) 645.1 (118.2) 650.7 (96.3) 0.28 640.7 (106.8) 659.6 (99.1) 673.0 (99.9) 0.00 1 10+ 485.7 (127.5) 497.2 (140.0) 511.2 (121.8) 0.0 3 486.5 (124.3) 514.8 (117.6) 544.7 (121.3) <0.001 30+ 299.5 (129.8) 313.7 (138.0) 329.4 (145.1) 0.01 298.3 (127.6) 328.4 (120.6) 366.7 (142.9) <0.001 60+ 151.3 (103.3) 162.9 (107.4) 179.3 (135.2) 0.00 4 148.9 (100.4) 167.5 (96.2) 209.0 (128.1) <0.001 Total physical activity 1+ 187.4 (68.2) 172.1 (73.9) 152.1 (67.0) <0.001 184.7 (65.0) 152.9 (57.8) 142.6 (60.3) <0.001 2+ 129.3 (52.3) 117.5 (57.0) 104.9 (54.1) <0.001 127.1 (52.8) 103.0 (46.0) 96.1 (47.1) <0.001 5+ 45.3 (26.7) 41.1 (26.4) 36.4 (27.1) <0.001 43.3 (26.7) 32.9 (22.6) 31.6 (22.0) <0.001 10+ 11.4 (11.8) 10.8 (10.3) 9.1 (10.8) 0.04 9.5 (11.0) 6.7 (8.1) 7.1 (8.0) 0.00 3 Low light physical activity 1+ 157.3 (53.6) 146.2 (57.6) 135.0 (56.5) <0.001 157.9 (54.0) 136.2 (48.0) 124.7 (51.2) <0.001 2+ 116.4 (47.0) 106.3 (50.0) 98.0 (49.2) <0.001 117.0 (47.0) 97.5 (42.8) 89.6 (44.3) <0.001 5+ 40.6 (25.5) 36.6 (24.0) 33.8 (25.8) 0.00 2 40.6 (31.7) 31.7 (22.2) 30.0 (21.6) <0.001 10+ 8.8 (10.5) 7.9 (7.6) 8.0 (10.3) 0. 30 8.4 (9.9) 6.4 (8.1) 6.4 (7.7) 0.0 1 High light physical activity 1+ 12.6 (8.6) 10.7 (8.1) 7.8 (6.6) <0.001 12.1 (9.1) 8.1 (6.2) 7.5 (5.8) <0.001 2+ 3.3 (3.1) 2.7 (2.5) 2.1 (2.6) <0.001 3.0 (3.1) 1.8 (1.9) 1.7 (2.0) <0.001 5+ 0.4 (1.0) 0.2 (0.9) 0.4 (0.9) 0. 76 0.2 (0.8) 0.1 (0.5) 0.1 (0.5) 0. 33 10+ 0.1 (0.6) <0.1 (0.3) 0.1 (0.4) 0. 40 <0.1 (0.4) <0.1 (0.2) <0.1 (0.4) 0. 95 Low moderate physical activity 1+ 13.5 (11.8) 11.6 (11.2) 7.8 (8.1) <0.001 12.0 (12.4) 7.3 (7.0) 7.5 (8.7) <0.001 2+ 6.8 (7.6) 5.8 (6.9) 3.8 (5.8) <0.001 5.5 (8.2) 2.9 (3.9) 3.4 (5.3) <0.001 5+ 2.4 (4.1) 2.3 (3.9) 1.5 (3.4) 0.0 3 1.6 (4.5) 0.7 (1.3) 1.1 (3.1) 0.0 7 10+ 1.2 (2.9) 1.2 (3.0) 0.7 (2.4) 0.1 8 0.6 (2.9) 0.1 (0.7) 0.5 (2.1) 0. 40 High moderate physical activity 1+ 4.0 (7.9) 3.6 (8.9) 1.6 (3.2) 0.00 3 2.6 (5.5) 1.3 (2.7) 2.4 (11.8) 0. 20 2+ 2.9 (6.7) 2.8 (8.0) 1.0 (2.8) 0.01 1.6 (4.6) 0.8 (2.1) 1.4 (7.5) 0. 20 5+ 1.9 (5.6) 2.0 (6.6) 0.6 (2.1) 0.0 5 0.9 (3.6) 0.3 (1.4) 0.4 (1.9) 0.0 7 10+ 1.4 (4.7) 1.6 (6.0) 0.3 (1.3) 0.0 8 0.5 (2.5) 0.2 (1.2) 0. 1 (0.1) 0.0 6 ^cumulative days spent in a hospital setting

PAGE 110

110 Table 5 5. Effect of hospital duration on daily accelerometer pattern change s (daily minutes) in mobility limited older adults ( beta coefficient ( SEM) )^ Physical activity (PA) Health education (HE) Combined p interaction 1 3 days 4+ days 1 3 days 4+ days 1 3 days 4+ days Sedentary time 1+ minute bouts 8.6 (4.3)* 15.5 (4.5)** 5.9 (4.4) 16.8 (4.9)*** 7.5 (3.1)* 16.2 (12.2)*** # 0.82 10+ minute bouts 10.1 (6.6) 18.8 (6.9)** 5.8 (6.6) 26.8 (7.4)*** 8.0 (4.7) 22.0 (5.1)*** 0.58 30+ minute bouts 10.5 (7.7) 17.0 (8.0)* 6.1 (7.8) 27.7 (8.8)** 8.5 (5.5) 21.3 (5.9)*** 0.58 60+ minute bouts 6.2 (7.0) 12.5 (7.3) 0.5 (7.1) 30.0 (8.0)*** 2.9 (5.0) 20.2 (5.4)*** 0.20 Total physical activity 1+ minute bouts 8.7 (4.0)* 17.4 (4.2)*** 3.8 (4.2) 13.7 (4.6)** 6.5 (2.9)* 15.9 (3.1)*** # 0.64 2+ minute bouts 7.4 (3.3) 12.6 (3.4)*** 2.6 (3.3) 9.5 (3.7)* 5.2 (2.3)* 11.4 (2.5)*** 0.56 5+ minute bouts 3.9 (2.0)* 5.3 (2.0)** 1.7 (2.0) 2.2 (2.3)) 2.9 (1.4)* 4.4 (1.5)** 0.58 10+ minute bouts 1.4 (1.1) 1.9 (1.1) 0.3 (1.0) 0.7 (1.2) 0.9 (0.8) 1.2 (0.9) 0.43 Low light physical activity 1+ minute bouts 7.2 (3.3)* 10.0 (3.5)** 2.1 (3.4) 11.9 (3.8)** 4.8 (2.4)* 10.7 (2.6)*** 0.57 2+ minute bouts 7.1 (3.0)* 8.5 (3.1)** 2.0 (3.1) 9.7 (3.4)** 4.7 (2.1)* 9.0 (2.3)*** 0.52 5+ minute bouts 4.2 (1.9)* 3.5 (2.0) 1.5 (2.0) 2.7 (2.2) 2.9 (1.3)* 3.4 (1.5)* 0.73 10+ minute bouts 2.0 (1.0) 1.0 (1.1) < 0.0 (1.0) 0.3 (1.1) 1.0 (0.7) 0.6 (0.8) 0.77 High light physical activity 1+ minute bouts 1.3 (0.6)* 2.6 (0.6)*** 0.9 (0.6) 2.0 (0.6)** 1.1 (0.4)** 2.4 (0.4)*** # 0.82 2+ minute bouts 0.6 (0.3)* 0.9 (0.3)** 0.3 (0.2) 0.5 (0.3) 0.4 (0.2)* 0.8 (0.2)*** 0.81 5+ minute bouts 0.2 (0.1) 0.1 (0.1) < 0.0 (0.1) < 0.0 (0.1) 0.2 (0.1) 0.1 (0.1) 0.72 10+ minute bouts 0.1 (0.1) 0.1 (0.1) < 0.0 (0.0) 0.0 (0.0) < 0.0 (0.0) < 0.0 (0.0) 0.72 Low moderate physical activity 1+ minute bouts 1.0 (0.8) 3.5 (0.8)*** 0.8 (0.8) 1.7 (0.8) 0.9 (0.6) 2.8 (0.6)*** 0.25 2+ minute bouts 0.6 (0.6) 2.3 (0.6)*** 0.4 (0.5) 0.7 (0.6) 0. 5 (0.4) 1.6 (0.4)*** 0.10 5+ minute bouts 0.2 (0.4) 1.0 (0.4)* 0.1 (0.3) 0.2 (0.3) 0.1 (0.3) 0.6 (0.3)* 0.06 10+ minute bouts 0.1 (0.3) 0.4 (0.3) < 0.0 (0.2) 0.4 (0.3) < 0.0 (0.5) 0.2 (0.2) 0.05 H igh moderate physical activity 1+ minute bouts 0.5 (0.6) 1.0 (0.6) 0.5 (0.5) 0.5 (0.6) < 0.1 (0.4) 0.2 (0.4) 0.12 2+ minute bouts 0.6 (0.5) 0.7 (0.5) 0.3 (0.4) 0.5 (0.4) 0.2 (0.3) 0.1 (0.3) 0.12 5+ minute bouts 0.6 (0.4) 0.5 (0.4) 0.2 (0.3) 0.2 (0.3) 0.3 (0.2) 0.3 (0.3) 0.29 10+ minute bouts 0.7 (0.4) 0.4 (0.4) 0.2 (0.2) < 0.1 (0.2) 0.2 (0.2) 0.3 (0.2) 0.38 ^Reference value is no hospitalizations (0 days). All models were adjusted for time (months), wear time, baseline accelerometer metric, age, sex, race/ethnicity, income, education, marital status, modified mini mental status exam score (3MSE where 100 is highest cognition score), self reported depression, body mass index (kg/m^2), smoked 100+ cigarettes ever, comorbidity status, previous hospitalizations in the past 6 months gait speed < 0.8 m/s, and clinical site. Intervention status was added as a covariate in pooled models. PA vs HE p value repr esents the interaction of hospitalization x intervention. *p<0 .05 **p<0.01 ***p<0.001 ; # p<0.05 for trend across hospital duration categories

PAGE 111

111 CHAPTER 6 CONCLUSIONS Summary The central objective of this dissertation was to examine the interaction between a long term structured physical activity intervention and the role intervening hospitalizations contribute to patterns of daily inactivity and activity in mobility impaired older adults at high risk of disability. Accelerome ters provide the ability to measure physical activity objectively in free living settings provid ing a unique opportunity to assess this interaction which is an important public health issue for one of the fastest growing age segments in the US population ( 109 ) No prior studies have evaluated patterns of daily inactivity and activity in mobility limited older adults. The first contribution this dissertation provides to the current literature is the charact erization of sedentary time and physical activity patterns typical to mobility limited older adults This is particularly important because mobility limited older adults are at high risk of physical disab ility and loss of independence ( 61 ) Our findings suggest that mobility limited older adults are generally sedentary through the day (77% of daily waking time) through the accumulation of short bouts (<30 minutes) rather than prolonged bouts of sedentary behaviors Physical activity accumulated by mobili ty limited older adults account for the rest of daily waking time (23%) and are largely accrued in short bouts (<5 minutes) of physical ac tivity at the lowest intensity which consists of 80% of total daily physical activity. An added contribution is the characteriz ation of long term activity patterns post hospitalization where b aseline sedentary patterns were higher and activity patterns were lower with higher time spent in hospital settings.

PAGE 112

11 2 The second contribution was the finding that mobility l imited older adults increased time spent in sedentary behaviors in mainly short bouts, but also long and prolonged bouts. Our findings suggest that changes in sedentary time partly explain declines in function, independent of physical activity engagement. Concurrently, total physic al activity decreased over time, while the composition of total physical activity shifted from higher intensity physical activity to the lowest intensity of physical activity. Hospitalizations accelerated the se aging effects primarily by further increasing time in short sedentary bouts and decreases shorter bouts of total physical activity Longer bouts of sedentary time were not affected unless hospitalized older adults spent 4 or more days in the hospital. This was similar t o longer bouts of higher intensity physical activity except for bouts patterns at the highest intensity of physical activity and 10+ minute bouts of activity at any intensity. This indicates that the capability to perform physical activity at any intensity remained intact after an intervening hospitalization but this adverse event conferred deleterious effects on patterns of daily activity. Coupled with our findings that increase in sedentary time contribute to functional decline and that hospitalizations t rigger a new trajectory towards overwhelming physical inactivity and shrinking activity beyond normal aging we add to current knowledge to help describe the degree to which hospitalizations precipitat e irreversible functional loss and future transition in to disability ( 41 43 ) The third contribution this dissertation provided was the evaluation of a structured, moderate intensity physical activity intervention on patterns of dai ly activity. The intervention had small but beneficial effects on patterns of activity in mobility limited older adults. This is important because sedentary behaviors have been

PAGE 113

113 associated with adverse health outcomes for which olde r adults are at high risk ( 29 51 62 64 75 80 ) Because we found t hat sedentary changes were associated with changes in physical function, it appears that the PA intervention may have beneficial effects on physical function through sedentary pattern reductions. Also, the PA intervention changed the composition of total d aily activity by increasing HLPA, LMPA, and HMPA, which previous literature suggest are bene ficial to mobility and health ( 52 83 86 89 ) Despite these beneficial effects of the intervention, age related decline in total activity w as not eliminated in mobility limited older adults. Lastly the PA intervention did not attenua te the added nega tive effects of hospitalizations on subsequent increase in sedentary time and decrease in physical activity but did not There are limitations to this study. First, ac tivity patterns were examined only in older adults with mobility limitations and not those with high physical function or severe cognitive deficits reducing the generalizability of the results to the entire older adult population Second, this is the firs t study to assess accelerometer derived outcomes but not all LIFE participants collected accelerometry for at least 3 days of 10+ hours of valid accelerometry After comparing those with and without valid accelerometry, we found no differences across age, sex, race/ethnicity, and number of hospitalizations. We did find those without valid accelerometry had slightly lower 3MSE scores ( 1.6 average score, p<0.001) and higher hosp italization stays (+ 2 3 average days, p=0.0 2 ) when compared to those with valid accelerometry However, cognition between those with and without valid accelerometry remained >90 (out of 100) and hospitalization stay differences were detected only during 12 to 24 months. Third, accelerometer outcomes were collected

PAGE 114

114 three times in 6 month intervals, and then a 12 month interval. These intervals may not fully capture in accelerometer determined activity changes due to less severe hospitalization but this bias is reduced using a comparison group Lastly accelerometer derived sedentary time does not distinguish between postural positions possibly capturing non sedentary activities such as standing and strengthening exercises but intervention comparisons reduce this estimation bias An important strength of this study was the inclusion of mobility limited older adults recruited from urban, suburban, and rural settings through 8 centers spanning across the United States Further, LIFE participants represent an u nderstudied portion of the population not typically examined in a randomized control trial setting. Though this study is not generalizable to adults 70 years of age and older, it focused on a vulnerable group of older adults at risk for mobility disability Lastly, this is the first study to repeat collection of accelerometry 4 times over a 2 year period. In conclusion, a structured, moderate intensity physical activity intervention known to improve mobility in older adults had small but beneficial effects on sedentary pattern reduction which appear to be associated with increases in physical function measures Also, the intervention increasing daily activity patterns by changing the intensity composition of daily activity patterns (mainly low intensity sho rt bouts) to higher intensities and long er bouts. However, this di d not eliminate overall decline in total physical activity. Lastly, the PA intervention was not capable in attenuating accelerated increased physical inactivity (short and long bouts) and de creases physical activity after hospitalization.

PAGE 115

115 Future Direction s Further work is needed to explore whether moderate intensity PA replaces sedentary time and how the type of PA influences sedentary time. Developing ways to increase PA among older adults r emains a key public health goal and future interventions should consider integrating a sedentary behavior reduction component to maximize health benefits in this vulnerable population. This dissertation help s describe the intersection between sedentary and physical activity patterns, exercise, and hospitalizations but only among mobility limited older adults. There still remains a gap towards patterns of physical activity in other populations, particularly amo ng those who are disabled who may be at higher risk for adverse health events such as hospitalizations The PA intervention was not meant to attenuate the effects of hospitalization. There needs to be future research which identifies specific reasons of h ospitalizations which may affect daily activity patterns differently. For example, a hospitalization due to a leg injury may decrease daily time usually spent in grocery shopping more than a hospitalization due to an arm injury. Future work should also cap ture types of activities which are affected by hospitalizations and identify those which are preserved despite hospitalizations. This dissertation provides the initial evidence that more work is needed to develop immediate post hospital care which specific ally promotes reducing sedentary behaviors and increasing total daily activity. Identifying patterns of objectively collected daily inactivity and activity in mobility limited older adults is unique to the literature. This work adds to the literature and h elps build future work to detect patterns of activity which may indicate trajectories toward disability. Although the LIFE study was powered to detect the capability of the PA

PAGE 116

116 intervention to reduce the risk of major mobility disability, this secondary dat a analysis showed potential beneficial effects on sedentary and activity patterns. Future work should build on these findings by developing physical activity interventions to integrate efficient strategies to preserve and even bolster physical activity in free living settings among those who are vulnerable to adverse health outcomes leading to loss of independence

PAGE 117

117 APPENDIX A R CODE USED TO GENERATE ACCELEROMETER VARIABLES ##call libraries library ( accelerometry ) library ( PhysicalActivity ) wide < data .frame () filelist < dir ( "E:/LIFE datasets/Updated LIFE data_R data format/second_data/" ) for ( i in filelist ) { load ( paste ( "E:/LIFE datasets/Updated LIFE data_R data format/second_data/" ,i,sep = "" )) mydata < AC.1s mydata1m = dataCollapser ( mydata, TS = "TimeStamp" col = "axis1" by = 60 ) data1m = wearingMarking ( dataset = mydata1m, frame = 90 perMinuteCts = 1 TS = "TimeStamp" cts = "axis1" streamFrame = NULL allowanceFrame = 2 dayStart = "00:00:00" dayEnd = "23:59:59" getMinuteMarking = TRUE newcolname = "wearing" ) #Prior to deriving the participant level daily average metrics, outliers were identified #and flagged. Two kinds of outliers were defined in the LIFE study. #first type outlierFlag1 < data1m $ axis1 > 10000

PAGE 118

118 #second type first component to outlier...correcting for max x of 0 secondMax < function ( x ) { if ( max ( x ) > 0 ) { max ( x [ x != max ( x )]) } else x < 0 } ag < aggregate ( data1m [ 2 ] list ( data1m $ days ) secondMax ) secondMedian < median ( ag $ x ) secondcutpoint < median ( ag $ x ) + 3500 outlierFlag2 < data1m $ axis1 > secondcutpoint #second type second component to outlier outlierFlag3 < FALSE #just a hack to make append work, not considered in the latter code for ( d in 1 : max ( data1m $ days )) { dayData < data1m [ data1m $ days == d, ] ; thirdcutpoint < ag [ ag $ Group.1 == d, ]$ x + 1000 ; outlierFlag3 < append ( x = outlierFlag3, values = dayData $ axis1 > thirdcutpoint ) } data1m $ outlierFlag < ( outlierFlag1 | outlierFlag2 | outlierFlag3 [ 2 : length ( outlierFlag3 )]) #created a 0/1 wear indicator data1m $ wearindicator [ data1m $ wearing == "nw" ] < 0 data1m $ wearindicator [ data1m $ wearing == "w" ] < 1 data1m $ wearindicator [ data1m $ outlierFlag == TRUE ] < 0 #Wear and non wear times summarized by day if ( is.element ( "w" ,data1m $ wearing ) == TRUE ){ mySummary < summaryData ( data1m, validCut = 600 perMinuteCts = 1 markingString = "w" ) validDayUse < ( names ( mySummary $ validWearTimeByDay )) numvalday < as.double ( validDayUse ) } else numvalday < 0 if ( is.element ( "nw" ,data1m $ wearing ) == TRUE ){ mynonSummary < summaryData ( data1m, perMinuteCts = 1 markingString = "nw" ) nonweartimes < mynonSummary $ wearTimeByDay } else nonweartimes < 0 ##Bout analysis #1+ bouts sed1total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v )

PAGE 119

119 if ( length ( v.sub $ axis1 ) > 1 ) { sed.1bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 1 thresh.upper = 99 ) sed1 < sed.1bouts.flag == "1" datased1 < v.sub [ sed1, ] sed1day < nrow ( datased1 ) } else sed1day < 0 sed1total < sed1total + sed1day rm ( sed.1bouts.flag, sed1, datased1, sed1day ) } sed1perday < as.double ( sed1total / length ( numvalday )) #10+ bouts sed10total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 10 ) { sed.10bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 10 thresh.upper = 99 ) sed10 < sed.10bouts.flag == "1" datased10 < v.sub [ sed10, ] sed10day < nrow ( datased10 ) } else sed10day < 0 sed10total < sed10total + sed10day rm ( sed.10bouts.flag, sed10, datased10, sed10day ) } sed10perday < as.double ( sed10total / length ( numvalday )) #30+ bouts sed30total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 30 ) { sed.30bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 30 thresh.upper = 99 ) sed30 < sed.30bouts.flag == "1" datased30 < v.sub [ sed30, ] sed30day < nrow ( datased30 ) } else sed30day < 0

PAGE 120

120 sed30total < sed30total + sed30day rm ( sed.30bouts.flag, sed30, datased30, sed30day ) } sed30perday < as.double ( sed30total / length ( numvalday )) #60+ bouts sed60total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 60 ) { sed.60bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 60 thresh.upper = 99 ) sed60 < sed.60bouts.flag == "1" datased60 < v.sub [ sed60, ] sed60day < nrow ( datased60 ) } else sed60day < 0 sed60total < sed60total + sed60day rm ( sed.60bouts.flag, sed60, datased60, sed60day ) } sed60perday < as.double ( sed60total / length ( numvalday )) #calculate low light bouts: 100 to 759 cpm #1+ bouts lowlight1total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 1 ) { lowlight.1bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 1 thresh.lower = 100 thresh.upper = 759 ) lowlight1 < lowlight.1bouts.flag == "1" datalowlight1 < v.sub [ lowlight1, ] lowlight1day < nrow ( datalowlight1 ) } else lowlight1day < 0 lowlight1total < lowlight1 total + lowlight1day rm ( lowlight.1bouts.flag, lowlight1, datalowlight1, lowlight1day ) } lowlight1perday < as.double ( lowlight1total / length ( numvalday )) #2+ bouts

PAGE 121

121 lowlight2total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 2 ) { lowlight.2bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 2 thresh.lower = 100 thresh.upper = 759 ) lowlight2 < lowlight.2bouts.flag == "1" datalowlight2 < v.sub [ lowlight2, ] lowlight2day < nrow ( datalowlight2 ) } else lowlight2day < 0 lowlight2total < lowlight2total + lowlight2day rm ( lowlight.2bouts.flag, lowlight2, datalowlight2, lowlight2day ) } lowlight2perday < as.double ( lowlight2total / length ( num valday )) #5+ bouts lowlight5total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 5 ) { lowlight.5bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 5 thresh.lower = 100 thresh.upper = 759 ) lowlight5 < lowlight.5bouts.flag == "1" datalowlight5 < v.sub [ lowlight5, ] lowlight5day < nrow ( datalowlight5 ) } else lowlight5day < 0 lowlight5total < lowlight5total + lowlight5day rm ( lowlight.5bouts.flag, lowlight5, datalowlight5, lowlight5day ) } lowlight5perday < as.double ( lowlight5total / length ( numvalday )) #10+ bouts lowlight10total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 10 ) {

PAGE 122

122 lowlight.10bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 10 thresh.lower = 100 thresh.upper = 759 ) lowlight10 < lowlight.10bouts.flag == "1" datalowlight10 < v.sub [ lowlight10, ] lowlight10day < nrow ( datalowlight10 ) } else lowlight10day < 0 lowlight10total < lowlight10total + lowlight10day rm ( lowlight.10bouts.flag, lowlight10, datalowlight10, lowlight10day ) } lowlight10perday < as.double ( lowlight10total / length ( num valday )) #calculate high light bouts: 760 to 1040 cpm #1+ bouts highlight1total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 1 ) { highlight.1bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 1 thresh.lower = 760 thresh.upper = 1040 ) highlight1 < highlight.1bouts.flag == "1" datahighlight1 < v.sub [ highlight1, ] highlight1day < nrow ( datahighlight1 ) } else highlight1day < 0 highlight1total < highlight1total + highlight1day rm ( highlight.1bouts.flag, highlight1, datahighlight1, highlight1day ) } highlight1perday < as.double ( highlight1total / length ( numvalday )) #2+ bouts highlight2total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 2 ) { highlight.2bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 2 thresh.lower = 760 thresh.upper = 1040 ) highlight2 < highlight.2bouts.flag == "1"

PAGE 123

123 datahighlight2 < v.sub [ highlight2, ] highlight2day < nrow ( datahighlight2 ) } else highlight2day < 0 highlight2total < highlight2total + highlight2day rm ( highlight.2bouts.flag, highlight2, datahighlight2, highlight2day ) } highlight2perday < as.double ( highlight2total / length ( numvalday )) #5+ bouts highlight5total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 5 ) { highlight.5bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 5 thresh.lower = 760 thresh.upper = 1040 ) highlight5 < highlight.5bouts.flag == "1" datahighlight5 < v.sub [ highlight5, ] highlight5day < nrow ( datahighlight5 ) } else highlight5day < 0 highlight5total < highlight5total + highlight5day rm ( highlight.5bouts.flag, highlight5, datahighlight5, highlight5day ) } highlight5perday < as.double ( highlight5total / length ( numvalday )) #10+ bouts highlight10total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 10 ) { highlight.10bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 10 thresh.lower = 760 thresh.upper = 1040 ) highlight10 < highlight.10bouts.flag == "1" datahighlight10 < v.sub [ highlight10, ] highlight10day < nrow ( datahighlight10 ) } else highlight10day < 0 highlight10total < highlight10total + highlight10day rm ( highlight.10bouts.flag, highlight10, datahighlight10, highlight10day )

PAGE 124

124 } highlight10perday < as.double ( highlight10total / length ( numvalday )) #calculate low moderate bouts: 1041 to 2019 cpm #1+ bouts lowmod1total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 1 ) { lowmod.1bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 1 thresh.lower = 1041 thresh.upper = 2019 ) lowmod1 < lowmod.1bouts.flag == "1" datalowmod1 < v.sub [ l owmod1, ] lowmod1day < nrow ( datalowmod1 ) } else lowmod1day < 0 lowmod1total < lowmod1total + lowmod1day rm ( lowmod.1bouts.flag, lowmod1, datalowmod1, lowmod1day ) } lowmod1perday < as.double ( lowmod1total / length ( numvalday )) #2+ bouts lowmod2total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 2 ) { lowmod.2bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 2 thresh.lower = 1041 thresh.upper = 2019 ) lowmod2 < lowmod.2bouts.flag == "1" datalowmod2 < v.sub [ lowmod2, ] lowmod2day < nrow ( datalowmod2 ) } else lowmod2day < 0 lowmod2total < lowmod2total + lowmod2day rm ( lowmod.2bouts.flag, lowmod2, datalowmod2, lowmod2day ) } lowmod2perday < as.double ( lowmod2total / length ( numvalday )) #5+ bouts

PAGE 125

125 lowmod5total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 5 ) { lowmod.5bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 5 thresh.lower = 1041 thresh.upper = 2019 ) lowmod5 < lowmod.5bouts.flag == "1" datalowmod5 < v.sub [ lowmod5, ] lowmod5day < nrow ( datalowmod5 ) } else lowmod5day < 0 lowmod5total < lowmod5total + lowmod5day rm ( lowmod.5bouts.flag, lowmod5, datalowmod5, lowmod5day ) } lowmod5perday < as.double ( lowmod5total / length ( numvalday )) #10+ bouts lowmod10total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 10 ) { lowmod.10bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 10 thresh.lower = 1041 thresh.upper = 2019 ) lowmod10 < lowmod.10bouts.flag == "1" datalowmod10 < v.sub [ lowmod10, ] lowmod10day < nrow ( datalowmod10 ) } else lowmod10day < 0 lowmod10total < lowmod10total + lowmod10day rm ( lowmod.10bouts.flag, lowmod10, datalowmod10, lowmod10day ) } lowmod10perday < as.double ( lowmod10total / length ( numvalday )) #calculate high moderate bouts: 2020+ cpm #1+ bouts highmod1total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v )

PAGE 126

126 if ( length ( v.sub $ axis1 ) > 1 ) { highmod.1bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 1 thresh.lower = 2020 ) highmod1 < highmod.1bouts.flag == "1" datahighmod1 < v.sub [ highmod1, ] highmod1day < nrow ( datahighmod1 ) } else highmod1day < 0 highmod1total < highmod1total + highmod1day rm ( highmod.1bouts.flag, highmod1, datahighmod1, highmod1day ) } highmod1perday < as.double ( highmod1total / length ( numvalday )) #2+ bouts highmod2total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 2 ) { highmod.2bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 2 thresh.lower = 2020 ) highmod2 < highmod.2bouts.flag == "1" datahighmod2 < v.sub [ highmod2, ] highmod2day < nrow ( datahighmod2 ) } else highmod2day < 0 highmod2total < highmod2total + highmod2day rm ( highmod.2bouts.flag, highmod2, datahighmod2, highmod2day ) } highmod2perday < as.double ( highmod2total / length ( numvalday )) #5+ bouts highmod5total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 5 ) { highmod.5bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 5 thresh.lower = 2020 ) highmod5 < highmod.5bouts.flag == "1" datahighmod5 < v.sub [ highmod5, ] highmod5day < nrow ( datahighmod5 ) } else highmod5day < 0

PAGE 127

127 highmod5total < highmod5total + highmod5day rm ( highmod.5bouts.flag, highmod5, datahighmod5, highmod5day ) } highmod5perday < as.double ( highmod5total / length ( numvalday )) #10+ bouts highmod10total = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) if ( length ( v.sub $ axis1 ) > 10 ) { highmod.10bouts.flag < accel.bouts ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, bout.length = 10 thresh.lower = 2020 ) highmod10 < highmod.10bouts.flag == "1" datahighmod10 < v.sub [ highmod10, ] highmod10day < nrow ( datahighmod10 ) } else highmod10day < 0 highmod10total < highmod10total + highmod10day rm ( highmod.10bouts.flag, highmod10, datahighmod10, highmod10day ) } highmod10perday < as.double ( highmod10total / length ( numvalday )) #number of sedentary breaks totalsedbreaks = 0 for ( v in validDayUse ) { v.sub < subset ( data1m, data1m $ days == v ) num.sedbreaks < accel.sedbreaks ( counts = v.sub $ axis1, weartime = v.sub $ wearindicator, thresh = 100 ) totalsedbreaks < totalsedbreaks + num.sedbreaks } sedbreaksperday < as.double ( totalsedbreaks / length ( numvalday )) #HID hid < unique ( na.omit ( as.numeric ( unlist ( strsplit ( unlist ( i ) "[^0 9]+" ))))) #dataframe to write wide format: HID, accel metrics, non wear, valid days wide < rbind ( wide, data.frame ( HID = hid [ 1 ] validdays = length ( numvalday ) wearmin = sum ( mySummary $ validWearTimeByDay )/( length ( numvalday )) nonwearmin = sum ( nonweartimes [ validDayUse ] na.rm = TRUE )/ length ( numvalday ) sedvol = sed1perday [ 1 ] sed10bouts = sed10perday [ 1 ] sed30bouts = sed30perday [ 1 ] sed60bouts = sed60perday [ 1 ] sedbreaks = sedbreaksperday [ 1 ] lowlightvol =

PAGE 128

128 lowlight1perday [ 1 ] lowlight2bouts = lowlight2perday [ 1 ] lowlight5bouts = lowlight5perday [ 1 ] lowlight10bouts = lowlight10perday [ 1 ] highlightvol = highlight1perday [ 1 ] highlight2bouts = highlight2perday [ 1 ] highlight5bouts = highlight5p erday [ 1 ] highlight10bouts = highlight10perday [ 1 ] lowmodvol = lowmod1perday [ 1 ] lowmod2bouts = lowmod2perday [ 1 ] lowmod5bouts = lowmod5perday [ 1 ] lowmod10bouts = lowmod10perday [ 1 ] highmodvol = highmod1perday, highmod2bouts = highmod2perday [ 1 ] highmod5bouts = highmod5perday [ 1 ] highmod10bouts = highmod10perday [ 1 ])) } ##load .RData I will use this file to merge everything else to load ( "E:/LIFE datasets/Updated LIFE data_R data format/second_data/PID_VC_HID.Rdata" ) total < REF #merging to d ataset combined < merge ( total,wide, by = "HID" ) write.table ( combined, file = "E:/LIFE sedentary_main/1 Sedentary time and bouts/R output/NEW LIFE M code/LIFE_accelerometry_all visits_streamNULL.csv" row.name = FALSE sep = "," )

PAGE 129

129 APPENDIX B R CODE USED TO GENERATE HOSPITAL VARIABLES # Functions ---------------------------findTheVisit < function ( ppt.ref, day ) { prev.visit < which.max ( which ( ppt.ref $ checklist_days < day )) as.character ( ppt.ref $ vc [ prev.visit ]) #+ 1 after prev.visit logical but was removed bc of the way vc is coded 1 = f6 2=f12 3=f24 4=sv2 } setwd ( "V:/Amal/LIFE/LIFE main/3 Hospitalization and accelerometry/Analysis/Version 2/Merging data/Matin/" ) #variable.df is a reminder that created variable is a dataframe # note that $ is to in dex columns, [row, column] indexes everything df.ref < read.csv ( "LIFE01b_ActivityStatus_v2_1_visit days_FU24_weardays.csv" ) df.ref $ events.hospitalevent < 0 df.ref $ events.lengthofstay < 0 df.ref $ events.days_recenthosp_fu < NA df.ref $ events.days_recenthosp_fu_v2 < 0 df.ref $ events.accel_hosp_overlap < 0 df.aux < read.csv ( "LIFE04_SAE_v2_1.csv" ) df.aux $ vc < "F24" # Sort df.ref to place SV2 visits on top of F06 df.ref $ maskid < factor ( df.ref $ maskid ) df.ref $ vc < factor ( df.ref $ vc, leve ls = c ( "SV2" "F06" "F12" "F24" )) df.ref2 < data.frame ( matrix ( nrow = 0 ncol = ncol ( df.ref ))) for ( maskid in levels ( df.ref $ maskid )) { ppt.ref.df < df.ref [ df.ref $ maskid == maskid, ] ppt.ref.df < ppt.ref.df [ with ( ppt.ref.df, order ( ppt.ref.df $ vc, decreasing = F )) ] df.ref2 < rbind ( df.ref2, ppt.ref.df ) } rm ( ppt.ref.df, maskid ) df.ref < df.ref2; rm ( df.ref2 ) ##create a variable of # of hospitalizations before each assessment for ( i in 1 : nrow ( df.ref )) { print ( paste ( "(" i, out of nrow ( df.ref ) ") checked." sep = "" )) maskid < df.ref $ maskid [ i ] visit < as.character ( df.ref $ vc [ i ]) upperBoundDay < df.ref $ checklist_days [ i ] if ((! is.na ( upperBoundDay )) && ( upperBoundDay > 0 )) { #if upperbound is not NA and greater than 0 lowerBoundDay < df.ref $ checklist_days [ i 1 ] #lowerbound is days from rand. 1

PAGE 130

130 df.ref $ events.days_recenthosp_fu_v2 [ i ] < df.ref $ checklist_days [ i ] df.ref $ checklist_days [ i 1 ] ppt.df < df.aux [ df.aux $ maskid == maskid, ] #selecting rows to preserve column data events.df < ppt.df [ ppt.df $ daysevntn_eevl > lowerBoundDay, ] #prune events.df < events.df [ which ( events.df $ daysevntn_eevl <= upperBoundDay ) ] # prune, which gives the true indices of a logical objective, all owing for array indices events.df < events.df [ which ( events.df $ hospital_eevl == 1 ) ] if ( nrow ( events.df ) == 0 ) { next () } df.ref $ events.hospitalevent [ i ] < sum ( events.df $ hospital_eevl [ events.df $ masked == 1 ] na.rm = T ) los < ( events.df $ disdays_ohos events.df $ admdays_ohos ) #added df.ref $ events.lengthofstay [ i ] < sum ( los, na.rm = T ) #added ##created a variable that gives days from recent hospitalization to future FU recent < NA if (! is.na ( max ( events.df $ disdays_ohos ))) { df.ref $ events.days_recenthosp_fu [ i ] < upperBoundDay max ( events.df $ disdays_ohos, fill = 0 ) #catches hospitalizations that are even after 24 months if they overlap the 24 month accel. collection df.ref $ events.days_recenthosp_fu _v2 [ i ] < df.ref $ events.days_recenthosp_fu [ i ] } else { #if(is.na(max(events.df$disdays_ohos))) { df.ref $ events.days_recenthosp_fu [ i ] < NA } } ##created 1 or 0 variable which marks any overlaps between accelerometry collection and hospitalization lowerDay < df.ref $ checklist_days [ i ] ##created new lower bounds based on FU visit upperDay < df.ref $ checklist_days [ i ] + df.ref $ validdays [ i ] ##created new upper bounds based on wear days if (! is.na ( lowerDay ) && (! is.na ( upperDay ))) { ppt.df < df.aux [ df.aux $ maskid == maskid, ] #selecting rows to preserve column data events.df < ppt.df [ ppt.df $ daysevntn_eevl > lowerDay, ] #prune events.df < events.df [ which ( events.df $ daysevntn_eevl <= upperDay ) ] if ( nrow ( events.df ) > 0 && sum ( events.df $ hospital_eevl [ events.df $ masked == 1 ] na.rm = T ) > 0 ){ df.ref $ events.accel_hosp_overlap [ i ] < 1 } } else { df.ref $ events.accel_hosp_overlap [ i ] < NA } }

PAGE 131

131 for ( i in 1 : nrow ( df.aux )) { print ( paste ( "(" i, out of nrow ( df.aux ) ") added." sep = "" )) maskid < df.aux $ maskid [ i ] ppt.ref < df.ref [ df.ref $ maskid == maskid, ] df.aux $ vc [ i ] < findTheVisit ( ppt.ref, df.aux $ daysevntn_eevl [ i ]) } df.aux.reordered < df.aux [ c ( 1 ncol ( df.aux ) 2 :( ncol ( df.aux ) 1 ))] write.csv ( df.ref, file = "LIFE01b_ActivityStatus_v2_1_visit days_hospitalevent_2.csv" row.names = F ) write.csv ( df.aux.reordered, file = "LIFE04_SAE_v2_1_vc_2.csv" row.names = F ) ##be careful bc events after f24 are labeled SV2 atm

PAGE 132

132 APPENDIX C HOSPITALIZATIONS BY INTERVENTION Table C 1. Reason for hospitalization across intervention group Physical activity Health education p value Hospitalization reason* n=299 n=247 n (%) Musculo skeletal system disorders 62 (20.7) 65 (26.3) 0.13 General cardiovascular disorders 48 (16.1) 31 (12.6) 0.25 Gastro intestinal system disorders 42 (14.1) 22 (8.9) 0.06 Central & peripheral nervous system disorders 26 (8.7) 19 (7.7) 0.67 Respiratory system disorders 21 (7.0) 22 (8.9) 0.42 Urinary system disorders 16 (5.4) 13 (5.3) 0.96 Vascular (extracardiac) disorders 13 (4.4) 12 (4.9) 0.78 Whole body general disord ers 14 (4.7) 11 (4.5) 0.90 Heart rate and rhythm disorders 10 (3.3) 12 (4.9) 0.80 Skin and appendages disorders 6 (2.0) 7 (2.8) 0.53 Metabolic and nutritional disord ers 7 (2.3) 3 (1.2) 0.33 Autonomic nervous system disorders 8 (2.7) 1 (0.4) 0.04 Endocrine disorders 7 (2.3) 2 (0.8) 0.16 Myo endo pericardial and val ve disorders 2 (0.7) 7 (2.8) 0.05 Liver and biliary system disorders 5 (1.7) 2 (0.8) 0.37 Platelet, bleeding & clotting disorders 2 (0.7) 4 (1.6) 0.29 Hearing and vestibular disorders 2 (0.7) 3 (1.2) 0.44 Red blood cell disorders 2 (0.7) 2 (0.8) 0.85 Female reproductive disorders 2 (0.7) 2 (0.8) 0.85 Neoplasms 2 (0.7) 2 (0.8) 0.85 Psychiatric disorders 1 (0.3) 1 (0.4) 0.89 Male reproductive disorders 0 (0.0) 2 (0.8) 0.12 Vision disorders 0 (0.0) 1 (0.4) 0.27 White cell and reticuloendothelial disorders 1 (0.3) 0 (0.0) 0.36 Secondary terms events 0 (0.0) 1 (0.4) 0.27 *Based on MedDRA system organ class, listed in decreasing order

PAGE 133

133 Table C 2. Baseline participant characteristics by hospitalization status and intervention over 24 months Hospitalized (n=358) Not hospitalized (n= 983 ) PA (n=187) HE (n= 171 ) p PA (n= 482 ) HE (n=501) p Age, mean(SD) >= 80, n(%) 79.0 (5.4) 79 (42.3) 79.8 (5.2) 87 (50.9) 0.15 0.10 78.3 (5.2) 192 (39.8) 78.7 (5.2) 206 (41.1) 0.31 0.68 Female, n(%) 126 (67.4) 107 (62.6) 0.34 305 (63.3) 354 (70.7) 0.01 Non Hispanic white, n(%) 139 (74.3) 144 (84.2) 0.02 358 (74.3) 376 (75.1) 0.81 College or higher education, n(%) 112 (65.5) 111 (59.4) 0.23 318 (63.5) 306 (63.5) 0.86 Married, n(%) 79 (42.3) 57 (33.3) 0.14 174 (36.1) 184 (36.7) 0.98 Annual income < $25,000 54 (28.9) 45 (26.3) 0.86 136 (28.2) 145 (28.9) 0.60 Body mass index, mean (SD) 30.3 (5.8) 30.1 (6.1) 0.75 30.1 (5.8) 30.6 (6.3) 0.26 Smoked 100+ cigarettes ever, n(%) 102 (54.6) 84 (49.1) 0.50 237 (49.2) 213 (42.5) 0.10 Modified Mini Mental State Examination score a mean(SD) 91.4 (5.3) 91.8 (5.4) 0.44 91.9 (5.5) 91.6 (5.3) 0.44 Sought medical advice for depression in past 5 years, n(%) 28 (15.0) 28 (16.4) 0.72 69 (14.3) 57 (11.4) 0.40 Self reported hospitalization in the past 6 months, n(%) 21 (11.2) 22 (12.9) 0.22 26 (5.4) 30 (6.0) 0.16 Comorbidities > 2, n(%) 57 (30.5) 63 (36.8) 0.25 115 (23.9) 114 (22.8) 0.87 Short Physical Performance Battery score < 8 b n(%) 88 (47.1) 103 (60.2) 0.01 198 (41.1) 206 (41.1) 0.99 400 meter walk < 0.8 m/sec, n(%) 111 (59.4) 77 (45.0) 0.97 287 (59.5) 299 (59.7) 0.001 Wear days, mean(SD) 8.0 (3.4) 8.1 (3.2) 0.73 8.0 (3.2) 7.8 (3.1) 0.47 Wear minute/day, mean (SD) 843.5 (120.2) 827.1 (85.8) 0.14 837.4 (109.8) 837 (116.3) 0.97 counts, mean (SD) 25.6 (20.2) 22.4 (20.4) 0.13 28.3 (23.7) 29.8 (27.3) 0.36 Note: PA physical a ctivity; HE health education a score range: 0 100 where higher scores indicate better performance b score range: 0 12 where higher scores indicate better performance; scoring < 8 indicates poor functioning

PAGE 134

134 APPENDIX D HOSPITALIZATION DESCRIPTIVES BY TIME INTERVAL Figure D 1. Flow diagram of participants hospitalized within each time in terval according to intervention Note: LOS length of stay (average daysSD)

PAGE 135

135 APPENDIX E CORRELATION ANALYSIS BETWEEN HOSPITALIZATION AND ACCELEROMETRY Figure E 1. Correlation between days from hospitalization and subsequent accelerometer collection and accelerometer outcome metrics

PAGE 136

136 Figure E 2. The physical activity i ntervention effect (beta coefficient, SEM) on sedentary and activity bouts when compared to a health education program independent of hospitalization over 2 years. Note: TPA total physical activity *p<0.05

PAGE 137

137 Figure E 3. The physical activity intervention effect (beta coefficient, SEM) on intensity specific activity bouts when compared to a health education program, independent of hospitalization over 2 yea rs. Note: LLPA low light physical activity; HLPA high light physical activity; LMPA low moderate physical activity; HMPA high moderate physical activity. *p<0.05

PAGE 138

138 LIST OF REFERENCES 1. Ainsworth BE, Haskell WL, Whitt MC, I rwin ML, Swartz AM, Strath SJ, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc 2000;32(9; SUPP/1):S498 S504. 2. Bandura A. Social cognitive theory: An agentic perspective. Annu Rev Psychol 2 001;52(1):1 26. 3. Bankoski A, Harris TB, McClain JJ, Brychta RJ, Caserotti P, Chen KY, et al. Sedentary activity associated with metabolic syndrome independent of physical activity. Diabetes Care 2011;34(2):497 503. 4. Bann D, Hire D, Manini T, Cooper R, Botoseneanu A, McDermott MM, et al. Light intensity physical activity and sedentary behavior in relation to body mass index and grip strength in older adults: cross sectional findings from the Lifestyle Interventions and Independence for Elders (LIFE) stu dy. PLoS One 2015;10(2):e0116058. 5. Benatti FB, Ried Larsen M. The Effects of Breaking up Prolonged Sitting Time: A Review of Experimental Studies. Med Sci Sports Exerc 2015;47(10):2053 61. 6. Bergmann MM, Byers T, Freedman DS, Mokdad A. Validity of self reported diagnoses leading to hospitalization: a comparison of self reports with hospital records in a prospective study of American adults. Am J Epidemiol 1998;147(10):969 77. 7. Biswas A, Oh PI, Faulkner GE, Bajaj RR, Silver MA, Mitchell MS, et al. Sedentary time and its association with risk for disease incidence, mortality, and hospitalization in adults: a systematic review and meta analysis. Ann Intern Med 2015;162(2):123 32. 8. Borg G. Me asuring perceived exertion and pain. Champaign, IL: Human kinetics; 1998. 9. Borg G. Psychophysical scaling with applications in physical work and the perception of exertion. Scand J Work Environ Health 1990:55 8. 10. Branch LG, Jette AM. A prospective st udy of long term care institutionalization among the aged. Am J Public Health 1982;72(12):1373 9. 11. Brown CJ, Redden DT, Flood KL, Allman RM. The underrecognized epidemic of low mobility during hospitalization of older adults. J Am Geriatr Soc 2009;57( 9):1660 5. 12. Buman MP, Hekler EB, Haskell WL, Pruitt L, Conway TL, Cain KL, et al. Objective light intensity physical activity associations with rated health in older adults. Am J Epidemiol 2010;172(10):1155 65.

PAGE 139

139 13. Butte NF, Ekelund U, Westerterp KR. A ssessing physical activity using wearable monitors: measures of physical activity. Med Sci Sports Exerc 2012;44(1 Suppl 1):S5 12. 14. Carson V, Wong SL, Winkler E, Healy GN, Colley RC, Tremblay MS. Patterns of sedentary time and cardiometabolic risk among Canadian adults. Prev Med 2014;65:23 7. 15. Caspersen CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions and distinctions for health related research. Public Health Rep 1985;100(2):126. 16. Chao D, Foy CG, Farmer D. Exercise adherence among older adults: challenges and strategies. Control Clin Trials 2000;21(5):S212 S7. 17. Chatterji S, Byles J, Cutler D, Seeman T, Verdes E. Health, functioning, and disability in older adults present status and future implications. The Lancet 2015;385(9967):563 75. 18. Choi L, Liu Z, Matthews CE, Buchowski MS. Validation of accelerometer wear and nonwear time classification algorithm. Med Sci Sports Exerc 2011;43(2):357 64. 19. Choi L, Ward SC, Schnelle JF, Buchowski MS. Assessment of wear/nonwear time classification algorithms for triaxial accelerometer. Med Sci Sports Exerc 2012;44(10):2009 16. 20. Church TS, Martin CK, Thompson AM, Earnest CP, Mikus CR, Blair SN. Changes in weight, waist circumference an d compensatory responses with different doses of exercise among sedentary, overweight postmenopausal women. PLoS One 2009;4(2):e4515. 21. The Journals of Gerontology Series A: Biological Sciences and Medical Sci ences 2008;63(8):829 34. 22. Control CfD, Prevention. Trends in aging -United States and worldwide. MMWR Morbidity and mortality weekly report 2003;52(6):101. 23. Copeland JL, Esliger DW. Accelerometer assessment of physical activity in active, healthy o lder adults. J Aging Phys Act 2009;17(1):17 30. 24. Covinsky KE, Palmer RM, Fortinsky RH, Counsell SR, Stewart AL, Kresevic D, et al. Loss of independence in activities of daily living in older adults hospitalized with medical illnesses: increased vulnera bility with age. J Am Geriatr Soc 2003;51(4):451 8. 25. Creditor MC. Hazards of hospitalization of the elderly. Ann Intern Med 1993;118(3):219 23.

PAGE 140

140 26. Deer RR, Dickinson JM, Fisher SR, Ju H, Volpi E. Identifying effective and feasible interventions to ac celerate functional recovery from hospitalization in older adults: A randomized controlled pilot trial. Contemp Clin Trials 2016;49:6 14. 27. Diaz KM, Howard VJ, Hutto B, Colabianchi N, Vena JE, Blair SN, et al. Patterns of Sedentary Behavior in US Middle Age and Older Adults: The REGARDS Study. Med Sci Sports Exerc 2016;48(3):430 8. 28. DiPietro L. Physical activity in aging changes in patterns and their relationship to health and function. The Journals of Gerontology Series A: Biological Sciences and Me dical Sciences 2001;56(suppl 2):13 22. 29. Dunstan DW, Howard B, Healy GN, Owen N. Too much sitting a health hazard. Diabetes Res Clin Pract 2012;97(3):368 76. 30. Dunstan DW, Kingwell BA, Larsen R, Healy GN, Cerin E, Hamilton MT, et al. Breaking up prol onged sitting reduces postprandial glucose and insulin responses. Diabetes Care 2012;35(5):976 83. 31. Evenson KR. Objective measurement of physical activity and sedentary behavior among US adults aged 60 years or older. Prev Chronic Dis 2012;9. 32. Fiel ding RA, Rejeski WJ, Blair S, Church T, Espeland MA, Gill TM, et al. The Lifestyle Interventions and Independence for Elders Study: design and methods. J Gerontol A Biol Sci Med Sci 2011;66(11):1226 37. 33. Fielding RA, Vellas B, Evans WJ, Bhasin S, Morle y JE, Newman AB, et al. Sarcopenia: an undiagnosed condition in older adults. Current consensus definition: prevalence, etiology, and consequences. International working group on sarcopenia. J Am Med Dir Assoc 2011;12(4):249 56. 34. Fitzgerald JD, Johnson L, Hire DG, Ambrosius WT, Anton SD, Dodson JA, et al. Association of Objectively Measured Physical Activity With Cardiovascular Risk in Mobility limited Older Adults. Journal of the American Heart Association 2015;4(2):e001288. 35. Fleg JL, Morrell CH, B os AG, Brant LJ, Talbot LA, Wright JG, et al. Accelerated longitudinal decline of aerobic capacity in healthy older adults. Circulation 2005;112(5):674 82. 36. Franco OH, de Laet C, Peeters A, Jonker J, Mackenbach J, Nusselder W. Effects of physical activ ity on life expectancy with cardiovascular disease. Arch Intern Med 2005;165(20):2355 60. 37. Freedson PS, Melanson E, Sirard J. Calibration of the Computer Science and Applications, Inc. accelerometer. Med Sci Sports Exerc 1998;30(5):777 81.

PAGE 141

141 38. Freedso n PS, Miller K. Objective monitoring of physical activity using motion sensors and heart rate. Res Q Exerc Sport 2000;71(sup2):21 9. 39. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phen otype. J Gerontol A Biol Sci Med Sci 2001;56(3):M146 56. 40. Gando Y, Yamamoto K, Murakami H, Ohmori Y, Kawakami R, Sanada K, et al. Longer time spent in light physical activity is associated with reduced arterial stiffness in older adults. Hypertension 2010;56(3):540 6. 41. Gill TM. Disentangling the disabling process: insights from the precipitating events project. The Gerontologist 2014;54(4):533 49. 42. Gill TM, Allore HG, Gahbauer EA, Murphy TE. Change in disability after hospitalization or restrict ed activity in older persons. JAMA 2010;304(17):1919 28. 43. Gill TM, Allore HG, Holford TR, Guo Z. Hospitalization, restricted activity, and the development of disability among older persons. JAMA 2004;292(17):2115 24. 44. Goran MI, Poehlman ET. Enduran ce training does not enhance total energy expenditure in healthy elderly persons. American Journal of Physiology Endocrinology And Metabolism 1992;263(5):E950 E7. 45. of hos AJN The American Journal of Nursing 2006;106(1):58 67. 46. Gregg EW, Pereira MA, Caspersen CJ. Physical activity, falls, and fractures among older adults: a review of the epidemiologic evidence. J Am Geriatr Soc 2 000;48(8):883 93. 47. Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower extremity function in persons over the age of 70 years as a predictor of subsequent disability. N Engl J Med 1995;332(9):556 61. 48. Guralnik JM, Simonsick EM, Ferru cci L, Glynn RJ, Berkman LF, Blazer DG, et al. A short physical performance battery assessing lower extremity function: association with self reported disability and prediction of mortality and nursing home admission. J Gerontol 1994;49(2):M85 M94. 49. Ha milton MT, Hamilton DG, Zderic TW. Role of low energy expenditure and sitting in obesity, metabolic syndrome, type 2 diabetes, and cardiovascular disease. Diabetes 2007;56(11):2655 67. 50. Health UDo, Services H, Health UDo, Services H. Physical activity guidelines for Americans. 2008.

PAGE 142

142 51. Healy GN, Dunstan DW, Salmon J, Cerin E, Shaw JE, Zimmet PZ, et al. Breaks in sedentary time beneficial associations with metabolic risk. Diabetes Care 2008;31(4):661 6. 52. Healy GN, Dunstan DW, Salmon J, Cerin E, Shaw JE, Zimmet PZ, et al. Objectively measured light intensity physical activity is independently associated with 2 h plasma glucose. Diabetes Care 2007;30(6):1384 9. 53. Healy GN, Matthews CE, Dunstan DW, Winkler EA, Owen N. Sedentary time and cardio metabo lic biomarkers in US adults: NHANES 2003 06. Eur Heart J 2011;32(5):590 7. 54. Healy GN, Wijndaele K, Dunstan DW, Shaw JE, Salmon J, Zimmet PZ, et al. Objectively measured sedentary time, physical activity, and metabolic risk: the Australian Diabetes, Obe sity and Lifestyle Study (AusDiab). Diabetes Care 2008;31(2):369 71. 55. Heckbert SR, Kooperberg C, Safford MM, Psaty BM, Hsia J, McTiernan A, et al. Comparison of self report, hospital discharge codes, and adjudication of cardiovascular events in the Wom Am J Epidemiol 2004;160(12):1152 8. 56. Hekler EB, Buman MP, Haskell WL, Conway TL, Cain KL, Sallis JF, et al. Reliability and validity of CHAMPS self reported sedentary to vigorous intensity physical activity in older adults. Jour nal of physical activity & health 2012;9(2):225. 57. Helmerhorst HHJ, Brage S, Warren J, Besson H, Ekelund U. A systematic review of reliability and objective criterion related validity of physical activity questionnaires. International Journal of Behavio ral Nutrition and Physical Activity 2012;9(1):1. 58. Hing E, Bloom B. Long term care for the functionally dependent elderly. Am J Public Health 1991;81(2):223 5. 59. Hirvensalo M, Rantanen T, Heikkinen E. Mobility Difficulties and Physical Activity as Pr edictors of Mortality and Loss of Independence in the Community Living Older Population. J Am Geriatr Soc 2000;48(5):493 8. 60. Janssen I, Heymsfield SB, Ross R. Low relative skeletal muscle mass (sarcopenia) in older persons is associated with functional impairment and physical disability. J Am Geriatr Soc 2002;50(5):889 96. 61. Katz S, Branch LG, Branson MH, Papsidero JA, Beck J C, Greer DS. Active life expectancy. N Engl J Med 1983;309(20):1218 24. 62. Katzmarzyk PT, Church TS, Craig CL, Bouchard C. Sitting time and mortality from all causes, cardiovascular disease, and cancer. Med Sci Sports Exerc 2009;41(5):998 1005.

PAGE 143

143 63. Knoops KT, de Groot LC, Kromhout D, Perrin A E, Moreiras Varela O, Menotti A, et al. Mediterranean diet, lifestyle factors, and 10 year mortality in elderly European men and women: the HALE project. JAMA 2004;292(12):1433 9. 64. Kortebein P, Ferrando A, L ombeida J, Wolfe R, Evans WJ. Effect of 10 days of bed rest on skeletal muscle in healthy older adults. JAMA 2007;297(16):1769 74. 65. Lee I M, Shiroma EJ. Using accelerometers to measure physical activity in large scale epidemiological studies: issues an d challenges. Br J Sports Med 2014;48(3):197 201. 66. Maeda K, Higashimoto Y, Honda N, Shiraishi M, Hirohata T, Minami K, et al. Effect of a postoperative outpatient pulmonary rehabilitation program on physical activity in patients who underwent pulmonary resection for lung cancer. Geriatrics & gerontology international 2015. 67. Man WD, Polkey MI, Donaldson N, Gray BJ, Moxham J. Community pulmonary rehabilitation after hospitalisation for acute exacerbations of chronic obstructive pulmonary disease: rand omised controlled study. BMJ 2004;329(7476):1209. 68. Manini TM, Pahor M. Physical activity and maintaining physical function in older adults. Br J Sports Med 2009;43(1):28 31. 69. Manns P, Ezeugwu V, Armijo Olivo S, Vallance J, Healy GN. Accelerometer D erived Pattern of Sedentary and Physical Activity Time in Persons with Mobility Disability: National Health and Nutrition Examination Survey 2003 to 2006. J Am Geriatr Soc 2015;63(7):1314 23. 70. Manton KG, Vaupel JW. Survival after the age of 80 in the U nited States, Sweden, France, England, and Japan. N Engl J Med 1995;333(18):1232 5. 71. Marsh AP, Applegate WB, Guralnik JM, Jack Rejeski W, Church TS, Fielding RA, et al. Hospitalizations During a Physical Activity Intervention in Older Adults at Risk of Mobility Disability: Analyses from the Lifestyle Interventions and Independence for Elders Randomized Clinical Trial. J Am Geriatr Soc 2016;64(5):933 43. 72. Marsh AP, Lovato LC, Glynn NW, Kennedy K, Castro C, Domanchuk K, et al. Lifestyle interventions and independence for elders study: recruitment and baseline characteristics. The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 2013;68(12):1549 58. 73. Matthew CE. Calibration of accelerometer output for adults. Med Sci Sports Exerc 2005;37(11 Suppl):S512 22. 74. Matthews CE, Chen KY, Freedson PS, Buchowski MS, Beech BM, Pate RR, et al. Amount of time spent in sedentary behaviors in the United States, 2003 2004. Am J Epidemiol 2008;167(7):875 81.

PAGE 144

144 75. Matthews CE, George SM, M oore SC, Bowles HR, Blair A, Park Y, et al. Amount of time spent in sedentary behaviors and cause specific mortality in US adults. The American journal of clinical nutrition 2012;95(2):437 45. 76. Mattlage AE, Redlin SA, Rippee MA, Abraham MG, Rymer MM, B illinger SA. Use of Accelerometers to Examine Sedentary Time on an Acute Stroke Unit. J Neurol Phys Ther 2015;39(3):166 71. 77. Meijer E, Westerterp K, Verstappen F. Effect of exercise training on physical activity and substrate utilization in the elderly Int J Sports Med 2000;21(07):499 504. 78. Meijer EP, Westerterp KR, Verstappen FT. Effect of exercise training on total daily physical activity in elderly humans. Eur J Appl Physiol Occup Physiol 1999;80(1):16 21. 79. Meijer G, Janssen G, Westerterp K, Verhoeven F, Saris W, Ten Hoor F. The effect of a 5 month endurance training programme on physical activity: evidence for a sex difference in the metabolic response to exercise. Eur J Appl Physiol Occup Physiol 1991;62(1):11 7. 80. Moore SA, Hallsworth K Pltz T, Ford GA, Rochester L, Trenell MI. Physical activity, sedentary behaviour and metabolic control following stroke: a cross sectional and longitudinal study. PLoS One 2013;8(1):e55263. 81. Mor V, Murphy J, Masterson Allen S, Willey C, Razmpour A, Jackson ME, et al. Risk of functional decline among well elders. J Clin Epidemiol 1989;42(9):895 904. 82. Morio B, Montaurier C, Pickering G, Ritz P, Fellmann N, Coudert J, et al. Effects of 14 weeks of progressive endurance training on energy expenditure in elderly people. Br J Nutr 1998;80(06):511 9. 83. Nelson ME, Rejeski WJ, Blair SN, Duncan PW, Judge JO, King AC, et al. Physical activity and public health in older adults: recommendation from the American College of Sports Medicine and the American He art Association. Circulation 2007;116(9):1094. 84. Newman AB, Simonsick EM, Naydeck BL, Boudreau RM, Kritchevsky SB, Nevitt MC, et al. Association of long distance corridor walk performance with mortality, cardiovascular disease, mobility limitation, and disability. JAMA 2006;295(17):2018 26. 85. Organization WH. World Report on Ageing and Health Geneva, Switzerland: World Health Organization; 2015 [cited 2016 October 9th]. Available from: http://www.who.int/ageing/publications/world report 2015/en/ 86. Osuka Y, Yabushita N, Kim M, Seino S, Nemoto M, Jung S, et al. Association between habitual light intensity physical activity and lower extremity performance: A cross sectional st udy of community dwelling older Japanese adults. Geriatrics & gerontology international 2015;15(3):268 75.

PAGE 145

145 87. Owen N, Healy GN, Matthews CE, Dunstan DW. Too much sitting: the population health science of sedentary behavior. Exerc Sport Sci Rev 2010;38(3 ):105. 88. Paffenbarger Jr RS, Kampert JB, Lee IM, Hyde RT, Leung RW, Wing AL. Changes in physical activity and other lifeway patterns influencing longevity. Med Sci Sports Exerc 1994;26(7):857 65. 89. Pahor M, Guralnik JM, Ambrosius WT, Blair S, Bonds DE Church TS, et al. Effect of structured physical activity on prevention of major mobility disability in older adults: the LIFE study randomized clinical trial. JAMA 2014;311(23):2387 96. 90. Paul DR, Kramer M, Moshfegh AJ, Baer DJ, Rumpler WV. Comparison of two different physical activity monitors. BMC Med Res Methodol 2007;7:26. 91. Plasqui G, Westerterp KR. Physical activity assessment with accelerometers: an evaluation against doubly label ed water. Obesity 2007;15(10):2371 9. 92. Prajapati SK, Mansfield A, Gage WH, Brooks D, McIlroy WE. Cardiovascular responses associated with daily walking in subacute stroke. Stroke research and treatment 2013;2013. 93. Prochaska JM. The transtheoretical model applied to the community and the workplace. J Health Psychol 2007;12(1):198 200. 94. Prochaska JO, Velicer WF. The transtheoretical model of health behavior change. Am J Health Promot 1997;12(1):38 48. 95. Rangan VV, Willis LH, Slentz CA, Bateman LA, Shields AT, Houmard JA, et al. Effects of an 8 month exercise training program on off exercise physical activity. Med Sci Sports Exerc 2011;43(9):1744 51. 96. Reid RD, Morrin LI, Pipe AL, Dafoe WA, Higginson LA, Wielgosz AT, et al. Determinants of phy sical activity after hospitalization for coronary artery disease: the Tracking Exercise After Cardiac Hospitalization (TEACH) Study. Eur J Cardiovasc Prev Rehabil 2006;13(4):529 37. 97. Robison J, Rogers MA. Adherence to exercise programmes. Sports Med 1994;17(1):39 52. 98. Roos MA, Rudolph KS, Reisman DS. The structure of walking activity in people after stroke compared with older adults without disability: a cross sectional study. Phys Ther 2012;92(9):1141 7. 99. Rosenkilde M, Auerbach P, Reichkendler MH, Ploug T, Stallknecht BM, Sjdin A. Body fat loss and compensatory mechanisms in response to different doses of aerobic exercise a randomized controlled trial in overweight sedentary males. American Journal of Physiology Regulatory, Integrative and Com parative Physiology 2012;303(6):R571 R9.

PAGE 146

146 100. Sallis JF, Saelens BE. Assessment of physical activity by self report: status, limitations, and future directions. Res Q Exerc Sport 2000;71(sup2):1 14. 101. Schoeller D, Van Santen E. Measurement of energy e xpenditure in humans by doubly labeled water method. J Appl Physiol 1982;53(4):955 9. 102. Schoeller DA. Measurement of energy expenditure in free living humans by using doubly labeled water. J Nutr 1988;118(11):1278 89. 103. Sedentary Behaviour Research N. Letter to the editor: standardized use of the terms "sedentary" and "sedentary behaviours". Appl Physiol Nutr Metab 2012;37(3):540 2. 104. Seguin R, Lamonte M, Tinker L, Liu J, Woods N, Michael YL, et al. Sedentary Behavior and Physical Function Decli ne in Older Women: Findings from the Women's Health Initiative. J Aging Res 2012;2012:271589. 105. Shiroma EJ, Freedson PS, Trost SG, Lee IM. Patterns of accelerometer assessed sedentary behavior in older women. JAMA 2013;310(23):2562 3. 106. Short ME, G oetzel RZ, Pei X, Tabrizi MJ, Ozminkowski RJ, Gibson TB, et al. How accurate are self reports? An analysis of self reported healthcare utilization and absence when compared to administrative data. Journal of occupational and environmental medicine/American College of Occupational and Environmental Medicine 2009;51(7):786. 107. Simons Morton DG, Calfas KJ, Oldenburg B, Burton NW. Effects of interventions in health care settings on physical activity or cardiorespiratory fitness. Am J Prev Med 1998;15(4):413 30. 108. Simpson ME, Serdula M, Galuska DA, Gillespie C, Donehoo R, Macera C, et al. Walking trends among US adults: the behavioral risk factor surveillance system, 1987 2000. Am J Prev Med 2003;25(2):95 100. 109. Statistics FIFoA R. Older Americans 2014 : Key Indicators of Well Being. Federal Interagency Forum on Aging Related Statistics 2014. 110. Stuck AE, Walthert JM, Nikolaus T, Bula CJ, Hohmann C, Beck JC. Risk factors for functional status decline in community living elderly people: a systematic li terature review. Soc Sci Med 1999;48(4):445 69. 111. Teng EL, Chui HC. The Modified Mini Mental State (3MS) examination. J Clin Psychiatry 1987;48(8):314 8. 112. Tieges Z, Mead G, Allerhand M, Duncan F, van Wijck F, Fitzsimons C, et al. Sedentary behavio r in the first year after stroke: a longitudinal cohort study with objective measures. Arch Phys Med Rehabil 2015;96(1):15 23.

PAGE 147

147 113. Tran B, Falster MO, Douglas K, Blyth F, Jorm LR. Health behaviours and potentially preventable hospitalisation: a prospecti ve study of older Australian adults. PLoS One 2014;9(4):e93111. 114. Troiano RP, Berrigan D, Dodd KW, Masse LC, Tilert T, McDowell M. Physical activity in the United States measured by accelerometer. Med Sci Sports Exerc 2008;40(1):181 8. 115. Van Domele n DR. accelerometry package. 2015. 116. Vogel T, Brechat PH, Lepretre PM, Kaltenbach G, Berthel M, Lonsdorfer J. Health benefits of physical activity in older patients: a review. Int J Clin Pract 2009;63(2):303 20. 117. Ward BW. Multiple chronic condition s among US adults: a 2012 update. Prev Chronic Dis 2014;11. 118. Wei JY. Age and the cardiovascular system. N Engl J Med 1992;327(24):1735 9. 119. Westerterp KR. Impacts of vigorous and non vigorous activity on daily energy expenditure. Proc Nutr Soc 2003;62(03):645 50. 120. Westerterp KR. Pattern and intensity of physical activity. Nature 2001;410(6828):539 121. Westerterp KR, Plasqui G. Physical activity and human energy expenditure. Curr Opin Clin Nutr Metab Care 2004;7(6):607 13.

PAGE 148

148 BI OGRAPHICAL SKETCH Amal Wanigatunga received his Ph.D. in e pidemiology from the University of Florida in the spring of 2016. H e also earned degree in public health. In parallel with his dissertation research, h e work ed as a graduate research associate with the Department of Aging and Geriatric Research at the Institute on Aging, University of Florida where he met with study participants, led clinical visits, trained incoming staff, and managed large data sets His interests lie in physical function, disability, and physical activity among older adult population s and has presented numerous research projects at both local and national conferences During his doctoral studies, his primary focus was on the measureme nt of physical activity, specifically with objective measures such as accelerometers.